text
stringlengths 66
377k
|
|---|
Uncovering the Mechanisms Responsible for Why Language Learning May Promote Healthy Cognitive Aging
One of the great challenges facing humankind in the 21st century is preserving healthy brain function in our aging population. Individuals over 60 are the fastest growing age group in the world, and by 2050, it is estimated that the number of people over the age of 60 will triple. The typical aging process involves cognitive decline related to brain atrophy, especially in frontal brain areas and regions that subserve declarative memory, loss of synaptic connections, and the emergence of neuropathological symptoms associated with dementia. The disease-state of this age-related cognitive decline is Alzheimer's disease and other dementias, which may cause older adults to lose their independence and rely on others to live safely, burdening family members and health care systems in the process. However, there are two lines of research that offer hope to those seeking to promote healthy cognitive aging. First, it has been observed that lifestyle variables such as cognitive leisure activities can moderate the risk of Alzheimer's disease, which has led to the development of plasticity-based interventions for older adults designed to protect against the adverse effects of cognitive decline. Second, there is evidence that lifelong bilingualism acts as a safeguard in preserving healthy brain function, possibly delaying the incidence of dementia by several years. In previous work, we have suggested that foreign language learning programs aimed at older populations are an optimal solution for building cognitive reserve because language learning engages an extensive brain network that is known to overlap with the regions negatively affected by the aging process. Here, we will outline potential future lines of research that may uncover the mechanism responsible for the emergence of language learning related brain advantages, such as language typology, bi-vs. multi-lingualism, age of acquisition, and the elements that are likely to result in the largest gains.
# Introduction
One of the great challenges facing humankind in the twenty-first century is dealing with the problems associated with an aging population. Over-60-year-olds are the fastest growing age group on earth. By 2050, the number of people over 60 is set to triple, eclipsing 2 billion worldwide (Department of . As the number of older adults increases, so too will the demands and costs associated with an aging population, placing increasing pressure on families, health systems, economies, and governments. The typical aging process is characterized by age-related decline in a number of cognitive subsystems [bib_ref] Models of visuospatial and verbal memory across the adult life span, Park [/bib_ref] [bib_ref] Aging of the brain and Alzheimer's disease, Drachman [/bib_ref]. Certain brain structures are particularly affected by the aging process, such as frontal areas, the hippocampus, and the entorhinal cortex [bib_ref] Profound loss of layer II entorhinal cortex neurons occurs in very mild..., Gómez-Isla [/bib_ref] [bib_ref] Age, executive function, and social decision making: a dorsolateral prefrontal theory of..., Macpherson [/bib_ref] [bib_ref] Neuronal death versus synaptic pathology in Alzheimer's disease, Bertoni-Freddari [/bib_ref]. Reduced function may be observed in working memory, declarative memory, as well as the interaction between declarative and procedural memory [bib_ref] Skill learning in the elderly: diminished implicit and explicit memory for a..., Harrington [/bib_ref] [bib_ref] Changes in memory processing with age, Grady [/bib_ref]. The disease state of cognitive decline is Alzheimer's disease (and other dementias), characterized by a gradual progressive difficulty with learning and retaining new information. Pharmacological trials have had little success in slowing down the progression of Alzheimer's disease [bib_ref] Two phase 3 trials of bapineuzumab in mild-to-moderate Alzheimer's disease, Salloway [/bib_ref]. This has led to increasing calls to treat the disease proactively using behavioral stimulations, ideally before symptoms manifest [bib_ref] Preventing Alzheimer's disease, Selkoe [/bib_ref].
Two promising lines of research have developed in parallel that offer some hope to combating age-related cognitive decline. On the one hand are studies demonstrating that environmental enrichment may result in positive brain changes. Studies of animals reared in standard vs. enriched enclosures have demonstrated the effects of environmental enrichment on the brain, namely denser dendritic connections resulting from stimulation [bib_ref] Rearing complexity affects branching of dendrites in the visual cortex of the..., Volkmar [/bib_ref] [bib_ref] Evidence for active synapse formation or altered postsynaptic metabolism in visual cortex..., Greenough [/bib_ref]. Such findings concerning environmental enrichment have been mirrored in investigations of lifestyle variables associated with healthy brain aging in humans: education, physical and mental stimulation, occupation, and leisure activities have all been linked to positive outcomes in cognitive aging [bib_ref] Environmental influences on cognitive and brain plasticity during aging, Kramer [/bib_ref] [bib_ref] What provides cerebral reserve?, Staff [/bib_ref] [bib_ref] Brain reserve and dementia: a systematic review, Valenzuela [/bib_ref] [bib_ref] Mapping the connections between education and dementia, Mcdowell [/bib_ref] [bib_ref] Education, the brain and dementia: neuroprotection or compensation?, Brayne [/bib_ref] [bib_ref] Education, occupation, leisure activities, and brain reserve: a population-based study, Foubert-Samier [/bib_ref]. These observations have led to the development of numerous plasticity-based interventions that aim to use environmental enrichment proactively by prescribing cognitively stimulating training regimens such as crossword puzzles [bib_ref] Leisure activities and the risk of dementia in the elderly, Verghese [/bib_ref] , math exercises [bib_ref] Reading aloud and arithmetic calculation improve frontal function of people with dementia, Kawashima [/bib_ref] , brain training [bib_ref] Effects of cognitive training interventions with older adults: a randomized controlled trial, Ball [/bib_ref] , and computerbased interventions [bib_ref] A cognitive training program based on principles of brain plasticity: results from..., Smith [/bib_ref] , and these cognitive improvements have been shown to persist over time [bib_ref] Memory enhancement in healthy older adults using a brain plasticity-based training program:..., Mahncke [/bib_ref]. The resulting improvements have been observed in healthy adults, and encouragingly, also in those with mild cognitive impairment [bib_ref] Training-related brain plasticity in subjects at risk of developing Alzheimer's disease, Belleville [/bib_ref] , and even in those diagnosed with Alzheimer's disease [bib_ref] Cognitive rehabilitation combined with drug treatment in Alzheimer's disease patients: a pilot..., Bottino [/bib_ref].
In parallel to the development and emergence of the cognitive training literature, evidence has been accumulating concerning the aging-related benefits of bilingualism. It was once thought that use of multiple languages (bilingualism) led to cognitive impairments [bib_ref] Racial differences in the intelligence of school children, Goodenough [/bib_ref]. However, carefully conducted scientific studies later showed it to result in cognitive improvements [bib_ref] The relation of bilingualism to intelligence, Peal [/bib_ref] , an outcome since reinforced by a further 30 years of research. The evidence now suggests that experience with two languages confers a general 'bilingual advantage, ' with improvement in executive function, metalinguistic awareness [bib_ref] Bilingualism and the development of metalinguistic awareness, Cummins [/bib_ref] , cognitive flexibility, creative thinking, and perhaps even several years' delay in the onset of dementia [bib_ref] Bilingualism as a protection against the onset of symptoms of dementia, Bialystok [/bib_ref]. Multilingualism is a better predictor of cognitive ability than age, age at immigration, education, or sex [bib_ref] Multilingualism and cognitive state in the oldest old, Kavé [/bib_ref]. These cognitive advantages that have been associated with bilingualism have neural correlates. For example, bilinguals demonstrate greater white matter integrity in old age compared to monolingual speakers [bib_ref] Lifelong bilingualism maintains white matter integrity in older adults, Luk [/bib_ref]. It has been suggested that this results in enhanced structural and functional connectivity that provides the neural basis for cognitive reserve. Bilingual older adults also show less steep cognitive decline than those who only speak one language [bib_ref] Bilingualism: the good, the bad, and the indifferent, Bialystok [/bib_ref]. However, in recent years, the robustness of a bilingual advantage has been hotly debated-questioned by some who have failed to replicate it [bib_ref] The bilingual advantage: acta est fabula?, Duñabeitia [/bib_ref] [bib_ref] Bilingual advantages in executive functioning either do not exist or are restricted..., Paap [/bib_ref] , but staunchly defended by its proponents . This debate itself highlights the absence of a detailed and systematic understanding of the factors that would underlie such an advantage. Interestingly, certain research laboratories consistently observe data patterns supporting a bilingual advantage, while other laboratories consistently find no advantage. It has even been suggested that bilingual advantages may reflect publication bias [bib_ref] Cognitive advantage in bilingualism: an example of publication bias?, De Bruin [/bib_ref] ; but both the significant and non-significant findings are so systematic that it is much more likely that other factors (e.g., linguistic, experiential) are involved [bib_ref] Publication bias and the validity of evidence: What's the connection?, Bialystok [/bib_ref].
In the sections that follow, we will review the bilingual cognitive aging literature with a view to exploring the potential mechanisms responsible for the emergence of language learning related brain advantages and how these may be investigated prospectively in longitudinal language learning studies in adult learners.
## Bilingualism and executive function
Many studies have reported that bilingualism yields advantages in executive function. [bib_ref] Lifelong bilingualism maintains neural efficiency for cognitive control in aging, Gold [/bib_ref] found that bilingual older adults showed better task-switching performance than monolinguals in a color-shape task where participants categorize images by their color (blue or red) and shape (square and circle). Furthermore, fMRI scans taken during this task revealed decreased activation in the bilinguals' left lateral frontal cortex and cingulate cortex, an indication of more efficient executive functioning, when compared to a monolingual control group, and this difference was consistent across both the younger and older participants. Additionally, bilinguals outperformed monolinguals in episodic memory recall and letter fluency, but not the categorical fluency task [bib_ref] A longitudinal study of memory advantages in bilinguals, Ljungberg [/bib_ref]. Learners of French as a second language outperformed monolinguals on a grammaticality judgment task (ignoring conflict introduced through misleading semantic content) and a non-verbal visual search task [bib_ref] Effects of short-term music and second-language training on executive control, Janus [/bib_ref]. Older adult bilinguals, including those who acquired their second language in adulthood, exhibited improved cognitive function (general intelligence and reading) compared to monolinguals . When exposed to a non-verbal switching task, monolinguals showed activation in the right inferior frontal cortex and the anterior cingulate whereas bilinguals showed activation in the left inferior frontal cortex and left striatum, both areas that underlie language control [bib_ref] Bridging language and attention: brain basis of the impact of bilingualism on..., Garbin [/bib_ref]. Older adult bilinguals processed distracting information more efficiently than their monolingual peers when completing the Flanker task [bib_ref] A diffusion model approach to analysing the bilingual advantage for the Flanker..., Ong [/bib_ref]. Aging bilinguals may also show less steep declines in executive function as they progress from healthy aging to mild cognitive impairment to probable Alzheimer's disease [bib_ref] Neuropsychological assessments of cognitive aging in monolingual and bilingual older adults, Anderson [/bib_ref]. Collectively, these studies suggest that bilinguals show an advantage for nonlinguistic cognitive abilities, particularly executive functions.
However, in recent years, a growing number of studies have raised questions regarding the robustness (and in some cases, the validity) of these bilingual advantage claims. This has led to attempts to understand the factors that give rise to the bilingual advantage in cognitive function and also explain its absence under certain conditions. Bilinguals' executive control abilities may be enhanced due to higher processing demands, and it has been argued that this may build cognitive reserve in the elderly [bib_ref] How does the bilingual experience sculpt the brain?, Costa [/bib_ref]. Although the exact mechanisms are not agreed upon, it is generally thought that bilinguals' cognitive and brain reserves share the same mechanism as executive control processing [bib_ref] Cognitive control, cognitive reserve, and memory in the aging bilingual brain, Grant [/bib_ref]. The complexity of the underlying cognitive processes may play a crucial role, with greater inhibitory demands resulting in greater benefit . If correct, bilingualism may ultimately delay clinical Alzheimer's disease symptoms by protecting brain regions that subserve executive control (frontostriatal and frontoparietal) rather than those that subserve memory (medial temporal lobe) per se [bib_ref] Lifelong bilingualism and neural reserve against Alzheimer's disease: a review of findings..., Gold [/bib_ref]. Further, the potentially mediating effect of age of second language acquisition on executive functions is not well understood, and thus neither is its potential impact on the structure of the brain [bib_ref] The bilingual advantage: acta est fabula?, Duñabeitia [/bib_ref] [bib_ref] Bilingual advantages in executive functioning either do not exist or are restricted..., Paap [/bib_ref]. Recently, Bialystok (2017) put forth an experience-dependent plasticity framework to evaluate the brain and cognitive modifications attributed to bilingualism. It was concluded that research broadly supports a relation between bilingualism and cognitive brain outcomes in infants and children, younger and older adults, and patients, however, behavioral studies with young adults, commonly fail to show these effects. This interpretation is consistent with findings in the executive function literature. Executive functions reach their peak in young adulthood [bib_ref] Models of visuospatial and verbal memory across the adult life span, Park [/bib_ref] , and thus greater variability in executive functions, as measured by behavioral tasks, are more likely to be observed in older adulthood (when cognitive functions decline; [bib_ref] Cognitive control and lexical access in younger and older bilinguals, Bialystok [/bib_ref] or in childhood (when the foundations of cognitive processing are being established; [bib_ref] The complexities of complex span: explaining individual differences in working memory in..., Bayliss [/bib_ref]. Thus, it seems reasonable that bilingual advantages would be easier to detect either in early or later life.
## Bilingualism and cognitive reserve
Cognitive reserve refers to the brain's resilience to neuropathological damage, resulting from experience-based neural changes associated with a physically and mentally stimulating lifestyle . [bib_ref] Cognitive reserve in ageing and Alzheimer's disease, Stern [/bib_ref] proposes two possible mechanisms for cognitive reserve: neural reserve, according to which differences in the resilience of already established networks, and neural compensation, according to which some individuals are better able to compensate for brain decline by using alternative networks. Evidence exists for both possibilities, and thus, the mechanisms responsible for cognitive reserve are a matter of ongoing research.
It is perhaps then unsurprising that the mechanism via which bilingualism improves the brain's resistance to neuropathology is not understood. Recent scholarly work has uncovered several potentially fruitful avenues concerning how bilingualism might build cognitive reserve focusing on the interactions between cognitive reserve and variables known to affect bilingualism [bib_ref] Bilingualism and cognitive reserve: a critical overview and a plea for methodological..., Calvo [/bib_ref] , as well as the brain networks that subserve memory [bib_ref] Cognitive control, cognitive reserve, and memory in the aging bilingual brain, Grant [/bib_ref] , brain metabolic connectivity [bib_ref] The impact of bilingualism on brain reserve and metabolic connectivity in Alzheimer's..., Perani [/bib_ref] , and the presence of Alzheimer's disease biomarkers in cerebrospinal fluid [bib_ref] Beneficial effect of bilingualism on Alzheimer's disease CSF biomarkers and cognition, Estanga [/bib_ref]. Nevertheless, numerous studies present evidence suggesting that bilingualism results in brain changes in healthy subjects. Higher degrees of bilingualism have been linked to better lexical memory performance [bib_ref] Bilingual proficiency and cognitive reserve in Persian-English bilingual older adults, Jafari [/bib_ref]. Bilinguals have higher white matter integrity than monolinguals in the corpus callosum extending to the superior and inferior longitudinal fasciculi, and also stronger anterior to posterior functional connectivity [bib_ref] Lifelong bilingualism maintains white matter integrity in older adults, Luk [/bib_ref]. Aging bilinguals outperformed monolinguals on the Flanker task, and had increased gray matter in the anterior cingulate cortex, whereas monolinguals showed decreased gray matter in the dorsolateral prefrontal cortex [bib_ref] Bilingualism provides a neural reserve for aging populations, Abutalebi [/bib_ref]. Further, brain regions that support executive control significantly overlap with brain regions recruited for language control [bib_ref] Neuroimaging of language control in bilinguals: neural adaptation and reserve, Abutalebi [/bib_ref]. The brain plasticity effects of lifelong bilingualism are thought to contribute to cognitive reserve and delay the onset of symptoms associated with dementia (Guzmán-Vélez and Tranel, 2015; [bib_ref] Bilingualism, dementia, cognitive and neural reserve, Perani [/bib_ref]. There is also evidence that bilingual brains are better able to accommodate anatomical and physiological brain changes and deterioration without exhibiting the expected increase in behavioral symptoms. Bilingual patients with Alzheimer's disease exhibited greater amounts of brain atrophy than monolingual patients (radial width of the temporal horn and the temporal horn ratio; [bib_ref] Bilingualism as a contributor to cognitive reserve: evidence from brain atrophy in..., Schweizer [/bib_ref]. Bilingual patients also showed substantially greater impairment of glucose uptake in frontotemporal and parietal regions (Brodmann areas 9, 47, 40, and 21) and in the left cerebellum relative to monolingual patients [bib_ref] Bilingualism as a contributor to cognitive reserve? Evidence from cerebral glucose metabolism..., Kowoll [/bib_ref]. This evidence supports the view that lifelong bilingualism may benefit the brain by making use of efficient or alternative neural networks in the event of age-related decline and that greater amounts of brain atrophy are required before the disease manifests, which may possibly delay the incidence of dementia.
## Does bilingualism protect against dementia?
The evidence for a protective effect of bilingualism on the incidence of dementia is considerable. Numerous studies have examined dementia incidence in hospital records and concluded that bilingualism exerts a protective effect. The first such study by [bib_ref] Bilingualism as a protection against the onset of symptoms of dementia, Bialystok [/bib_ref] revealed that lifelong bilinguals showed a delay in the onset of symptoms of dementia by 4 years compared to monolinguals. Similarly, [bib_ref] Delaying the onset of Alzheimer disease, Craik [/bib_ref] reported that bilingual patients had been diagnosed with Alzheimer's disease 4.3 years later and had reported the onset of symptoms 5.1 years later than the monolingual patients. Additionally, [bib_ref] Bilingualism delays clinical manifestation of Alzheimer's disease, Woumans [/bib_ref] found that bilingual patients had been diagnosed with Alzheimer's disease 4.8 years later and presented symptoms 4.6 years later than monolingual patients. Similarly, speakers of two or more languages had a delayed onset of Alzheimer's disease by up to 5 years and a protective effect was significant when speaking at least two to four languages . Looking at specific dementia subtypes, bilingualism delayed the age at onset in the behavioral but not in the aphasic variants of Frontotemporal Dementia [bib_ref] Bilingualism delays the onset of behavioral but not aphasic forms of frontotemporal..., Alladi [/bib_ref] , a finding consistent with the observation that bilingualism has positive effects on behavioral syndromes but not on language disorders. Indeed the effects of bilingualism on language functions are not always beneficial (e.g., smaller vocabulary size in a single language, slower lexical processing, reduced verbal fluency etc.). Further, a similar study by [bib_ref] Impact of bilingualism on cognitive outcome after stroke, Alladi [/bib_ref] comparing monolingual and bilingual stroke patients found that bilinguals had a significantly lower frequency of poststroke dementia and mild cognitive impairment but the same frequency of post-stroke aphasia. Moreover, [bib_ref] Does bilingualism delay the development of dementia?, Atkinson [/bib_ref] reviewed nine papers and concluded that frequent use of two languages over a lifetime may be protective against dementia, and that inconsistencies arise due to study design or definitions of bilingualism. This evidence supports the protective effect of bilingualism against the symptoms of dementia , as well as the later onset of symptoms of mild cognitive impairment compared to monolinguals . Bilingual individuals diagnosed with singledomain amnesic mild cognitive impairment demonstrated a later age of diagnosis than did monolinguals [bib_ref] The effect of bilingualism on amnestic mild cognitive impairment, Ossher [/bib_ref]. Cerebral hypometabolism was more severe in the left hemisphere in bilinguals with Alzheimer's dementia compared to monolinguals, but nevertheless bilinguals outperformed monolinguals on memory tasks, suggesting that bilinguals are better able to compensate for the loss of brain structure and function [bib_ref] The impact of bilingualism on brain reserve and metabolic connectivity in Alzheimer's..., Perani [/bib_ref]. Furthermore, exposure to foreign language instruction during childhood and adolescence has been associated with lower risk of developing mild cognitive impairment in old age [bib_ref] Early life instruction in foreign language and music and incidence of mild..., Wilson [/bib_ref]. Bilingualism has been associated with delayed onset of dementia and is also observed in illiterate patients [bib_ref] Bilingualism delays age at onset of dementia, independent of education and immigration..., Alladi [/bib_ref]. Taken together, this body of work suggests that bilingual experience delays the onset of neurodegenerative disease.
However, an increasing number of studies have failed to detect a bilingual advantage in dementia incidence. A cohort design with non-immigrant samples found no significant differences in the onset of dementia between mono-and bilingual subjects . No significant association was found between non-native English speakers and the incidence of dementia or Alzheimer's disease [bib_ref] Non-native language use and risk of incident dementia in the elderly, Sanders [/bib_ref]. In that study, nonnative English speakers with at least 16 years of education had a fourfold increased risk for dementia compared to those with less education, which is an unusual finding and inconsistent with past literature on the protective effect of education. [bib_ref] Is bilingualism associated with a lower risk of dementia in community-living older..., Yeung [/bib_ref] found no association between dementia diagnoses for bilinguals (English as a second language and bilingual English) and monolinguals. [bib_ref] Bilingualism does not alter cognitive decline or dementia risk among Spanishspeaking immigrants, Zahodne [/bib_ref] reported that adult learners of English had better memory and executive function than monolinguals, but that bilingualism was not associated with cognitive decline or dementia. Fuller-Thomson (2015) has claimed that the support for a bilingual advantage in dementia onset is questionable, and has attributed the current state of the literature to the file drawer problem, a bias against publishing non-significant findings from small studies with low to medium statistical power, a selection bias due to use of patients from a memory clinic, potential recall bias in caregivers' reporting of age of onset of dementia and confounding by immigration status. Indeed, [bib_ref] Bilingualism, executive control, and age at diagnosis among people with early-stage Alzheimer's..., Clare [/bib_ref] did not observe any advantage for delay in Alzheimer's onset in Welsh-English bilinguals over English monolinguals (but see Bak, 2016 for a discussion of how this finding is conflated by the unusual situation of monolingual migration). A recent meta-analysis concluded that bilingualism offers no protection against cognitive decline [bib_ref] The relationship of bilingualism compared to monolingualism to the risk of cognitive..., Mukadam [/bib_ref] , and that retrospective studies supporting the bilingual protective effect against dementia are marred by methodological confounds. Note, however, that this meta-analysis has already been criticized as misleading and incomplete [bib_ref] Bilingualism and cognitive decline: a story of pride and prejudice, Woumans [/bib_ref]. In sum, these studies have led to questions regarding the robustness (or in some cases the validity) of the bilingual dementia advantage.
In order to resolve the debate, attempts have been made to understand the role of any potential mediating factors and experimental confounds. [bib_ref] Degree of bilingualism predicts age of diagnosis of Alzheimer's disease in low-education..., Gollan [/bib_ref] claim that higher degrees of bilingualism are associated with increasingly later age of diagnosis and symptom onset, but this may be obscured by interactions between education and bilingualism, and a failure to obtain objective measures of bilingualism. [bib_ref] Can being bilingual affect the onset of dementia?, Bak [/bib_ref] highlight that although there exists support that bilingualism has a positive effect on cognition throughout the lifespan, common misconceptions concerning the nature of bilingualism persist, including that bilingualism is an unusual phenomenon, the holistic nature of bilingualism and its effects on cognition and bilingual diversity. Further, Fuller-Thomson's (2015) and assertions that monolinguals and bilinguals do not differ in the onset of dementia have been criticized as overly simplistic. [bib_ref] Bilingualism, dementia and the tale of many variables: why we need to..., Bak [/bib_ref] point out that it is necessary to study the effects of bilingualism separately from those of immigration and education, and to use data from both community-based approaches and memory clinics. [bib_ref] The impact of bilingualism on cognitive aging and dementia: finding a path..., Bak [/bib_ref] further highlights the importance of addressing confounding variables in bilingualism, aging and dementia research which include heterogeneity, migration, social factors, differences in general intelligence and the related issue of reverse causality.
The above literature review has demonstrated that bilingualism yields executive functioning advantages, and these may contribute to building cognitive reserve, which may ultimately delay the onset of dementia. The exact mechanisms are not agreed upon, and there exists counterevidence that limits the generalisability of these claims. A possible fruitful avenue is the recent suggestion that sustained activation of noradrenergic signaling pathways associated with bilingualism could provide a possible mechanism linking current and previous results supporting a delayed onset of dementia in bilinguals [bib_ref] Biology enters the scene-a new perspective on bilingualism, cognition, and dementia, Bak [/bib_ref]. The following sections of this article are devoted to proposing additional possible explanations and mechanisms that may provide parsimonious explanations for the seemingly conflicting findings currently in the literature.
## Age of acquisition
The majority of studies examining a bilingual advantage in cognitive aging have considered the effects of lifelong experience on cognitive function and decline. Consequently, very little attention has been paid to the age of acquisition of the second language. Age of acquisition of the second language positively correlates with cortical thickness in the left inferior frontal gyrus and a thinner cortex in the right inferior frontal gyrus [bib_ref] Age of language learning shapes brain structure: a cortical thickness study of..., Klein [/bib_ref]. Encouragingly, there is evidence of a positive effect of language experience on individuals who acquired their languages later in life. Both early and late bilinguals were found to have more efficient executive networks than monolinguals. Proficient late bilinguals showed the greatest advantage in conflict resolution, whereas early bilinguals showed enhanced monitoring processes [bib_ref] The efficiency of attentional networks in early and late bilinguals: the role..., Tao [/bib_ref]. Interestingly, [bib_ref] The neuroprotective effects of bilingualism upon the inferior parietal lobule: a structural..., Abutalebi [/bib_ref] found that age of acquisition did not correlate with gray matter volumes in the left or right inferior parietal lobules in aging bilinguals.
Age of acquisition is a complex variable in that it not only represents the level of input experienced by a learner, where early age of acquisition results in more years of exposure, but also potentially differing patterns of language use between speakers who acquired their second language in early or in later life. Such differences in language use may modulate the cognitive advantages associated with bilingualism. For instance, balanced bilinguals showed age-related decline in their inhibition abilities (as indexed by the Simon task), whereas dominant bilinguals showed no evidence of age-related decline [bib_ref] Language dominance and inhibition abilities in bilingual older adults, Goral [/bib_ref]. Further, when looking purely at amount of input, age of acquisition may need to be evaluated differently in older adulthood than it is for younger adults. For example, [bib_ref] The efficiency of attentional networks in early and late bilinguals: the role..., Tao [/bib_ref] define early acquisition as occurring by an average of 4.0 years, and late acquisition as occurring by an average of 12.3 years. Although this difference in years of second language input might be marked for young adults, it is possible that this difference is negligible for those over the age of 65. Additionally, age of acquisition may result in executive control differences, not because of biological or maturational constraints on language learning, but because age of acquisition may be a proxy for a set of environmental differences that are necessarily associated with early vs. late second language learning [bib_ref] The efficiency of attentional networks in early and late bilinguals: the role..., Tao [/bib_ref]. Indeed, those learning a language later due to migration will necessarily use their languages differently than someone learning a heritage language at an early age. Future longitudinal language training studies are needed to determine how age of acquisition modulates any cognitive improvements resulting from language learning, and whether it truly is never too late to begin language learning.
## Neuroimaging studies of language learning in adults
A large neuroscientific literature has demonstrated that lifelong bilingualism alters the structure of the brain. Recent work has confirmed that brain changes may also be observed in healthy adults following relatively short periods of language training, and a picture is emerging concerning the brain changes that subserve dynamic uses of language (see [fig_ref] TABLE 1 |: Summary of empirical studies investigating brain changes in healthy adults due to... [/fig_ref] for a summary of these findings). Interpreters who learned a foreign language intensively for 3 months showed increases in hippocampus volume and in cortical thickness in the left middle frontal gyrus, inferior frontal gyrus, and superior temporal gyrus, relative to controls; those with high proficiency showed structural malleability (right hippocampus and the left superior temporal gyrus) and struggling interpreters presented larger gray matter increases in the middle frontal gyrus [bib_ref] Growth of language-related brain areas after foreign language learning, Mårtensson [/bib_ref]. Foreign language training in students increased white matter including pathways in the right hemisphere, and correlated with gain in second language ability not observed in controls [bib_ref] Dynamic neural network reorganization associated with second language vocabulary acquisition: a multimodal..., Hosoda [/bib_ref]. English natives who spent 5 months learning Swiss German showed structural changes in the left inferior frontal gyrus which correlated with increased second language proficiency [bib_ref] Structural plasticity in the language system related to increased second language proficiency, Stein [/bib_ref]. Moreover, structural changes in gray matter (inferior parietal cortex and left inferior frontal gyrus) and white matter (anterior corpus callosum) have been repeatedly linked to second language proficiency [bib_ref] Structural brain changes related to bilingualism: does immersion make a difference?, Stein [/bib_ref]. Successful learners of a tonal language showed significant differences in language-related regions in the brain and a more coherent, integrated multi-path brain network compared to less successful learners, whereas monolinguals relied on different brain networks to process tonal and lexical information .
In sum, support has been found for second language experience-induced brain changes via increased gray matter density and white matter integrity in children, young adults, and the elderly; with such changes occurring rapidly following short-term language training. Further, these changes are sensitive to age, age of acquisition, proficiency or performance level, language-specific characteristics, and individual differences .
## The role of language typology
One factor that has so far hardly played a role in the bilingual advantage debate concerns the languages that bilinguals use. A powerful factor might be the match or mismatch in typology (types of language structure, e.g., where verbs occur, use of affixes, etc.). Such structural features indeed influence learning of a foreign language [bib_ref] The additive effect of bilingualism on third language acquisition: a review, Cenoz [/bib_ref] [bib_ref] The bilingual advantage in phonetic learning, Antoniou [/bib_ref] , so they may also affect the likelihood of language-related advantages emerging, though this has not yet been examined. There are two possible, and contrasting, roles that language typology could play in determining whether cognitive advantages result from foreign language learning. First, typologically different languages might be more demanding to learn because they share few linguistic commonalities. It has recently been proposed that tasks that are more cognitively engaging will yield greater cognitive benefits (e.g., photographic training is superior to watching documentaries; [bib_ref] The impact of sustained engagement on cognitive function in older adults: the..., Park [/bib_ref] , and for language learners, the benefits may be greatest when demands exceed their available cognitive resources [bib_ref] Cognitive consequences of trilingualism, Schroeder [/bib_ref]. Second, typologically similar languages could lead to rapid learning because they share linguistic commonalities (and cognates), leading, after attainment of some proficiency, to competition between the languages exceeding that between distant languages [bib_ref] Lexical competition in non-native spoken-word recognition, Weber [/bib_ref] [bib_ref] Competition dynamics of secondlanguage listening, Broersma [/bib_ref] [bib_ref] Representation of second language phonology, Cutler [/bib_ref]. This in turn would require greater suppression, placing greater demands on the executive function system and its associated brain structures (prefrontal cortex; [bib_ref] The role of the left putamen in multilingual language production, Abutalebi [/bib_ref]. These alternatives predict opposite results regarding the appearance of cognitive benefit, but in both cases, the relationship between a learner's native and target languages would influence the demands on their cognitive resources.
We refer to the first possibility as the processing complexity effect, according to which greater cognitive improvements will ensue from learning typologically differing languages, as these require more effort to learn and existing nativelanguage knowledge cannot be relied on. The alternative is the interference inhibition effect, according to which greater cognitive improvements will ensue from learning typologically similar languages, because similar languages interfere more, increasing demands on executive control systems in the brain.
As noted above, studies to date have typically reported advantages for individuals using multiple languages for many years. But rigorous investigation of how language learning affects cognitive function must: (a) measure the cognitive abilities of interest prior to, as well as post, language learning, and (b) experimentally manipulate who learns what language. Neither can be done with individuals who are already bilingual. Systematic investigation of the effects outlined above will require longitudinal experiments in which participants will be cognitively assessed both prior to and after completing language training. This allows (1) control of extraneous variables (e.g., education, [bib_ref] Degree of bilingualism predicts age of diagnosis of Alzheimer's disease in low-education..., Gollan [/bib_ref] ; quality and quantity of language input, Sorace and Serratrice, 2009), (2) assignment of participants to target language, and (3) experimental manipulation of the relationship between the target language and the learner's native language (i.e., typological similarity).
Although these possibilities have not yet been tested systematically, there is some evidence that typological distance will result in reliable brain differences. For example, [bib_ref] The neuroprotective effects of bilingualism upon the inferior parietal lobule: a structural..., Abutalebi [/bib_ref] observed greater gray matter volumes in the brains of aging Chinese bilinguals relative to monolinguals, specifically in the left and right inferior parietal lobules. Importantly, when comparing Cantonese-English and Cantonese-Mandarin bilinguals, both groups showed greater gray matter volumes for the right inferior parietal lobule, but only Cantonese-Mandarin bilinguals showed greater gray matter volumes for the left inferior parietal lobule. Although preliminary, this observation is consistent with our hypothesized interference inhibition effect, suggesting that two similar languages result in greater competition and place greater demands on the executive control system, requiring more inhibition to avoid language interference. This may result in brain differences that are more prominent for speakers of typologically similar languages. This interference inhibition effect could potentially go some way to providing a parsimonious explanation for conflicting findings in the literature.
## The cognitive benefits of additional language learning: bilinguals vs. multilinguals
Experience with multiple languages is thought to yield cognitive advantages that promote healthy cognitive aging, although the mechanisms responsible are not fully understood. If our proposed interference inhibition effect is correct, then conditions that increase interference will yield the greatest benefits. Therefore, knowledge of a greater number of languages would likely increase competition and the required inhibitory control and thus yield a greater cognitive benefit. However, we note that presently it is not clear if experience with a greater number of languages results in an additive benefit. There is some evidence suggesting that this may indeed be the case. Aging bilinguals and multilinguals maintain higher levels of cognitive functioning than monolinguals, irrespective of immigration and education levels [bib_ref] Multilingualism and cognitive state in the oldest old, Kavé [/bib_ref]. [bib_ref] Multilingualism (but not always bilingualism) delays the onset of Alzheimer disease: evidence..., Chertkow [/bib_ref] found evidence for a later age of onset of Alzheimer's disease symptoms in multilinguals as compared with monolinguals, whereas a limited effect was found between bilinguals compared to monolinguals. Participants who practiced more than two languages presented a lower risk of cognitive impairment without dementia, compared to bilinguals. Progressing from two to three languages was associated with a sevenfold protection against cognitive impairment without dementia [bib_ref] Lifelong exposure to multilingualism: new evidence to support cognitive reserve hypothesis, Perquin [/bib_ref]. There is some evidence that this multilingual benefit is mediated by age. Older trilingual adults showed larger advantages on cognitive reserve than bilinguals. However, younger trilingual adults and children showed the same advantages as bilinguals on inhibitory control measures. Trilingual infants and toddlers performed worse than bilinguals on memory generalization tasks [bib_ref] Cognitive consequences of trilingualism, Schroeder [/bib_ref]. In sum, it is not clear if multilingualism brings about greater cognitive benefits than bilingualism, although the present evidence suggests that it is likely to emerge under certain circumstances.
## Language learning studies with older adults
We have established that there is evidence to suggest that lifelong bilingualism may enhance executive functions, contribute to cognitive reserve, and possibly protect against Alzheimer's disease. However, whether language learning initiated in older adulthood could yield cognitive improvements remains an open research question. Previous research has shown that both healthy older adults and those at risk of neural dysfunction have demonstrated positive brain changes to training [bib_ref] Memory training alters hippocampal neurochemistry in healthy elderly, Valenzuela [/bib_ref]. This indicates that the benefits of mental stimulation are not limited to younger adults, but that even the aging brain retains its neuroplasticity, and thus training-related benefits may still be observed in older participants. Given that language learning engages an extensive neural network [bib_ref] Neurophysiological mechanisms involved in language learning in adults, Rodríguez-Fornells [/bib_ref] that overlaps with the network affected by age-related cognitive decline [bib_ref] Aging of the brain and its impact on cognitive performance, Raz [/bib_ref] , a tantalizing possibility is that language learning may promote healthy brain aging in older adults (see [bib_ref] Foreign language training as cognitive therapy for age-related cognitive decline: a hypothesis..., Antoniou [/bib_ref] for a full review). Although there is currently very little research in this area, there are positive signs that this may well be the case [fig_ref] TABLE 2 |: Summary of studies investigating language learning benefits in older adults [/fig_ref]. For example, found language learning advantages for task switching using the elevator task with reversal. In this task, participants listen to a sequence of three different tones: low, mid, and high. Participants are required to count the mid tones, add one for the high tones, and subtract one for the low tones. The advantage demonstrated on this task was found after only 1 week of intensive language training, composed of 14 h of formal classes in Scottish Gaelic. Participants were also offered Gaelic entertainment in the evenings including concerts, films and conversation circles. In addition, it was found that those individuals who continued practicing Gaelic for at least 5 h per week following the cessation of the course retained their improvement at the 9 month follow-up. Finally, and perhaps most importantly, improved attentional switching was observed across all age groups, ranging from 18 to 78 years of age, indicating that just 1 week of foreign language training can provide some cognitive benefit even for older learners. In contrast, [bib_ref] Does learning a language in the elderly enhance switching ability?, Ramos [/bib_ref] did not observe an improvement in non-verbal task switching ability in older Spanish monolinguals who participated in 8 months of Basque language classes for 5.5 h per week. However, it is not clear why these authors elected to examine task switching as an outcome measure, rather than for example inhibitory control, which would be expected to show some training-related changes following several months of language learning. The domain-general brain circuitry that subserves task switching will not necessarily be affected by language learning per se, but rather will depend critically on a bilingual's pattern of language use. For instance, constant switching between languages (as in the case of codeswitching), or the need to constantly monitor the environment for both languages would be expected to yield improvements in task switching more broadly [bib_ref] Language control in bilinguals: the adaptive control hypothesis, Green [/bib_ref]. There are three key differences between the studies conducted by and [bib_ref] Does learning a language in the elderly enhance switching ability?, Ramos [/bib_ref] that may give way to differing outcomes in terms of task switching improvement. The first is the intensity of the initial training. Participants in the [bib_ref] Does learning a language in the elderly enhance switching ability?, Ramos [/bib_ref] study had less intense training in the initial week. That is, they participated in three sessions totalling 5.5 h.
In comparison, the participants in the study completed approximately 14 h of language classes, and additional Gaelic language activities. Beyond the initial week, participants in both studies completed at least 5 h of practice in their respective languages. The second difference is that each study measured switching with a different task. [bib_ref] Does learning a language in the elderly enhance switching ability?, Ramos [/bib_ref] used the Color-Shape Task, a task commonly used to measure shifting between mental sets, whereas used the Elevator Task with Reversal, a measure of attentional switching from the Test of Everyday Attention. This can be problematic as these two tasks stem from different theoretical perspectives (the former from working memory, and the latter from attention research) it is not known whether these two tasks measure comparable constructs (see [bib_ref] Cognitive control and attentional functions, Mackie [/bib_ref] for a review on defining cognitive control and attentional functions). The third difference between these studies that may give way to improvement differences, is the context of subsequent language use. While participants in the [bib_ref] Does learning a language in the elderly enhance switching ability?, Ramos [/bib_ref] study continued formal classes for 5.5 h per week, those in the no longer continued their intensive language training. However, it was only those that continued to use Gaelic for at least 5 h per week that improved from their baseline switching performance. Given that language switching is expected to provide improvements to switching more broadly, it is likely that formal classes such as those used in the [bib_ref] Does learning a language in the elderly enhance switching ability?, Ramos [/bib_ref] study provide less opportunity for codeswitching, compared to the Gaelic learners who were practicing Gaelic in their everyday lives. Finally, a recent second language training study aimed to determine whether an English learning program implemented with French-speaking seniors would improve cognition, as well as subjective levels of loneliness and social isolation. Scores on these measures did not improve significantly, perhaps due to the small sample size or short study duration, including the length of the language learning sessions themselves. However, the study did demonstrate that a 2-h per week, technology-based language learning intervention is feasible for seniors to participate in [bib_ref] Maintaining cognitive functioning in healthy seniors with a technologybased foreign language program:..., Ware [/bib_ref]. Given that determined that 5 h per week is the minimum level of language use required for cognitive advantages to arise, future research in this area needs to determine whether this also extends to technologybased interventions. Additionally, further research investigating our proposed processing complexity and interference inhibition effects will assist researchers in determining if typological similarity can be used to maximize language training for aging populations. Whether language learning can yield cognitive improvements in older adults, and if so, under what specific conditions, remain open research questions. Answers to these questions are being pursued by research laboratories around the world.
## Language use in individuals with alzheimer's disease
One final research question concerns the potential role of language learning in individuals with mild cognitive impairment or Alzheimer's disease. Studies with Alzheimer's patients often suffer from design inconsistencies and small sample sizes.
However, a picture is starting to emerge regarding bilingual language use in Alzheimer's disease. Bilinguals diagnosed with Alzheimer's disease may exhibit cognitive impairment and lapses in attention, decreased language control ability and increased unwanted code-switching [bib_ref] Language mixing in bilingual speakers with Alzheimer's dementia: a conversation analysis approach, Friedland [/bib_ref]. Bilingual individuals with Alzheimer's disease show linguistic decrements in both their dominant and non-dominant languages [bib_ref] Language changes in bilingual individuals with Alzheimer's disease, Stilwell [/bib_ref]. English dominant bilinguals with Alzheimer's disease were more likely to name pictures in the non-dominant language than controls; and Spanish-dominant bilinguals with Alzheimer's disease were equally likely to name pictures in their nondominant language than controls [bib_ref] Degree of bilingualism predicts age of diagnosis of Alzheimer's disease in low-education..., Gollan [/bib_ref]. A case study of two bilingual patients presented early symptoms of dementia after regressing to their primary language [bib_ref] Language preference and development of dementia among bilingual individuals, Mcmurtray [/bib_ref]. Bilinguals with mild to moderate dementia had impaired retrieval of their first language (Frisian) and L2 (Dutch) naming ability, with a significant effect of age of acquisition. Earlier acquired words were better preserved and retrieved. Qualitatively, inappropriate code switching occurred within the Frisian test setting [bib_ref] Age of acquisition and naming performance in Frisian-Dutch bilingual speakers with dementia, Veenstra [/bib_ref]. These studies provide a glimpse of the effects of Alzheimer's disease on bilingual language use. Whether language training benefits Alzheimer's disease patients warrants future investigation.
# Conclusion
In this review, we have outlined the benefits of bilingualism on executive functioning and how this may increase cognitive reserve in older adults. Additionally, we have discussed how foreign language learning programs may potentially promote healthy aging and protect against cognitive decline including Alzheimer's disease, as a result of the overlap between the brain networks involved in language learning and those that decline in older age. It is proposed that future research in this area should aim to uncover the mechanisms responsible for language learning related brain advantages, and determine how language learning can be optimized to reap the maximum cognitive gains. Specifically, to achieve these aims, future research should determine the role that language typology plays in promoting healthy cognitive aging by systematically manipulating typological similarity in foreign language learning studies. In doing so, language learning programs can be customized to provide maximal cognitive advantage in line with either the processing complexity or interference inhibition effects. We also suggest that more rigorous investigation in this field could be achieved by measuring cognitive abilities prior to, as well as post language learning. Further, we have discussed the potential advantages of bilingualism vs. multilingualism, and suggest that studies that compare cognitive advantages of language learning between monolinguals learning a second language, and bilinguals learning a third, could reveal whether learning additional languages provides an additive effect. Finally, future research needs to determine the optimum language learning conditions that will provide maximum cognitive benefits in older populations. The findings from these lines of research would provide convincing evidence as to whether language learning might promote healthy cognitive aging in older adulthood and, if so, provide guidelines for how these programs should be developed to provide the greatest cognitive advantage.
[table] TABLE 1 |: Summary of empirical studies investigating brain changes in healthy adults due to language training. [/table]
[table] TABLE 2 |: Summary of studies investigating language learning benefits in older adults. [/table]
|
Factors associated with a SARS-CoV-2 recurrence after hospital discharge among patients with COVID-19: systematic review and meta-analysis* #
Background: The proportion of recurrences after discharge among patients with coronavirus disease 2019 was reported to be between 9.1% and 31.0%. Little is known about this issue, however, so we performed a meta-analysis to summarize the demographical, clinical, and laboratorial characteristics of non-recurrence and recurrence groups. Methods: Comprehensive searches were conducted using eight electronic databases. Data regarding the demographic, clinical, and laboratorial characteristics of both recurrence and non-recurrence groups were extracted, and quantitative and qualitative analyses were conducted. Results: Ten studies involving 2071 COVID-19 cases were included in this analysis. The proportion of recurrence cases involving patients with COVID-19 was 17.65% (between 12.38% and 25.16%) while older patients were more likely to experience recurrence (weighted mean difference (WMD)=1.67, range between 0.08 and 3.26). The time from discharge to recurrence was 13.38 d (between 12.08 and 14.69 d). Patients were categorized as having moderate severity (odds ratio (OR)=2.69, range between 1.30 and 5.58), while those with clinical symptoms including cough (OR=5.52, range between 3.18 and 9.60), sputum production (OR=5.10, range between 2.60 and 9.97), headache (OR=3.57, range between 1.36 and 9.35), and dizziness (OR=3.17, range between 1.12 and 8.96) were more likely to be associated with recurrence. Patients presenting with bilateral pulmonary infiltration and decreased leucocyte, platelet, and CD4 + T counts were at risk of COVID-19 recurrence (OR=1.71, range between 1.07 and 2.75; WMD=−1.06, range between −1.55 and −0.57, WMD=−40.39, range between −80.20 and −0.48, and WMD=−55.26, range between −105.92 and −4.60, respectively). Conclusions: The main factors associated with the recurrence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) after hospital discharge were older age, moderate severity, bilateral pulmonary infiltration, laboratory findings including decreased leucocytes, platelets, and CD4 + T counts, and clinical symptoms including cough, sputum production, headache, and dizziness. These factors can be considered warning indicators for the recurrence of SARS-CoV-2 and might help the development of specific management strategies.
# Introduction
Subsequent to the first report of coronavirus disease 2019 in December 2019, this disease has spread rapidly worldwide [bib_ref] Epidemiological and clinical characteristics of 99 cases of 2019 novel coronavirus pneumonia..., Chen [/bib_ref]. We know that the clinical spectrum of COVID-19 ranges widely from asymptomatic infections to death . As we have developed a deeper understanding of COVID-19, progress has been made in treatment. In China, 83 022 cases were reported as of June 4, 2020, and 78 319 of these were cured and patients were discharged (National Health Commission of the People's Republic of China, 2020). According to the World Health Organization's guidelines and scheme of diagnosis and treatment of COVID-19 (Trial 7th Edition), patients who meet discharge criteria should have at least two consecutive negative nucleic acid test results conducted over an interval of at least 24 h (Medical Administration and Hospital Authority, 2020; [bib_ref] The epidemiology and pathogenesis of coronavirus disease (COVID-19) outbreak, Rothan [/bib_ref]. The recurrence of a positive severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) RNA test result was found in some discharged patients [bib_ref] Recurrence of positive SARS-CoV-2 RNA in COVID-19: a case report, Chen [/bib_ref] [bib_ref] Recurrence of positive SARS-CoV-2 in patients recovered from COVID-19, Hoang [/bib_ref]. This phenomenon raises the concern that discharged patients may be at risk of viral reactivation and should be considered potential sources of SARS-CoV-2 infection.
The proportion of recurrence cases after discharge among patients with COVID-19 is between 9.09% and 30.77% [bib_ref] Positive result of Sars-Cov-2 in faeces and sputum from discharged patient with..., Li [/bib_ref] [bib_ref] False negative of RT-PCR and prolonged nucleic acid conversion in COVID-19: rather..., Xiao [/bib_ref] [bib_ref] Clinical characteristics of severe acute respiratory syndrome coronavirus 2 reactivation, Ye [/bib_ref]. The short inter-episode period from discharge to first recurrence was 5 d in these cases [bib_ref] The correlation between viral clearance and biochemical outcomes of 94 COVID-19 infected..., Yuan [/bib_ref]. The reasons why SARS-CoV-2 RNA has been re-detected in some discharged patients remain unclear. A number of studies have explored the factors involved in this phenomenon. [bib_ref] Factors influencing the outcome of 34 patients with COVID-19, Zhuo [/bib_ref] reported that fever on admission is as an independent risk factor that can be used to predict recurrence outcome. A number of demographical and clinical factors including age, body mass index (BMI), and low levels of some blood parameters have been found to be driving forces in a recurrence prediction model . Data show that specific characteristics exhibit substantial differences between non-recurrence and recurrence groups . We therefore performed a systematic review and meta-analysis to summarize data on de-mographic characteristics, clinical features, and laboratory findings in non-recurrence and recurrence groups as well as to explore critical factors associated with COVID-19 recurrence. These might provide more evidence for implementing preventive and controllable strategies for recurrence cases.
# Methods
## Search strategy and selection criteria
Studies published between May 1st, 2020 and May 29th, 2020 were comprehensively interrogated via the Web of Science, PubMed, medRxiv, bioRxiv, China National Knowledge Infrastructure (CNKI), SinoMed, VIP, and Wangfang databases. Search terms were "COVID-19," "2019-nCoV," "SARS-CoV-2," "severe acute respiratory syndrome coronavirus 2," "recurrence," "reinfection," "reactivation," "re-positive," and related terms. Details are presented in [fig_ref] Table 1: Studies and characteristics of COVID-19 patients included in this meta-analysis [/fig_ref].
## Inclusion and exclusion criteria
The inclusion criteria used in this analysis were:
(1) negative on two consecutive reverse transcriptasepolymerase chain reaction (RT-PCR) tests at least 24 h apart before being discharged and later presented with recurrence of SARS-CoV-2; (2) recurrence of SARS-CoV-2 among patients with COVID-19 as confirmed using RT-PCR; (3) studies with sufficient data to describe epidemiological and clinical characteristics of both recurrence and non-recurrence cases.
The exclusion criteria used in this analysis were: (1) studies with recurrence cases only; (2) duplicated publication data; (3) studies missing key data; (4) case reports, reviews, and studies lacking original data.
## Data extraction and quality assessment
The following information was extracted from each study: first author, title, journal name, publication date, study periods, regions, sample size, demographic data (e.g., gender, age), chronic medical conditions (e.g., comorbidities, smoking history), clinical features (e.g., fever, headache), laboratory results (e.g., leukocytes, neutrophils), and radiographical features (e.g., ground-glass opacity, bilateral pulmonary infiltration). Three components, including selection, comparability, and exposure or outcome, were assessed for each study using the Newcastle-Ottawa Scale [bib_ref] Systematic review and meta-analysis of the efficacy and safety of immunosuppressive pulse..., Xu [/bib_ref]. This consisted of eight items with a full score of nine stars [bib_ref] Critical evaluation of the Newcastle-Ottawa scale for the assessment of the quality..., Stang [/bib_ref]. EndNote (Version X9) was used to manage the articles and citations.
# Statistical analysis
The online tool developed by [bib_ref] Optimally estimating the sample mean from the sample size, median, mid-range, and/or..., Luo [/bib_ref] was used to convert median and interquartile range (IQR)/range to mean and standard deviation for continuous variables. Statistical differences between continuous variables were then evaluated using the weighted mean difference (WMD) and a 95% confidence interval (95% CI), while the odds ratio (OR) and its 95% CI were calculated to estimate whether, or not, categorical variables increased the risk of SARS-CoV-2 recurrence. Heterogeneity amongst each study was then assessed using I 2 ; thus, when I 2 >50%, a random-effect model was chosen, otherwise a fixed effect model was used [bib_ref] The interpretation of random-effects meta-analysis in decision models, Ades [/bib_ref]. Potential publication bias was appraised using a funnel plot and Egger weighted regression [bib_ref] Comparison of two methods to detect publication bias in meta-analysis, Peters [/bib_ref]. The software R (Version 3.2.3) was used for data clearing and analyses.
# Results
# Search results
A total of 999 records were identified via a database search. Titles and abstracts of 481 published studies were then screened after discarding duplicates and 56 articles were selected for detailed assessment. Subsequent to applying exclusion criteria, 10 studies consisting of 1803 non-recurrence and 268 recurrence cases were included in a meta-analysis [fig_ref] Figure 1: Flow diagram of the study selection process for meta-analysis between −105 [/fig_ref]. Details are presented in [fig_ref] Table 1: Studies and characteristics of COVID-19 patients included in this meta-analysis [/fig_ref].
## Demographical and time interval features of recurrence cases
The overall proportion of recurrence cases of SARS-CoV-2 among the patients with COVID-19 was 17.65% (95% CI: range between 12.38% and 25.16%; Figs. S1 and 2). Patients in recurrence cases were found to be older than those in non-recurrence cases (WMD=1.67, 95% CI: range between 0.08 and 3.26), Data show that BMI was lower in recurrence than in non-recurrence cases (WMD=−1.20, 95% CI: range between −2.05 and −0.35). No statistical dif-ferences were seen in the time from onset to negative SARS-CoV-2 RNA test results between recurrence and non-recurrence groups. The time from discharge to recurrence of SARS-CoV-2 was 13.38 d (95% CI: range between 12.08 and 14.69 d; [fig_ref] Table 2: WMD and 95% CI of demographical and time interval between recurrence and... [/fig_ref].
## Clinical data on recurrence cases
Patients were categorized as having moderate severity conditions (OR=2.69, 95% CI: range between 1.30 and 5.58) while those with clinical symptoms including cough (OR=5.52, 95% CI: range between 3.18 and 9.60), sputum production (OR=5.10, 95% CI: range between 2.61 and 9.97), headache (OR=3.57, 95% CI: range between 1.36 and 9.35), and dizziness (OR=3.17, 95% CI: range between 1.12 and 8.96) were more likely to experience recurrence. Patients who presented with bilateral pulmonary infiltration were also at risk of recurrence (OR=1.71, 95% CI: range between 1.07 and 2.75; Figs. S2a and 3a). Laboratory findings show that leucocytes (WMD=−1.06, 95% CI: range between −1.55 and −0.57), platelets (WMD=−40.39, 95% CI: range between −80.20 and −0.48), and CD4 + T (WMD=−55.26, 95% CI: range [bib_ref] Transmission scenarios for Middle East respiratory syndrome coronavirus (MERS-CoV) and how to..., Cauchemez [/bib_ref] [bib_ref] Origin and evolution of pathogenic coronaviruses, Cui [/bib_ref]. This is important because as CoVs are an emerging infectious agent, a number of notable studies have been conducted to explore the biological, epidemiological, and clinical characteristics of the new CoV [bib_ref] Epidemiology of the 2020 pandemic of COVID-19 in the state of Texas:..., Khose [/bib_ref] [bib_ref] Clinical observation and management of COVID-19 patients, Li [/bib_ref] [bib_ref] Distribution of ACE2, CD147, CD26, and other SARS-CoV-2 associated molecules in tissues..., Radzikowska [/bib_ref]. Throughout the early COVID-19 epidemic, the recurrence of positive SARS-CoV-2 RNA test results after hospital discharge
## Fig. 2 overall proportion of recurrence cases among the covid-19 patients
appeared in at least seven provinces in China while approximately 14% of discharged patients in Guangdong Province experienced recurrences . A comprehensive understanding of this phenomenon is lacking, however. Patient outcomes are often influenced by numerous factors. The proportion of recurrence cases among patients with COVID-19 was 17.65% while the median time between discharge and the recurrence of a positive SARS-CoV-2 RNA test result was 13.38 d. In order to detect recurrence cases early, the diagnosis and treatment of COVID-19 (Trial 7th Edition) should follow a revised management strategy of the discharged patient from "should continue 14 d self-health condition monitoring" to "continue 14 d isolation management and health condition monitoring" . Results show that older aged people are at higher risk of positive SARS-CoV-2 RNA recurrence. Previous research has shown that older people are more likely to develop severe COVID-19 and even die [bib_ref] Characteristics of and important lessons from the coronavirus disease 2019 (COVID-19) outbreak..., Wu [/bib_ref]. Thus, due to poor self-health status, lower immunity, and chronic medical conditions, older patients have relatively weak virus clearance rates and tolerance, and may lead to recurrence. Moderate severity cases easily developed into recurrence cases, a conclusion supported by previous work . Data show statistical differences between cough and sputum production both recurred and did not in the study of. These cases are associated with a higher risk of recurrence. Headaches and dizziness are the most common neurological manifestations associated with COVID-19; however, detailed mechanisms have yet not to be reported and so further research is needed [bib_ref] Neurological manifestations and complications of COVID-19: a literature review, Ahmad [/bib_ref]. Integrated into larger-scale sample sizes from each research, headaches and dizziness were also associated with recurrence. Thus, in the study of [bib_ref] Clinical course and risk factors for recurrence of positive SARS-CoV-2 RNA: a..., Chen [/bib_ref] , elevated interleukin-6 (IL-6) levels and increased lymphocytes were also shown to be risk factors associated with recurrence. In our study, patients with bilateral pulmonary infiltration and decreased leucocyte, platelet, and CD4 + T counts were more likely to experience recurrence. Indeed, in multivariate regression analysis, patients who presented with bilateral pulmonary infiltration turned out to be at serve recurrence risk for positive SARS-CoV-2 RNA. [bib_ref] Clinical course and risk factors for recurrence of positive SARS-CoV-2 RNA: a..., Chen [/bib_ref] assumed that these results may be attributed to potentially undetectable amount of SARS-CoV-2 which persists in respiratory epithelia during patient recovery. SARS-CoV-2 infection can also lead to a severe dysfunctional immune response [bib_ref] Tropism, replication competence, and innate immune responses of the coronavirus SARS-CoV-2 in..., Hui [/bib_ref]. Studies have also shown that patients with abnormal, over-activated immunity and "cytokine storms," characterized as immune cell infiltration and elevated pro-inflammatory cytokine release, are also at high risk [bib_ref] COVID-19 and asthma: reflection during the pandemic, Liu [/bib_ref] [bib_ref] Immunological and inflammatory profiles in mild and severe cases of COVID-19, Song [/bib_ref]. Immune cells such as white blood cells enable the body to defend itself against external infections and play an important role in response to viruses (Liu W
## Fig. 3 clinical characteristic and laboratory findings associated with recurrence cases
OR, odds ratio; CI, confidence interval; WMD, weighted mean difference et al., 2020). [bib_ref] Clinical features predicting mortality risk in patients with viral pneumonia: the MuLBSTA..., Guo [/bib_ref] previously reported that CD3 + T, CD3 + CD4 + T, and CD3 + CD8 + T cells significantly decreased in fatal COVID-19 cases [bib_ref] Clinical features predicting mortality risk in patients with viral pneumonia: the MuLBSTA..., Guo [/bib_ref] [bib_ref] Cause analysis and treatment strategies of "recurrence" with novel coronavirus pneumonia (COVID-19)..., Zhou [/bib_ref]. Over-activated T and T-cell inhibitory factor expression may also lead to decreased T cell exhaustion [bib_ref] SARS-CoV-2 T cell immunity: specificity, function, durability, and role in protection, Altmann [/bib_ref] [bib_ref] Increased CD95 (Fas) and PD-1 expression in peripheral blood T lymphocytes in..., Bellesi [/bib_ref] [bib_ref] Severe immunosuppression and not a cytokine storm characterizes COVID-19 infections, Remy [/bib_ref]. Decreased cellular immune function might also lead to incomplete virus clearance and recurrences. Indeed, in study of [bib_ref] Thrombocytopenia is associated with severe coronavirus disease 2019 (COVID-19) infections: a meta-analysis, Lippi [/bib_ref] , a lower platelet count was significantly correlated with severe COVID-19 and death cases while thrombocytopenia in COVID-19 implies serious organ malfunction or physiologic decompensation. The progression of COVID-19 promotes the activation of platelets and alters their function, which means that SARS-CoV-2 interacts with platelets during the infection process [bib_ref] Platelet activation and platelet-monocyte aggregate formation trigger tissue factor expression in patients..., Hottz [/bib_ref]. These results may explain the lower platelet count, also found in recurrence cases. Several other possible factors could also induce a SARS-CoV-2 recurrence. Firstly, false-negative RT-PCR results are due to sample sources, sampling procedures, and sensitivity/specificity of a test kit. Secondly, SARS-CoV-2 RNA shed in feces from at least two weeks to more than one month, and up to a maximum of 83 d has also been reported [bib_ref] A case series of children with 2019 novel coronavirus infection: clinical and..., Cai [/bib_ref] [bib_ref] Prolonged SARS-CoV-2 RNA shedding: not a rare phenomenon, Li [/bib_ref]. Thirdly, when the body is infected with a virus, a specific antibody is produced and therefore a body will have a certain period of protection against secondary infections. It remains unknown whether, or not, recovered patients will be immune when exposed again to the virus.
# Conclusions
A total of 1803 cases of non-recurrence as well as 268 recurrences were included in this study to analyze the factors contributing to SARS-CoV-2 reappearances. Older age, moderately severe cases, and clinical manifestations including cough, sputum production, headache, and dizziness are all recurrence risk factors. Radiographic signs and laboratory findings including bilateral pulmonary infiltration as well as decreased numbers of leucocytes, platelets, and CD4 + T cells are more likely associated with SARS-CoV-2 recurrences. These factors should be consid-ered as early warning signs for recurrence. It is therefore necessary to self-quarantine for 14 d after discharge, while process management from diagnosis, isolation, and treatment to discharge requires strict control.
## Contributors
Zhi CHEN, Meng-qi YAO, and Qiu-xian ZHENG designed the analysis. Meng-qi YAO, Qiu-xian ZHENG, and Jia XU finished the analysis and prepared the manuscript. Jing-wen DENG, Tian-tian GE, Hai-bo ZHOU, and Feng-tian WU prepared the figures and tables. Xin-yu GU, Qin YANG, Yan-li REN, and Gang WANG prepared the tables and edited this manuscript. Zhi CHEN supervised the study. All authors reviewed and approved the final version of the manuscript.
## Compliance with ethics guidelines
Meng-qi YAO, Qiu-xian ZHENG, Jia XU, Jing-wen DENG, Tian-tian GE, Hai-bo ZHOU, Feng-tian WU, Xin-yu GU, Qin YANG, Yan-li REN, Gang WANG, and Zhi CHEN declare that they have no conflict of interest.
This article does not contain any studies with human or animal subjects performed by any of the authors.
## List of electronic supplementary materials
[fig] Figure 1: Flow diagram of the study selection process for meta-analysis between −105.92 and −4.60) all decreased in recurrence cases (Figs. S2b and 3b). The publication bias of the recurrence case proportion among the COVID-19 patients was illustrated in Fig. S3. 4 Discussion Subsequent to the outbreak of SARS in 2002 and Middle East respiratory syndrome (MERS) in 2012, coronavirus (CoV) transmission between animals and humans has been confirmed [/fig]
[table] Table 1: Studies and characteristics of COVID-19 patients included in this meta-analysis [/table]
[table] Table 2: WMD and 95% CI of demographical and time interval between recurrence and non-recurrence of COVID-19 patientsWMD, weighted mean difference; CI, confidence interval; OR, odds ratio; BMI, body mass index [/table]
[table] Table S1: Search strategy of the recurrence cases with COVID-19 Fig. S1 Forest plots of meta-analysis on the proportion of the asymptomatic individuals among the COVID-19 infections Fig. S2 Forest plots of meta-analysis on clinical characteristics and laboratory findings of the recurrence cases among the COVID-19 infections Fig. S3 Funnel plot assessing publication bias in studies reporting proportion of recurrence cases among the COVID-19 patients中文概要 题 目：新型冠状病毒肺炎患者出院后 SARS-CoV-2 复阳 的危险因素：系统评价和 meta 分析 概 要：通过检索筛选到 10 项涉及 2071 例新型冠状病毒 肺炎(COVID-19)患者出院后复查严重急性呼 吸综合征冠状病毒 2 型(SARS-CoV-2)核酸检 测的研究，总结其流行病学、临床症状和辅助检 查特征。COVID-19 患者出院后病毒检测复阳病 例占比为 17.65%， 而年龄较大的患者更有可能病 毒复阳，临床症状合并咳嗽、咳痰、头晕症状的 患者有 SARS-CoV-2 复阳的风险。此外，辅助检 查结果呈双侧肺浸润且白细胞、 血小板和 CD4 + T 计数降低的患者有 SARS-CoV-2 病毒复阳的风 险。这些因素可以被视为 SARS-CoV-2 复阳的预 警指标，并可能有助于临床制定个体化管理策 略。 关键词：复阳病例；严重急性呼吸综合征冠状病毒 2 型 (SARS-CoV-2)；危险因素；meta 分析 [/table]
|
Allele dynamics plots for the study of evolutionary dynamics in viral populations
[fig_ref] Table S3: Test for directional evolution of protein sequences [/fig_ref] [fig_ref] Table S4: Test [/fig_ref]
## Steinbrueck and mchardy
ACC67219, ACC67220, ACC67221, ACC67222, ACC67223, ACC67224, ACC67225, ACC67226, ACC67227, ACC67229, ACC67230, ACC67231, ACC67232 , ACC67479, ACC67486, ACC67491, ACC67494, ACC67495, ACC67496, ACC67510, ACC67511, ACC67513, ACC67516, ACC67521, ACC67523, ACC67525, ACC67526, ACC67527, ACC67530, ACC67541, ACC67542, ACC67546, ACC67578, ACC67580, ACC67590, ACC67619, ACC67620, ACC67633, ACC67715, ACC67716, ACC67718, ACC67721, ACC67722, ACC67723, ACC67724, ACC67725, ACC67726, ACC67727, ACC67728, ACC67729, ACC67730, ACC67731, ACC67732, ACC67733, ACC67734, ACC67735, ACC67736, ACC67737, ACC67738, ACC67739, ACC67740, ACC67741, ACC67742, ACC67743, ACC67744, ACC67745, ACC67746, ACC67747, ACC67748, ACC67749, ACC67750, ACC67751, ACC67752, ACC67753, ACC67754, ACC67755, ACC67756, ACC67757, ACC67758, ACC67759, ACC67760, ACC67761, ACC67762, ACC67763, ACC67764, ACC67765, ACC67766, ACC67767, ACC67768, ACC67769, ACC67770, ACC67771, ACC67772, ACC67773, ACC67774, ACC67775, ACC67776, ACC67777, ACC67778, ACC67779, ACC67780, ACC67781, ACC67782, ACC67783, ACC67784, ACC67785, ACC67786, ACC67787, ACC67788, ACC67789, ACC67790, ACC67791, ACC67792, ACC67793, ACC67794, ACC67795, ACC67796, ACC67797, ACC67798, ACC67799, ACC67800, ACC67801, ACC67802, ACC67803, ACC67804, ACC67805, ACC67806, ACC67807, ACC67808, ACC67809, ACC67810, ACC67811, ACC67812, ACC67813, ACC67814, ACC67815, ACC67816, ACC67817, ACA33679, ACA33681, ACA33743, ACD64992, ACF36378, ACF36413, ACF36414, ACF36421, ACF36422, ACF36427, ACF36434, ACF36435, ACF36439, ACF36441, ACF36443, ACF36445, ACF76751, ACF76752, ACF76753, ACF76754, ACF76755, ACF76756, ACF76760, ACI25143, ACI25147, ACI25148, BAH20654, BAH20655, BAH20656, BAH20657, BAH20658, BAH20659, BAH20660, BAH20661, BAH20662, BAH09174, BAH09175, BAH09176, BAH09177, BAH09178, BAH09179, BAH09180, BAH09181, BAH09182, BAH09183, BAH09184, BAH09185, BAH09186, BAH09187, BAH09188, BAH09189, BAH09190, BAH09191, BAH09192, BAH09193, BAH09194, BAH09195, BAH09196, BAG72222, BAG72243, ABR68692, ABR68693, ABR68694, ABR87638, ABR87639, ABR87640, ABR87641, ABR87642, ABR87643, ABR87644, ABR87645, ABR87646, ABR87647, ABR87648, ABR87649, ABR87650, ABR87651, ABR87652, ABR87653, ABR87654, ABR87655, ABS00288, ABS00289, ABS00290, ABS00291, ABS00292, ABS00293, ABS00294, ABS00295, ABS00296, ABS00297, ABS00298, ABS00299, ABS00300, ABS00301, ABS00302, ABS00303, ABS11181, ABS11182, ABS11183, ABS11184, ABS11185, ABS11186, ABS11187, ABS11188, ABS11189, ABS11190, ABS11191, ABS11192, ABV29623, ABV29711, ABV29722, ABV29810, ABV29821, ABV29898, ABV29909, ABV29931, ABV30063, ABV30074, ABV30118, ABV30206, ABV30217, ABV30228, ABV30239, ABV30250, ABV30261, ABV30272, ABV30382, ABV30393, ABV30404, ABV30415, ABV30426, ABV30437, ABV30448, ABV30470, ABV30481, ABV30514, ABV30635, ABV30646, ABV30657, ABW38363, ABV45860, ABV45904, ABV45970, ABV45981, ABV45992, ABV82562, ABW36168, ABW39872, ABW39938, ABW40026, ABW40081, ABW40136, ABW40169, ABW40191, ABW40246, ABW40268, ABW40323, ABW40378, ABW40587, ABW40631, ABW40653, ABW71349, ABW86563, ABW91196, ABW91207, ABW91229, ABW91240, ABW91251, ABW91262, ABW91394, ABW91438, ABW91548, ABW91647, ABY51028, ABY51127, ABY51226, ABY51270, ABY81371, ACA24174, ACA24175, ACA24176, ACA24177, ACA24178, ACA24179, ACA24180, ACA24181, ACA24182, ACA24183, ACA24184, ACA24185, ACA24186, ACA24502, ACA24689, ACA24700, ACA28722, ACA28723, ACD56258, ACD56269, ACD62296, ACD62299, ACD62302, ACD62305, ACD62308, ACD62311, ACD62314, ACD62317, ACD62320, ACD62323, ACD62326, ACD62332, ACD62335, ACD62338, ACD62341, ACD62344, ACD62347, ACD62350, ACD62353, ACD62356, ACD62359, ACD62362, ACD62365, ACD62368, ACD69065, ACD69066, ACD69067, ACD69068, ACD69069, ACD69070, ACD69071, ACD69078, ACD69079, ACD85429, ACD93579, ACD85440, ACD85451, ACD85462, ACD85473, ACD85484, ACD85495, ACD85506, ACD85517, ACD85528, ACD85539, ACD85550, ACD85561, ACD85572, ACD85583, ACD87453, ACF22408, ACF41669, ACF54532, ACF54543, ACG70142, ACG70153, ACI26318, ACI26395, ACI26406, ACI26417, ACJ09818, ABU50560, ABU50579, ABU50581, ABU50582, ABU50583, ABU50584, ABW23422, ABW23424, ABW23241, ABW23244, ABW23245, ABW23246, ABW23247, ABW23248, ABW23249, ABW23250, ABW23251, ABW23252, ABW23253, ABW23254, ABW23255, ABW23256, ABW23257, ABW23258, ABW23259, ABW23260, ABW23261, ABW23263, ABW23264, ABW23265, ABW23266, ABW23267, ABW23268, ABW23269, ABW23270, ABW23271, ABW23272, ABW23330, ABW23331, ABW23332, ABW23333, ABW23334, ABW23344, ABW23345, ABW23346, ABW23347, ABW23348, ABW23349, ABW23350, ABW23351, ABW23352, ABW23353, ABW23354, ABW23355, ABW23356, ABW23357, ABW23358, ABW23359, ABW23360, ABW23361, ABW23362, ABW23363, ABW23364, ABW23365, ABW23366, ABW23367, ABW90787, ABW90790, ABZ85889, ABZ85892, ABZ85893, ABZ85894, ABZ85895, ABZ85896, ABZ85899, ABZ85900, ABZ85901, ABZ85903, ABZ85904, ABZ85915, ABZ85916, ABZ85917, ABZ85918, ACA33674, ACA33680, ACA33682, ACA33683, ACA33684, ACA33685, ACA33686, ACA33687, ACA33688, ACA33689, ACA33690, ACA33691, ACA33692, ACA33693, ACA33694, ACA33695, ACA33696, ACA33697, ACA33698, ACA33699, ACA33700, ACA33706, ACA33707, ACA33708, ACA33709, ACA33710, ACA33711, ACA33714, ACA33715, ACA33716, ACA33717, ACA33718, ACA33719, ACA33720, ACA33721, ACA33745, ACA33749, ACA33753, ACA33754, ACA33755, ACA33756, ACA33757, ACA33758, ACA33759, ACA33760, ACA33761, ACA33762, ACA33763, ACA33764, ACA33765, ACA33766, ACA33767, ACA33768, ACA33769, ACA33770, ACB11766, ACB11788, ACB11801, ACB11805, ACD64989, ACD64990, ACD64991, ACD47213, ACD47214, ACD47215, ACF40108, ACG80394, ACG80395, ACG80396, ACG80397, ACG80398, ACG80399, ACG80400, ACG80401, ACI25138, ACI25142, ACI25146, ACI22582, ACI22583, ACI22584, ACI22585, ACI22586, ACI22587, ACI22588, ACI22589, ACI22590, ACI41220, BAG85041, BAG85042, BAG85043, BAH09309, ACA24187, ACA24188, ACA24189, ACA24190, ACA24191, ACA24192, ACA24193, ACA24194, ACA24195, ACA24197, ACA24198, ACA24199, ACA24200, ACA24201, ACA24202, ACA24203, ACA24204, ACA24205, ACA24206, ACA24207, ACA24208, ACA24209, ACA24210, ACA24211, ACA24212, ACA24213, ACA24214, ACA24215, ACA24216, ACA24217, ACA24218, ACA24219, ACA24220, ACA24221, ACA24222, ACA24223, ACA24224, ACA24196, ACA65920, ACA65905, ACA65906, ACA65907, ACA65908, ACA65909, ACA65910, ACA65911, ACA65912, ACA65913, ACA65914, ACA65915, ACA65916, ACA65917, ACA65918, ACA65919, ACA65921, ACA65922, ACA65923, ACD13250, ACD13251, ACD13252, ACD13253, ACD13254, ACD13255, ACD13256, ACD13257, ACD13258, ACD13259, ACD13260, ACD13261, ACD13262, ACD13263, ACD13264, ACD13795, ACD69080, ACD69081, ACD69082, ACD69083, ACD69084, ACD69085, ACD69086, ACD69087, ACD69088, ACD69089, ACD69090, ACD69091, ACD69092, ACD69093, ACD69094, ACD69095, ACD69096, ACD69097, ACD69098, ACD69099, ACD69100, ACD69101, ACD69102, ACD69103, ACD69104, ACD69105, ACD69106, ACD69107, ACD69108, ACD69109, ACD69110, ACD69111, ACD69112, ACD69113, ACD69114, ACD69115, ACD69116, ACD69117, ACD69118, ACD69119, ACD69120, ACD69121, ACD69122, ACD69123, ACD69124, ACD69125, ACD69126, ACD69127, ACD69128, ACD69129, ACD69130, ACD69131, ACD69132, ACD69133, ACD69134, ACD69135, ACD69136, ACD69137, ACD69138, ACD69139, ACD69140, ACD69141, ACD69142, ACD69143, ACD69144, ACD69145, ACD69146, ACD69147, ACD69148, ACD69149, ACD69150, ACD69151, ACD69152, ACD69153, ACD69154, ACD69155, ACD69156, ACD69157, ACD69158, ACD69159, ACD69160, ACD69161, ACD69162, ACD69163, ACD69164, ACD69165, ACD69166, ACD69167, ACD69168, ACD69169, ACD69170, ACD69171, ACD69172, ACD69173, ACD69174, ACD69175, ACD69176, ACD69177, ACD69178, ACD69179, ACD69181, ACD69182, ACD69184, ACD69185, ACD69186, ACD69187, ACD69188, ACD69190, ACD69191, ACD69192, ACD69193, ACD69194, ACD69196, ACD69197, ACD69199, ACD69200, ACD69201, ACD69202, ACD69203, ACD69205, ACD69209, ACD69213, ACD69215, ACD69216, ACD69217, ACD69218, ACD69219, ACD69220, ACD69221, ACD69222, ACD69223, ACD69224, ACD69225, ACD69226, ACD69227, ACD69228, ACD69229, ACD69230, ACD69231, ACD69232, ACD69233,
## Steinbrueck and mchardy
ACD69234, ACD69235, ACD69236, ACD69237, ACD69238, ACD69239, ACD69240, ACD69241, ACD69242, ACD69243, ACD69244, ACD69245 q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q
## 137f
## 658a
1408T *658A* Steinbrueck and McHardy
[fig] Figure S3: Sampling bias of the influenza A (H3N2) hemagglutinin sequences. Numbers of available isolate sequences by season from 1988 to 2008. [/fig]
[fig] Figure S4: AD-plot for the major surface protein and antigenic determinant of the new influenza A (H1N1) based on genomic sequences sampled between April and December of 2009 without allele frequency correction. Alleles that reach a prevalence of more than 95% and are subsequently fixed are shown in color; all other alleles are shown in gray. Substitutions are enumerated according to H3 HA1 numbering. [/fig]
[fig] Figure S5: Sampling bias of the swine-origin influenza A (H1N1) hemagglutinin sequences. Numbers of available isolate sequences between April 2009 and December 2009. [/fig]
[fig] , GQ162183 ,: GQ162191, GQ162197, GQ162200, GQ162204, CY043094, GQ160607, GQ232010, CY043086, GQ117067, GQ117097, GQ160606, GQ377093, CY044251, FJ984385, GQ232002, GQ232076, CY041549, FJ974026, FJ984360, FJ984364, FJ984397, FJ984401, FJ998207, FJ998209, GQ117024, GQ117051, GQ117059, GQ117062, GQ117100, GQ160579, GQ160601, GQ168617, GQ168628, GQ168633, GQ168642, GQ168644, GQ168661, GQ200221, GQ200250, GQ221814, GQ221818, GQ232019, GQ232028, GQ232049, GQ338349, CY040637, CY040888, CY041122, CY041130, CY041541, CY041613, CY044163, FJ969509, FJ973557, FJ984337, FJ984347, FJ984355, FJ984375, FJ984394, GQ117032, GQ117040, GQ117082, GQ117086, GQ117091, GQ117103, GQ117116, GQ160527, GQ160543, GQ160556, GQ160567, GQ160602, GQ160613, GQ168650, GQ168652, GQ168655, GQ168668, GQ168671, GQ168673, GQ168851, GQ168886, GQ205434, GQ205436, GQ205438, GQ221786, GQ231995, GQ231996, GQ231999, GQ232017, GQ232038, GQ232064, GQ377061, GQ377103, GQ457497, CY039893, CY041050, CY041058, CY041066, CY041098, CY041170, CY041530, CY041605, CY041734, GQ117043, GQ117056, GQ117079, GQ117112, GQ160538, GQ160574, GQ168620, GQ168676, GQ221788, GQ221794, GQ221805, GQ221820, GQ231981, GQ265528, GQ265529, GQ265530, GQ338409, GQ457482, GQ465677, CY039901, CY039999, CY040007, CY040015, CY040023, CY040042, CY040822, CY040838, CY041082, CY041514, CY041522, CY041565, GQ117119, GQ160526, GQ160550, GQ160591, GQ160599, GQ160605, GQ168664, GQ183612, GQ200237, GQ200268, GQ200273, GQ221809, GQ221812, GQ232003, GQ232023, GQ265531, GQ338364, GQ338368, GQ338391, GQ338414, GQ402191, GQ465678, GQ894929, CY040629, CY040701, CY040709, CY040717, CY041090, CY041106, CY041114, CY041146, CY041154, CY041186, CY041202, CY041490, CY041498, CY041506, CY041557, CY041573, CY041581, CY041589, CY041830, CY046331, CY047318, CY047326, CY050067, CY053071, GQ160531, GQ160541, GQ160578, GQ160582, GQ160586, GQ160594, GQ166661, GQ168623, GQ168631, GQ168636, GQ200208, GQ221791, GQ221823, GQ221826, GQ232037, GQ232044, GQ232057, GQ265533, GQ323564, GQ338355, GQ338394, GQ338400, GQ402192, GQ465673, CY039527, CY040031, CY040830, CY040854, CY040862, CY041162, CY041194, CY041482, CY041726, CY046299, CY046307, CY047310, FJ982430, GQ150335, GQ160565, GQ200231, GQ200255, GQ200270, GQ214151, GQ232014, GQ247724, GQ265534, GQ265535, GQ265536, GQ338339, GQ338372, GQ338375, GQ338416, GQ365658, GQ377087, GQ402189, GQ402190, GQ402193, GQ465674, GQ465675, CY040645, CY040693, CY041621, CY046259, CY046355, CY046387, CY046411, CY046419, CY046451, CY046475, CY046515, CY046843, CY047350, CY049828, GQ166194, GQ168606, GQ214138, GQ214144, GQ232009, GQ232033, GQ232052, GQ232054, GQ323443, GQ323451, GQ323483, GQ338358, GQ338402, GQ377045, GQ377075, GQ377101, GQ465676, GQ465681, GQ894862, CY039986, GQ144465, GQ152374, GQ258462, GQ377035, CY040457, CY041178, CY041637, CY046187, CY046291, CY046315, CY046427, CY046459, CY046491, CY046971, CY047342, CY053079, CY053087, CY053095, CY053103, GQ166195, GQ166196, GQ166197, GQ214156, GQ231987, GQ232067, GQ249333, GQ323473, GQ323560, GQ338378, GQ377085, GQ457517, GQ465679, CY041074, CY041597, CY041806, CY046195, CY046203, CY046235, CY046267, CY046339, CY046379, CY046395, CY046435, CY046851, CY049859, CY049867, CY049875, GQ131023, GQ365666, GQ894905, CY046211, CY046227, CY046243, CY046275, CY046283, CY046347, CY046371, CY046883, GQ251035, CY041645, CY041782, CY043171, CY043203, CY045482, CY046219, CY046251, CY046323, CY046363, CY046403, CY046443, CY046467, CY046483, CY046499, CY046507, CY046523, CY046531, CY046859, CY046867, CY046875, CY046891, CY046915, CY049843, CY050142, CY053142, CY053150, GQ232007, GQ232035, GQ232070, GQ232073, GQ323464, GQ323470, GQ365674, GQ377043, GQ402196, CY049820, GQ231990, GQ231993, GQ323520, GQ457501, GQ457511, CY041750, CY041814, CY041822, CY043195, CY045490, CY046555, CY047334, GQ169382, GQ329088, GQ329106, GQ377064, GQ402198, CY046603, CY050198, GQ160611, GQ202695, GQ323495, GQ323509, GQ323551, GQ338361, GQ402195, GQ465682, GQ465683, CY040880, CY041766, CY043123, CY046539, CY046571, CY046579, CY046587, CY050206, CY050214, CY050222, CY050879, GQ165814, GQ165815, GQ227545, GQ338405, GQ402197, GQ465680, CY045946, CY046563, CY046611, CY050174, CY050230, CY050238, CY050246, CY050254, CY050847, CY050863, CY050871, GQ323446, CY046547, CY046595, CY050166, CY050444, CY050887, GQ166223, GQ183633, GQ200287, CY041758, CY043163, CY043219, GQ232060, GQ249337, GQ323455, CY043155, CY043179, CY043299, CY050328, CY050855, GQ323574, GQ323576, CY043211, CY043283, CY046675, CY050372, CY051903, CY052959, CY053055, CY053063, CY041790, CY043131, CY046651, CY046659, CY046667, CY046683, CY044973, CY046627, CY048925, CY050380, CY050388, GQ225373, CY041774, CY043187, CY043235, CY043243, CY043251, CY043291, CY044981, CY046619, CY046635, CY046643, CY046691, CY048933, CY051984, GQ183617, GQ183625, GQ219578, GQ219579, GQ219580, CY041798, CY043227, CY043307, CY044128, CY045085, CY050396, GQ219575, GQ219576, GQ219581, GQ221694, GQ329100, GQ402199, CY044088, CY044909, CY051911, CY052967, CY052983, CY052991, GQ223440, GQ261272, GQ268003, GQ377069, CY043139, CY043267, CY044064, CY044136, CY044196, CY044885, CY044917, CY045143, CY046715, CY046739, CY046755, CY046771, CY047358, CY050412, CY052999, CY053007, GQ243751, GQ287621, GQ323486, GQ323489, CY043147, CY043259, CY044048, CY044056, CY044869, CY044893, CY044933, CY046707, CY046723, CY046731, CY053015, CY053023, CY053174, CY053182, GQ225381, GQ243753, GQ290106, GQ402200, GQ414764, AB514227, CY044877, CY044901, CY044965, CY046963, CY051919, CY052975, CY053031, CY053039, CY053047, CY053190, CY053198, GQ184630, GQ219586, GQ223408, GQ232099, GQ243761, GQ261275, GQ267839, GQ402203, CY044204, CY044212, CY044949, CY045093, CY046747, CY046763, CY053206, CY053214, GQ225365, GQ255897, GQ261277, GQ329082, GQ377047, GQ402202, CY049907, CY049915, CY050404, CY050420, CY053127, CY053135, CY053222, CY053230, GQ225349, GQ225357, GQ232093, GQ243749, GQ323530, GQ329066, GQ338335, CY041960, CY041983, CY044957, CY045938, CY046699, GQ253492, GQ323579, GQ402187, GQ402201, CY043275, CY044925, CY050428, GQ243759, GQ243760, GQ293441, CY043102, CY043110, CY044072, CY044941, CY045151, CY046779, CY046955, CY051247, GQ243755, GQ247726, GQ255900, GQ283488, GQ375284, GQ385300, GU014804, CY046787, CY046811, CY046835, CY049931, CY050436, CY050895, CY051985, CY053238, CY053246, CY053254, CY053901, GQ232085, GQ283484, GQ368664, CY044220, CY045005,CY046795, CY046827, CY046899, CY046907, CY050150, CY050903, CY050911, CY053261, CY053309, GQ265537, GQ283493, GQ288372, GQ368665, GQ402204, GQ411908, CY044080, CY046946, CY049068, CY050919, CY050927, CY050935, CY051255, GQ223435, GQ243757, GQ250161, GQ359765, GQ377095, GQ402205, GQ414765, GQ414766, GQ414767, CY044228, CY045962, CY046803, CY046819, CY046949, CY053269, CY053277, GQ360060, GQ402206, CY044861, CY045029, CY045183, GQ253498, GQ293077, GQ377107, CY045226, CY045234, CY045242, GQ359758, GU369647, CY044989, CY045175, CY047366, CY047374, CY050943, CY050951, CY050959, CY050967, CY052349, CY053285, CY053910, FN423713, GQ338381, GQ356787, GQ365368, CY044997, CY045013, CY045021, CY045159, CY045167, CY049076, CY050975, CY050983, CY051223, CY051263, CY051271, CY051279, CY053293, CY053301, GQ287625, GQ329070, GQ329076, GQ329093, GQ334346, GU014776, AB514226, CY044096, CY045495, CY051231, GQ265538, GQ330645, GQ411897, GU014808, CY043334, CY043342, CY051015, CY051023, CY051031, CY051039, CY051047, CY051287, GQ255901, CY044104, CY050991, CY052138, CY053896, GQ280797, GQ368667, GQ396732, GQ287623, GQ351319, GQ411907, CY053119, GQ287619, GQ334330, GQ334338, CY045207, CY051055, CY051063, CY051071, CY051295, CY051303, CY051311, CY051319, CY051327, CY051495, [/fig]
[fig] , CY051543 ,: CY052098, CY053936, GQ392029, GQ455032, GQ866959, GU014750, CY043118, CY044179, CY044187, CY051167, CY051447, CY051455, CY053928, GQ280264, GQ293444, GQ365428, GU014778, GU014794, CY045069, CY049084, CY049364, CY049987, CY051559, CY053916, GQ359771, CY046061, CY049060, CY049140, CY051175, CY051191, GQ494354, CY051183, CY051535, CY053945, CY049092, CY049356, CY049372, CY049851, CY051199, CY053939, CY053947, CY053971, GQ387429, GU014764, GU014810, CY049292, CY049300, CY049308, CY051479, CY051551, CY053950, GQ365418, GQ387430, GU014800, CY049332, CY051471, CY052146, CY053942, CY053953, CY044147, CY049100, CY049148, CY049156, CY049316, CY049324, CY049380, CY049388, CY049396, CY049939, CY049979, CY050182, CY051527, CY052154, CY053968, GQ355303, GU014788, CY044155, CY049108, CY049164, CY049172, CY049180, CY049188, CY049340, CY049348, CY049404, CY052114, CY053166, CY053956, GQ355304, GQ355305, CY049196, CY049204, CY049212, CY049220, CY051207, CY053959, GU290047, CY046952, CY043078, CY049995, CY053907, GQ355306, GQ355307, GQ355309, GQ355310, GQ457519, GU014768, CY049963, CY051215, CY051487, CY051647, CY052045, CY053962, GQ392017, GU014784, CY047744, CY050363, GQ866928, GQ866929, GU117765, GU369646, GU369648, GU369659, CY047110, CY051567, CY051575, CY053913, CY053919, GQ421201, GQ465672, GU014782, CY049947, CY049955, CY052046, CY048994, GQ414768, GQ502906, GU014786, CY049228, CY049236, CY049244, CY049252, CY049419, CY049427, CY049435, CY049443, AB530161, CY049116, CY049451, CY049459, CY045511, CY049260, CY049268, CY049276, CY049284, CY045954, CY047090, CY053158, GU014748, GU014766, CY050190, CY051591, CY051599, CY051607, GU014756, CY049483, CY049491, CY049539, CY051583, CY051615, GQ392022, CY049531, CY049899, CY050075, CY050083, GU014752, CY047105, CY049467, CY049475, CY049547, CY052346, GQ421203, AB530249, CY049499, CY049507, CY049515, CY049523, CY049555, CY049563, GU014770, CY049571, CY049579, CY049587, CY049595, CY051623, CY051631, GQ499336, GU014758, CY047123, CY047125, CY047128, CY049603, CY049611, GU290055, CY047131, CY047134, CY051639, CY046064, CY046065, CY052048, GU014806, GQ915017, GU433025, CY047137, CY047143, CY047145, CY052049, CY052050, CY053904, CY046066, CY047150, GU014754, GU433033, CY047140, GQ496142, GU014760, GU014762, AB530461, AB530462, AB530463, CY052347, CY054279, AB530464, AB530465, AB530466, AB530467, CY047155, CY051655, GQ496149, GU290188, GU290190, GU292341, AB530468, CY051663, CY047071, GU057010, GU057011, GU057012, GU057013, GU057014, GU057015, GU057016, GU057017, GU057018, GU057019, GU057020, GU057021, AB530469, AB530470, CY051986, CY053922, GQ915076, GU226572, CY047158, CY047161, CY047167, CY047170, CY047173, CY047185, CY051671, CY054280, GQ499334, AB530471, AB535738, CY047074, CY047182, CY051987, CY052348, CY054282, GU226571, AB530472, AB530473, AB530474, AB530475, AB530476, AB535739, CY047164, CY047176, CY047179, CY051679, GU226570, CY047077, CY049619, CY049627, CY050339, CY050347, CY050355, GQ866930, GQ866931, GU369650, GU369652, GU369655, GU369657, GU369661, GU369662, GU369666, GQ915018, GQ915019, GQ915020, GQ915022, GQ915021, AB530477, AB530478, AB530479, AB530480, AB530481, CY047194, GU220604, AB530482, AB535740, CY047191, CY047197, CY047206, CY047215, CY049635, CY049643, GQ915023, GQ915024, CY047200, CY047209, CY049659, CY049667, CY049675, CY049683, CY049691, CY049699, CY051687, GU045582, CY047218, GU045583, GU045584, GQ915025, AB530483, AB530484, CY047230, CY051989, CY054281, CY047224, CY051988, CY051991, CY047221, CY047235, CY047240, CY047243, CY047248, CY051695, GU057022, GU057023, GU057024, GU057025, GU057026, GU057027, GU057028, GU057029, GU057030, GU057031, AB530485, AB530486, CY047251, CY047254, CY047257, CY051703, CY047260, CY047263, CY047266, CY047293, CY051990, AB530487, AB530488, CY051719, CY047284, CY051711, CY051727, CY051743, CY052162, CY047269, CY047272, CY047275, CY047278, CY047281, CY051735, CY047287, CY047290, CY047301, CY047307, CY048997, CY051993, CY052004, CY049000, CY049003, CY049006, CY051992, CY049009, CY051994, AB530489, GU369654, GU369656, GU369660, CY049015, CY052003, CY052019, CY050019, CY051751, CY049012, CY050003, CY051759, CY051767, CY051775, CY050035,CY051783, GU108486, CY052007, GU112090, GU189649, CY051998, CY049018, CY049021, CY050043, CY051791, CY051799, CY052005, CY052006, CY052170, AB535742, CY050051, CY050059, CY051807, CY050011, CY052178, CY049651, AB539047, CY050027, CY049024, CY053474, CY047713, CY049027, CY049030, CY049033, CY053482, CY053490, CY053498, CY053506, CY047714, CY047712, CY052008, CY049036, CY049045, CY049039, GU211227, CY049042, CY049048, CY049051, CY049054, GU369663, GU369664, GU369665, CY052009, CY049704, CY049707, AB535743, CY049710, CY049713, AB535744, AB535745, CY052011, CY049716, CY049719, CY049725, CY049722, CY052010, CY052015, GU198201, CY049728, CY049730, CY052012, CY049733, CY049739, CY049742, AB539046, CY049736, CY052014, CY052013, AB536768, CY050298, CY050301, CY050304, CY050307, CY050310, CY050844, AB535746, AB535747, AB536769, AB535741, CY050313, CY050316, CY050319, CY050322, AB535748, AB536770, CY050325, CY050845, CY050846, GU369674, CY053338, CY053348, CY053353, CY053416, CY053681, CY053689, CY053728, CY053736, CY053744, CY053752, CY051931, CY051934, CY051937, CY051943, CY051955, CY051958, CY051946, CY051949, GU323343, CY051927, CY051961, CY051964, CY051967, CY051970, CY051973, CY051976, CY051979, CY051982, CY052366, CY053627, GU369671, CY052032, CY052038, GU369672, CY052028, CY052030, CY052034, CY052036, CY052040, CY052042, GU369673, GU371270, CY053373, CY053376, CY053379, CY053382, CY053385, CY053388, CY053391, CY053394, GU371271, CY052026, GU371272, CY053397, CY053401, CY053409, AB539048, CY053405, CY053706, CY053710, CY053369, CY053714, CY053718, CY053722, CY053695, CY053699 [/fig]
[table] Table S3: Test for directional evolution of protein sequences (DEPS) and traditional dN/dS test (FEL) on the influenza A (H3N2) hemagglutinin sequences and according phylogenetic tree under the influenza A substitution model. Only positions associated with substitutions that rise to predominance inFigure 2are shown (HA1 numbering). If the empirical Bayes factor (DEPS EBF) is high enough (> 20) the position shows evidence for directional selection. Positions also detected to be under positive selection are shown in bold. [/table]
[table] Table S4: Test [/table]
|
Is metformin use associated with low mortality in patients with type 2 diabetes mellitus hospitalized for COVID-19? a multivariable and propensity score-adjusted meta-analysis
BackgroundCoronavirus disease 2019 (COVID-19) is a new pandemic that the entire world is facing since December of 2019. Increasing evidence has shown that metformin is linked to favorable outcomes in patients with COVID-19. The aim of this study was to address whether outpatient or inpatient metformin therapy for type 2 diabetes mellitus is associated with low inhospital mortality in patients hospitalized for COVID-19.MethodsWe searched studies published in PubMed, Embase, Google Scholar and Cochrane Library up to November 1, 2022. Raw event data extracted from individual study were pooled using the Mantel-Haenszel approach. Odds ratio (OR) or hazard ratio (HR) adjusted for covariates that potentially confound the association using multivariable regression or propensity score matching was pooled by the inverse-variance method. Random effect models were applied for meta-analysis due to variance among studies.ResultsTwenty-two retrospective observational studies were selected. The pooled unadjusted OR for outpatient metformin therapy and in-hospital mortality was 0.48 (95% CI, 0.37-0.62) and the pooled OR adjusted with multivariable regression or propensity score matching was 0.71 (95% CI, 0.50-0.99). The pooled unadjusted OR for inpatient metformin therapy and in-hospital mortality was 0.18 (95% CI, 0.10-0.31), whereas the pooled adjusted HR was 1.10 (95%
# Introduction
including meta-analysis. Full texts were then retrieved and reviewed. The meta-analysis was in adherence to the principles outlined by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA). The study was registered on the PROSPERO (CRD42021284113).
## Study selection and risk of bias assessment for meta-analysis
Two reviewers independently extracted the following data from each study: first author, year of publication, study population characteristics (age and gender), study size, study design, country, definition of mortality, number of metformin users, effect size including odds ratios (ORs) or hazard ratios (HRs), confounding adjustment, and timing of metformin use. Studies included in the analysis met: 1) comparing the effect of metformin use prior to admission or during hospitalization on COVID-19 patients with type 2 diabetes; 2) the raw data on events can be extracted from the manuscript or supplementary materials or effect sizes (OR, HR) were reported; and 3) outcomes include in-hospital mortality. Studies were discarded, if: 1) duplicated studies from the same Registry or dataset; 2) studies without a control group or without in-hospital mortality reported. 3) total patients with type 2 diabetes in the study < 50; 4) studies that were not published in English. Discrepancies were resolved by discussion. The risk of bias in the included studies was assessed using the Newcastle-Ottawa Scale (NOS) [bib_ref] Critical evaluation of the Newcastle-Ottawa scale for the assessment of the quality..., Stang [/bib_ref].
## Data syntheses and statistical analysis
To assess whether metformin use is associated with low in-hospital mortality by comparing and integrating the results of different studies, meta-analysis was carried out as described previously [bib_ref] Does sleeve lobectomy concomitant with or without pulmonary artery reconstruction (double sleeve)..., Ma [/bib_ref]. OR was used as the main summary statistics. The OR represents the odds of a death event occurring in the metformin group in comparison to the non-metformin group. The point of estimate of the OR is considered statistically significant at the P < 0.05 level if the 95% confidence interval does not include the value 1. Raw event data extracted from individual study were pooled using the Mantel-Haenszel approach and effect size (OR, HR) was pooled using the inverse-variance method. In our study, random effect models were employed due to variance among studies. Heterogeneity between studies was investigated by the standard chi-squared Q-test. Publication bias was assessed by Funnel plots and Egger's test. A funnel plot is a type of scatter plot that resembles an inverted funnel (the 95% confidence interval), in which the treatment effects estimated from individual studies on the horizontal axis (OR), against a measure of study size on the vertical funnel (SE [log OR]). In addition, sensitivity analyses were performed by excluding one study at a time or subgroup to assess the robustness of the results. To better limit confounding bias due to baseline characteristics, several studies conducted analyses with propensity score matching approaches, in which propensity score was derived as the probability of being treated with metformin based on the participant's observed covariates between metformin and non-metformin users. All analyses were conducted with R (version 4.1.2) and R package 'meta'.
# Results
## Study selection and characteristics
After studies were carefully reviewed based on the inclusion and exclusion criteria, 22 retrospective observational studies were selected for further meta-analysis of metformin use and inhospital mortality. The flowchart for the study selection is shown in [fig_ref] Fig 1: Flowchart illustration of the study selection [/fig_ref] Of these 22 studies, 17 studies reported on the association of outpatient metformin use and in-hospital mortality [bib_ref] Outpatient metformin use is associated with reduced severity of COVID-19 disease in..., Bramante [/bib_ref] [bib_ref] Metformin and risk of mortality in patients hospitalised with COVID-19: a retrospective..., Bramante [/bib_ref] [bib_ref] Phenotypic characteristics and prognosis of inpatients with COVID-19 and diabetes: the CORONADO..., Cariou [/bib_ref] [bib_ref] Clinical Characteristics and Outcomes of Patients With Diabetes and COVID-19 in Association..., Chen [/bib_ref] [bib_ref] Metformin Use Is Associated With Reduced Mortality in a Diverse Population With..., Crouse [/bib_ref] [bib_ref] Metformin is associated with lower hospitalizations, mortality and severe coronavirus infection among..., Ghany [/bib_ref] [bib_ref] Metformin use is associated with a reduced risk of mortality in patients..., Lalau [/bib_ref] [bib_ref] Metformin Use in Diabetes Prior to Hospitalization: Effects on Mortality in Covid-19, Li [/bib_ref] [bib_ref] Impact of Comorbidities and Glycemia at Admission and Dipeptidyl Peptidase 4 Inhibitors..., Mirani [/bib_ref] [bib_ref] Metformin use and risk of COVID-19 among patients with type II diabetes..., Oh [/bib_ref] [bib_ref] Mortality and other adverse outcomes in patients with type 2 diabetes mellitus..., Perez-Belmonte [/bib_ref] [bib_ref] Glucose-lowering drugs and outcome from COVID-19 among patients with type 2 diabetes..., Luk [/bib_ref] [bib_ref] Metformin use is associated with a decrease in the risk of hospitalization..., Ojeda-Fernandez [/bib_ref] [bib_ref] Association Between Metformin Use and Mortality among Patients with Type 2 Diabetes..., Ong [/bib_ref] , 6 studies investigated the link between inpatient metformin use and in-hospital mortality [bib_ref] Association Between Metformin Use and Mortality among Patients with Type 2 Diabetes..., Ong [/bib_ref] [bib_ref] Metformin Is Associated with Higher Incidence of Acidosis, but Not Mortality, in..., Cheng [/bib_ref] [bib_ref] Association of metformin with mortality or ARDS in patients with COVID-19 and..., Jiang [/bib_ref] [bib_ref] Inpatient use of metformin and acarbose is associated with reduced mortality of..., Li [/bib_ref] [bib_ref] Metformin Treatment Was Associated with Decreased Mortality in COVID-19 Patients with Diabetes..., Luo [/bib_ref] [bib_ref] Outcome and death risk of diabetes patients with Covid-19 receiving pre-hospital and..., Tamura [/bib_ref] , and 2 study probed the association of both outpatient and inpatient metformin use and in-hospital mortality [bib_ref] Association Between Metformin Use and Mortality among Patients with Type 2 Diabetes..., Ong [/bib_ref] [bib_ref] Outcome and death risk of diabetes patients with Covid-19 receiving pre-hospital and..., Tamura [/bib_ref]. There were 16 studies with extractable raw event data. According to adjustment for confounding bias, 16 studies performed multivariable regression or propensity scoring matching. The characteristics and quality assessment of the included studies are summarized in [fig_ref] Table 1: Characteristics of studies comparing mortality in metformin users vs non-users in patients... [/fig_ref].
## Outpatient metformin use and in-hospital mortality
Due to the inherent nature of retrospective observational studies, to better probe the in-hospital mortality benefits and risks for outpatient metformin use, we set out to perform the meta- analysis using two approaches. First, meta-analysis of 16 studies with extractable raw event data from 30,891 metformin and 25,770 non-metformin users revealed that outpatient metformin use was associated with a significant reduction of in-hospital death with a pooled unadjusted OR of 0.48 (95% CI, 0.37-0.62; I 2 = 87%). Publication bias was assessed using a
## Plos one
Metformin in patients with Type 2 diabetes mellitus hospitalized for COVID-19
funnel plot which showed the 16 studies used in our meta-analysis. Egger's test was performed to quantitively assess publication bias and revealed no significant publication bias (p = 0.86). Second, meta-analysis of 10 studies with adjusted OR using either multivariable regression or propensity score matching demonstrated a pooled OR of 0.71 (95% CI, 0.50-0.99; I 2 = 68%), suggesting that there was a significant association between decreased in-hospital mortality and outpatient metformin use . The funnel plot for the included 10 studies showed an approximate symmetric distribution . Egger's test revealed no statistically significant publication bias (p = 0.54).
In the subgroup analysis of propensity scored-matched studies, outpatient metformin use was not associated with a statistically significant but a trend toward a reduction of in-hospital mortality with a pooled OR of 0.75 (95% CI, 0.53-1.05; I 2 = 66%) . In contrast, subgroup analysis of 5 studies with adjusted HR revealed outpatient metformin use was associated with a statistically significant decrease of in-hospital mortality with a pooled HR of 0.78 (95% CI, 0.69-0.89; I 2 = 43%) . In the subgroup analysis of studies with extractable raw events, excluding studies with NOS < 7 or sample size < 100 revealed a similar significant association between outpatient metformin use and in-hospital mortality (OR 0.50; 95% CI, 0.37-0.68; I 2 = 90%) (S1 In addition, to determine the impact of individual studies on pooled effects, sensitivity analysis was performed by excluding one study a time. The pooled ORs in studies with extractable raw event data showed constantly significant differences with the upper limit of 95% CI less than 1 (S1B whereas studies with adjusted OR demonstrated unstable results in the sensitivity analysis (S1C These results indicate that outpatient metformin use is associated with a significant reduction of in-hospital death in unadjusted raw data with stability of meta-analysis and that the association was still statistically significant after adjustments for confounding but with instability in the sensitivity analysis.
## Inpatient metformin use and in-hospital mortality
There were a few studies mainly from China investigating the association between inpatient metformin use and in-hospital mortality in patients with type 2 diabetes hospitalized for COVID-19. We performed the similar analyses with extractable raw data and adjusted OR. Meta-analysis of 5 studies with extractable raw event data from 396 metformin and 743 nonmetformin users showed that inpatient metformin use was associated with a significant reduction of in-hospital death [fig_ref] Fig 3: The association of inpatient metformin use and in-hospital mortality in patients with... [/fig_ref] with a pooled unadjusted OR of 0.18 (95% CI, 0.10-0.31; I 2 = 0%). The funnel plot for the included 5 studies demonstrated a good symmetric distribution [fig_ref] Fig 3: The association of inpatient metformin use and in-hospital mortality in patients with... [/fig_ref]. However, meta-analysis of 2 studies with adjusted HR demonstrated a pooled HR of 1.10 (95% CI, 0.38-3.15; I 2 = 43%), suggesting there is no statistically significant association between inpatient metformin use and in-hospital mortality [fig_ref] Fig 3: The association of inpatient metformin use and in-hospital mortality in patients with... [/fig_ref]. No sensitivity or subgroup studies were performed due to small number of studies available.
# Discussion
Diabetes has been linked to poor outcomes in patients with COVID-19, so treatments that are effective, safe, and available immediately are urgently needed. Metformin is a the first-line therapeutic agent that is safe, effective, and inexpensive for type 2 diabetes mellitus. In the light of host-directed anti-viral properties and anti-inflammatory effects of metformin, many retrospective studies including our previous studyhas explored whether metformin is beneficial in patients with COVID-19 and preexisting type 2 diabetes mellitus. To further compare and integrate the results of different studies and to increase statistical power, we set out to perform a meta-analysis. In this meta-analysis study, we found that outpatient metformin use was associated with a significantly lower in-hospital mortality in patients with type 2 diabetes hospitalized for COVID-19 in the raw and adjusted analyses, compared to non-metformin use. We also found that there was a significant association between decreased in-hospital mortality and inpatient metformin use in the unadjusted analysis, but this significant association was not observed after adjustments for confounding factors.
Prior to this meta-analysis, several meta-analyses [bib_ref] Metformin in Patients With COVID-19: A Systematic Review and Meta-Analysis, Li [/bib_ref] [bib_ref] Metformin therapy, severity and mortality of SARS-CoV-2 infection: a meta-analysis, Oscanoa [/bib_ref] [bib_ref] The effect of metformin on mortality and severity in COVID-19 patients with..., Yang [/bib_ref] of the association between metformin therapy and severity and mortality were published, but they have issues. First, metformin therapy was not differentiated between outpatient and inpatient use, as oral hypoglycemic medications including metformin are usually held during hospitalization. Second, while study endpoints such as pre-defined mortality were somewhat different among studies, effect sizes (OR or HR) of different mortality outcomes were combined in previous meta-analysis studies. Third, due to nature of retrospective studies, various effect sizes with adjustments for baseline characteristics were reported in several studies. However, unadjusted and adjusted effect sizes were mixed in different meta-analyses. Fourth, using pre-calculated standard error instead of raw event data may not be a good estimator of the precision of the binary effect size (OR or HR). Nevertheless, only pre-calculated OR was used in the previous meta-analyses.
In this study, we specifically addressed these issues by investigating the association between outpatient or inpatient metformin therapy and in-hospital mortality in patients with diabetes hospitalized for COVID-19. We also used extractable raw data instead of pre-calculated ORs or HRs to determine the pooled OR for the unadjusted analysis. We found that there was a significant association between the reduction of in-hospital mortality and outpatient or inpatient metformin therapy for type 2 diabetes mellitus in COVID-19 patients in the unadjusted analysis, which is in line with previous meta-analyses [bib_ref] Metformin in Patients With COVID-19: A Systematic Review and Meta-Analysis, Li [/bib_ref] [bib_ref] Metformin therapy, severity and mortality of SARS-CoV-2 infection: a meta-analysis, Oscanoa [/bib_ref] [bib_ref] The effect of metformin on mortality and severity in COVID-19 patients with..., Yang [/bib_ref]. Furthermore, evidence obtained from observational studies is an of great importance source for clinical practice where randomized clinical trials are unavailable or infeasible [bib_ref] Overlap Weighting: A Propensity Score Method That Mimics Attributes of a Randomized..., Thomas [/bib_ref]. Unlike clinical trials through randomization to ensure patient characteristics are comparable across treatment and control groups, observational studies usually attempt to adjust for confounding by multivariable regression and propensity score matching. We took the advantage of these approaches and applied adjusted ORs or HRs into our meta-analysis. Similarly, we found that outpatient metformin therapy was associated with decreased in-hospital mortality in the meta-analysis of studies with adjusted OR or adjusted HR. However, when only the propensity score-matched studies [bib_ref] Outpatient metformin use is associated with reduced severity of COVID-19 disease in..., Bramante [/bib_ref] [bib_ref] Metformin and risk of mortality in patients hospitalised with COVID-19: a retrospective..., Bramante [/bib_ref] [bib_ref] Metformin use is associated with a reduced risk of mortality in patients..., Lalau [/bib_ref] [bib_ref] Mortality and other adverse outcomes in patients with type 2 diabetes mellitus..., Perez-Belmonte [/bib_ref] [bib_ref] Metformin use is associated with a decrease in the risk of hospitalization..., Ojeda-Fernandez [/bib_ref] were considered for outpatient metformin use and in-hospital mortality in patient with diabetes hospitalized for COVD-19, there was no significant association with reduced in-hospital mortality. A similar result was reached for inpatient metformin use and in-hospital mortality between the unadjusted and adjusted analyses. These discordant findings could be due to confounding bias, between-study heterogeneity and a small number of studies in the subgroup analysis. Therefore, further randomized clinical trials are needed to investigate the clinical benefits of metformin therapy in patient with diabetes hospitalized for COVD-19.
Several limitations are present in our study. First, owing to lack of randomized clinical trials, all studies included for analysis were retrospective studies. Although efforts were made to balance and control for potential confounding factors by multiple variable adjustments and propensity score matching, due to the inherent nature of retrospective observational studies, residual confounders are likely to exist and could not be balanced. Therefore, even meta-analysis of studies with adjusted ORs or HRs could be misleading. Second, the size of the retrospective studies that were included in the meta-analysis varied considerably, resulting in moderateto-high between-study heterogeneity, as revealed by high I 2 . Third, there were multiple sources of bias, resulting in high risk of bias in the meta-analysis. Confounding bias could be resulted from different patient characteristics, such as comorbidities, age, gender, and countries. Baseline of HbA1c, metformin dosage, duration of diabetes and treatment were often not reported in studies, contributing to heterogeneity and bias. Furthermore, different reported outcomes in studies could lead to selection and measurement bias. Fourth, there were very few studies investigating inpatient metformin use and in-hospital mortality. More studies are needed to perform a robust meta-analysis for inpatient metformin therapy and in-hospital mortality in patient with diabetes hospitalized for COVD-19.
In conclusion, our findings demonstrate that there is a significant association between the reduction of in-hospital mortality and outpatient metformin therapy for type 2 diabetes mellitus in patients hospitalized for COVID-19 and that inpatient metformin use is associated with a significant decrease of in-hospital mortality in the unadjusted analysis, but this mortality association does not retain after adjustments for confounding bias. Therefore, further randomized clinical trials are needed to provide clinical evidence regarding metformin therapy and inhospital mortality in patient with diabetes hospitalized for COVD-19.
## Supporting information
[fig] Fig 1: Flowchart illustration of the study selection. https://doi.org/10.1371/journal.pone.0282210.g001 PLOS ONE [/fig]
[fig] 8 PS: Propensity score; IPTW: Inverse probability of treatment weighting; NA: Not accessed; MSA: Metformin-specific adjustment (The probability of a patient being treated with metformin as dependent variable was estimated using a logistic regression model); ACS, acute coronary syndrome; CHF, congestive heart failure; CAD, coronary artery disease; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; DCCI, Drug-Derived Complexity Index; ALT, alanine aminotransferase; BUN, blood urea nitrogen; TBILI, total bilirubin; NAFLD: Nonalcoholic fatty liver disease; NASH: Nonalcoholic steatohepatitis; NOS: The Newcastle-Fig 2. The association of outpatient metformin use and in-hospital mortality in patients with COVID-19 and pre-existing T2DM. (A) Meta-analysis of 16 selected studies with extractable raw event data using the Mantel-Haenszel approach. (B) Funnel plot test of 16 studies included in panel A. (C) Meta-analysis of 10 selected studies with multivariable or propensity score-adjusted OR using the inverse-variance method to pool effect sizes. (D) Funnel plot of 10 studies included in panel C. (E) Meta-analysis of OR for in-hospital mortality in 6 studies with propensity score matched for metformin use by the inverse-variance method. (F) Meta-analysis of 5 studies with adjusted HR for in-hospital mortality using the inverse-variance method. https://doi.org/10.1371/journal.pone.0282210.g002 [/fig]
[fig] Fig 3: The association of inpatient metformin use and in-hospital mortality in patients with COVID-19 and pre-existing T2DM. (A) Meta-analysis of 5 selected studies with extractable raw event data using the Mantel-Haenszel approach. (B) Funnel plot of 5 studies included in panel A. (C) Meta-analysis of 2 selected studies with multivariable and propensity score-adjusted HR using the inverse-variance method to pool effect sizes. https://doi.org/10.1371/journal.pone.0282210.g003 [/fig]
[fig] S1: Checklist. PRISMA 2020 checklist. (DOCX) Fig. Sensitivity analysis of the studies included in meta-analysis in Fig 2A and 2C. (TIF) Author Contributions Conceptualization: Zhiyuan Ma, Mahesh Krishnamurthy. Data curation: Zhiyuan Ma, Mahesh Krishnamurthy. Formal analysis: Zhiyuan Ma, Mahesh Krishnamurthy. Investigation: Zhiyuan Ma, Mahesh Krishnamurthy. Methodology: Zhiyuan Ma. Project administration: Mahesh Krishnamurthy. Resources: Mahesh Krishnamurthy. Supervision: Mahesh Krishnamurthy. Validation: Zhiyuan Ma. Writing -original draft: Zhiyuan Ma. Writing -review & editing: Zhiyuan Ma, Mahesh Krishnamurthy. [/fig]
[table] Table 1: Characteristics of studies comparing mortality in metformin users vs non-users in patients with type 2 diabetes hospitalized for COVID-19. [/table]
|
Retailer opinions about and compliance with family smoking prevention and tobacco control act point of sale provisions: a survey of tobacco retailers
Background: The objectives of this study were to document retailer opinions about tobacco control policy at the point of sale (POS) and link these opinions with store level compliance with sales and marketing provisions of the Tobacco Control Act. Methods: This study conducted interviews of 252 tobacco retailers in three counties in North Carolina and linked their opinions with in-person observational audit data of their stores' compliance with POS policies. We conducted analyses examining retailer factors associated with noncompliance using Generalized Estimating Equations (GEE) controlling for individual, store, neighborhood, and county factors. Results: Over 90 % of retailers support minors' access provisions and a large minority (over 40 %) support graphic warnings and promotion bans. Low levels of support were found for a potential ban on menthol cigarettes (17 %). Store noncompliance with tobacco control policies was associated with both more reported retailer barriers to compliance and less support for POS policies. Awareness of and source of information about tobacco control regulations were not associated with compliance when accounting for neighborhood and county characteristics. Conclusions: Retailers expressed some support for a wide range of POS policies. Advocates and government agencies tasked with enforcement can work with retailers as stakeholders to enhance support, mitigate barriers, and promote compliance with tobacco control efforts at the point of sale.
# Background
In 2009, the Family Smoking Prevention and Tobacco Control Act ("Tobacco Control Act") (Public Law 111-31) instituted new sales and marketing provisions at the point of sale (POS). Public policy theory suggests that the extent of policy implementation and compliance with new policy rests largely with 'street level bureaucrats'implementers on the ground (in this case, tobacco retailers). Currently, tobacco retailers are often viewed as tobacco industry allies because their economic self-interest is tied to tobacco sales. Convenience store associations have also served as front groups for the industry to block or blunt the effects of POS policy [bib_ref] Breaking the alliance: defeating the tobacco industry's allies and enacting youth access..., Andersen [/bib_ref]. However, a number of studies suggest that the majority of tobacco retailers are compliant with Tobacco Control Act POS sales and marketing provisions [bib_ref] Retailer adherence to family smoking prevention and tobacco control Act, North Carolina, Rose [/bib_ref] [bib_ref] Tobacco advertising and sales practices in licensed retail outlets after the food..., Frick [/bib_ref] [bib_ref] Smokeless tobacco marketing and sales practices in Appalachian Ohio following federal regulations, Klein [/bib_ref]. For example, two studies in Ohio examined the compliance of retailers with four Tobacco Control Act POS provisions, finding violation rates under 10 % [bib_ref] Tobacco advertising and sales practices in licensed retail outlets after the food..., Frick [/bib_ref] [bib_ref] Smokeless tobacco marketing and sales practices in Appalachian Ohio following federal regulations, Klein [/bib_ref]. In 2012, the US Food and Drug Administration (FDA) conducted compliance checks in 37 states and the District of Colombia and issued warning letters or civil penalty letters to retailers for violations of sales and marketing restrictions (not including sales to minors) in only 1 % of the checks. Our prior study in North Carolina identified a 15.7 % violation rate of any of 12 provisions of the Tobacco Control Act [bib_ref] Retailer adherence to family smoking prevention and tobacco control Act, North Carolina, Rose [/bib_ref]. Taken together these studies suggest that it may be possible to engage tobacco retailers as stakeholders in tobacco control efforts, rather than adversaries.
Engaging with tobacco retailers in order to ensure implementation of Tobacco Control Act policies requires an understanding of the factors associated with retailer compliance. However, to date, little research has examined retailer characteristics and opinions that may be associated with compliance practices. The current study sought to understand how retailer opinions may be associated with compliance through interviews with 252 retailers whose stores had been audited for compliance with Tobacco Control Act POS provisions [bib_ref] Retailer adherence to family smoking prevention and tobacco control Act, North Carolina, Rose [/bib_ref]. We assess four constructs related to Mazmanian and Sabatier's Framework of Analysis for Policy Implementation [bib_ref] The implemention of public policy: a framework of analysis, Sabatier [/bib_ref] : (1) retailer barriers to complying with regulations, (2) awareness of policy, [bib_ref] Breaking the alliance: defeating the tobacco industry's allies and enacting youth access..., Andersen [/bib_ref] source of information about policies; and (4) retailer support for policies. Each of these constructs was then examined in relation to retailer compliance with Tobacco Control Act POS provisions.
# Methods
## Data source and study population
This study linked interview data from retailers to data on store compliance with POS provisions as measured by store audits conducted by the research team (Healthy Stores, Healthy Communities) [bib_ref] Retailer adherence to family smoking prevention and tobacco control Act, North Carolina, Rose [/bib_ref] [bib_ref] Field validation of secondary data sources for enumerating retail tobacco outlets in..., D'angelo [/bib_ref]. We selected the three counties, Buncombe, Durham, and New Hanover, to represent distinct geographic regions of North Carolina (mountain, central, and coastal); all have a high cancer burden. [fig_ref] Figure 1: Study sample response diagram [/fig_ref] shows the response for each data collection activity for this study. We identified all tobacco retailers within these three counties (n = 671) in Summer/Fall 2011 through driving all primary and secondary roads. In Fall 2011, we conducted in-person store audits of 347 retailers selected through stratified random sampling proportionate to the number of retailers in each county. Of these, we completed 324 audits; 14 stores were ineligible; 5 stores refused; and 4 stores were incomplete for safety or other reasons.
We then conducted retailer interviews in Summer 2012. Of the 324 stores with audit data, 4 stores were ineligible (out of business or could not be located) and 56 retailers refused to complete the interview. Interviews were incomplete in 12 stores -1 store due to data collector error and 11 where the interview could not be completed in 3 attempts. For the interview, we achieved a response rate of 78 % of stores. The final sample of stores with both interview and audit data was 252.
## Respondent eligibility criteria
Respondents in the store were eligible if they were the owner, on-site manager/assistant manager, or store clerk of an audited tobacco retail outlet. If multiple potential respondents were in the store, we interviewed the "highest" ranking participant. Eligible respondents needed to speak English. Only one respondent, who also refused the interview, did not speak English. Respondents received a $20 gift card to a chain store as an incentive.
## Data collection and measures
We conducted all retailer audits electronically using data collection forms programmed in Pendragon on an iPod touch. The University of North Carolina at Chapel Hill Public Health-Nursing institutional review board determined that the audit did not constitute human subjects research and did not require approval.
For the interview, we piloted the questionnaire at 6 retailers in a county not selected for the study. Data collectors were trained in a half-day session on study procedures and were certified in using the interview instrument which was programmed in Qualtrics for use on an iPad. The interview was approved by the UNC-CH Public Health-Nursing institutional review board (Study Number #12-0548).
## Compliance measure
The primary measure of compliance was from the audit. We assessed compliance with 12 POS provisions of the Tobacco Control Act implemented at the time of the store audit [bib_ref] Retailer adherence to family smoking prevention and tobacco control Act, North Carolina, Rose [/bib_ref]. A store was non-compliant if there were any: (1) sale of flavored cigarettes, (2) "light" or "low tar" labeled cigarettes, (3) loose cigarettes, (4) loose smokeless tobacco, (5) branded non-tobacco products, (6) self-service of cigarettes or smokeless tobacco, (7) tobacco vending machines, (8) audio advertisements with sound effects, (9) video advertisements with sound effects, music, or color, (10) gifts given with cigarette or smokeless tobacco purchase, (11) availability of promotions offering gifts with proof of purchase (e.g., tobacco 'reward' programs), or [bib_ref] Attitudes towards smoking policies and tobacco control measures in relation to smoking..., Schumann [/bib_ref] promotion of tobacco brand name event sponsorship.
## Retailer opinions
Based on the Mazmanian and Sabatier framework, we used three measures focused on factors that influence the extent of change needed to comply with regulations: (1) Awareness, (2) Source of information, and (3) Barriers to Compliance. These factors are relevant because, first, if retailers are unaware of regulations, they may be unable to comply. Second, their source of information about regulations may influence the extent to which they receive timely and accurate information about compliance requirements. Finally, retailers may find it difficult to comply with regulations because of structural or logistic barriers.
Awareness was measured as whether retailers were aware of the Tobacco Control Act as a dichotomous item from from the 2009 International Tobacco Control US survey [bib_ref] Smokers' reactions to FDA regulation of tobacco products: findings from the 2009..., Fix [/bib_ref].
Source of information was measured through a series of yes/no/does not apply questions asking about usual source of information about tobacco control regulations. We assessed nine different sources including both formal sources (government, tobacco industry, corporate, boss/manager, and trade associations) and informal sources (media, family and friends, customers, and other retailers). Respondents could indicate multiple 'usual sources.' Respondents could also note if a source was not applicable to them. From these variables, we created a dichotomous variable of whether a retailer cited any formal source of information.
Barriers to compliance was assessed through four items measured on a five-point Likert scale from strongly disagree to strongly agree. Items were coded so that higher values indicated stronger agreement with barriers. Barriers items (i.e., hurts my business, too costly, takes too much time, too hard to redo displays/shelves) were only asked of owners and managers (n = 165), and not of clerks.
We assessed level of support for POS regulations by retailers through 10 items measured on a five-point Likert scale from strongly disagree to strongly agree. These items were drawn from or adapted from multiple sources [bib_ref] Smokers' reactions to FDA regulation of tobacco products: findings from the 2009..., Fix [/bib_ref] [bib_ref] Attitudes towards smoking policies and tobacco control measures in relation to smoking..., Schumann [/bib_ref] [bib_ref] Demographic differences in support for smoking policy interventions, Doucet [/bib_ref] [bib_ref] The development and initial validation of the smoking policy inventory, Velicer [/bib_ref] [bib_ref] Public attitudes regarding limits on public smoking and regulation of tobacco sales..., Cummings [/bib_ref] [bib_ref] US attitudes about banning menthol in cigarettes: results from a nationally representative..., Winickoff [/bib_ref]. Items were scored so that higher scores represented more support for POS regulations. We assessed support for POS provisions that had been enacted such as a ban on flavored cigarettes, and those delayed by litigation including graphic warning labels and black and white tobacco advertisements. This scale had two items addressing each of five different aspects of the provisions:
bans on flavored and menthol cigarettes (product), graphic warning labels on packs and ads (counter advertising), black and white text ads and packaging (advertising and labeling), bans on gifts with tobacco sale and branded non-tobacco items (promotion), fines for retailers that sold tobacco to minors and increased fines for repeat sales (minor's access).
## Controls
We used control variables at the individual, store, neighborhood, and county levels as factors associated with either compliance, other tobacco marketing disparities, or with support for policy in prior studies.
Individual Individual level variables included respondent current smoking status (everyday/some days vs. not at all) and respondent role (owner, manager, or clerk).
Store We controlled for store proximity to school measured as a dichotomous variable of whether the store is within 1000 ft of a public school. We categorized store type (e.g., pharmacy, supermarket, convenience store) using definitions from the North American Industry Classification System (NAICS) definitions, coded with supermarkets as the reference category. Total amount of tobacco marketing material was derived from the store audit data and included counts of tobacco ads on the store interior and exterior, branded functional items (e.g. change mats), and tobacco moveable displays.
Neighborhood We used census tracts as neighborhood. Using the latitude and longitude of the store taken at the front entrance, we linked store location to the following neighborhood characteristics: the percentage of black and Hispanic residents, derived from 2010 US census data; and the percentage of families living below federal poverty guidelines, based on the 2006-2010 American Community Survey 5-year estimates .
County Our prior study found that odds of compliance varied significantly by county [bib_ref] Retailer adherence to family smoking prevention and tobacco control Act, North Carolina, Rose [/bib_ref].
# Data analysis
We calculated descriptive statistics to characterize the study sample and patterns of source of information about regulations, awareness of, and barriers to compliance with the Tobacco Control Act, support for POS policy, and compliance with POS policy. Confirmatory Factor Analysis (CFA) of the Barriers items indicated that the 4 items formed a unidimensional scale with excellent fit based on established cutoff criteria (χ 2 =.92 df = 2 p = .63; RMSEA .00 90 % CI 0.00, .12 CFI = 1.00; TLI = 1.03; SRMR = .014) [bib_ref] Cutoff criteria for fit indexes in covariance structure analysis: conventionall criteria versus..., Hu [/bib_ref]. All four items loaded significantly onto the latent factor. For the Level of Support measure, CFA found that the items formed a unidimensional scale with excellent model fit (χ 2 =10.97 df = 16 p = .81; RMSEA 0.0 90 % CI 0.0, 0.037; SRMR = .02; CFI = 1.000; TLI = 1.01). Residuals for each of the pairs of items for each POS domain were correlated. All items loaded significantly onto a single support for POS factor with the exception of the two minor's access items which had over 90 % support and thus little variance. They were dropped from the scale for adjusted analyses.
We conducted bivariate analyses (not shown) using Fisher's exact test and logistic regression for binary variables and ANOVA and Pearson correlations for continuous variables. Covariates that were significant at p < .05 were included in adjusted analyses. We conducted analyses looking at the relationships of Awareness, Source of information, and Barriers, and Support for POS policies with likelihood of noncompliance using GEE using PROC GENMOD in SAS 9.3. We used GEE because the intraclass correlation (ICC) of the null model using census tract as the clustering variable showed that 11 % of the variance in noncompliance was due to neighborhood (census tract). Because of this clustering, independence of stores cannot be assumed and GEE calculates robust standard errors using an exchangeable covariance structure to ensure appropriate confidence intervals. Additionally, we adjusted for sample weights at the county level to account for the sampling design.
For these analyses we separately modeled the factors associated with extent of change required (operationalized as awareness, source of information, and barriers) and retailer support for policies. These separate analyses were conducted to understand these facts as separate theoretical constructs. Empirically, barriers items were only asked of managers and owners but not of clerks, also supporting separate analyses.
We added covariates (not shown) in hierarchical sets [bib_ref] Retailer adherence to family smoking prevention and tobacco control Act, North Carolina, Rose [/bib_ref]. We entered the control variables in order from the most 'proximal' to the most 'distal' influences: theory driven, individual, store-level, neighborhood, and then county-level factors. The final model with all hierarchical sets of variables had the best model fit based on the quasi-likelihood under the independence model criterion (QIC) [bib_ref] Akaike's information criterion in generalized estimating equations, Pan [/bib_ref].
# Results
## Sample description
Among the 252 stores in the study, 16 % were noncompliant with Tobacco Control Act provisions based on our store audits. The predominant type of violations shown in [fig_ref] Table 1: Descriptive characteristics of interview respondents [/fig_ref] were sales of modified risk labeled cigarettes (e.g., "light" or "low tar") (13 % of stores) and self-service displays of cigarettes or smokeless tobacco (2 % of stores). Interview respondents, shown in [fig_ref] Table 1: Descriptive characteristics of interview respondents [/fig_ref] , were predominantly store managers or assistant managers (54 %), followed by clerks (35 %) and owners . Smoking prevalence among respondents (some day or everyday) was 40 %, higher than the 21.8 % smoking rate in North Carolina in 2011. The predominant store type was gas station or gas station with convenience stores (53 %). On average, stores had 34 tobacco marketing materials and 16 % were within 1000 ft of a K-12 public school. Stores in the sample were in 116 neighborhoods (i.e., census tracts). On We conducted analyses for non-response bias by examining bivariate relationships between respondents (n = 252) and non-respondents (n = 72) using chisquare tests for categorical variables and t-tests for continuous variables. These analyses showed no significant differences by any store, neighborhood or county characteristic among stores that completed the interview and those that did not participate (analyses not shown). Stores also did not differ on noncompliance between responders (16.3 %) and non-responders (16.8 %) (χ 2 (1) = .0064 p = .94).
## Descriptive statistics awareness of the tobacco control act
Fewer than half of respondents (43 %) were aware of the Tobacco Control Act 3 years after its implementation.
## Source of information about tobacco control regulations
Percent of respondents listing each source of information from those who answered yes or no to that source is shown in [fig_ref] Figure 2: Percent of respondents citing category as a 'usual source' of Information about... [/fig_ref]. Almost all respondents (97 %) had at least one formal source of information. The least common usual source of information about tobacco control regulations was government agencies (24 %). In contrast, boss/store managers were cited by 86 % of respondents; corporate offices for chain stores were also highly cited (71 %). Almost 70 % of retailers cited tobacco companies as a usual source of information about regulations.
## Barriers
Overall, 41 % of owner and manager respondents noted at least one barrier to complying with regulations, with the most common that making changes to how tobacco is sold hurts their business (29 %). Less than one-quarter agreed that it was too hard to redo store space/displays (24 %), took too much time to make required changes (23 %), or was too costly (22 %).
## Support for pos regulations
As shown in [fig_ref] Figure 3: Percent agree or strongly agree with each POS provision [/fig_ref] , respondents varied in the percent who agreed or strongly agreed with each particular POS provision. At least 90 % of respondents supported minor's access provisions while the least support was found for a menthol ban, at 17 %. A large minority (over 40 %) support graphic warnings and promotion bans.
In looking at correlates of compliance, we ran GEE models for theoretical factors -Awareness, Source of information, and Barriers, (Model A, [fig_ref] Table 2: Retailer opinions associated with store noncompliance with POS provisions [/fig_ref] (n = 161 owners and managers only) and a separate model using Support for POS policies (Model B, [fig_ref] Table 2: Retailer opinions associated with store noncompliance with POS provisions [/fig_ref] (n = 249). Awareness of regulations was significantly related to non-compliance only in model 3 when controlling for individual respondent and store covariates. Stores without formal sources of information about regulations were more likely to be non-compliant in models 2 and 3, including individual and store covariates. When adding neighborhood and county characteristics to the model, this result was no longer statistically significant. Sensitivity analyses using Fisher's exact test (not shown) examining source of information by neighborhood demographic quintile found that lacking a formal source did significantly differ with the percent of black residents (p = .04) and with education level of residents (p = .03) in store neighborhoods. Lack of formal source appeared to be more common in areas with a lower percent of black residents and a lower percent of residents with a bachelor's degree. In all models, retailers expressing higher levels of barriers had significantly higher odds of noncompliance. In model 5, accounting for individual, store, retailer neighborhood and county characteristics, stores with higher levels of barriers had 5.6 times the odds of non-compliance (AOR = 5.56, 95 % CI 2.24, 12.26). In model 5, stores in neighborhoods with more African-American residents had 8 % less likelihood of a violation (AOR = 0.92, 95 % CI 0.87, 0.97). But in neighborhood with more family poverty, there was 11 % increased likelihood of a violation (AOR: 1.11, 95 % CI 1.05, . No other covariates were significantly related to noncompliance (analyses not shown). [fig_ref] Table 2: Retailer opinions associated with store noncompliance with POS provisions [/fig_ref] , Model B shows the results for using Support for POS policies as a correlate of noncompliance. In each model, greater support for POS provisions was associated with decreased odds of noncompliance. In the final model accounting for individual, store, neighborhood, and county covariates, for every one-unit increase in level of support for POS provisions, store likelihood of a violation decreased by 41 % (AOR = .59, 95 % CI: .36, .97). In this model, among covariates only pharmacies had higher odds of noncompliance compared with grocery stores (AOR = 3.43, 95 % CI: 1.05, . No other covariate was significantly related to noncompliance (not shown).
# Discussion
This study suggests that store noncompliance with FDA POS provisions is significantly related to both barriers and lack of support for POS provisions among retailers. However, compliance was unrelated to awareness of the Tobacco Control Act or having formal sources of information about tobacco control regulations. In the final model, higher levels of barriers were positively associated with over 5 times the odds of noncompliance with Tobacco Control Act policies. This finding was consistent with the theoretical framework that the more change that the policy requires of implementers (operationalized as barriers), policy compliance is reduced [bib_ref] The implemention of public policy: a framework of analysis, Sabatier [/bib_ref]. However, this study could not assess the causal direction of this relationship. Findings about barriers are in line with prior studies of tobacco retailers indicating that barriers such as use of false identification made it difficult to comply with restricting sales to minors [bib_ref] Policy alternatives for reducing tobacco sales to minors: results from a national..., Altman [/bib_ref] and that lack of space is a barrier to displaying anti-tobacco messages [bib_ref] Tobacco advertising in retail stores, Cummings [/bib_ref].
Other authors have found that compliance with smokefree air [bib_ref] Do businesses comply with a Nosmoking Law? Assessing the self-enforcement approach, Rigotti [/bib_ref] or minor's access regulations [bib_ref] Changing retailer knowledge, attitudes, and behaviors related to cigarette sales to minors, Woodruff [/bib_ref] is related to awareness of regulations. In this study, awareness of the Tobacco Control Act among retailers was relatively low (43 %), but did not correspond with violations of POS provisions. However, the study was conducted in a period prior to FDA inspections of compliance with the Tobacco Control Act in North Carolina, which may increase both compliance and awareness over time. Moreover, in some cases retailers could be unknowingly compliant because their wholesaler stopped providing them with restricted products (e.g., no more flavored cigarettes, cigarettes without 'lights' descriptors).
Our study also found that having formal sources of information was unrelated to compliance once controlling for store neighborhood and county characteristics. Our results contradict a prior study, which found that compliance with smokefree air legislation at worksites was related to citing formal sources of information about regulations, rather than informal sources like friends or family [bib_ref] Evaluating compliance with Australia's first smokefree public places legislation, Goodin [/bib_ref]. Our null finding in models controlling for neighborhood and county characteristics may be related to the fact that most stores had a formal source of information about regulations. However, the fact that results were attenuated when controlling for neighborhood and county covariates perhaps suggests that some communities may differentially lack a formal source and may need more targeted outreach. Even so, in this study, the formal source least cited was government agencies (24 %), which are tasked with enforcement of these regulations. In contrast, almost 70 % of stores received information about regulations from tobacco companies, who have used prior retailer programs to build ties with retailers to undermine tobacco control efforts [bib_ref] Tobacco industry youth smoking prevention programs: protecting the industry and hurting tobacco..., Landman [/bib_ref]. Boss/stores managers were the most cited source of information and may be a valuable conduit for relaying tobacco control information.
Support for policies was also significantly related to retailer compliance. A national telephone survey found that 66 % of retailers thought it should be illegal for retailers to sell tobacco to minors [bib_ref] Policy alternatives for reducing tobacco sales to minors: results from a national..., Altman [/bib_ref]. However, no prior study examined the relationship between retailer support for policy and compliance. We found higher support for provisions in the Tobacco Control Act that had been enacted the longest. Over 90 % of retailers supported minor's access provisions enacted under the 1992 Synar Amendment, and over 40 % supported promotion restrictions implemented under the 1999 Master Settlement Agreement. For newer or proposed policies, less than a quarter to a third of respondents agreed with tobacco advertising restrictions or bans on flavored or menthol cigarettes.
The exception was relatively high levels of support among retailers for graphic warning labels on cigarette packs (59 %) and graphic warnings on advertisements in stores (43 %). Overall, the level of support among retailers for most of the provisions more closely resembled the level of support we found in a similar study among the general public (smokers and nonsmokers) than among smokers who had significantly lower support for every provision compared with non-smokers [bib_ref] Public support for Family Smoking Prevention and Tobacco Control Act point of..., Rose [/bib_ref].
Overall, the findings from this study best support Mazmanian and Sabatier's "Effective Implementation" scenario which describes a time of rapidly rising compliance after policy implementation followed by high levels of compliance maintained over time. For the POS provisions that have already been implemented, compliance was relatively high even prior to enforcement. It is likely to improve further once active inspections, warnings, and fines for noncompliance begin. Additionally, as noted, support for enacted provisions was high among retailers, suggesting few barriers to implementation success over time. However, to see "effective implementation" with provisions that have not yet been implemented, will need more effort. Public health advocates would do well to work with retailers to improve support for these provisions and enhance the climate for implementation over time. Fruitful strategies may include direct inclusion of supportive retailers in policy advocacy action; engagement of retail trade associations or corporate chains; media campaigns aimed at tobacco control in the point-of-sale environment; or direct outreach, education, or retailer training activities (e.g., FDA's Break the Chain retailer education program).
# Limitations and strengths
We note several limitations and strengths of the research. First, the temporal sequence of data collection of audit followed by interview, limits our ability to make causal inferences. We cannot determine whether more supportive retailers and those with fewer barriers are more likely to be compliant, or whether retailer compliance enhances retailer support and reduces perceived barriers. Nevertheless this research shows that these factors are associated over and above individual, store, neighborhood, and county characteristics. Future longitudinal studies are needed to separate out these effects. However, we do note that the research audits that were part of this study were not conducted for compliance or enforcement purposes. During the time period of the study, North Carolina had not yet been awarded a compliance and enforcement contract by the FDA. Thus, no enforcement inspections were active at the time of the study that may have influenced retailer compliance or attitudes toward regulations during the study period.
Second, the study was conducted in only three counties in North Carolina, which limits generalizability. Counties were selected to include diverse geographic areas of the state (a mountain, coastal, and central county of the state) and stores were randomly selected from a comprehensive list of stores within each county.
Measuring awareness of the Tobacco Control Act with only one item may have also been a limitation. Measuring awareness or knowledge of specific provisions as conducted in prior studies may have better correlated with compliance [bib_ref] Changing retailer knowledge, attitudes, and behaviors related to cigarette sales to minors, Woodruff [/bib_ref]. Additionally, there was little variance in the dichotomous measure of whether retailers had a formal source of information about tobacco control regulations. Additional research about retailers' trust in different sources about tobacco control regulations may be more salient in improving compliance, as has been found in other areas [bib_ref] Factors affecting food safety compliance within small and medium-sized enterprises: implication for..., Yapp [/bib_ref]. Social desirability of responses is possible in an interviewer-administered survey, but appears unlikely as personal information about respondents was limited and they were not asked about compliance with regulations.
The study also has several strengths. It is one of the only studies that includes tobacco retailer opinion about policies and links it to observations of retailer compliance [bib_ref] Changing retailer knowledge, attitudes, and behaviors related to cigarette sales to minors, Woodruff [/bib_ref] [bib_ref] Store tobacco policies: a survey of store managers, Weinbaum [/bib_ref] and the only one, thus far, which does so in relation to compliance with newer sales and marketing provisions of the Tobacco Control Act. It also uses theoretically derived factors that may influence policy implementation. Finally, it has a relatively large sample size and high response rate for retailer interviews.
# Conclusions
Understanding the relationship between retailer opinions and retailer compliance with POS provisions is important to helping implement Tobacco Control Act provisions. The Mazmanian and Sabatier framework would predict that policies that are 'easier' to implement or require few changes on the part of the retailer (e.g., passive retailer compliance due to changes in manufacturing or product availability) will have more compliance, while those that require active change (e.g., no sales or loose cigarettes, no self-service displays) will require more effort to gain compliance. Thus, helping retailers to address and overcome barriers, such as time and cost of implementing regulations, may enable them to become more compliant with these provisions. In most analyses, retailer awareness of the Tobacco Control Act and FDA authority over tobacco products was not necessary for them to comply with regulations. Instead guidance on specific provisions and how to successfully implement them as well as how to train staff to achieve compliance may be more valuable to retailers in overcoming barriers. This research also shows that few tobacco retailers were getting information about tobacco control regulations from government agencies. As such provisions are enforced, government agencies tasked with enforcement can do a better job communicating with and educating retailers about regulatory changes. Working through bosses and store managers and with small stores without corporate support can be a valuable approach to gaining support for tobacco control measures among retail staff who are ultimately responsible for implementing these policies.
This research also documents retailer support for specific POS measures. It is encouraging that some retailers are supportive of many POS policies. For proposed provisions with little support, advocates need to work with retailers to mitigate opposition to controversial provisions such as banning menthol cigarettes. While in some instances retail trade associations and some retailers have been opponents of tobacco control regulations and allies of the tobacco industry [bib_ref] Submit your next manuscript to BioMed Central and take full advantage of:..., Bloom [/bib_ref] , this research demonstrates that individual retailers have more varied opinions toward tobacco control regulations and may be engaged as stakeholders in tobacco control efforts at the POS.
Competing interests KM Ribisl is the Executive Director of Counter Tools (http://countertools.org), a nonprofit from which they receive compensation. Counter Tools provides technical assistance on point of sale tobacco control issues and distributes store mapping and store audit tools. KM Ribisl also has a royalty interest in a store audit and mapping system owned by the University of North Carolina at Chapel Hill. Neither the store audit tool nor the audit mapping system was used in this study. Dr. Ribisl is donating all of his royalties in 2012 and 2013 to student scholarships. Dr. Ribisl is also a special government employee for the FDA Center for Tobacco Products and a member of the Tobacco Products Scientific Advisory Committee; the views expressed in this article are his and not those of the FDA. All other authors declare that they have no competing interests.
Authors' contributions SWR originated the study, developed the analysis plan, performed the statistical analyses, and led the writing of the article. SLE, SE, JS, and HLMR provided substantive guidance and feedback on study design, instrument development and data analysis . KMR provided guidance and oversight to all aspects of the study and was the co-PI of the parent studies. All authors contributed to the development of the article and provided substantive feedback on the development of the manuscript. All authors read and approved the final manuscript.
[fig] Figure 1: Study sample response diagram [/fig]
[fig] Figure 2: Percent of respondents citing category as a 'usual source' of Information about tobacco control regulations [/fig]
[fig] Figure 3: Percent agree or strongly agree with each POS provision (n = 252) [/fig]
[table] Table 1: Descriptive characteristics of interview respondents [/table]
[table] Table 2: Retailer opinions associated with store noncompliance with POS provisions [/table]
|
Pre-impact fall detection
BackgroundFalls are a major safety concern, especially in older people. Approximately 28-35 % of people aged of 65 and above fall at least once every year[4,10,36]. In the US, 70 % of all emergency department visits by people over the age of 75 years were related to falls[39]. The consequences of falls are often devastating. Physical injuries caused by falls, such as hip fracture and head injuries, are often associated with high mortality and morbidity among older people[37]. Besides, older people are likely to develop "fear of repeated falls" after a fall-related incident. This often leads to the loss of mobility and independence[45]. For these and many more reasons, developing an effective fall prevention strategy is imperative to mitigate the harm of falls, especially in older people.Fall detection, whose main idea is to detect the occurrence of a fall event automatically[16,33,46], has been proposed to be an effective fall prevention strategy. Fall detection can be generally classified as post-fall mobility detection and pre-impact detection. Postfall mobility detection is expected to initiate timely medical assistance for fall victims, and thus avoid unnecessary losses caused by 'long-lie'[29,46]. However, this type of fall detection has an inherent limitation. Falls can be only detected after impacts, so injuries directly caused by fall impacts cannot be prevented.Unlike post-fall mobility detection, pre-impact fall detection is able to overcome the limitations mentioned above. Pre-impact fall detection refers to the technique that allows falls to be detected before the body hits against the ground (i.e., the body-ground impact). Thus, it can not only help initiate timely medical assistance for fall victims, butAbstractPre-impact fall detection has been proposed to be an effective fall prevention strategy. In particular, it can help activate on-demand fall injury prevention systems (e.g. inflatable hip protectors) prior to fall impacts, and thus directly prevent the fall-related physical injuries. This paper gave a systematical review on pre-impact fall detection, and focused on the following aspects of the existing pre-impact fall detection research: fall detection apparatus, fall detection indicators, fall detection algorithms, and types of falls for fall detection evaluation. In addition, the performance of the existing pre-impact fall detection solutions were also reviewed and reported in terms of their sensitivity, specificity, and detection/lead time. This review also summarized the limitations in the existing pre-impact fall detection research, and proposed future research directions in this field.
## Review procedure
The search of articles was based on the following database: Google Scholar, IEEE Xplore, PubMed, Web of Science and Science Direct. To avoid unnecessary omission, the searching criteria was deliberately broadened. We included the keywords of "fall detection", "fall detector", "fall event detection", "fall recognition", "fall protection", "fall prevention", "detect falls", and "detecting falls". This initial search yielded an initial collection of 683 papers. After that, screening was carried out to manually identify the studies on pre-impact fall detection only. Further actions were carried out to examine the citations of resulted papers to identify the relevant studies. We excluded papers that gave insufficient information, such as the conference abstract. In addition, one study on pre-impact detection and protection of motorcyclist falls [bib_ref] A simple fall detection algorithm for powered two wheelers, Boubezoul [/bib_ref] was also excluded as it is irrelevant to falls in older people.
A total of 23 studies on pre-impact fall detection were identified, including 19 journal articles and 4 full papers from conference proceedings. These studies were reviewed in details chronically. The review results were summarized in [fig_ref] Table 1: Summary on pre-impact fall detection studies MCS motion capture system, THD threshold,... [/fig_ref].
## Fall detection apparatus
The apparatus used in the existing pre-impact fall detection can be generally classified into the context aware systems and wearable sensors [bib_ref] Challenges, issues and trends in fall detection systems, Igual [/bib_ref]. Five studies were based on the context aware technology. Among them, four studies used the motion capture system [bib_ref] Detecting falls using a fall indicator defined by a linear combination of..., Hu [/bib_ref] [bib_ref] An individual-specific fall detection model based on the statistical process control chart, Hu [/bib_ref] [bib_ref] Pre-impact fall detection: optimal sensor positioning based on a machine learning paradigm, Martelli [/bib_ref] [bib_ref] Distinguishing fall activities from normal activities by velocity characteristics, Wu [/bib_ref] , and one was based on the radio frequency signal analysis [bib_ref] Leveraging RF signals for human sensing: fall detection and localization in human-machine..., Kianoush [/bib_ref].
In the fall detection studies using the motion capture system, kinematic fall detection indicators were determined by reflective markers placed at anatomic landmarks of the human body. The trajectories of these reflective markers were tracked by motion capture cameras mounted in fixed locations. The biomechanical convention to calculate the body kinematic measures based on the motion capture system has been well established. Therefore, the major advantage of using the motion capture system for fall detection is that fall detection indicators can be accurately determined. However, it is practically impossible to implement the motion capture system in real life applications. Therefore, the motion capture system can be only used for fall detection model development and evaluation.
Kianoush et al. [bib_ref] Leveraging RF signals for human sensing: fall detection and localization in human-machine..., Kianoush [/bib_ref] introduced a novel context-aware technology to detect falls in preimpact phase. Their technology was based on tracking the fluctuation of wireless radiofrequency (RF) signals. They found that fall events would cause abnormal perturbations on the RF signal which can be discerned by a hidden Markov model. The wireless RF signal was ubiquitous, and no attachable markers or wearable sensors were required. These features made this fall detection technology nonintrusive.
A limitation of the context-aware based fall detection is that it is restricted by space in the application. The motion capture system always has limited capture volume. Kianoush's RF based technology also required wireless network deployment within the fall detection area. In addition, both the motion capture system and the RF based technology are expensive. Thus, related research and applications were still restricted to experimental settings. Lower cost context-aware technologies for fall detection have been introduced recently, such as the vision-based technology [bib_ref] An intelligent emergency response system: preliminary development and testing of automated fall..., Lee [/bib_ref] , camera image processing technology [bib_ref] A customized human fall detection system using omni-camera images and personal information, Miaou [/bib_ref] , acoustic based technology [bib_ref] A microphone array system for automatic fall detection, Li [/bib_ref] , and Microsoft Kinect [bib_ref] Fall detection in homes of older adults using the microsoft kinect, Stone [/bib_ref]. However, to our knowledge, no one has used these technologies in pre-impact fall detection. This can definitely be a direction for future research.
The majority of existing pre-impact fall detection systems (17 out of 21) are wearablesensor-based. Due to the advancement in microelectronics and wireless communication technology, wearable micro-electro-mechanical systems (MEMS), such as accelerometers and gyroscopes, become small, light-weight and low-cost [bib_ref] A review of wearable sensors and systems with application in rehabilitation, Patel [/bib_ref]. They are capable of capturing body movement unobtrusively and allow kinematic measurements to be monitored over extended space and time. This makes them suitable for pre-impact fall detection.
A few studies implemented fall detection by using a single type of wearable sensors. For example, fall detection solutions from Bourke et al. [bib_ref] The identification of vertical velocity profiles using an inertial sensor to investigate..., Bourke [/bib_ref] ; Lindemann et al. [bib_ref] Evaluation of a fall detector based on accelerometers : a pilot study, Lindemann [/bib_ref] ; Shan and Yuan [bib_ref] A wearable pre-impact fall detector using feature selection and support vector machine, Shan [/bib_ref] and Tong et al. [bib_ref] HMM-based human fall detection and prediction method using tri-axial accelerometer, Tong [/bib_ref] relied on accelerometers only, while in Nyan et al. [bib_ref] A wearable system for pre-impact fall detection, Nyan [/bib_ref] and Bourke and Lyons [bib_ref] A threshold-based fall-detection algorithm using a bi-axial gyroscope sensor, Bourke [/bib_ref] , gyroscopes were the only measuring device for fall detection. The use of a single type of sensors can significantly reduce complexity and computational demand of the fall detection system. However, it also has adverse effects on fall detection. For instance, it has been argued that acceleration signals alone might not be able to detect falls accurately due to that such signals cannot effectively and efficiently differentiate falls from fall-like activities (e.g., successful balance recovery from external perturbations, and jumping) [bib_ref] Differentiating slip-induced falls from normal walking and successful recovery after slips using..., Hu [/bib_ref].
Compared to using a single type of wearable sensors, integrated inertial measurement units (IMUs), which typically consist of a tri-axial accelerometer and a tri-axial gyroscope, have become more popular in pre-impact fall detection applications. There might be three reasons behind this trend. First, the recent development in MEMS technology facilitated the development of low-cost, small-sized IMU chips with low energy consumption. This makes the implementation of wearable IMUs much easier than before. Second, the IMUs can provide more data than a single type of sensors and allow using multiple fall detection indicators, which could improve the fall detection accuracy. Third, integrated IMUs allow the sensor fusion algorithm, such as the Kalman Filter [bib_ref] Exploration and implementation of a pre-impact fall recognition method based on an..., Zhao [/bib_ref] or extended Kalman Filter [bib_ref] Prior-to-and post-impact fall detection using inertial and barometric altimeter measurements, Sabatini [/bib_ref] , to be used to correct errors in both accelerometer and gyroscope data, and thus could result in more accurate estimation of fall detection indicators.
There is an emerging trend of using wearable sensors in fall detection research [bib_ref] Challenges, issues and trends in fall detection systems, Igual [/bib_ref]. The wearable sensors have some advantages over the context-aware systems. First, the wearable sensors can be used to detect falls in extended space. Second, wearable sensors have much lower cost than context-aware systems. Third, wearable sensors are easier to implement than the context aware systems. In particular, wearable sensors often have wireless communication features which allow them to communicate easily with smart phones or other internet-enabled devices, and do not require additional infrastructure installation.
However, there are still some limitations in wearable-sensor-based fall detection. First, people under the fall detection surveillance are required to wear the sensors all the time. This might result in low user compliance because people sometimes might forget to wear them. Second, the data stability are lower than the context-ware system. This can be affected by many factors like insufficient battery power, and the alarm transmission might be affected by data lost in wireless communication. Third, some fall detection indicators cannot be measured directly by the wearable sensors, and have to be obtained on an estimation basis. For instance, body segment velocity variables have to be estimated by the integration of acceleration signals from accelerometers. There are inevitably some errors resulted from such estimation procedure.
In a recent study, Liu and Lockhart [bib_ref] Automatic individual calibration in fall detection: an integrative ambulatory measurement framework, Liu [/bib_ref] proposed an integrative ambulatory measurement framework that combined the motion capture data and IMU data to detect falls in the pre-impact phase. This was the first attempt to incorporate both the context-aware based system and wearable sensors simultaneously in pre-impact fall detection. In this framework, the data from the motion capture system was used as a reference to facilitate the development of an individual-calibrated fall discriminant function. Data from wearable IMUs was used as the input of this function to search the optimal thresholds that can distinguish falls from activities of daily living (ADLs). The evaluation results showed that the fall detection performance was enhanced with such an integrative system [bib_ref] Development and evaluation of a prior-to-impact fall event detection algorithm, Liu [/bib_ref].
## Fall detection indictors
Fall detection indicators refer to the variables selected to discriminate falls from non-fall activities in fall detection algorithms. Among the reviewed articles, all except Kianoush et al. [bib_ref] Leveraging RF signals for human sensing: fall detection and localization in human-machine..., Kianoush [/bib_ref] defined fall detection indicators by using human body kinematics. The kinematic measures used to define fall detection indicators are typically classified into two categories: segment translational measures and segment rotational measures. Among the translational measures, trunk velocity [bib_ref] Fall-detection through vertical velocity thresholding using a tri-axial accelerometer characterized using an..., Bourke [/bib_ref] [bib_ref] A threshold-based fall-detection algorithm using a bi-axial gyroscope sensor, Bourke [/bib_ref] [bib_ref] Detecting falls using a fall indicator defined by a linear combination of..., Hu [/bib_ref] [bib_ref] An individual-specific fall detection model based on the statistical process control chart, Hu [/bib_ref] ; Lee et al. [bib_ref] Inertial sensing-based pre-impact detection of falls involving near-fall scenarios, Lee [/bib_ref] [bib_ref] Distinguishing fall activities from normal activities by velocity characteristics, Wu [/bib_ref] [bib_ref] Portable preimpact fall detector with inertial sensors, Wu [/bib_ref] and trunk accelerations [bib_ref] The effect of window size and lead time on pre-impact fall detection..., Aziz [/bib_ref] [bib_ref] Pre-impact fall detection: optimal sensor positioning based on a machine learning paradigm, Martelli [/bib_ref] [bib_ref] A wearable pre-impact fall detector using feature selection and support vector machine, Shan [/bib_ref] [bib_ref] Mobile human airbag system for fall protection using MEMS sensors and embedded..., Shi [/bib_ref] [bib_ref] A wearable airbag to prevent fall injuries, Tamura [/bib_ref] [bib_ref] HMM-based human fall detection and prediction method using tri-axial accelerometer, Tong [/bib_ref] [bib_ref] Exploration and implementation of a pre-impact fall recognition method based on an..., Zhao [/bib_ref] were the most widely used fall detection indicators. Head acceleration [bib_ref] An individual-specific fall detection model based on the statistical process control chart, Hu [/bib_ref] [bib_ref] Evaluation of a fall detector based on accelerometers : a pilot study, Lindemann [/bib_ref] and upper arm velocity [bib_ref] An individual-specific fall detection model based on the statistical process control chart, Hu [/bib_ref] were also used to define fall detection indicators. Fall detection indicators defined by rotational measures included angular rate of the sternum [bib_ref] Distinguishing fall activities from normal activities by angular rate characteristics and high-speed..., Nyan [/bib_ref] , angular rate of the waist [bib_ref] A wearable system for pre-impact fall detection, Nyan [/bib_ref] [bib_ref] Mobile human airbag system for fall protection using MEMS sensors and embedded..., Shi [/bib_ref] [bib_ref] A wearable airbag to prevent fall injuries, Tamura [/bib_ref] and trunk [bib_ref] Development and evaluation of a prior-to-impact fall event detection algorithm, Liu [/bib_ref] [bib_ref] Automatic individual calibration in fall detection: an integrative ambulatory measurement framework, Liu [/bib_ref] , and segment orientation of the trunk [bib_ref] Development and evaluation of a prior-to-impact fall event detection algorithm, Liu [/bib_ref] [bib_ref] Automatic individual calibration in fall detection: an integrative ambulatory measurement framework, Liu [/bib_ref] and thigh [bib_ref] Application of motion analysis system in pre-impact fall detection, Nyan [/bib_ref] [bib_ref] A wearable system for pre-impact fall detection, Nyan [/bib_ref]. Most existing studies used a single kinematic measure to define the fall detection indicator. Only a few monitored more than one kinematic measures in their fall detection applications [bib_ref] Detecting falls using a fall indicator defined by a linear combination of..., Hu [/bib_ref] [bib_ref] A wearable system for pre-impact fall detection, Nyan [/bib_ref] [bib_ref] Distinguishing fall activities from normal activities by velocity characteristics, Wu [/bib_ref]. In particular, Wu [bib_ref] Distinguishing fall activities from normal activities by velocity characteristics, Wu [/bib_ref] and Nyan et al. [bib_ref] A wearable system for pre-impact fall detection, Nyan [/bib_ref] both implemented ' AND' logic to link the selected kinematic measures, where falls were considered to be detected only when all the selected kinematic measures were beyond predefined thresholds. Hu and Qu [bib_ref] Detecting falls using a fall indicator defined by a linear combination of..., Hu [/bib_ref] used a linear combination of trunk vertical velocity and shank frontal velocity as the fall detection indicator. The linear combination was considered as the simplest format of combination that made the fall detection implementation easy. Their results showed that such combination improved fall detection performance as compared to that using a single kinematic measure.
The selection of fall detection indicators was closely related to where to place fall detection sensors. The most commonly chosen body site was the waist area. Seven different research groups attached the wearable sensors on either the anterior or posterior side of the waist area to monitor the lower trunk kinematics. The chest (or upper trunk) area was another popular choice. Four research groups attached the sensor to the chest. The upper and lower trunk became preferred body sites for fall detection sensors, most probably because these two body areas are close to the whole-body center of mass. Researchers also considered the head [bib_ref] Evaluation of a fall detector based on accelerometers : a pilot study, Lindemann [/bib_ref] , underarm [bib_ref] Distinguishing fall activities from normal activities by angular rate characteristics and high-speed..., Nyan [/bib_ref] and thigh [bib_ref] A wearable system for pre-impact fall detection, Nyan [/bib_ref] for fall detector placement.
Some fall detection indicators cannot be measured directly by the wearable sensors. Instead, they were estimated from the sensor output. For example, the accelerometers can only measure the acceleration with the effect of gravity, which needs to be calibrated with the sensor orientation information to obtain the free acceleration. Besides, body segment velocity and orientation can only be calculated by integration of acceleration data (from accelerometer) and angular rate data (from gyroscope), respectively. For such integration procedure, there is always the drift problem due to the error accumulation in the integral results. Many approaches have been proposed to solve the drift problem, such as Kalman filters [bib_ref] Measuring orientation of human body segments using miniature gyroscopes and accelerometers, Luinge [/bib_ref] , the advanced integration methodology [bib_ref] Gait analysis using a shoe-integrated wireless sensor system, Bamberg [/bib_ref] , and the optimization approach [bib_ref] Kinematics of gait: new method for angle estimation based on accelerometers, Djurić-Jovičić [/bib_ref]. However, these methods have only been tested for normal activities. To our knowledge, no studies have attempted to use the aforementioned methods to estimate body kinematics during acute activities like falls. Future studies need to be carried out in this direction.
## Fall detection algorithms
Threshold-based algorithms appeared to be the simplest algorithm utilized in pre-impact fall detection research. With the application of threshold-based algorithms, a fall is considered to be detected if the selected fall detection indicators are beyond a pre-defined threshold. Otherwise, the activity is classified as a non-fall activity. Threshold-based algorithms are computationally efficient which allows them to be easily implemented in real-time applications. However, setting an appropriate threshold was always difficult. Typically, a higher threshold would result in fewer false alarms but more misdetection of falls; whereas a lower threshold would lead to less misdetection but more false alarms. Almost all the current threshold-based techniques face this dilemma. Some researchers defined the threshold by using the maximum readings of fall detection indicators during non-fall activities [bib_ref] Fall-detection through vertical velocity thresholding using a tri-axial accelerometer characterized using an..., Bourke [/bib_ref]. However, the selected non-fall activities in the experimental settings cannot be inclusive of all possible non-fall activities, and thus the threshold determined in this way may not sufficiently address the problem.
Machine learning algorithms, including support vector machine [bib_ref] The effect of window size and lead time on pre-impact fall detection..., Aziz [/bib_ref] , hidden Markov model [bib_ref] Mobile human airbag system for fall protection using MEMS sensors and embedded..., Shi [/bib_ref] ; hidden Markov model [bib_ref] Leveraging RF signals for human sensing: fall detection and localization in human-machine..., Kianoush [/bib_ref] [bib_ref] HMM-based human fall detection and prediction method using tri-axial accelerometer, Tong [/bib_ref] and artificial neural networks [bib_ref] Prior-to-and post-impact fall detection using inertial and barometric altimeter measurements, Sabatini [/bib_ref] , were also used in pre-impact fall detection research. A training period was required in the machine learning algorithms. In the training period, the data collected during non-fall activities were usually used to facilitate the feature extraction and activities classification. Machine learning algorithms are more computationally intensive compared to the threshold-based algorithms, and thus might lead to longer detection time.
Recently, Hu and Qu [bib_ref] Detecting falls using a fall indicator defined by a linear combination of..., Hu [/bib_ref] [bib_ref] An individual-specific fall detection model based on the statistical process control chart, Hu [/bib_ref] have presented a novel fall detection algorithm that was based on the statistical process control chart. In this model, statistical process control chart that was specified by upper and lower control limits was used to monitor fall detection indicator time series. The control limits were individual-specific as they were based on the data collected from each individual. Similar as the threshold-based approach, a fall was considered to be detected if the monitored fall detection indicator went beyond the range defined by the upper and lower control limits. Otherwise, the activity was considered to be normal. The limitation of this study, however, was that only slip-induced falls were tested.
## Types of falls for fall detection evaluation
Most studies used simulated falls, where the participants were asked to fall voluntarily. Forward, backward and sideway falls were the most common simulated falls. Falls during sit-to-stand transition [bib_ref] Application of motion analysis system in pre-impact fall detection, Nyan [/bib_ref] [bib_ref] A wearable system for pre-impact fall detection, Nyan [/bib_ref] , and falls during stair negotiation [bib_ref] A wearable pre-impact fall detector using feature selection and support vector machine, Shan [/bib_ref] were also simulated in the laboratory for fall detection research. Most of fall accidents in real life are unexpected and involuntary in nature. Movement patterns during simulated falls must be different from the realistic unexpected falls. Thus, using simulated falls may result in over-rated fall detection evaluation results. Bagalà et al. [bib_ref] Evaluation of accelerometerbased fall detection algorithms on real-world falls, Bagalà [/bib_ref] reported that the fall detection sensitivity with real-life fall data were much lower than that using simulated fall data. This was partially due to that the threshold calibrated on simulated fall signals might not be suitable for real-life fall scenarios.
Instead of simulated falls, some researchers used involuntary falls for fall detection evaluation [bib_ref] The effect of window size and lead time on pre-impact fall detection..., Aziz [/bib_ref] [bib_ref] Detecting falls using a fall indicator defined by a linear combination of..., Hu [/bib_ref] [bib_ref] An individual-specific fall detection model based on the statistical process control chart, Hu [/bib_ref] [bib_ref] Development and evaluation of a prior-to-impact fall event detection algorithm, Liu [/bib_ref] [bib_ref] Automatic individual calibration in fall detection: an integrative ambulatory measurement framework, Liu [/bib_ref] [bib_ref] Pre-impact fall detection: optimal sensor positioning based on a machine learning paradigm, Martelli [/bib_ref]. For example, realistic slip-induced falls were used in Hu and Qu [bib_ref] Detecting falls using a fall indicator defined by a linear combination of..., Hu [/bib_ref] [bib_ref] An individual-specific fall detection model based on the statistical process control chart, Hu [/bib_ref]. However, this study was limited by only one type of fall. It is worth noting that Aziz [bib_ref] The effect of window size and lead time on pre-impact fall detection..., Aziz [/bib_ref] ; Sabatini et al. [bib_ref] Prior-to-and post-impact fall detection using inertial and barometric altimeter measurements, Sabatini [/bib_ref] ; Lee et al. [bib_ref] Inertial sensing-based pre-impact detection of falls involving near-fall scenarios, Lee [/bib_ref] used both voluntary falls (e.g. sitto-stand fall and fall from fainting) and involuntary falls (e.g. falls from slips, trips and hit/bump). Their results can be better generalized into real-life situations than the other studies. The direction of future work in terms of selecting proper types of falls for fall detection evaluation is to establish a database about the involuntary falls in real life for fall detection development and evaluation.
It is also important to note that fall data were typically collected from younger adults instead of older adults in fall detection research due to safety and ethics concern. However, older adults have different postural control characteristics when being exposed to external perturbations initiating falls [bib_ref] Age-related joint moment characteristics during normal gait and successful reactive-recovery from unexpected..., Liu [/bib_ref] , and may hit the ground sooner than younger adults during falls [bib_ref] Distinguishing fall activities from normal activities by velocity characteristics, Wu [/bib_ref]. Therefore, the model evaluation results based on younger adults was in general overestimated.
## Fall detection performance
The performance of pre-impact fall detection is mainly assessed from two perspectives: accuracy and efficiency. The fall detection accuracy was typically measured by sensitivity and specificity. Sensitivity is determined by the ratio of the number of successfully detected falls over the total number of falls. Specificity is determined by the ratio of the number of successfully detected non-fall activities over the total number of non-fall activities.
The sensitivities reported in the reviewed studies ranged from 80 to100 %. The sensitivity (=80 %) in Sabatini et al. [bib_ref] Prior-to-and post-impact fall detection using inertial and barometric altimeter measurements, Sabatini [/bib_ref] was the lowest. The rest studies can achieve sensitivity above 90 %. A few studies reported 100 % sensitivities, including Lindemann et al. [bib_ref] Evaluation of a fall detector based on accelerometers : a pilot study, Lindemann [/bib_ref] , Nyan et al. [bib_ref] Distinguishing fall activities from normal activities by angular rate characteristics and high-speed..., Nyan [/bib_ref] , Bourke et al. [bib_ref] The identification of vertical velocity profiles using an inertial sensor to investigate..., Bourke [/bib_ref] , Wu and Xue [bib_ref] Portable preimpact fall detector with inertial sensors, Wu [/bib_ref] , Shi et al. [bib_ref] Mobile human airbag system for fall protection using MEMS sensors and embedded..., Shi [/bib_ref] , Shan and Yuan [bib_ref] A wearable pre-impact fall detector using feature selection and support vector machine, Shan [/bib_ref] , Tong et al. [bib_ref] HMM-based human fall detection and prediction method using tri-axial accelerometer, Tong [/bib_ref] and Liu and Lockhart [bib_ref] Development and evaluation of a prior-to-impact fall event detection algorithm, Liu [/bib_ref] [bib_ref] Automatic individual calibration in fall detection: an integrative ambulatory measurement framework, Liu [/bib_ref]. The specificities reported in the reviewed articles were from 85.6-100 %. The lowest specificity was reported in Aziz et al. [bib_ref] The effect of window size and lead time on pre-impact fall detection..., Aziz [/bib_ref]. A few studies reported 100 % specificities, including Bourke et al. [bib_ref] Fall-detection through vertical velocity thresholding using a tri-axial accelerometer characterized using an..., Bourke [/bib_ref] [bib_ref] The identification of vertical velocity profiles using an inertial sensor to investigate..., Bourke [/bib_ref] ; Nyan et al. [bib_ref] Application of motion analysis system in pre-impact fall detection, Nyan [/bib_ref] [bib_ref] A wearable system for pre-impact fall detection, Nyan [/bib_ref] ; Shi et al. [bib_ref] Mobile human airbag system for fall protection using MEMS sensors and embedded..., Shi [/bib_ref] ; Shan and Yuan [bib_ref] A wearable pre-impact fall detector using feature selection and support vector machine, Shan [/bib_ref] and Sabatini et al. [bib_ref] Prior-to-and post-impact fall detection using inertial and barometric altimeter measurements, Sabatini [/bib_ref].
Note that it is not reasonable to conclude the best fall detection models just based on the reported sensitivity and specificity, as the fall data for evaluation are quite different across studies. In particular, the number of participants and the number of falls and other activities in model evaluation varied in different studies. These factors did affect the reported accuracy. Some studies, like Lindemann et al. [bib_ref] Evaluation of a fall detector based on accelerometers : a pilot study, Lindemann [/bib_ref] and Kianoush [bib_ref] Leveraging RF signals for human sensing: fall detection and localization in human-machine..., Kianoush [/bib_ref] involved only one or two participants, and reported relatively high fall detection accuracy. Besides, as mentioned earlier, the types of fall selected in the model evaluation also affect the reported fall detection accuracy. For the sake of comparison among different fall detection models, it is important to unify the standard such as the number of falls and other activities involved and the selection of fall types in the model evaluation.
Lead time and/or detection time are typically used to assess the efficiency of fall detection. Lead time was defined by the time interval between when the fall was detected and fall impact, and accounts for the time for protective measures to be activated to protect the fall victims from fall impacts. Thus, the longer the lead time is, the better the fall detection performance is. The reported lead time were from 40 to 750 ms. Shi et al. [bib_ref] Mobile human airbag system for fall protection using MEMS sensors and embedded..., Shi [/bib_ref] proposed airbag technology for preventing fall impacts. The inflation time of the airbag was reported to be approximately 130 ms. In order to be effective in avoiding a fall impact, the lead time of fall detection should be longer than 130 ms.
Detection time was the time difference between the fall initiation and fall detection. This parameter is also used to indicate how rapid the fall detection system responds to a fall. A better fall detection performance is associated with a smaller detection time. Liu and Lockhart [bib_ref] Development and evaluation of a prior-to-impact fall event detection algorithm, Liu [/bib_ref] [bib_ref] Automatic individual calibration in fall detection: an integrative ambulatory measurement framework, Liu [/bib_ref] and Martelli et al. [bib_ref] Pre-impact fall detection: optimal sensor positioning based on a machine learning paradigm, Martelli [/bib_ref] reported a mean detection time of 255 ms and 351 ms, respectively. Hu and Qu [bib_ref] An individual-specific fall detection model based on the statistical process control chart, Hu [/bib_ref] reported a mean detection time range between 620 and 710 ms. As the time interval between heel-strike and fall impact can be estimated to be around 900 ms [bib_ref] Earliest gait deviations during slips: implications for recovery, Beschorner [/bib_ref] [bib_ref] A proposal for the classification and evaluation of fall detectors Une proposition..., Noury [/bib_ref] , these fall detection models can provide sufficient time for triggering fall protection device.
[table] Table 1: Summary on pre-impact fall detection studies MCS motion capture system, THD threshold, ACCM accelerometer, GYRO gyroscope, SMF simulated falls, SEN sensitivity, SPE specificity, ADLs activities of daily living [/table]
|
Shewanella knowledgebase: integration of the experimental data and computational predictions suggests a biological role for transcription of intergenic regions
Shewanellae are facultative g-proteobacteria whose remarkable respiratory versatility has resulted in interest in their utility for bioremediation of heavy metals and radionuclides and for energy generation in microbial fuel cells. Extensive experimental efforts over the last several years and the availability of 21 sequenced Shewanella genomes made it possible to collect and integrate a wealth of information on the genus into one public resource providing new avenues for making biological discoveries and for developing a system level understanding of the cellular processes. The Shewanella knowledgebase was established in 2005 to provide a framework for integrated genome-based studies on Shewanella ecophysiology. The present version of the knowledgebase provides access to a diverse set of experimental and genomic data along with tools for curation of genome annotations and visualization and integration of genomic data with experimental data. As a demonstration of the utility of this resource, we examined a single microarray data set from Shewanella oneidensis MR-1 for new insights into regulatory processes. The integrated analysis of the data predicted a new type of bacterial transcriptional regulation involving co-transcription of the intergenic region with the downstream gene and suggested a biological role for co-transcription that likely prevents the binding of a regulator of the upstream gene to the regulator binding site located in the intergenic region.
# Introduction
Shewanellae inhabit a wide range of niches; thriving as free-living cells in fresh and marine waters and sediments in conditions of atmospheric to high pressure, and at low to moderate temperatures as well as in association with higher life forms such as squid and fish . Shewanellae's ability to inhabit these diverse environments is likely enabled by diversification in gene content among different species and utilization of an expansive regulatory network to monitor and respond to the surrounding environment (3). By far, the most thoroughly researched aspect of their physiology has been their ability to conduct extracellular electron transfers to iron and manganese (hydr) oxides and to reduce radionuclides [bib_ref] Extracellular respiration, Gralnick [/bib_ref] with more recent studies directed at their use in microbial fuel cells [bib_ref] The influence of acidity on microbial fuel cells containing Shewanella oneidensis, Biffinger [/bib_ref] [bib_ref] Growth with high planktonic biomass in Shewanella oneidensis fuel cells, Lanthier [/bib_ref]. The surge in research on this organism over the last 10 years is in part due to the current public availability of 19 complete and two partial genome sequences from cultured isolates of Shewanella and to Department of Energy support for a distributed group of investigators collectively known as the Shewanella Federation (SF) to systematically investigate the metabolic potential of these species. The integrated genome-based studies of Shewanella ecophysiology performed by this group require significant collaborative efforts in the annotation of the genomes, designing and conducting experiments, and in sharing of data, information, and resources. The Shewanella knowledgebase (SKB) was initiated as a web resource tailored to these tasks as well as a means of integrating experimental data with genomic data.
The implementation of a systems biology approach to study ecophysiology of an organism relies on efficient compilation, integration and sharing of available experimental, ecological, and genomic data on this organism. Most biological databases, however, limit their content to genomic or experimental data associated with a specific cellular function or biological process, such as metabolic pathways [bib_ref] The MetaCyc Database, Karp [/bib_ref] , regulation [bib_ref] RibEx: a web server for locating riboswitches and other conserved bacterial regulatory..., Abreu-Goodger [/bib_ref] or genome sequences [bib_ref] IMG ER: a system for microbial genome annotation expert review and curation, Markowitz [/bib_ref]. Thus, information associated with a single organism is scattered across different resources. Several recently developed databases address this problem by bringing diverse information into one resource for in-depth characterization of a microbe with a specific phenotype. Examples of such databases are Pseudomonas Genome Database [bib_ref] Pseudomonas genome database: facilitating user-friendly, comprehensive comparisons of microbial genomes, Winsor [/bib_ref] , Systomonas [bib_ref] SYSTOMONAS-an integrated database for systems biology analysis of Pseudomonas, Choi [/bib_ref] , The Candida Genome Database [bib_ref] The Candida Genome Database (CGD), a community resource for Candida albicans gene..., Arnaud [/bib_ref] , EchoBASE [bib_ref] EchoBASE: an integrated post-genomic database for Escherichia coli, Misra [/bib_ref] and EcoCyc [bib_ref] EcoCyc: a comprehensive view of Escherichia coli biology, Keseler [/bib_ref]. This integrative approach has proved helpful in analyzing experimental data and in promoting biological discoveries. Similar to these databases, the SKB combines many independent data sources for the purpose of using existing knowledge to support efficient analysis of the experimental data. This infrastructure combines experimental data generated on members of the Shewanella genus, biological knowledge from the published literature, computational predictions made by specialized Internet databases, and analytical tools to produce new scientific insights. The inset 'Resources' in the SKB main menu summarizes the Internet resources employed by the knowledgebase for annotation of the Shewanella genomes, and for characterization of specific bacterial systems, including metabolism, transport, signaling, proteolysis, adaptive evolution and different types of regulation [bib_ref] RibEx: a web server for locating riboswitches and other conserved bacterial regulatory..., Abreu-Goodger [/bib_ref] [bib_ref] The SUPERFAMILY database in 2007: families and functions, Wilson [/bib_ref] [bib_ref] MBGD: a platform for microbial comparative genomics based on the automated construction..., Uchiyama [/bib_ref] [bib_ref] PEDANT genome database: 10 years online, Riley [/bib_ref] [bib_ref] TransportDB: a comprehensive database resource for cytoplasmic membrane transport systems and outer..., Ren [/bib_ref] [bib_ref] MEROPS: the peptidase database, Rawlings [/bib_ref] [bib_ref] NCBI Reference Sequence (RefSeq): a curated non-redundant sequence database of genomes, transcripts..., Pruitt [/bib_ref] [bib_ref] The comprehensive microbial resource, Peterson [/bib_ref] [bib_ref] Tractor_DB (version 2.0): a database of regulatory interactions in gamma-proteobacterial genomes, Perez [/bib_ref] [bib_ref] ODB: a database of operons accumulating known operons across multiple genomes, Okuda [/bib_ref] [bib_ref] Virtual Footprint and PRODORIC: an integrative framework for regulon prediction in prokaryotes, Munch [/bib_ref] [bib_ref] CRISPRFinder: a web tool to identify clustered regularly interspaced short palindromic repeats, Grissa [/bib_ref].
Detailed annotation of a sequenced genome is essential for a comprehensive analysis of experimental data and for linking an organism's genotype and phenotype. Although computational predictions can produce large quantities of information for sequenced organisms, the false positive/ negative rate of such predictions is rather high [bib_ref] Evaluation of annotation strategies using an entire genome sequence, Iliopoulos [/bib_ref]. Manual annotation and the use of a reference genome can significantly improve the quality of computational predictions. Most databases for model organisms [bib_ref] Pseudomonas genome database: facilitating user-friendly, comprehensive comparisons of microbial genomes, Winsor [/bib_ref] [bib_ref] The Candida Genome Database (CGD), a community resource for Candida albicans gene..., Arnaud [/bib_ref] [bib_ref] EcoCyc: a comprehensive view of Escherichia coli biology, Keseler [/bib_ref] have manually curated content. Manual curation significantly improves the quality and value of annotations, but it is a time consuming job for biologists, especially if the curation targets several genomes. An important feature of the SKB is, therefore, a multi-genome annotation environment consisting of ortholog and genome editors to facilitate manual curation of all sequenced Shewanella genomes. The editing environment uses the Shewanella oneidensis MR-1 (MR1) genome as a reference and makes use of computational predictions and experimental data to support both automatic and manual curation.
The SKB is a compilation of experimental and computational analyses of all sequenced Shewanella strains, with the majority associated with the model organism, MR1. The combination of these analyses and the SKB tools provides a unique capability allowing researchers to view experimental data in the wider context of computational predictions and to integrate the predictions into analysis. Such integration has already led to the development of a novel approach for linking phenotype and genotype of organisms and for deciphering the cellular mechanism underlying Shewanella cold tolerance [bib_ref] Conserved synteny at the protein family level reveals genes underlying Shewanella species'..., Karpinets [/bib_ref].
In this study, we demonstrate the value of the SKB using one of the many SF studies involving MR1 Affymetrix microarrays designed to measure not only the expression of genes, but also the intergenic regions (IG) located on the same strand as the adjacent genes [bib_ref] Many Microbe Microarrays Database: uniformly normalized Affymetrix compendia with structured experimental metadata, Faith [/bib_ref]. This design supports the study of changes in expression of genes and IGs under different growth conditions. In prokaryotes, IG transcription has not been studied before by the expression profiling. Most previous studies of IG expression have used the tiling arrays or high throughput sequencing in model eukaryotic organisms [bib_ref] Exosome complex and pervasive transcription in eukaryotic genomes, Belostotsky [/bib_ref]. Such studies have revealed that genomes of eukaryotes including yeast, mouse, human and plants are pervasively transcribed. The transcription generates numerous untranslated transcripts that associate with different parts of an IG and have different length and stability. Biological roles of the intergenic transcription are unclear. Experimental studies in Saccharomyces cerevisiae have shown that the intergenic transcripts can mediate the transcriptional gene silencing by generation of antisense transcripts [bib_ref] A cryptic unstable transcript mediates transcriptional trans-silencing of the Ty1 retrotransposon in..., Berretta [/bib_ref] or by continued transcription of gene regulatory regions [bib_ref] Intergenic transcription is required to repress the Saccharomyces cerevisiae SER3 gene, Martens [/bib_ref]. Untranslated transcripts can also play a role in maintaining an open chromatin structure at eukaryotic promoters [bib_ref] Bidirectional promoters generate pervasive transcription in yeast, Xu [/bib_ref]. A recent study of the Bacillus anthracis transcriptome using high-throughput sequencing [bib_ref] Structure and complexity of a bacterial transcriptome, Passalacqua [/bib_ref] has also shown a significant transcriptional activity in the bacterial IGs. Motivated by this observation, expression profiling of the MR1 IGs under different growth conditions in combination with SKB annotations and tools was used to explore the extent and the biological role of bacterial IG transcription. In this report, we describe an integrated analysis of experimental data from a study of the transition from aerobic to anaerobic growth in the wild-type strain
of MR1 and in its Crp(À) mutant. This analysis suggests that overall IGs in the MR1 genome are extensively transcribed and detects a significant change in the level of IG expression in the Crp(À) mutant versus wild-type strain. A significant number of the transcribed IGs, which likely implement complex regulatory functions, cannot be attributed to bacterial operons. One of the functions suggested by our analysis using the SKB involves co-transcription of an IG with one of its adjacent genes in order to prevent binding of transcription factors regulating the other adjacent gene.
## Knowledgebase description a combination of physiological and omics data in skb
In order to provide a system level characterization of MR1, the SF performed more than 40 experimental studies generating transcriptome, proteome, metabolome and biolog (a phenotype microarray for testing thousands of cellular phenotypes at once) data of wild-type and mutant strains cultivated under different physiological conditions. The SKB includes data produced at SF labs or published in the literature. The data is available as individual projects under the tab 'Projects' in the main menu. Relevant metadata, including a brief description of the experiment, its main objectives, experimental design and the principal investigator of the project, is provided for each project. In addition to a table presentation of the data, graphical data associated with experiments, growth curves and other images are also available for more complete descriptions of some projects, like 'Physiology' or 'Biolog'.
The majority of the accumulated data (25 projects) was produced with microarray technology, Affymetrix or two-color microarray, for the model organism MR1. These projects usually employed a factorial design to test different culture and light conditions, growth stages, media components, a diverse set of carbon sources, e-donors, acceptors and their combination, as well as varying aeration, heat, pH, salt and exposure to heavy metals. A unique feature of the Affymetrix microarray data is the measurement of gene and IG expression as discussed earlier. The Affymetrix data sets were obtained from the Many Microbe Microarrays Database [bib_ref] Many Microbe Microarrays Database: uniformly normalized Affymetrix compendia with structured experimental metadata, Faith [/bib_ref]. Two SF experiments, referred to as FedEx, address metabolic pathways and regulatory networks involved in aerobic and oxygen-limited growth of MR1 and include transcriptional and proteome analyses.
Screening of the MR1 transcriptome is accompanied by a comprehensive physiological and phenotypic characterization of several of the Shewanella strains using two complementary technologies: biolog phenotype microarrays and different culture conditions. The physiological studies conducted include the response to environmental factors, such as temperature and salinity, response to different carbon sources, such as fatty acids and sugars, and response to different electron acceptors. The project 'MR1CloneSet' provides a comprehensive collection of MR1 genes constructed using the lambda recombinase (Gateway) cloning system. The MR1CloneSet includes 3584 individual ORFs (85%) cloned into the entry plasmids. Another project named 'Mutant' provides information on over 200 mutants generated by SF members. Mutants of genes involved in redox reactions (most c-type cytochromes, terminal reductases and hydrogenases), secretion, central metabolism, signaling and regulation are included in the list. Many of the mutants are bar coded (tagged with short synthetic oligonucleotides). Some mutants contain multiple mutations (arcA/ etrA, arcA/crp, etrA/crp, etc.). SKB also contains three data sets with 3-D structures of Shewanella proteins. Structural information on Shewanella proteins generated at Argonne National Laboratory or by the Joint Center for Structural Genomics (JCSG) using high-throughput X-ray crystallographic technology is available under projects 'Protein structure target-Argonne' and 'Protein structure target-JCSG'. A third data set with 3D structures of proteins that have been deposited in Protein Data Bank (PDB) is available as a separate project. Protein sequences are mapped to locus tags from the recent annotations integrating the protein structure data with the rest of the data in SKB.
## Multi-genome annotation solution: integration of ortholog and genome editors
To facilitate annotation of the 21 sequenced Shewanella genomes the SKB provides an editing environment for manual curation of protein-encoding gene models and functional annotations. A distinctive feature of this environment is the availability of two interfaces by which a curator can edit the genome annotations, one that enables editing of fields associated with either the gene models or functions encoded by a single genome and another that enables manual curation of functional annotations for orthologous sets of genes and the gene membership within an orthologous group. The ortholog editor interface enables simultaneous annotation of genes from multiple genomes, thereby dramatically reducing the amount of manual curation necessary. It immerses the curator in a comparative genome environment that facilitates, for example, identification of inaccuracies in the gene model, identification of laterally acquired genes and discovery of key functional differences between the sequenced strains.
Both editor interfaces enable curation of product descriptors, a more lengthy functional description, gene names, EC numbers, predicted subcellular localization and functional subsystem assignment, thereby enabling transfer of information recorded in these fields from the ortholog editor to the same fields in the appropriate genome editors. Both interfaces also enable the curator to add new records so that new genes or new ortholog groups can be appended to the database. Additional curatable fields unique to the ortholog editor include one that denotes paralogous group IDs and others that indicate the locus tags of proteins that belong to the orthologous group [fig_ref] Figure 1: Interplay between the ortholog and genomes editors in SKB [/fig_ref]. In addition, the curator is provided with links to pre-computed alignments of protein sequences and tables listing protein size and domain content associated with orthologous proteins to facilitate curation of group membership and to suggest inaccuracies in the gene model (e.g. missing genes or inaccurate start codons).
The individual genome editors uniquely include curatable fields for adjusting the gene location and assigning new locus tag thereby enabling curation of gene models [fig_ref] Figure 1: Interplay between the ortholog and genomes editors in SKB [/fig_ref]. By also including curatable fields that track the nature of disruptions to genes (e.g. frameshifts, point mutations, 5 0 -or 3 0 -truncations) and the occurrence of translational reprogramming (e.g. ribosomal slippage, selenocysteine), and by indicating the positions of frameshifts within the gene location field it was possible to develop a script that automatically produces a current FastA file of all proteins and genes (including pseudogenes) at the click of a button for use by researchers wishing to conduct other analyses (e.g. prediction of protein localization, identification of peptides in proteome data sets, interpretation and design of microarrays). As an additional feature for researchers using the site, both the genome and ortholog editor interfaces provide a means to generate up-to-date tab-delimited tables listing values of the fields present within each editor database (ortholog or genome specific) and information indicating when and by which curator fields for a gene or an orthologous group was edited.
Because MR1 was the first genome sequenced and has the most experimental data (transcriptomics and proteomics) associated with it, the genome has also received the most attention regarding curating its gene model. Proteome data has been used to determine whether hypothetical genes encode proteins [bib_ref] Confirmation of the expression of a large set of conserved hypothetical proteins..., Elias [/bib_ref] [bib_ref] Global profiling of Shewanella oneidensis MR-1: expression of hypothetical genes and improved..., Kolker [/bib_ref] [bib_ref] Validation of Shewanella oneidensis MR-1 small proteins by AMT tag-based proteome analysis, Romine [/bib_ref] and to search for evidence supporting predicted N-termini of the encoded proteins [bib_ref] Whole proteome analysis of post-translational modifications: applications of mass-spectrometry for proteogenomic annotation, Gupta [/bib_ref]. Mobile elements (insertion sequences and miniature inverted repeat transposable elements) have been mapped to the chromosome to identify disrupted genes [bib_ref] Identification of mobile elements and pseudogenes in the Shewanella oneidensis MR-1 genome, Romine [/bib_ref]. In addition, comparative sequence analysis across orthologous proteins in Shewanella and with more distantly related bacteria have been used to manually adjust gene start codons and to identify genes that were missed by automated gene calling methods. The current gene model is consequently significantly improved with numerous genes dropped (596), new genes identified (89) and start positions changed (597) and enabled us to better assess transcriptome data for non-coding RNAs.
## Characterization of mr1 regulation and its overlay with the experimental data in shewregdb
Understanding regulatory processes that control metabolic, structural, and behavioral adaptations to changes in the environment are at the heart of systems biology research. Numerous experimental and bioinformatics analyses focused on characterizing S. oneidensis MR-1 regulatory networks have been published in the literature and/or compiled in various databases. ShewRegDb was developed as a component of the SKB to collect this information and to visualize it in the context of other types of experimental and genomic data. Computational predictions of the regulatory elements were collected from the published literature and from internet resources such as Rfam, RibEx, TractorDB, RegTransBase, BioCyc and PromScan . The collected information was analyzed to identify the types of regulatory data that were available so that appropriate structural presentation could be developed in the relational database. The data types identified include transcriptional regulator and sigma factor binding sites, operon structures, promoters, regulons and regulatory RNAs. The last category encompasses a diverse class of regulators including non-coding RNAs, tRNA, rRNA, translational and transcriptional terminators, antiterminators, anti-antiterminators and riboswitches. Visualization of the information was implemented by configuring the generic genome browser from the Generic Model Organism Database (GMOD) Toolkitfor specific features of the collected data, like the expression of the IG. Users can overlay the regulatory information with the experimental data by selecting them from a list of preprocessed data sets. This option ('select experimental data set') is available from the genome browser page. The organization of the database and the genome browser has been described in more detail. An example of the regulatory information and experimental data integration is considered in the case study.
## Shewcyc
Support for comparative analysis of metabolism across multiple Shewanella species has been implemented by the construction of ShewCyc, which contains pathway genome databases (PGDBs) of the species and pathway tools [bib_ref] The MetaCyc database of metabolic pathways and enzymes and the BioCyc collection..., Caspi [/bib_ref] to query the databases and overlay the metabolic information with experimental data. ShewCyc hosts a manually curated PGDB, SheonCyc, for MR1 and automatically constructed PGDBs for 16 species. To improve quality of the automatic reconstruction, we have used the curated MR1 PGDB as a reference database and our manually curated orthologs table of the species described above to annotate the species with enzyme names and EC numbers. ShewCyc tools provide a diverse set of options to query and to visualize information in the databases using the Pathway Tools software [bib_ref] Pathway Tools version 13.0: integrated software for pathway/genome informatics and systems biology, Karp [/bib_ref]. An example of overlaying the metabolic information with experimental data is given below. Revealing the biological role of the IGs transcription using knowledgebase: a case study Several SF transcriptomic and proteomic analyses were conducted to investigate MR1 cellular adaptations that occur during transitioning from aerobic to anaerobic growth on fumarate. While the data collected was initially examined only to identify differentially expressed genes and proteins, further biological insights into MR1 behavior can be derived from the same data set through integration with other data types within the SKB, enabled by SKB tools. Below, we provide an example of such an analysis of the data from one of the SF studies (Project 'PNLcrp'). The experimental data chosen was produced using Affymetrix microarray's designed to probe transcripts derived from both genes and IGs [bib_ref] Identification of diverse carbon utilization pathways in Shewanella oneidensis MR-1 via expression..., Driscoll [/bib_ref]. Probes of IGs included on the array were intended to probe a transcription of regions between two closest genes with a unidirectional orientation. IGs for convergent and divergent orientations of genes were not included in the microarray design. Some of the IGs probed by the microarray were rather long and included opposite strands of two or more genes, i.e. they represented antisense transcription. For example, the IG labeled as SO0016_SO0022 was downstream of SO0022 and upstream of SO0016, and transcription of the IG was assumed from the 'plus' strand, i.e. in the same direction as the up-and downstream genes. In the example, the IG transcription was from the stop codon of SO0016 to the start codon of SO0022 and included antisense sequences of five predicted genes transcribed from the 'minus' strand. If two genes were transcribed from the 'minus' strand, then the IG between them was also considered to be transcribed in the same direction. For example, transcription of the IG labeled as SO0197_SO0196 was assumed from the negative strain. This IG did not include antisense sequences of genes, but its transcription might prevent binding of a regulator of the upstream gene, which was SO0196. The microarray was used to measure abundances of transcripts extracted from wild-type and a Crp(À) mutant [bib_ref] Involvement of a membrane-bound class III adenylate cyclase in regulation of anaerobic..., Charania [/bib_ref]. The RNA was collected at various time points (0, 20, 40, 60, 90, 120 min, 4, 8, 12, 24 h, steady-state) during the transition from aerobic growth with lactate to anaerobic growth with lactate and fumarate. The focus of our study was to examine changes in expression of IGs and how these changes correlated with changes in expression of the genes that were adjacent to the IG and transcribed in the same direction. Using SKB tools, we also addressed questions about a biological role of the IG transcription, namely, if generation of antisense transcripts in the studied conditions plays a role in gene silencing or in changing accessibility of regulators to their binding sites in gene promoters.
Almost half of the IGs and more than half of the genes significantly change their expression level in the Crp(À) mutant relative to the wild-type during the transition from aerobic to anaerobic growth To identify IGs and genes with significant changes in gene expression the statistical analysis of the data was implemented as a two class unpaired experiment, with one class referring to the Crp(À) mutant and the other to the wild-type strain of MR1. Biological replicates and time points were combined because of the high pair-wise correlation between them (87-99%) across genes and IGs within either the wild-type or Crp(À) mutant expression profiles (Supplementary Data 1). Stanford Significance Analysis of Microarray (SAM) software (53) was used for the analysis with the recommended parameters and a threshold for false discovery rate of 0.06%. Results of the analysis (Supplementary Data 2) revealed that almost half of the IGs and more than half of the genes significantly changed, either decreased or increased, the level of their expression in the Crp(À) mutant relative to the wild-type. The percentage of IGs yielding significantly decreased expression was about the same (28%) as the percentage of genes that were repressed (31%). About 15% of IGs and 21% of genes had a significantly increased level of expression. Thus, a change in the growth condition in combination with the knockout of Crp, a regulator of the MR1 response to the change, induced significant transcriptional reprogramming of not only genes, but also of IGs in the MR1 genome.
Many IGs are transcribed with only up-or downstream gene, but not as a part of the operon One potential explanation of significant changes in transcription of an IG is that these changes may be coordinated with the transcription of the adjacent genes. This was the initial hypothesis of the study underlying the microarray design. Each IG probed by the array had adjacent genes transcribed in the same direction. The IG and its up-and downstream genes could belong to the same operon and were transcribed and changed their expression together. To validate the hypothesis we have examined whether changes in IG transcription levels correlate with the expression of the adjacent genes (up-and downstream). We averaged the expression levels across time points and replicates for each strain and calculated the log(2) ratio of the average values for IGs/genes so that changes in the expression level of each IG and its adjacent genes in the mutant strain versus the wild-type strain could be compared. Microarray data was re-mapped to the current MR1 gene models, which have been extensively manually curated (see 'knowledgebase description' for details), to update the assignment of the IGs/genes and then compiled in a table comprised of the IG log(2) ratio along with current annotations and log(2) ratios of its up-and downstream genes. Pearson correlation coefficients between changes in expression of the IG and its up-and downstream genes were calculated ('Bioinformatics TOOLBOX' in 'Tools') [fig_ref] Figure 2: Pearson correlations of changes in the expression [log [/fig_ref].
We found a significant correlation between the expression of IGs and either the expression of its upstream or its downstream gene [fig_ref] Figure 2: Pearson correlations of changes in the expression [log [/fig_ref] across all IGs. Even higher correlation (R = 0.77-0.79) was found among the 548 IG and genes that were selected by SAM, i.e. those that showed significant changes in the expression level. Surprisingly, we did not find the same level of correlation between up-and downstream genes in the compiled table. In fact this correlation was twice as low as between an IG and either its up-or downstream gene [fig_ref] Figure 2: Pearson correlations of changes in the expression [log [/fig_ref]. About 50% of the analyzed IGs had a different direction of change in the expression [Crp(À) mutant versus wild-type strain] relative to one or both adjacent genes , even if they belonged to known operons in ShewRegDB('Operon Search').
## Gene silencing by generation of antisense transcripts is not a likely regulatory mechanism in the studied conditions
The transcription of long IGs containing antisense sequences of one or more genes might exert a silencing effect on the genes. Such regulatory effect has been reported for eukaryotes [bib_ref] A cryptic unstable transcript mediates transcriptional trans-silencing of the Ty1 retrotransposon in..., Berretta [/bib_ref]. To check the hypothesis, we selected the long IGs spanning antisense sequences for 1, 2, 3 or 4 genes and analyzed correlation between changes in the expression level of the genes and changes in the expression level of the corresponding antisense IGs . If the antisense IG transcription exerted a silencing effect on the corresponding genes in the studied condition, we would observe a negative correlation between the IGs and genes in the data set. In fact, we found a low positive correlation (R = 0.20) indicating that antisense transcripts do not silence genes transcribed from the opposite strand in the studied conditions. Thus, the regulation of gene transcription by generation of antisense transcripts is not a likely regulatory mechanism in bacteria.
## Changes in the ig transcription have a regulatory effect on one of the adjacent genes
Another potential biological role of IGs transcription proposed for eukaryotes is its regulatory effect on adjacent genes through continued transcription of the gene promoters [bib_ref] Intergenic transcription is required to repress the Saccharomyces cerevisiae SER3 gene, Martens [/bib_ref]. To examine if this regulatory mechanism takes place in MR1, we subdivided the IGs into three subsets according to the consistency of changes in the levels of IG, up-and downstream gene expressions. Subset I was comprised of IGs with the same direction of change (up-or downregulation) as their adjacent genes. Subset A was comprised of IGs with changes in expression that similar to that of their downstream genes and opposite that of their upstream genes. Subset B was comprised of IGs with change in expression similar to that of their upstream genes and opposite that of their downstream genes. We expected that Subset I would be comprised of IGs that are part of the same transcript as both adjacent genes and that consequently there would be high pair-wise correlations between levels of IG, up-and downstream gene expression. Subset A consists of IGs that are transcribed with their downstream genes but not with the upstream gene, so we expected to find correlation between an IG and its downstream gene but not between an IG and its upstream gene. Similarly, in the Subset B we expected to find correlation only between an IG and its upstream gene. For each of the subsets we also analyzed Pearson correlations by further selecting those IGs and adjacent genes with significant changes found by the SAM analysis. We did not consider those IGs with changes of expression opposite to that of both adjacent genes; only three IGs were in this category.
Results of the correlation analysis are presented in [fig_ref] Figure 2: Pearson correlations of changes in the expression [log [/fig_ref] and reveal some unexpected regularities in the IG transcription. The correlation pattern proposed for Subset I, namely, about the same level of high pair-wise correlation among an IG, its up-and downstream gene, was confirmed by the analysis. This pattern suggests that most IGs are transcribed with their adjacent genes as one transcription unit. The correlation patterns in Subsets A and B were unexpected. On one hand, we do find high correlation of the IGs with downstream genes in Subset A and with upstream genes in Subset B, which suggests that in these subsets the IG is transcribed with its downstream gene (Subset A) or with its upstream gene (Subset B). But, on the other hand, in each of the subsets, there was a significant negative correlation between expression levels of the IG and its other adjacent gene, and also between the adjacent genes. According to the correlation pattern in the Subset A, the transcription of the IG occurs together with the downstream gene, and an increase in the level of the transcription corresponds to a proportional decrease in the transcription of the upstream gene. In the Subset B, the IG is transcribed together with the upstream gene, and an increase in the transcription corresponds to a proportional decrease in the transcription of the downstream gene. Thus, in both cases the IG transcription has an obvious regulatory effect on the adjacent gene with the direction of the effect that is opposite to direction of changes in the IG transcription. It means that, similar to eukaryotes, continued transcription of gene regulatory regions is a likely mechanism of bacterial transcriptional gene regulation.
## Discovering a biological role of the ig transcription using skb tools
To understand details of the regulatory mechanism implemented through the IG transcription, we employed SKB tools and explored a specific example from the Subset A of a significantly downregulated [in the Crp(À) mutant] IG (SO2490_SO2491) located downstream of the genes encoding the transcriptional regulator, HexR (SO2490) and upstream to the gene encoding pyruvate kinase II, PykA (SO2491). HexR is also significantly downregulated in the mutant, but pykA is significantly upregulated. To find out more about the roles of the genes, we first employed ShewCyc tools. Using the manually curated PGDB for MR1 in ShewCyc , we noted that the pyruvate kinase catalyzed the glycolytic conversion of phosphoenolpyruvate into pyruvate coupled to the synthesis of ATP. The enzyme participates in different metabolic pathways that include glycolysis, mixed acid fermentation and anaerobic respiration. Thus, PykA has a rather general metabolic function and is important for the bacterial growth on different carbon sources [bib_ref] Participation of the Entner-Doudoroff pathway in Escherichia coli strains with an inactive..., Ponce [/bib_ref]. Furthermore, the enzyme is under control of HexR, its upstream neighbor, which can repress pykA transcription. Further querying of ShewCyc reveals that HexR can repress not only pykA but also enzymes of the Entner-Doudoroff (ED) pathway transcribed from the four gene operon (zwf-eda-edd-pgl), which is located upstream of the hexR, but on the opposite strand. In Escherichia coli, enzymes of the ED pathway are induced by growth on gluconate, glucuronate, or galacturonate and participate in glucose catabolism of strains devoid of a key metabolic enzyme [bib_ref] Responses of the central metabolism in Escherichia coli to phosphoglucose isomerase and..., Hua [/bib_ref] [bib_ref] Global metabolic response of Escherichia coli to gnd or zwf gene-knockout, based..., Zhao [/bib_ref]. Considering the fact that pykA has more general metabolic functions than the ED operon, the activation of the ED suppressor hexR should affect the activity of the ED enzymes operon, but not pykA activity, which may be important for other cellular functions. Transcription of the pykA promoter, which contains the HexR binding site, together with the hexR gene is a straightforward way to implement this type of regulation.
The information from ShewCyc helped us to understand why the transcription of HexR occurs together with the downstream IG harboring the inhibitor binding site. However, it did not explain the observed downregulation of the inhibitor in the Crp(À) mutant. Additional information on regulation of the genes and an overlay of the regulatory information with the gene expressions in the study can be obtained querying ShewRegDB and using Gbrowser of the database to visualize the results . The figure shows that hexR, pykA and the ED operon have predicted Crp binding sites in their promoters listed in .. TractorDB, therefore, the absence of Crp in the mutant can explain the observed changes in the expression of genes and IGs in the genomic locus.
Integrating the information on ED pathway, HexR and PykA functioning and regulation collected using SKB tools we can suggests that co-transcription of SO2490_SO2491 IG with its downstream gene SO2490 (hexR) likely prevents binding of HexR, a transcriptional regulator of the upstream gene SO2491 (pykA), to HexR binding site located in the IG.
## Predicting regulator binding sites using skb
The ortholog editor of the knowledgebase provides further information on how conserved the regulatory mechanism involving co-transcription of hexR and its downstream IG is across Shewanella species and allows identification of the consensus binding site for the regulator. The SKB orthologs table (Tools!Ortholog Editor!Show All) sorted by the MR1 genome order shows that the ED operon, hexR and pykA have orthologs in all sequenced species and represent a conserved synteny, i.e. the order of genes in all genomes is the same. The 'Genome Editor' tool in the SKB allowed us to extract the IGs discussed above in all sequenced species, align them using Muscle [bib_ref] MUSCLE: multiple sequence alignment with high accuracy and high throughput, Edgar [/bib_ref] and then visualize by Jalview (58) using the 'Bioinformatics TOOLBOX' (Tab 'Tools'). Using the alignment we reveal a conserved HexR binding site, which is perfectly matched to the Pseudomonas aeruginosa binding site, and, in addition, . Regulation of pykA, hexR and enzymes of the ED pathway in MR1 and overlay of the regulatory information with the experimental data. The figure indicates that enzymes of the ED pathway (SO2486_ SO2489) as well as pykA are significantly upregulated in the Crp(À) mutant, but hexR and the intergenic region between hexR and pykA are downregulated. The regulatory information provided in the first track of the browser (TractorDB) gives an explanation for the observed changes in the gene expressions. It indicates that the genes, hexR and pykA and the operon with the enzymes of the ED pathway have binding sites in their promoters predicted by TractorDB. Thus, both genes and the operon are likely regulated by Crp. In the wild-type strain of MR1, the ED pathway is not active because of upregulation of hexR by Crp during the transition from aerobic to anaerobic conditions. The Crp(À) mutant strain does not produce Crp; therefore, the transcriptional repressor HexR is absent in the mutant, and enzymes of the ED pathway, which were suppressed by HexR in the wild-type strain, are upregulated in the Crp(À) strain.
.. the consensus binding site for ArcA, which likely activates the ED operon in aerobic conditions [bib_ref] Probing regulon of ArcA in Shewanella oneidensis MR-1 by integrated genomic analyses, Gao [/bib_ref].
# Conclusion
The unique combination of experimental data, computational predictions, curated genome annotations and visualization tools in SKB provides a unique opportunity for researches to discover new biological regularities not only related to Shewanella species, but also to all bacterial organisms. The presented case study demonstrates only some of the available opportunities and capabilities. Although the biological role of IG transcription suggested in the study is only a hypothesis and will require further experimental validation, the discovery of the potential role was significantly facilitated by the collected information and the tools provided in SKB.
## Supplementary data
Supplementary data are available at Database Online. . The alignment of a conserved regulatory locus in the Shewanella genomes in the intergenic region between hexR and zwf genes. The locus is located from À56 to À103 nc upstream of zwf and has three adjacent binding sites for ArcA, HexR and Crp, respectively. The Crp binding site (TAAAAATAGATGCATTTAACAAATTC) was extracted from ShewRegDB. It was predicted by TractorDB [bib_ref] Tractor_DB (version 2.0): a database of regulatory interactions in gamma-proteobacterial genomes, Perez [/bib_ref]. The HexR binding site is identified by similarity to HexR binding site consensus in Pseudomonas species [bib_ref] Regulation of glucose metabolism in Pseudomonas: the phosphorylative branch and Entner-Doudoroff enzymes..., Daddaoua [/bib_ref]. The HexR binding site in Shewanella perfectly matches the Pseudomonas aeruginosa binding site (TGTTGTttaattACtACA) with the same location (À71 nt 5 0 of the transcriptional start site of zwf. Binding of HexR to this binding site in P. aeruginosa was experimentally confirmed by gel shift assays. The HexR binding site in Shewanella has a perfect consensus TGTTGTTATATTACAACA across the species. The alignment also indicates that in Shewanella spp. the HexR binding site overlaps the Crp binding site. Thus, in some growth conditions, the ED pathway may be activated by Crp, but HexR binding to the site can prevent the pathway activation. In the Crp(À) mutant, hexR is downregulated and releases the ED operon for activation by other transcription factors, like ArcA, for example. One of the binding sites for the two-component response regulator ArcA is demonstrated with the consensus (5 0 -GTTAgctagAATGTTA-3 0 ).
..
[fig] .: ß The Author(s) 2010 Published by Oxford University Press This is Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http:// creativecommonsorg/licenses/by-nc/25), which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited [/fig]
[fig] Figure 1: Interplay between the ortholog and genomes editors in SKB (a) and the interface for curation of the Shewanella orthologous table (b). The interface provides access to different types of product annotations, to sequence alignments and to results of the automatic checking of the consistency on protein length and protein families across products of the orthologous group.... [/fig]
[fig] Figure 2: Pearson correlations of changes in the expression [log (2) ratio of the average expression level in Crp(À) mutant versus wild-type strain] of IG regions with changes in the expression of upstream and downstream genes. The analyzed data sets are provided in the Supplementary Data 3. Subset I (1466): IG regions with the same direction of change in gene expression as their neighboring genes. Subset I_S (449): A selection of IG regions from Subset I with significant changes in gene expression in the IG regions and in the neighboring genes. Subset A (805): IG regions with directions of changes in gene expression that are opposite to upstream genes. Subset A_S (54): A selection of IG regions from Subset A with significant changes in gene expression in the IG regions and in the neighboring genes. Subset B (820): IG regions with directions of changes in gene expression that are opposite to downstream genes. Subset B_S (48): A selection of IG regions from Subset B with significant changes in gene expression in the IG regions and in the neighboring genes.... [/fig]
|
Utilizing the Second-Meal Effect in Type 2 Diabetes: Practical Use of a Soya-Yogurt Snack
OBJECTIVE -To evaluate the effect of a prebreakfast high-protein snack upon postbreakfast hyperglycemia.RESEARCH DESIGN AND METHODS -We studied 10 men and women with dietand/or metformin-controlled type 2 diabetes. Metabolic changes after breakfast were compared between 2 days: breakfast taken only and soya-yogurt snack taken prior to breakfast.RESULTS -There was a significant lower rise in plasma glucose on the snack day. The incremental area under the glucose curve was 450 Ϯ 55 mmol ⅐ min/l on the snack day compared with 699 Ϯ 99 mmol ⅐ min/l on the control day (P ϭ 0.013). The concentration of plasma free fatty acids immediately before breakfast correlated with the increment in plasma glucose (r ϭ 0.50, P ϭ 0.013).CONCLUSIONS -Consuming a high-protein prebreakfast snack results in almost 40% reduction of postprandial glucose increment. The second-meal effect can be applied simply and practically to improve postbreakfast hyperglycemia in people with type 2 diabetes.
P revious studies [bib_ref] The second-meal phenomenon is associated with enhanced muscle glycogen storage in humans, Jovanovic [/bib_ref] have shown a considerable reduction in hyperglycemia after the second meal of the day, provided that breakfast had been taken. The preservation of this effect in type 2 diabetes was not confirmed until recently [bib_ref] The second-meal phenomenon in type 2 diabetes, Jovanovic [/bib_ref]. Postprandial hyperglycemia acts as an independent risk factor for cardiovascular disease (4 -6), a major cause of death in subjects with type 2 diabetes [bib_ref] Hyperglycemia and cardiovascular disease in type 2 diabetes, Laakso [/bib_ref].
We hypothesized that postbreakfast hyperglycemia in subjects with type 2 diabetes could be improved nonpharmacologically by using a high-protein, lowcarbohydrate prebreakfast snack.
## Research design and
METHODS -Ten subjects (four female, six male) with type 2 diabetes treated by diet and/or metformin were recruited (aged 61 Ϯ 2.3 years, waist-to-hip ratio 0.96 Ϯ 0.01, BMI 28.8 Ϯ 1.0 kg/m 2 , body fat 33.4 Ϯ 2.6%, and A1C 6.7 Ϯ 0.1%). The Newcastle and North Tyneside Research Ethics Committee No. 2 approved the study. All participants gave written informed consent.
## Design and protocol
Every subject was studied on 2 separate days in random order with 2-4 week intervals. On both days, breakfast (51 g carbohydrate; 4.8 g fat; 5.8 g protein) was given at 10:00 A.M. On the snack day only, a snack (30 g soya beans; 75 g yogurt) was given 2 h before the breakfast. The details of metabolic testing, hormone, and metabolites assays were as previously described [bib_ref] Acute inhibition of lipolysis does not affect postprandial suppression of endogenous glucose..., Carey [/bib_ref].
# Statistical analysis
Data are expressed as means Ϯ SEM. Paired Student t test was performed by SPSS (version 17.0; SPSS, Chicago). Incremental area under the curve (IAUC) was computed by the trapezoidal rule.
# Results
## Plasma glucose
At baseline, 2 h before breakfast, fasting plasma glucose was similar on the 2 study days (6.6 Ϯ 0.3 and 6.6 Ϯ 0.4 mmol/l). The snack induced a small peak in plasma glucose at Ϫ60 min (7.3 Ϯ 0.4 mmol/l). The concentrations of plasma glucose immediately before breakfast (t ϭ 0 min) did not differ significantly between the 2 days (6.6 Ϯ 0.3 vs. 6.4 Ϯ 0.3 mmol/l, P ϭ 0.45) [fig_ref] Figure 1 -: Metabolic changes between the 2 study days in 10 subjects with type... [/fig_ref].
Following breakfast, the rise in plasma glucose was lower on the snack day. IAUC was 450 Ϯ 55 mmol ⅐ min/l compared with 699 Ϯ 99 mmol ⅐ min/l for snack and control day, respectively (P ϭ 0.013) [fig_ref] Figure 1 -: Metabolic changes between the 2 study days in 10 subjects with type... [/fig_ref]. The 2-h postprandial plasma glucose concentration was significantly lower on the snack day compared with the control day (9.8 Ϯ 0.3 vs. 10.8 Ϯ 0.4 mmol/l, P ϭ 0.001). Plasma glucose levels were identical in the two groups 5 h after breakfast.
## Plasma free fatty acid
The fasting plasma free fatty acid (FFA) concentrations on both study days were similar (0.57 Ϯ 0.05 and 0.57 Ϯ 0.04 mmol/l, P ϭ 0.9). After ingestion of the snack, plasma FFA concentration declined, falling to 70% of the fasting levels just before breakfast (0.41 Ϯ 0.04 mmol/ l). After breakfast, plasma FFA levels in both groups were suppressed (0.16 Ϯ 0.02 vs. 0.14 Ϯ 0.02 mmol/l for snack and control days, respectively) [fig_ref] Figure 1 -: Metabolic changes between the 2 study days in 10 subjects with type... [/fig_ref]. FFA concentration before breakfast was positively correlated with the postprandial IAUC for plasma glucose (r ϭ 0.50, P ϭ 0.013).
## Serum insulin and c-peptide
Fasting mean serum insulin levels were similar on the 2 study days (13.4 Ϯ 2.4 and 13.2 Ϯ 2.7 mU/l). On the snack day, insulin concentration rose promptly after ingestion of the snack to 26.5 Ϯ 4.0 mU/l at Ϫ60 min and fell before breakfast [fig_ref] Figure 1 -: Metabolic changes between the 2 study days in 10 subjects with type... [/fig_ref]. Following breakfast, insulin levels increased similarly on both days (58.5 Ϯ 7.9 vs. 51.2 Ϯ 7.6 mU/l) at 60 min. The mean insulin-to-C-peptide ratios from the area under the curves (t ϭ 0 -5 h) were similar on both days (13.6 Ϯ 3.2 vs. 16. 9 Ϯ 1.4 mU/nmol, P ϭ 0.36) implying similar hepatic insulin extraction.
## Substrate oxidation
Fasting glucose oxidation rates were not significantly different on the 2 study days (0.55 Ϯ 0.18 vs. 0.77 Ϯ 0.25 mg/kg ⅐ min, P ϭ 0.37). On the snack day, glucose oxidation rate increased to 1.38 Ϯ 0.24 mg/kg ⅐ min at 90 min after breakfast compared with 1.09 Ϯ 0.25 mg/kg ⅐ min on the control day (P ϭ 0.11). Fasting lipid oxidation rates were similar (0.7 Ϯ 0.1 vs. 0.63 Ϯ 0.12 mg/kg ⅐ min) and 90 min after breakfast decreased to 0.39 Ϯ 0.09 vs. 0.51 Ϯ 0.09 mg/kg ⅐ min, respectively (P ϭ 0.09).
CONCLUSIONS -This study demonstrated for the first time that the provision of a practical, high-protein, lowcarbohydrate snack prior to breakfast reduced by 40% the postprandial plasma glucose increment in people with type 2 diabetes. These findings confirm a potent expression of the second-meal effect in people with type 2 diabetes similar to that resulting from intravenous arginine infusion [bib_ref] The second-meal phenomenon in type 2 diabetes, Jovanovic [/bib_ref]. The importance of the present observation is that a more practical means of improving glucose tolerance could potentially be of therapeutic benefit in people with type 2 diabetes.
We observed no effect of the prior snack on insulin secretion after breakfast.
The mechanism underlying the secondmeal effect has been shown to be due to suppression of plasma FFA, allowing greater storage of glucose as muscle glycogen [bib_ref] The second-meal phenomenon is associated with enhanced muscle glycogen storage in humans, Jovanovic [/bib_ref]. We previously demonstrated a strong negative correlation between the decrease of preprandial plasma FFA levels and the postmeal glucose increment. In the present study, a significant positive correlation was found between prebreakfast plasma FFA and the rise in postprandial plasma glucose concentration. This observation is analogous to the acute effect of the antilipolytic nicotinic acid analog acipimox [bib_ref] Effect of the antilipolytic nicotinic acid analogue acipimox on whole-body and skeletal..., Vaag [/bib_ref].
The snack used in the present study was empirically designed. It will be important to optimize both the composition of the snack and the interval before breakfast to maximize the benefit of this approach. In everyday life, the gap between snack and breakfast would have to be ac- commodated, for instance, by delaying breakfast until mid-morning. Although the snack induced a small increase in plasma glucose, it was minimal and unlikely to contribute to the hyperglycemic burden. The sample size was dictated by prior power calculation (80% power with 10 subjects).
We have demonstrated that a highprotein, low-carbohydrate snack before breakfast attenuates postbreakfast hyperglycemia, and further studies must determine whether long-term use is associated with improvement in A1C.
[fig] Figure 1 -: Metabolic changes between the 2 study days in 10 subjects with type 2 diabetes. A: Plasma glucose concentrations. B: Incremental area under the plasma glucose curves during the 5-h postbreakfast period. C and D: Change in FFA and insulin concentrations, respectively. *P Ͻ 0.05; **P Ͻ 0.02; ***P Ͻ 0.01; Ⅺ, snack day; F, control day. [/fig]
|
Understanding the Current Health Services Research Workforce and Maximizing its Future
In 2016, AcademyHealth continued its longstanding efforts to understand the health services research (HSR) workforce, to inform its changing needs through the commissioning of several papers and an invitational conference. This paper serves to summarize the commissioned studies that appear in the current issue of this journal.Defining the evolving boundaries of health services research and its workforce remains a challenging and complex exercise. As a multidisciplinary field, health service research has, at times, struggled to remain broad enough to encompass the diverse interests of its members while trying to become more precise for the purposes of branding and advocacy work. This tension has created obstacles for understanding the current stock of the health services research workforce, as well as its future needs and direction. In the context of this background, AcademyHealth hosted a strategic conference of key stakeholders of the health services research workforce to plan for the future of the field. A description of the planning process and resulting recommendations from the conference is available in the current issue (Menachemi, Wolfe, and Simpson 2018). In addition, AcademyHealth commissioned a variety of analyses that were shared with conference participants and appear as full manuscripts in the current supplement issue of the journal. Below is a description of This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
each of these featured papers which collectively seek to address many of the above challenges, as well as some opportunities for the field of health services research to advance training programs, increase diversity, and ultimately produce higher quality, more impactful research.
## Overview of featured papers
The featured papers examine (1) the current stock of health services researchers [bib_ref] Update on the Stock and Supply of Health Services Researchers in the..., Frogner [/bib_ref] ; (2) employment demand for health services researchers [bib_ref] Current and Future Demand for Health Services Researchers: Perspectives from Diverse Research..., Rich [/bib_ref] ; (3) trends in health services research funding [bib_ref] Show Me the Money! Trends in Funding for Health Services Research, Simpson [/bib_ref] ; (4) funding trends for the training of health services researchers [bib_ref] Funding the Training of Future Health Services Researchers, Mor [/bib_ref] ; (4) issues and trends in the global health policy and systems research workforce [bib_ref] Building a Workforce for Future Health Systems: Reflections from Health Policy and..., Javadi [/bib_ref] ; (6) updates to the health services research doctoral core competencies [bib_ref] Update on the Health Services Research Doctoral Core Competencies, Burgess [/bib_ref] ; (7) updates to the Canadian health services and policy research core competencies [bib_ref] Development of Enriched Core Competencies for Health Services and Policy Research, Bornstein [/bib_ref] ; and (8) recommendations for supporting the growth and evolution of the health services research workforce.
One of the primary issues within the field of health services research is identifying its bounds. Frogner's work seeks to estimate the number of health services researchers and the industries within which they work. By leveraging numerous data sources, including new social media sources, and building on the work of McGinnis and Moore (2009), Frogner is able to provide an updated estimate for the number of individuals who identify as health services researchers.
In the current supplement issue, authors examine the ways in which the field of health services research measures and understands the pipeline of new researchers into the field. Because the field of health services research is inherently interdisciplinary, students entering the field do not just come from traditional health services research doctoral programs. Researchers in the field hold advanced degrees in public health, medicine, law, and other social and basic sciences, making it difficult to predict how the field of health services research will grow and evolve. Frogner looks more closely at health services research-related training programs and provides guidance on how the field might change and grow in the foreseeable future.
Whereas Frogner examines the current stock of health services researchers, as well as future supply of students entering the field, Rich and Collins examine the current and future demand for health services researchers. Through interviews with leaders working in diverse sectors that employ health services researchers, Rich and Collins were able to characterize the current and future needs of employers in the field and identify skills gaps that training programs in the field might address. Many common themes emerged from the interviews despite the informants representing different settings and a focus on different topics of health services research.
The set of papers from Simpson and colleagues, and from Mor and Wallace provide an overview of current federal funding trends in health services research and funding for health services research training programs, respectively. From these it is clear that the vast majority of public funding for health services research comes from the National Institutes of Health and that recent pressures on federal discretionary spending as well as enduring questions about the value of funding health services research have slowed growth in funding.
In examining funding for training programs, Mor and Wallace provide an overview on the current funding climate and explore alternative funding and training models that health services research training programs may consider adopting to address some of the skills gaps raised by Rich and Collins. In particular, they discuss the importance of training in applied setting and in diverse fields and sectors to provide new researchers with the pragmatic and business-minded experience that employers seek. Mor and Wallace's commentary should serve as a starting point for health services research training programs that are looking to innovate and grow during a time where funding is more uncertain and potentially more precarious.
Javadi and colleagues provide a global perspective on the health policy and systems research workforce, its relationship to health services research, and its importance to achieving the Sustainable Development Goals adopted by countries in 2015. Outlining similar challenges to those faced by the USbased workforce, the authors reflect on challenges and strategies in managing the global health policy and systems research workforce in order to stimulate dialogue and learning across similar fields.
Burgess and colleagues make a significant contribution to health services research training programs in their commentary, which provides an update of the doctoral core competencies to reflect advances in methods and topic-specific proficiencies that have emerged over the past several years. While a number of these remain unchanged from the previous version [bib_ref] Health Services Research Doctoral Core Competencies, Forrest [/bib_ref] , they also speak to the many new developments in the field. Much like Burgess et al., Brown and colleagues worked with various stakeholders in the health services and policy research field, including leaders of training programs, health policy and health services practitioners, and funders, to update the core competencies for doctoral HSPR trainees in Canada. Drawing many parallel conclusions, the authors of both papers propose changes to the doctoral competencies that emphasize applied skillsets and equip trainees with a range of proficiencies, including analyzing complex problems using a variety of methods, interdisciplinary work skills, and knowledge translation.
The final paper in the current supplement issues outlines a set of recommendations that emerged from a consensus-building process of a group of stakeholders that was facilitated by AcademyHealth. Menachemi and colleagues share this process and the resulting action areas that AcademyHealth will engage in to continue to monitor and report on the status of the health services research workforce.
## Summary
The topics and trends described in this paper are strikingly similar to conclusions drawn by fields related to health services research, such as the biomedical workforce and the clinical workforce [bib_ref] Snapshot of the Biomedical Workforce, Meggeness [/bib_ref]. These fields are also experiencing radical shifts including the expansion of types of employers, changes in training and career trajectories, and a diversifying workforce.
In April 2018, the National Academies of Science, Engineering and Medicine (NASEM) released a report on the biomedical research workforce, including health services research . It documents the findings of earlier reports and includes a number of recommendations which align with the recommendations put forth from the AcademyHealth Workforce Initiative Task Force that are outlined in this supplement. For example, the report notes the continued decrease in the number of Ph.D.'s who secure tenure-track positions and the increasing average age at which an R01 is secured, both indicators of the growing difficulty of establishing independent research careers in academic settings. The committee also notes the mismatch between the training that is received and the available career opportunities outside of academic settings, a key finding of Collins and Rich. The NASEM report goes on to recommend that "Congress and the National Institutes of Health (NIH) should create and expand existing entrepreneurial and private-sector opportunities", echoing the recommendations in this supplement. In another example, the report calls for more transparency and accountability for monitoring the research workforce, including its inclusion of underrepresented racial/ethnic minorities, another recommendation consistent with those for health services research.
The papers in the current issue provide a snapshot of the challenges and opportunities facing the health services research workforce, and fundamentally, our field. Some of these are not new and must continued to be grappled with, such as identifying the bounds of the field and supporting a diverse pipeline of researchers, while others signal shifts in the realities of funding availability and responses to the changing needs of employers. The environment is different due in part to the explosion of new data, a stronger focus on social determinants and population health, far greater demand for stakeholder engagement, and renewed and perhaps loudercalls for reducing the costs and increasing the quality of healthcare. With these and other changes in the overall health care and research ecosystems, AcademyHealth and the field must continue to ensure that important research questions are being addressed so that evidence can inform policy and practice changes designed to ultimately improve health and health care.
ACKNOWLEDGMENTSJoint Acknowledgment/Disclosure Statement: The authors thank the Agency for Healthcare Research and Quality (HHSP233201600155P), the Robert Wood Johnson Foundation, and the Patient-Centered Outcomes Research Institute for generous funding and guidance. The views expressed here do not necessarily reflect the views of these organizations.Disclosure: None. Disclaimer: None.Appendix SA1: Author Matrix.3926HSR: Health Services Research 53:5, Part II (October 2018) |
Cost‐effectiveness of second‐line empagliflozin versus liraglutide for type 2 diabetes in the United States
Aim: To estimate the cost-effectiveness of sequential use of the sodium-glucose cotransporter-2 inhibitor empagliflozin and glucagon-like peptide-1 receptor agonist liraglutide after metformin in patients with type 2 diabetes (T2D) from the US payer perspective.Materials and Methods: An economic simulation model with a lifetime horizon was developed to estimate T2D-related complications (including cardiovascular[CV] death, myocardial infarction, stroke, and renal outcomes) using EMPA-REG OUTCOME data or UK Prospective Diabetes Study risk equations, in patients with or without a history of cardiovascular disease (CVD), respectively. Evidence synthesis methods were used to provide effectiveness inputs for empagliflozin and liraglutide.Population characteristics, adverse event rates, treatment escalation, costs ($2019), and utilities (both discounted 3%/year) were taken from US sources.Results: Compared with second-line liraglutide in the overall T2D population, second-line empagliflozin was dominant as it was associated with lower total lifetime cost ($11 244/patient less) and resulted in a quality-adjusted life-year (QALY) gain (0.32/patient). Second-line empagliflozin was associated with reductions in CV death (by 5%) and lower cumulative complication rates in patients with CVD (by 2%), relative to second-line liraglutide. These findings were consistent among patients with co-morbid CVD, with gains in incremental QALYs (0.43/ patient) and lower lifetime cost (by $10 175/patient) relative to second-line liraglutide. Scenario analyses consistently showed dominance for second-line empagliflozin.Conclusion: For patients with T2D, use of second-line empagliflozin combined with metformin was a dominant strategy for US payers, associated with extended survival, improved QALYs, and lower costs compared with second-line liraglutide. K E Y W O R D S cardiovascular disease, cost-effectiveness, empagliflozin, GLP-1, liraglutide, SGLT-2 inhibitor
# | introduction
Type 2 diabetes (T2D) is a leading cause of premature morbidity and mortality in the adult population within the United States (US).Alone, or combined with risk factors (e.g. high blood glucose, overweight/ obesity, and high blood pressure), patients with T2D are at a high risk of microvascular and macrovascular complications, including cardiovascular disease (CVD), kidney disease, blindness, and lower-limb Clinical trials do not capture the potential sequences of therapies that physicians could recommend in real-world practice, and there is a lack of published data documenting the economic implications of sequential use of these therapies to ensure healthcare decision-makers recommend therapies that bring value to patients, healthcare systems, and payers. This analysis estimated the costeffectiveness of adding empagliflozin as a second-line agent followed by liraglutide as third line compared with liraglutide as a second-line agent followed by empagliflozin as third line, in patients with T2D from the US payer perspective. A counterfactual comparison of projected lifetime health outcomes and costs of these two treatment pathways was performed, where the main difference was the drug added as second-line therapy, to understand the costs and benefits of initiating empagliflozin earlier in the treatment pathway rather than later. By combining clinical trial data and economic modelling methods, this study contributes a novel analysis of two alternative treatment escalation pathways in T2D that have not been directly compared in clinical research, focusing on long-term patient outcomes and direct medical costs for US payers. This study was conducted using a previously published model for the United States that assessed sequential treatment with second-line empagliflozin versus sitagliptin, with appropriate modifications to reflect clinical efficacy, drug adverse events (AEs), health-related quality of life (QoL), and costs with liraglutide. [bib_ref] Cost-effectiveness analysis of empagliflozin versus sitagliptin as second-line therapy for treatment in..., Reifsnider [/bib_ref] 2 | MATERIALS AND METHODS clinical event history, using a 3% annual discount rate for costs and QALYs as recommended for US cost-effectiveness analysis. [bib_ref] Recommendations for conduct, methodological practices, and reporting of cost-effectiveness analyses: second panel..., Sanders [/bib_ref] The
## | model overview
[formula] F I G U R E 1 [/formula]
## | patient population
## | treatment initiation
Treatment initiation was based on a mean rate of initiations for third-line . These parameters were derived from an indirect treatment comparison (ITC) performed in a network meta-analysis, in which the relative clinical efficacy of treatment was compared with placebo in adult patients with T2D who were inadequately controlled with metformin (Appendix S1). The model considered treatment effects during the first year of treatment and assumed that patients progressed subsequently according to the UKPDS Outcomes Model 1 (UKPDS-OM1) [bib_ref] A model to estimate the lifetime health outcomes of patients with type..., Clarke [/bib_ref] equations for evolving risk factors. The efficacy of these treatments as third-line therapy was assumed to be the same as for second-line therapy.
## | clinical and treatment efficacy inputs
In patients with CVD, published event-free survival (EFS) curves developed from EMPA-REG OUTCOME CVOT data were applied to estimate the time to non-fatal MI, non-fatal stroke, HF, UA, TIA, revascularization, macroalbuminuria, renal injury, renal failure, and CV death using baseline (i.e. at the time CVD is established) and time-dependent covariates. [bib_ref] Cost-effectiveness analysis of empagliflozin treatment in people with type 2 diabetes and..., Kansal [/bib_ref] These complications were defined as reported by Kansal et al. [bib_ref] Cost-effectiveness analysis of empagliflozin treatment in people with type 2 diabetes and..., Kansal [/bib_ref] The shape of the EFS curves were assumed to implicitly capture HbA1c and other evolving risk factors that contribute to changing rates of complications and disease progression over time. Thus, the treatment benefit of glycaemic control was not directly modelled in patients with CVD. The equations were supplemented with non-CV death according to US age-and sex-adjusted life table data. [bib_ref] United States life tables, Arias [/bib_ref] Time to each T2D-related complication was estimated with empagliflozin treatment as a covariate to capture efficacy shown in the EMPA-REG OUTCOME CVOT. The relative clinical efficacy of liraglutide was taken from an ITC,in which liraglutide was compared with empagliflozin in treating adult patients with T2D and established CVD based on EMPA-REG OUTCOME and published LEADER CVOT data. A hazard ratio (HR) for liraglutide versus empagliflozin on each complication except UA, TIA, and revascularization was obtained . 14 Data on these complications were not available from the LEADER CVOT publication, 5 therefore risks were assumed to be similar to the placebo arm of the EMPA-REG OUTCOME CVOT. The HRs were applied to the EFS functions for empagliflozin to estimate outcomes among simulated patients treated with liraglutide.
## Risks of treatment-related aes for empagliflozin and liraglutide
were based on US prescribing labels; AEs with 5% or more incidence or key adverse reactions were selected, specifically urinary tract infection, genital mycotic infection, nausea, hypoglycaemia, and injection site reaction.AE occurrence was assumed to be mutually exclusive and not impact treatment use.
## | utility inputs
Utility scores that captured the QoL of modelled patients with different health status were used to adjust survival and estimate QALYs . Utility values were sourced from published literature and the algorithm developed by Sullivan and Ghushchyan 17 was used (Appendix S1). The same study reported US-specific values for patient utility at baseline and disutilities for most clinical events. [bib_ref] EQ-5D scores for diabetes-related comorbidities, Sullivan [/bib_ref] Other .
[formula] T A B L E [/formula]
Costs for the overall population were estimated by weighting costs for a commercially-insured (55% <65 years of age) and Medicare (45% ≥65 years of age) payer. Refer to Appendix S1 for additional details. , SA5, and SA7.
## | sensitivity analysis inputs
## | model validation and verification
Model validation and verification included evaluation of face validity, technical validity, and predictive validity. The validation approach and results are provided in Tables SA8 and SA9; Appendix S1.
# | results
## | population characteristics
Baseline characteristics of the simulated population have been previously reported. [bib_ref] Cost-effectiveness analysis of empagliflozin versus sitagliptin as second-line therapy for treatment in..., Reifsnider [/bib_ref] The mean age was 61.4 ± 13.3 years, 50% were female, mean body mass index was 31.0 ± 7.0 kg/m 2 , mean HbA1c was 9.4% ± 3.5% (79 ± 15 mmol/mol), and mean SBP was 144.8 ± 27.1 mmHg. At baseline, 20% of patients had CVD.
## | base case results
Base case results are shown in
## | sensitivity analyses results
In DSA, all scenarios showed that second-line empagliflozin provides QALY improvement with cost savings compared with second-line liraglutide . Empagliflozin also provides survival benefits at .
## T a b l e 2 scenario analyses results
## | scenario analyses results
The results of scenario analyses are shown in The strengths of the model include that outcomes in patients with CVD were based on hard endpoint data from CVOTs to more accurately capture observed CV event rates. Furthermore, analyses were based on US-specific inputs for the population, non-CV death rates, treatment escalation rates, utilities, and costs.
The model included several assumptions. The occurrence of diabetes-related complications estimated from the UKPDS data was assumed to reflect rates observed in T2D patients without CVD in the United States. As the UKPDS-OM2 equations have been broadly used in diverse T2D populations worldwide, the applicability of these equations to a US population without CVD is reasonable. [bib_ref] Economic models in type 2 diabetes, Yi [/bib_ref] Next, the treatment effect of liraglutide on revascularization, UA, and TIA in patients with CVD was assumed to be similar to the placebo arm in the EMPA-REG OUTCOME CVOT, because the LEADER CVOT publication did not report treatment benefit on these outcomes. This was challenged in sensitivity analyses by assuming: (a) liraglutide had the same effect as empagliflozin on these outcomes; and (b) the effect of UA and revascularization to be the same as MI, and the effect of TIA to be the same as stroke. Results were insensitive to this assumption.
## Peer review
The peer review history for this article is available at https://publons. com/publon/10.1111/dom.14625.
# Data availability statement
Our study data (which is based on de-identified data from a clinical trial) is not in a repository, but is available upon reasonable request from the corresponding author.
## Orcid
Odette S. Reifsnider https://orcid.org/0000-0003-0714-5619
Sarah Brand https://orcid.org/0000-0003-3116-5951
[table] 1: Base case model results Abbreviations: AE, adverse event; CV, cardiovascular; CVD, cardiovascular disease; ICER, incremental cost-effectiveness ratio; LYs, life years; PY, personyears; QALY, quality-adjusted life year; T2D, type 2 diabetes; USD, United States dollar. [/table]
|
Hospital case volume and outcomes for proximal femoral fractures in the USA: an observational study
Objective: To explore whether older adults with isolated hip fractures benefit from treatment in highvolume hospitals.Design: Population-based observational study. Setting: All acute hospitals in California, USA. Participants: All individuals aged ≥65 that underwent an operation for an isolated hip fracture in California between 2007 and 2011. Patients transferred between hospitals were excluded.Primary and secondary outcomes: Quality indicators (time to surgery) and patient outcomes (length of stay, in-hospital mortality, unplanned 30-day readmission, and selected complications).Results: 91 401 individuals satisfied the inclusion criteria. Time to operation and length of stay were significantly prolonged in low-volume hospitals, by 1.96 (95% CI 1.20 to 2.73) and 0.70 (0.38 to 1.03) days, respectively. However, there were no differences in clinical outcomes, including in-hospital mortality, 30-day re-admission, and rates of pneumonia, pressure ulcers, and venous thromboembolism.Conclusions: These data suggest that there is no patient safety imperative to limit hip fracture care to high-volume hospitals.Metcalfe D, et al. BMJ Open 2016;6:e010743.
# Introduction
There are approximately 250 000 hip fractures in the USA 1 every year, which are a major cause of mortality and morbidity. High provider case volumes have been associated with improved outcomes across a range of surgical procedures, including major arterial vascular surgery, 2 oesophageal resection, [bib_ref] Surgical volume and quality of care for esophageal resection: do high-volume hospitals..., Dimick [/bib_ref] and elective arthroplasty. A small number of studies have explored the effect of hospital volume on hip fracture outcomes. [bib_ref] Hemiarthroplasty in hip fracture care: effects of surgical volume on short-term outcome, Lavernia [/bib_ref] [bib_ref] Ninety-day mortality after intertrochanteric hip fracture: does provider volume matter?, Forte [/bib_ref] [bib_ref] Hip fracture outcomes: does surgeon or hospital volume really matter?, Browne [/bib_ref] [bib_ref] Does practice make perfect? Examining the relationship between hospital surgical volume and..., Hamilton [/bib_ref] [bib_ref] Hemiarthroplasty for femoral neck fracture in the elderly surgeon and hospital volume-related..., Shah [/bib_ref] [bib_ref] Modeling the volume-effectiveness relationship in the case of hip fracture treatment in..., Sund [/bib_ref] However, these reports reached inconsistent conclusions, with only two identifying a relationship between hospital volume and outcomes. [bib_ref] Ninety-day mortality after intertrochanteric hip fracture: does provider volume matter?, Forte [/bib_ref] [bib_ref] Hemiarthroplasty for femoral neck fracture in the elderly surgeon and hospital volume-related..., Shah [/bib_ref] These studies predominantly used cross-sectional data sets that could not measure longitudinal outcomes such as readmission to hospital and complications following discharge. [bib_ref] Hemiarthroplasty in hip fracture care: effects of surgical volume on short-term outcome, Lavernia [/bib_ref] They also included cases from over 15 years ago 7 that are unlikely to reflect modern hip fracture management. Contemporary hip fracture treatment emphasises the use of standardised clinical pathways, [bib_ref] Implementing a clinical pathway for hip fractures; effects on hospital length of..., Burgers [/bib_ref] formal geriatric assessment [bib_ref] Orthogeriatric care models and outcomes in hip fracture patients: a systematic review..., Grigoryan [/bib_ref] and early operation to expedite mobility. [bib_ref] Does timing of surgery matter in fragility hip fractures?, Leung [/bib_ref] It is possible that the increasing standardisation of hip fracture management will have influenced any relationship between clinical outcomes and provider volume.
A recent systematic review called for more studies aimed at characterising volumeoutcome relationships for specific orthopaedic patient populations. [bib_ref] Orthopaedic procedure volume and patient outcomes: a systematic literature review, Shervin [/bib_ref] This is necessary to determine the optimal setting for hip fracture patients and to inform both prehospital triage and interhospital referral pathways.
The aim of this study was to explore associations between case volume and outcomes using a comprehensive population database.
# Methods
## Data source
Hospital discharge records were analysed from the California State Inpatient Database (SID) 2007-2011. [bib_ref] Systematic review of comorbidity indices for administrative data, Sharabiani [/bib_ref] and has been shown to predict resource utilisation 18 and mortality [bib_ref] Predicting 30-day mortality following hip fracture surgery: evaluation of six risk prediction..., Karres [/bib_ref] in hip fracture populations. Hospital characteristics included trauma centre level (1-4, with level 1 representing large regional trauma centres), teaching status (defined as hosting a physician training programme accredited by the Accreditation Council for Graduate Medical Education (ACGME)), and hospital bed size (categorised as <200 and ≥200 beds).
Unique identifiers within the SID were used to determine the annual hip fracture case volume of each hospital. Visual inspection of a histogram (number of hip fracture patients vs the annual volume at each treating hospital) revealed four distinct groups (figure 1). The four volume groups were defined as: low <20, intermediatelow 20-99, intermediate-high 100-215, and high >215 cases per year. Although data from all categories are reported, the principal comparison in this paper was between high and low-volume hospitals.
## Outcome measures
Outcome measures included length of stay (LOS), in-hospital mortality, unplanned readmission, and selected complications experienced as an inpatient or within 30 days postdischarge from the hospital. Both days to operation and LOS were measured from time of admission rather than time of injury, which is not available from the SID. Complications were identified by ICD-9-CM codes as venous thromboemboli (deep vein thrombosis 453.4, pulmonary embolus 415.1, pneumonias (480-488) and decubitus ulcers (707.0). These complications were also considered together as a single composite outcome. Only patients discharged alive from the hospital were eligible for calculating LOS and 30-day readmission. Readmissions and sequelae were captured even if the patient presented to a different (non-index) hospital in the state of California rather than the institution that treated their hip fracture.
# Sensitivity analysis
As comparatively few patients (and associated adverse events) were anticipated in the low-volume category, a sensitivity analysis was planned with low and intermediate-lowvolume categories combined before comparison with the two higher volume groups.
# Statistical analysis
Outcome variables were compared between the volume categories using χ 2 tests for categorical variables and Kruskall-Wallis one-way analysis of variance for nonnormally distributed continuous variables. Multivariable logistic regression models were used to examine the risk-adjusted associations of case volume with mortality, unplanned 30-day readmission, and postoperative complications. Covariates included in regression models were age, sex, race, payment source, weighted CCI (as a continuous covariate), discharge destination, hospital bed size, teaching status and trauma centre level. All models accounted for clustering of patients within hospitals and used robust SEs.
LOS presented as right-skewed data and so riskadjusted associations were explored using generalised linear regression with a γ distribution 20 and link log followed by postestimation of average marginal effects to attain predicted mean differences in LOS. The threshold for statistical significance was set at two-sided p<0.05. Statistical analyses were performed using Stata V.13.0. The Partners Human Research Committee approved the study protocol (IRB 2014P002072/BWH). Within California, there were 257 individual hospitals that treated hip fractures, characteristics of which are described in table 2. They were predominantly teaching institutions (77.0%) without trauma centre designation (73.2%) and located in a non-rural setting (87.9%). The mean annual case volume was 79.1 (SD 72.6). However, this varied significantly between the categories: low 5.2 (SD 6.0), intermediate-low 59.0 (SD 22.6), intermediatehigh 150.0 (SD 34.5), and high volume 276.0 (SD 37.5) cases per year ( p<0.001). A higher proportion of lowvolume hospitals were rurally situated (23.5% vs 0.0%, p<0.001) and maintained an accredited residency programme (79.4% vs 73.3%, p<0.001) but a lower proportion were designated as a trauma centre (26.8% vs 33.3%, p<0.001). Low-volume hospitals were also smaller in size, ranging from a mean of 109.3 beds in the low volume to 348.8 in the high volume category ( p<0.001).
# Results
## Patient and hospital characteristics
## Clinical outcomes
The unadjusted outcomes are summarised in table 3 and results of the multivariable regression analyses in table 4.
## Time to operation
The overall median time to the operating theatre was 1.0 days (IQR 0.0-2.00). In the unadjusted analysis, lowvolume hospitals had a longer time to theatre (median 1, 90th centile 3 days) than high-volume hospitals (median 1, 90th centile 2 days) ( p<0.001). Within a generalised linear regression model, adjusted surgical delay was inversely associated with hospital volume ( p<0.001). This was a stepwise association with a higher predicted mean difference observed in each successive volume category relative to high- theatre almost 2 days later than those in the highest volume category ( p<0.001).
## Length of stay
Median LOS was 5.0 (IQR 4.0-6.0) days but this was inversely associated with case volume in unadjusted and adjusted analyses. In the multivariable regression model, there was no significant difference between the two highest volume categories. However, LOS in the intermediate-low volume and low-volume groups were 0.32 and 0.70 days longer respectively. A higher proportion of patients were discharged to another care facility from high-volume hospitals than from low-volume hospitals (86.9% vs 79.2%, p<0.001).
## In-hospital mortality
There were 1663 in-hospital deaths in the cohort, with an overall mortality of 1.8%. There were no significant differences between volume categories in terms of in-hospital mortality, either in the unadjusted ( p=0.585) or adjusted ( p=0.380) analyses. The sensitivity analysis (high volume vs combined low and intermediate-lowvolume hospitals) also did not detect any difference between the combined low volume and high-volume categories ( p=0.964).
Thirty-day unplanned readmissions A total of 9888 (11.0%) patients required unplanned readmission to hospital within 30 days of discharge.
Rates of readmission varied between the groups with the highest rate observed in the low-volume category (12.6%, p=0.042). In the multivariable analysis, there was no consistent association between case volume and likelihood of readmission (OR 1.06, 95% CI 0.65 to 1.73). This finding was confirmed by the sensitivity analysis with combined low-volume categories (OR 0.88, 0.74 to 1.05).
## Hip fracture sequelae
Major hip fracture sequelae (venous thromboembolism (VTE), decubitus ulcers, and pneumonia) were reported in 9513 cases (10.4%). They occurred more commonly in the lowest volume category (11.0 vs 9.7%, p<0.001). However, this difference was not significant in the multivariable regression analysis [fig_ref] Table 4: Adjusted hip fracture outcomes by hospital case volume* [/fig_ref]. There were 2194 patients with venous thromboembolism (2.4%), 6237 with decubitus ulcers (6.8%), and 1866 with pneumonia (2.0%). In the unadjusted analyses, decubitus ulcers and pneumonia occurred more commonly in low-volume hospitals while VTE occurred in high-volume hospitals (all p<0.001). However, there were no differences in either the primary adjusted (table 4) or sensitivity analyses.
# Discussion
This study found evidence of less efficient hip fracture treatment (delayed operation and prolonged LOS) in low-volume hospitals. However, it did not identify any relationship between volume and clinical outcomes for patients with hip fractures. This is the first study to examine the relationship between hospital case volume and hip fracture outcomes using a contemporary population data set. Importantly, postdischarge complications could be identified if they required admission to any hospital in the state. Previous studies are dated or used cross-sectional databases that could not facilitate longitudinal follow-up of patients between institutions. [bib_ref] Hemiarthroplasty in hip fracture care: effects of surgical volume on short-term outcome, Lavernia [/bib_ref] [bib_ref] Ninety-day mortality after intertrochanteric hip fracture: does provider volume matter?, Forte [/bib_ref] [bib_ref] Hip fracture outcomes: does surgeon or hospital volume really matter?, Browne [/bib_ref] [bib_ref] Does practice make perfect? Examining the relationship between hospital surgical volume and..., Hamilton [/bib_ref] [bib_ref] Hemiarthroplasty for femoral neck fracture in the elderly surgeon and hospital volume-related..., Shah [/bib_ref] [bib_ref] Modeling the volume-effectiveness relationship in the case of hip fracture treatment in..., Sund [/bib_ref] In this study, the mean hip fracture case volume was 79.1 per year; with a relatively high number of lowvolume (<20 per year) hospitals. Although the mean annual case volume in California was higher than previously described across the USA, hip fracture cases may be more concentrated in other settings. For example, in the UK, only six hospitals reported annual case volumes <100 in 2014.Although two previous studies have reported an inverse relationship between hospital volume and hip fracture mortality, 7 10 no such finding emerged from this comprehensive population data set. This also conflicts with reports from other distinct surgical populations. [bib_ref] Provider volume and outcomes for abdominal aortic aneurysm repair, carotid endarterectomy, and..., Killeen [/bib_ref] [bib_ref] Surgical volume and quality of care for esophageal resection: do high-volume hospitals..., Dimick [/bib_ref] [bib_ref] Relation between surgeon volume and risk of complications after total hip arthroplasty:..., Ravi [/bib_ref] [bib_ref] Relationship of surgical volume to short-term mortality, morbidity, and hospital charges in..., Lavernia [/bib_ref] One possibility is that hip fractures are commonly encountered during orthopaedic training [bib_ref] Trauma experience in the UK and Ireland: analysis of orthopaedic training using..., Jameson [/bib_ref] and so surgeons should therefore be familiar with the needs of this patient group, even in low-volume centres. This might explain why hip fractures do not exhibit the volume-outcome relationship that has been identified for more specialised populations, for example, those undergoing revision arthroplasty surgery. Hip fracture treatment is also increasingly driven by protocols and pathways, which might reduce variation between hospitals. There was, however, evidence of higher quality care in high-volume centres. These include reduced delay to the operating theatre and LOS. One possible explanation for this finding is that staff expertise and clinical pathways improve with the experience that results from treating high numbers of similar patients. For example, pathways and processes might have been more developed at higher volume centres. It is important that, although LOS was shorter at high-volume centres, patients were more likely to be discharged to another healthcare facility than their own home. This suggests that relationships with other institutions (such as skilled nursing facilities) may contribute to achieving a shorter LOS. It is also a reminder that discharge from hospital does not necessarily represent the end of each patient journey.
An alternative explanation is proposed by the 'selective referrals' hypothesis, which claims that high-quality hospitals are referred a greater number of patients. [bib_ref] Does practice make perfect? Examining the relationship between hospital surgical volume and..., Hamilton [/bib_ref] This reverses the presumed direction of causation between volume and outcome. In this study, we controlled for some fixed hospital characteristics (eg, trauma centre status) but unknown hospital-level founding factors might have persisted. However, patients transferred between institutions were excluded to minimise the 'selective referrals' effect.
Although prolonged operation time has been associated with hip fracture sequelae (venous thromboembolism, pneumonia, and decubitus ulcers), [bib_ref] Length of stay, mortality, morbidity and delay to surgery in hip fractures, Lefaivre [/bib_ref] these were not over-represented in the lower volume centres.
This study is not without limitations. As the SID is a retrospective data set, unknown confounding factors might have been omitted from our multivariable regression models. In particular, it was not possible to determine the role of individual surgeon case volume. Previous studies have suggested that surgeon volume may be even more important than hospital volume on patient outcomes. In one series of 173 508 elderly patients undergoing hip hemiarthroplasty for fracture, surgeons in the lowest volume category had an 18% increased mortality compared with those in the highest. [bib_ref] Hemiarthroplasty for femoral neck fracture in the elderly surgeon and hospital volume-related..., Shah [/bib_ref] It also is known that low-volume surgeons cluster in low-volume hospitals across a range of surgical procedures. However, we accounted for clustering of fixed hospital-level characteristics in the multivariable regression analysis, which should have controlled for such differences. Although we found no hospital-level effect, it is still possible that low-volume surgeons could have worse mortality outcomes, even in the absence of hospital-level differences. Although the California SID does not include the unique surgeon identifiers that would be necessary to calculate surgeon volume, this may be available in other data sets. For example, other SIDs include unique surgeon identifiers that could be used to explore interactions between surgeon and hospital volume. The California SID was selected for this study because its unique patient identifier variable permitted analysis of readmissions to all hospitals in the state. Further data sets may also be sought that can be linked to public death records so as to track deaths occurring outside of hospital. This is important because our study was unable to identify systematic differences in long-term outcomes (eg, 12-month survival) that might be more important to patients than 30-day readmission.
# Conclusion
In light of these findings, there is no patient safety imperative to discourage low-volume hospitals from treating patients with hip fractures. However, our data did suggest that patients treated in low-volume hospitals are less likely to undergo prompt operation than those in high-volume institutions. Further work should attempt to determine whether volume could be associated with process differences, costs or long-term outcomes for older adults with hip fractures.
[fig] Figure 1: A histogram showing the frequency of hospitals in California by annual hip fracture case volume and selected category thresholds. [/fig]
[table] Table 2: Characteristics of hospitals treating patients in each volume category [/table]
[table] Table 3: Unadjusted hip fracture outcomes by hospital case volume [/table]
[table] Table 4: Adjusted hip fracture outcomes by hospital case volume* [/table]
|
Association between pre-diabetes and microvascular and macrovascular disease in newly diagnosed type 2 diabetes
Objective The associated risk of vascular disease following diagnosis of type 2 diabetes in people previously identified as having pre-diabetes in real-world settings is unknown. We examined the presence of microvascular and macrovascular disease in individuals with newly diagnosed type 2 diabetes by glycemic status within 3 years before diagnosis.Research design and methods We identified 159 736 individuals with newly diagnosed type 2 diabetes from the UK Clinical Practice Research Datalink database in England between 2004 and 2017. We used logistic regression models to compare presence of microvascular (retinopathy and nephropathy) and macrovascular (acute coronary syndrome, cerebrovascular and peripheral arterial disease) disease at the time of type 2 diabetes diagnosis by prior glycemic status. Results Half of the study population (49.9%) had at least one vascular disease, over one-third (37.4%) had microvascular disease, and almost a quarter (23.5%) had a diagnosed macrovascular disease at the time of type 2 diabetes diagnosis. Compared with individuals with glycemic values within the normal range, those detected with pre-diabetes before the diagnosis had 76% and 14% increased odds of retinopathy and nephropathy (retinopathy: adjusted OR (AOR) 1.76, 95% CI 1.69 to 1.85; nephropathy: AOR 1.14, 95% CI 1.10 to 1.19), and 7% higher odds of the diagnosis of acute coronary syndrome (OR 1.07, 95% CI 1.03 to 1.12) in fully adjusted models at time of diabetes diagnosis. Conclusions Microvascular and macrovascular diseases are detected in 37%-24% of people with newly diagnosed type 2 diabetes. Pre-diabetes before diagnosis of type 2 diabetes is associated with increased odds of microvascular disease and acute coronary syndrome. Detection of pre-diabetes might represent an opportunity for reducing the burden of microvascular and macrovascular disease through heightened attention to screening for vascular complications.
# Introduction
The effectiveness of identifying individuals with pre-diabetes to prevent type 2 diabetes has been intensely debated.Clinical trials demonstrating that lifestyle modification and drug-based interventions could prevent or delay progression to type 2 diabetes provide some robust evidence. However, critics argue that only a subset of individuals with pre-diabetes will develop type 2 diabetes, and the population benefits of intervention are outweighed by the potential negative effects due to over-testing, unnecessary medicalization, and uncertainties in benefits of prevention strategies outside the research environment, among other factors. There is also lack of consensus on diagnostic tests and glycemic thresholds for the detection of pre-diabetes.Various diagnostic criteria have been adopted by different organizations that have been repeatedly revised over time.Different diagnostic criteria identify different groups of individuals who differ in
## Significance of this study
What is already known about this subject? ► Pre-diabetes has been shown to contribute to the pathogenesis of vascular dysfunction that might partly explain the increased risk of vascular morbidity and mortality in pre-diabetes and type 2 diabetes.
What are the new findings?
► Half of the study population (49.9%) had at least one vascular disease. ► Compared with individuals with glycemic values within the normal range, those detected with prediabetes before the diagnosis had 76% and 14% increased odds of retinopathy and nephropathy, and 7% higher odds of the diagnosis of acute coronary syndrome at time of diabetes diagnosis.
How might these results change the focus of research or clinical practice?
► Detection of pre-diabetes might represent an opportunity for reducing the burden of microvascular and macrovascular disease through heightened attention to screening for vascular complications.
progression rates to type 2 diabetes and risk of associated morbidity,raising the question of whether those who might benefit the most from tight clinical management are effectively identified as high risk by each criterion.Pre-diabetes has been shown to contribute to the pathogenesis of macrovascular dysfunction that might partly explain the increased risk of cardiovascular disease morbidity and mortality in pre-diabetes and type 2 diabetes. Pre-diabetes is also linked to generalized microvascular dysfunction similar to the vascular damage typical to type 2 diabetes. This suggests that the development of type 2 diabetes-associated microvascular disease may precede the clinical diagnosis of type 2 diabetes.Early stages of retinopathy, neuropathy, and nephropathy, that are generally milder forms compared with that seen in established type 2 diabetes, have been reported in people with pre-diabetes, and prevention studies have demonstrated that their risk can be reduced with lifestyle interventions.This may have important implications for preventive strategies, and whether glycemic testing and detection of pre-diabetes in real-world settings affect the development of microvascular and macrovascular disease among individuals who subsequently develop type 2 diabetes.
Our study aims to examine whether the occurrence of vascular disease differs in individuals newly diagnosed with type 2 diabetes with prior pre-diabetes compared with those with normal glycemic levels in real-world settings. We assessed this using different diagnostic criteria currently applied to pre-diabetes.
# Methods study population
We used data from the UK Clinical Practice Research Datalink (CPRD), one of the largest databases of electronic medical records globally.CPRD holds anonymized routinely collected longitudinal primary care records, covering approximately 7% of the UK population and it is representative in terms of age and sex.We defined a cohort of individuals who were newly diagnosed with type 2 diabetes between January 1, 2004 and September 30, 2017. Participants included in the cohort were registered with one of the 75% of CPRD practices with linked hospital admission and mortality data. Participants also had to have been continuously registered with a practice for at least 1 year before the diagnosis of type 2 diabetes and with availability of historical data in clinical records. Diagnoses of type 2 diabetes were identified using both diagnostic (C10) and management (66A) codes for type 2 diabetes in primary care records based on an established method.Individuals who were diagnosed with type 2 diabetes under the age of 35 years who were prescribed insulin within 3 months of diagnosis and who were not prescribed oral hypoglycemic agents for longer than 3 months were excluded because these individuals were likely to have type 1 diabetes.As our study focuses on detection of pre-diabetes before type 2 diabetes diagnosis, individuals who had glycemic values within the diabetes range recorded more than 3 months before the date of the first diagnosis of type 2 diabetes were excluded from this study, as in this case testing might not be attributable to the diagnostic process but to misclassification. A study diagram summarizing inclusion and exclusion criteria is shown in online supplementary figure S1.
detection of pre-diabetes in primary care settings For the detection of pre-diabetes, we used diagnostic criteria published by WHO and International Expert Committee (IEC): fasting plasma glucose (FPG) 6.1-6.9 mmol/L; oral glucose tolerance test (OGTT) 7.8-11.1 mmol/L; HbA1c 42 to 47 mmol/mol or 6.0%-6.4%. To identify glycemic values within the prediabetes range, we used all available clinical data within 3 years before the date of type 2 diabetes diagnosis considering that among the majority of individuals progressing to type 2 diabetes, a marked increase in glycemic levels is observed within 2 to 3 years before the diagnosis of type 2 diabetes. Individuals with multiple glycemic recordings were classified as having pre-diabetes if at least one measurement met the detection criteria for pre-diabetes. We also used diagnostic codes in primary care records for impaired fasting glucose, impaired glucose tolerance, or other (eg, "Non-diabetic Hyperglycaemia" or "Intermediate Hyperglycaemia") to identify pre-diabetes cases. In case of multiple records over time, the date of the earliest available pre-diabetes detection was used. Individuals were classified as (1) glycemic values within the normal range before type 2 diabetes diagnosis, (2) prediabetes detected before type 2 diabetes diagnosis, or (3) no glycemic measures recorded within 3 years before type 2 diabetes diagnosis. study outcomes Study outcomes included the diagnosis of microvascular (retinopathy and nephropathy) and macrovascular diseases (cerebrovascular disease, acute coronary syndrome, and peripheral arterial disease) at the time of diagnosis of type 2 diabetes. Diagnoses were defined based on the combination of laboratory tests, diagnostic codes in primary care records, and ICD-10 codes on hospital admissions (see online supplementary table S2). Microvascular disease at the time of the diagnosis was defined by the recording of a microvascular disease within 5 years before or 15 months after the diagnosis of type 2 diabetes.The 15-month period was defined based on the time periods of specific process of care indicators of the Quality and Outcomes Framework (QOF) for diabetic retinopathy screening and urine microalbumin testing (12 months and 15 months, respectively).Macrovascular disease at time of diagnosis was defined by the recording of a macrovascular disease any time before or within 1 year of the diagnosis of type 2 diabetes. study covariates Study covariates included age, sex, ethnicity, smoking status, blood pressure (systolic and diastolic), body mass index (BMI), total cholesterol, number of diagnosed comorbid conditions (considering a previously published list, prescription of anti-hypertensive (ACE inhibitor (ACEi) or angiotensin receptor blocker (ARB); others), anti-platelet, lipid-lowering, and antidiabetic medications (biguanides, sulfonylureas, insulin, others) and number of primary care visits during the year before the diagnosis of type 2 diabetes and quintile of the index of multiple deprivation (IMD) at practice level.Information on covariates was collected in the year following the diagnosis of type 2 diabetes. In case of multiple measurements for the same individual, the mean value was calculated for continuous variables and the latest data recorded within the year were used for binary variables. To reduce missing data for study covariates in the year following the diagnosis of type 2 diabetes, we used the latest clinical recording for individuals with missing values within 5 years before the start of the study period.Individuals with missing data on smoking were classified as non-smokers if there was no indication in the past of the individual being a smoker. secondary analyses We undertook two secondary analyses. The first secondary analysis compared results obtained adopting diagnostic criteria for the detection of pre-diabetes published by WHO/IEC with those obtained using those published by the American Diabetes Association (ADA; FPG 5.6-6.9 mmol/L; OGTT 7.8-11.1 mmol/L; HbA1c 39 to 47 mmol/mol or 5.7%-6.4%) and the UK National Institute for Health and Care Excellence 8 (NICE; FPG 5.5-6.9 mmol/L; OGTT 7.8-11.1 mmol/L; HbA1c 42 to 47 mmol/mol (6.0%-6.4%)) (see . An additional secondary analysis was undertaken to explore whether among individuals with pre-diabetes clinical outcomes differed according to whether diagnostic coding for pre-diabetes in electronic health records was assigned following the detection of pre-diabetes. Assigning a diagnostic code for prediabetes might be associated with a more intensive clinical management of cardiovascular risk factors among individuals who were classified as having pre-diabetes. Analysis was undertaken using various diagnostic criteria (WHO/IEC, ADA, and NICE) for pre-diabetes.
# Statistical analysis
Missing data were present for blood pressure (0.5%), BMI (2.8%), total cholesterol (7.0%), and HbA1c (24.1%). We conducted a missing pattern analyzing employing logistic regression analyses and using graphical tools and we concluded that missing data were at random (data not shown).Therefore, we used multiple imputation by chained equations (10 copies) to estimate missing data for these variables. We included the following variables in the imputation model as likely to be associated with recording of risk factors: age, sex, ethnicity, smoking status, number of diagnosed comorbid conditions, number of primary care consultations in the year before the diagnosis of type 2 diabetes, general practice IMD, presence of acute coronary syndrome, cerebrovascular and peripheral arterial disease, and prescription of ACEi or ARB, lipid-lowering, and anti-diabetic medications.
We compared population characteristics according to glycemic status before the diagnosis of type 2 diabetes (pre-diabetes, normal glycemic status, and not recorded) using χ 2 test and ANOVA, as appropriate. Multivariable logistic regression models were employed to assess the odds of having microvascular and macrovascular disease at the time of type 2 diabetes diagnosis in individuals with pre-diabetes and those without glycemic measures recorded compared with individuals with normal glycemic values before the diagnosis of type 2 diabetes. HbA1c was excluded as covariate from the statistical models because the pre-diabetes definition is based on glycemic values. Other variables were tested for multicollinearity calculating the variance-inflation factor and correlation coefficients. Thus, BMI and the use of medications were also excluded due to collinearity with pre-diabetes and other risk factors. Therefore, statistical models were adjusted for age, sex, ethnicity, smoking status, mean systolic and diastolic blood pressure, total cholesterol, and number of diagnosed comorbid conditions, number of primary care visits in the year before the diagnosis of type 2 diabetes, and general practice IMD. Statistical models were further adjusted for the year of diagnosis of type 2 diabetes during the study period . National guidance in England published in 2012 set out a proactive approach to type 2 diabetes prevention through identification and improved clinical management of pre-diabetes.We used receiver operating characteristic (ROC) curves to assess whether the inclusion of a dummy variable defining whether diagnosis of type 2 diabetes occurred before or after 2012 would improve our models. Considering that ROC curve values did not differ from the models only adjusted for year of diagnosis, this latter model was preferred (data not shown). Adjusted ORs and 95% CIs were estimated and results were considered significant if p value <0.05. Statistical analyses were conducted using Stata SE V.15.1.
# Results
In the 3 years before the diagnosis of type 2 diabetes, 65 787 individuals (41.2% of the study population) had at least one glycemic measure recorded and 43 885 individuals (66.7%) reached detection thresholds for pre-diabetes. Moreover, 74.4% of individuals detected with pre-diabetes had recorded at least one FPG measurement, 58.2% at least one HbA1c measurement, and 22.6% at least one OGTT measurement, while 53.0% had recorded a combination of these glycemic measures. During the 3-year period before the diagnosis of type 2 diabetes, the time interval fromCharacteristics of the study population in the year following the diagnosis of type 2 diabetes stratified by whether individuals were tested and reached detection thresholds for pre-diabetes before the diagnosis of type 2 diabetes Results are presented using WHO/International Expert Committee criteria for the definition of pre-diabetes. Clinical data within 3 years before the diagnosis of type 2 diabetes were used to define the detection of pre-diabetes. P values from χ 2 , ANOVA, and Kruskal-Wallis tests, as appropriate, are reported for comparison between the three groups defined by testing and detection of pre-diabetes. WHO/ International Expert Committee criteria to define pre-diabetes: FPG: 6.1-6.9 mmol/L; OGTT 7.8-11.1 mmol/L; HbA1c 42 to 47 mmol/mol or 6.0%-6.4%. *χ 2 test was performed to assess the unadjusted difference between groups. †ANOVA test was performed to assess the unadjusted difference between groups. ‡If an individual was prescribed multiple medications from different anti-diabetic classes, each class was considered (eg, for an individual who was prescribed biguanides and sulfonylureas in the year following the diagnosis of type 2 diabetes, data were recorded as follows: anti-diabetic YES; biguanides YES; sulfonylureas YES; insulin NO; other anti-diabetic NO).
§Kruskal-Wallis test was performed to assess the unadjusted difference between groups. ACEi, ACE inhibitor; ARB, angiotensin II receptor blocker; BMI, body mass index; DBP, diastolic blood pressure; FPG, fasting plasma glucose; OGTT, oral glucose tolerance test (2 hours after 75 g glucose load); SBP, systolic blood pressure; T2D, type 2 diabetes. The prevalence of diagnosed nephropathy present at time of diagnosis of type 2 diabetes was similar between those with normal glycemic values and those without glycemic values recorded (20.7% and 19.0%, respectively), while the prevalence was higher (23.8%) for those with pre-diabetes. After adjusting for confounders, those with pre-diabetes had 14% increased odds of having nephropathy at diagnosis (OR 1.14, 95% CI 1.10 to 1.19), compared with those with prior normal glycemic values.
The prevalence of both microvascular disease being present at time of diagnosis was 3.9%, 6.4%, and 5.2% in individuals with prior normal glycemic values, prediabetes, and those without glycemic measures recorded, respectively. Individuals who reached detection thresholds for pre-diabetes had 53% increased odds of having both diseases at time of diagnosis (OR 1.53, 95% CI 1.41 Prevalence of microvascular (retinopathy and nephropathy) and macrovascular (acute coronary syndrome, cerebrovascular, and peripheral arterial disease) diseases present at time of the diagnosis of type 2 diabetes according to glycemic status in the 3 years before the diagnosis of type 2 diabetes. A microvascular disease was considered being present at time of type 2 diabetes diagnosis if the condition was diagnosed between 5 years before and 15 months after the diagnosis of type 2 diabetes. A macrovascular disease was considered being present at time of type 2 diabetes diagnosis if the condition was diagnosed any time before the diagnosis and during the year following the diagnosis of type 2 diabetes. For the detection of pre-diabetes, the WHO/International Expert Committee diagnostic criteria were adopted (FPG: 6.1-6.9 mmol/L, OGTT 7.8-11.1 mmol/L, HbA1c 42 to 47 mmol/mol or 6.0%-6.4%). FPG, fasting plasma glucose; OGTT, oral glucose tolerance test; T2D, type 2 diabetes. to 1.65), while those without glycemic measures recorded had 35% increased odds (OR 1.35, 95% CI 1.25 to 1.46), as compared with those with normal glycemic values (figure 2).
## Macrovascular disease
At time of type 2 diabetes diagnosis, 23.5% of the study population had at least one diagnosed macrovascular disease. Using the WHO/IEC criteria, 26.9% of individuals with normal glycemic values, 29.8% of those with prior pre-diabetes, and 19.7% of those without glycemic measures recorded had a macrovascular disease. Individuals with glycemic measures within the normal range had the highest unadjusted prevalence of cerebrovascular disease (6.9%), while those who reached detection thresholds for pre-diabetes had the highest prevalence of acute coronary syndrome at time of diagnosis of type 2 diabetes (24.2%) (figure 1). When adjusting for confounders, individuals with prior pre-diabetes had 7% higher odds of previous diagnosis of acute coronary syndrome (OR 1.07, 95% CI 1.03 to 1.12), 12% lower odds of diagnosis of cerebrovascular events (OR 0.88, 95% CI 0.82 to 0.94), and peripheral arterial disease (OR 0.88, 95% CI 0.81 to 0.96), as compared with those with normal glycemic values recorded before the diagnosis of type 2 diabetes. Those without glycemic measures recorded had 27% lower odds of diagnosis of acute coronary syndrome (OR 0.73, 95% CI 0.70 to 0.77), 9% lower odds of diagnosis of cerebrovascular disease (OR 0.91, 95% CI 0.85 to 0.97), 10% lower odds of diagnosis of peripheral arterial disease (OR 0.90, 95% CI 0.84 to 0.98), and 23% lower odds of diagnosis of any macrovascular disease (OR 0.77, 95% CI 0.74 to 0.80;.
## Secondary analyses
Results obtained using NICE and ADA criteria for the detection of pre-diabetes were broadly similar to the findings for the WHO/IEC criteria (see online supplementary figure S6-9). However, when using the NICE and ADA criteria, -individuals without glycemic measures recorded had 12% and 9% increased odds of nephropathy present at time of the diagnosis of type 2 diabetes, respectively, as compared with those with prior normal glycemic values (see online supplementary figure S8).
## Epidemiology/health services research
Furthermore, individuals who reached the detection thresholds for pre-diabetes had 8% increased odds of having any macrovascular disease at diagnosis, as compared with those with normal glycemic values (see online supplementary figure S9).
When further classifying individuals with prior prediabetes into two groups based on having or not having a diagnostic code recorded for pre-diabetes, the OR for having vascular diseases were lower among individuals with a pre-diabetes diagnostic code assigned for most study outcomes. For instance, compared with individuals with normal glycemic values recorded, individuals with a diagnostic code for pre-diabetes had 43% higher odds of any microvascular disease at the time of type 2 diabetes diagnosis, while those without diagnostic codes had a 51% increased odds (diagnostic assigned: OR 1.43, 95% CI 1.37 to 1.49; without diagnostic code: OR 1.51, 95% CI 1.45 to 1.57). Full results are shown in online supplementary figures 10 and 11.
# Discussion
In this large retrospective study of a cohort of individuals newly diagnosed with type 2 diabetes in England, we found that the presence of microvascular and macrovascular disease varied substantially by glycemic status before diagnosis. Compared with individuals with glycemic levels within the normal range within 3 years before type 2 diabetes diagnosis, individuals with prior pre-diabetes and those without glycemic testing were significantly more likely to have microvascular disease at the time of type 2 diabetes diagnosis. Individuals with prior prediabetes were also more likely to have a previous diagnosis of acute coronary syndrome at the time of diagnosis. Conversely, individuals with prior pre-diabetes were less likely to have cerebrovascular and peripheral arterial disease compared with those who had glycemic values within the normal range before type 2 diabetes diagnosis. There were only small variations in these findings across various pre-diabetes diagnostic criteria including WHO/IEC, ADA, and NICE. Individuals who had a diagnostic label for pre-diabetes in their health records had lower odds of microvascular and macrovascular diseases compared with those with pre-diabetes without a diagnostic code.
In this study, we specifically focused on individuals who eventually progressed to type 2 diabetes to examine how prior glycemic testing and status is linked to vascular disease in a population which would benefit the most from preventative interventions. Over one-third of individuals in this study were diagnosed with either retinopathy or nephropathy at the time of type 2 diabetes diagnosis, one-fifth had at least one macrovascular disease, and half of them had at least one microvascular or macrovascular disease. These findings correspond with previous studies reporting a high burden of vascular disease among newly diagnosed individuals with type 2 diabetes, 39-41 and support existing evidence suggesting that in individuals with pre-diabetes vascular disease might occur even before progressing to type 2 diabetes. Pre-diabetes is associated with an excess risk for the development of both macrovascular and microvascular diseases with a continuum of risk across the glycemic range of prediabetes. In the majority of people who progress to type 2 diabetes, an abrupt increase in glycemic measures has been described within 2 to 3 years before diagnosis. In our study, the higher burden of retinopathy among individuals with pre-diabetes compared with individuals with normoglycemia might be explained by prolonged exposure to mild hyperglycaemia.
Accordingly, individuals with glycemic values within the normal range might include a subgroup of individuals with a more rapid progression to type 2 diabetes or could represent people with a similar glycemic trajectory leading to diabetes but with a diagnosis earlier in the natural history of the disease or most likely a combination of these mechanisms. It is also important to note that while more than 80% of this group had recorded measures of FPG in the 3 years before the diagnosis of type 2 diabetes, only less than a quarter (23.7%) had recorded measures of more than one type of glycemic test, which might suggest that this group was less frequently and accurately tested. Considering the predominant use of FPG, more prone to intra-person variability as compared with HbA1c, and the intermittent nature of glycemic values during the pre-diabetes status and in general in the 3 years before progression to type 2 diabetes, a proportion of this group might have been misclassified with glycemic values within the normal range but this reflects detection patterns in real-world settings.
Importantly, 59% of individuals did not have a recorded glycemic measurement in the 3 years before type 2 diabetes diagnosis. Individuals without glycemic measurements had a notably higher mean HbA1c following type 2 diabetes diagnosis compared with those with glycemic testing (with or without pre-diabetes), potentially indicating late diagnosis of type 2 diabetes and leading to delayed treatment. The clinical characteristics of these people are compatible with at least two explanations: First, these people may be less health conscious and have a worse adherence with preventive procedures reflected by their higher prevalence of smoking and a lower number of primary care visits. Second, individuals without previous glycemic testing before their diagnosis of type 2 diabetes might have had lower vascular risk, as reflected by lower unadjusted prevalence of macrovascular disease at time of diabetes diagnosis, and lower number of other comorbid conditions. This may have resulted in fewer contacts with primary care and missed opportunities for screening. The considerably higher mean HbA1c at the time of diagnosis in this group may reflect a diagnosis at a later stage compared with the other groups and explain the higher prevalence of retinopathy. This is also in line with previous findings showing that glycemic testing and clinical management might potentially delay the diagnosis of diabetes and potentially its complications.
## Epidemiology/health services research
The differences for nephropathy were less pronounced between the glycemic groups compared with that for retinopathy. These findings suggest that the association between pre-existing pre-diabetes and nephropathy might not be as strong as that for retinopathy. This is in line with the findings of a recent meta-analysis that concluded that the association between pre-diabetes and nephropathy was significant but modest, and this might be partially explained by underlying confounding or common causes contributing to both hyperglycemia and kidney disease. A small proportion of individuals (5.4% considering the whole sample) had both retinopathy and nephropathy in this study, potentially indicating prolonged exposure to chronic hyperglycemia or nondiabetic glomerular disease in some individuals that, at least partly, may explain the different patterns of renal involvement.We only found small variations in our findings across pre-diabetes subgroups defined by WHO/IEC, NICE, and ADA, suggesting that when focusing on the proportion of the pre-diabetes population who eventually progresses to type 2 diabetes, differences are less pronounced then what has been found in studies focusing on the whole pre-diabetes population, including those who will never progress to type 2 diabetes.We found that individuals with pre-diabetes detected before the diagnosis of type 2 diabetes were more likely to have a previous diagnosis of acute coronary syndrome but were less likely to have cerebrovascular and peripheral arterial disease at time of type 2 diabetes diagnosis. These findings correspond with previous studies showing that chronic hyperglycemia contributes to the pathogenesis of macrovascular dysfunction. Hyperglycemia has been shown to be strongly associated with an increased risk of acute coronary syndrome. Individuals with acute coronary syndrome have increased prevalence of pre-diabetes and the risk of mortality for those hospitalized with acute coronary syndrome with hyperglycemia is also higher. However, the association between prediabetes and cerebrovascular disease is less clear. For instance a meta-analysis found that the associated risk of cerebrovascular disease among those having pre-diabetes is modest.These findings are also compatible with a potential surveillance bias: people with any cardiovascular disease are more likely to be screened for type 2 diabetes and intermediate hyperglycemia compared with the general population, potentially resulting in an earlier diagnosis of type 2 diabetes. This bias might have had a greater influence on findings where the link between pre-diabetes and the outcome is weaker and may have changed the direction of the association (ie, cerebrovascular disease as compared with acute coronary syndrome).
# Strengths and limitations
To our knowledge, this is the first large populationbased study to examine associations between glycemic status before type 2 diabetes diagnosis and the presence of microvascular and macrovascular disease in individuals newly diagnosed with type 2 diabetes. Our findings provide further evidence that pre-diabetes has significant clinical implications for microvascular and macrovascular diseases and type 2 diabetes outcomes. We used routinely collected primary and secondary care data representative of the English population to better understand these associations in real-world settings. While the implementation of a national retinopathy screening program in the UK and the increased surveillance within the QOF ensured good quality data for the diagnosis of retinopathy and nephropathy, we did not include diabetic neuropathy in our analyses because the diagnosis and coding of this condition appears suboptimal in primary care settings in England.We could not assess the duration individuals remained in the pre-diabetes state before progressing to type 2 diabetes, and the focus of the study was to identify the recording of pre-diabetes before type 2 diabetes diagnosis. Additional study limitations include the presence of missing data for clinical variables such as blood pressure, BMI, total cholesterol, and HbA1c. However, we overcame the latter issue by using multiple imputation by chained equations. It was not possible to assess differences in adherence to lifestyle interventions, as we did not have data on diet and physical activity. Finally, when using routinely collected data, concerns have been raised about miscoding, misclassification, and misdiagnosis. However, CPRD is subject to regular quality checks and is widely used for health research.Implications for clinical practice Microvascular and macrovascular diseases were present in 37%-24% of people with newly diagnosed type 2 diabetes, with over half not having any glycemic measurement within 3 years before their diagnosis. While there are many unanswered questions regarding its detection, pre-diabetes has significant clinical implications for microvascular and macrovascular diseases and type 2 diabetes outcomes. A major consideration is whether targeted preventive strategies that identify individuals with pre-diabetes for interventions would provide opportunities for vascular risk reduction, considering that major benefits are likely to occur from early diagnosis and treatment.While discussions on the pathophysiological differences between pre-diabetes subtypes continue, there have been calls to move away from a glucocentric definition toward a multifactorial detection strategy for pre-diabetes that reflects the presence of other risk factors for type 2 diabetes as well as early manifestation of vascular disease. COnClusIOn Our large observational study using real-world data has shown that both microvascular and macrovascular diseases were frequently detected at the time of type 2 diabetes diagnosis. Microvascular disease was more frequent among individuals with newly diagnosed Epidemiology/Health Services Research type 2 diabetes who were previously detected with prediabetes. The identification of pre-diabetes and specific clustering of pre-diabetes with other risk factors for type 2 diabetes might prompt earlier assessment for risk factors and tailored cardiovascular risk reduction strategies during the pre-diabetes phase to reduce the burden of vascular disease, but further research is needed to confirm this. |
mRNA expression of somatostatin receptor subtypes SSTR-2, SSTR-3, and SSTR-5 and its significance in pancreatic cancer
Background: The aim of this study is to investigate the expressions of somatostatin receptor (SSTR), SSTR-2, SSTR-3, and SSTR-5, in pancreatic tissue and non-cancerous tissue and elucidate their clinical significance. Methods: The expression of somatostatin receptor subtypes SSTR-2, SSTR-3, and SSTR-5 messenger RNA (mRNA) in 108 cases of cancer tissue and adjacent tissue in patients with pancreatic cancer was detected by reverse transcriptase polymerase chain reaction (RT-PCR). Expression of SSTR-2, SSTR-3, and SSTR-5 mRNA was evaluated after specimens were taken from selected patients who underwent surgical resection by Whipple's operation. We speculated the clinical significance of the expression of somatostatin receptor (SSTR) subtype genes SSTR-2, SSTR-3, and SSTR-5 in pancreatic tissue and non-cancerous tissue and further elucidated their clinical significance. Results: The expression rates of SSTR-2 mRNA in cancer and adjacent tissue of 108 patients with pancreatic cancer were 81.5% (88/108) and 97.2% (105/108), respectively; SSTR-3 mRNA expression rates were 69.4% (75/108) and 55.6% (60/108). SSTR-5 mRNA expression rates were 13.0% (14/108) and 18.5% (20/108). Conclusion: We propose that SSTR-2 plays an important role in clinical implications for patients with pancreatic cancer undergoing somatostatin or its analog therapy.
# Background
Somatostatin is a type of peptide hormone which has a wide range of biological functions. Recent studies have indicated that somatostatin and its analogs have obvious anti-proliferative effects on various solid tumors, such as gastroenteropancreatic neuroendocrine tumors and breast cancer, which are mediated through somatostatin receptors [bib_ref] Somatostatin analogs therapy in gastroenteropancreatic neuroendocrine tumors: current aspects and new perspectives, Baldelli [/bib_ref] [bib_ref] Combined effects of melatonin and all-trans retinoic acid and somatostatin on breast..., Margheri [/bib_ref] , but some clinical research have indicated that the impact of somatostatin and its analog receptor expression levels on outcomes in patients with pancreatic neuroendocrine tumors (PNETs) has not been evaluated [bib_ref] Clinicopathologic characteristics of pancreatic neuroendocrine tumors and relation of somatostatin receptor type..., Okuwaki [/bib_ref] [bib_ref] Prognostic value of somatostatin receptor-2 positivitiy in gastroenteropancreatic neuroendocrine tumors in reference..., Yenıay [/bib_ref]. Thus, whether or not there is expression of somatostatin receptor (somatostatin receptor, SSTR) in tumor tissue is an important underlying factor affecting the efficacy of somatostatin therapy. In this study, we detected the expression of somatostatin receptor subtypes SSTR-2, SSTR-3, and SSTR-5 messenger RNA (mRNA) from 108 cases of cancer tissue and adjacent tissue in patients with pancreatic cancer using RT-PCR method. Thereafter, we explored the clinical implications of somatostatin therapy for patients with pancreatic cancer. Our study is aimed at exploring the cause from different therapeutic effects of somatostatin and its analogs on human pancreatic cancer and their relationship between changes of SSTR-2 gene expression and expression of SSTR-2, SSTR-3, and SSTR-5 genes [bib_ref] Expression of somatostatin receptor (SSTR) subtypes (SSTR-1, 2A, 3, 4 and 5)..., Mizutani [/bib_ref].
# Methods
## Patient population and data collection
Our study was a retrospective analysis of patients who had been diagnosed pathologically as having pancreatic ductal adenocarcinoma by the Department of Pathology in Qilu hospital, between January 2010 and June 2013. The demographic, clinical, laboratory, and radiological data for all patients admitted or transferred to our hospital with a diagnosis of pancreatic ductal adenocarcinoma were reviewed retrospectively for this study. Included in this study, patients with pancreatic cancer must have received surgical resection as the initial treatment modality without major perioperative complications, have adequate archived tissue kept, and have complete clinicopathologic data obtained. This resulted in a collection of tissue from 108 patients, 72 males and 36 females with a median age of 55.7 years and an age range of 33 to 71 years. Among those, 85 had cases of carcinoma of the head of the pancreas and 23 cases of carcinoma of the body and tail of the pancreas. For UICC TNM clinical stages: stages I to II had 33 cases, and stages III to IV had 75 cases. We also collected the non-cancerous pancreatic tissue from 10 patients (all by excision of the normal pancreas were taken from surgical specimens) in order to compare the mRNA expression rate with the cancerous pancreatic tissues and adjacent normal tissue. The pathologic tumornode-metastasis (TNM) classification was based on the criteria of the International Union Against Cancer (2009). The study protocol was approved by the Ethics Committee of Qilu Hospital affiliated to Shandong University. Informed consent was obtained from each patient prior to The expression of SSTR-2 was significantly relevant with the survival rate. The survival rate of patients with positive expression was higher than those with negative expression (P < 0.05 * ). The expressions of SSTR-2 mRNA and SSTR-3 mRNA were relevant with gender. Female positive rate was significantly higher than male (P < 0.05 ** ). These three kinds of mRNA expression had no obvious relationships with age, TNM stage, and differentiation (P > 0.05).
study enrollment at the time of hospital admission, and the detailed case characteristics are summarized in [fig_ref] Table 1: The basic characteristics of patients in this study [/fig_ref].
## Reagents
Total RNA extraction reagents were bought from Promega (Madison, WI, USA). Specific primers, RT-PCR reagents and restriction enzymes, were provided by TaKaRa (Bio Company, Kyoto, Japan).
## Specimen collection
Non-necrotic cancer tissue and adjacent normal tissue (3 cm from cancer tissue) were taken from surgical specimens (non-cancerous pancreatic tissues were taken from the normal tissue). All were frozen in liquid nitrogen immediately after removal and placed in −80°C.
## Rt-pcr
Complementary DNA (cDNA) was synthesized by reverse transcription of 20 μl total RNA in a reaction system. Following reaction system was used: MgCl 2 5 mmol/L, dNTPs 5 mmol/L, RNAsin 20U, AMV reverse transcriptase XL2.5U, oligo d (T) adaptor primers 2.5 μmol/L. The reaction was carried out at 65°C for 10 min; after cooling at 4°C for 5 min, AMV reverse transcriptase Xl was added and then incubated at 42°C for 60 min, 99°C heating for 5 min to inactivate the reverse transcriptase. PCR was performed using the following schedule: 30 cycles of denaturation at 94°C for 30 s; annealing of SSTR-2 and SSTR-5 at 57°C for 30 s and SSTR-3 at 63°C for 60 s; positive control at 60°C for 30 s; extension at 72°C for 1 min; frozen at 4°C for 5 min. Special primer pairs were listed as supplementary.
# Statistical method
All statistical data was evaluated by χ2 test. The P < 0.05 was considered statistically significant while three samples were compared. While comparing two samples, we used partitions of χ2 method to significant level α'. α' = α/((k − 1) × k/2 + 1) (k, numbers of combination of any two). We could calculate the outcome of α' is 0.0125. Thus, P < 0.0125 was considered statistically significant. Survival analyses were conducted by the life table method and the log-rank test. P values <0.05 were considered to be statistically significant. Survival curves were generated through the life table method.
# Results
In 108 cases of pancreatic cancer patients, the expression rate of SSTR-2 receptor mRNA in adjacent cancer tissues was 97.2% (105/108) and in cancer tissue was 81.5% (88/108). The rate of positive SSTR-2 mRNA expression among carcinous tissue, adjacent tissue of cancer, and non-carcinous pancreatic tissues were significantly different via analysis (P < 0.01). When compared between any two groups, the difference between carcinous tissue and adjacent tissue was also significant. Positive expression in carcinous tissue was also higher than adjacent tissue (P < 0.0125). Analysis from other groups did not show obvious differences (P > 0.0125). The expression rate of SSTR-3 receptor mRNA in adjacent cancer tissue was 55.6% (60/108). In cancer tissue, it was 69.4% (75/108). The expression rate of SSTR-5 receptor mRNA in adjacent cancer tissues was 18.5% (20/108). In cancer tissues, it was 13% (14/108) [fig_ref] Table 4: The χ 2 test outcomes of SSTR-2, SSTR-3, and SSTR-5 mRNA expression [/fig_ref]. SSTR-2, SSTR-3, and SSTR-5 mRNA expression rate of ten cases of non-cancerous pancreatic tissue were 100.0% (10/10), 70.0% (7/10), and 10.0% (1/10). No significant differences were showed among three groups or the combination of any two groups via χ2 tests. Thus, we could draw a conclusion that the SSTR-3 mRNA expression among carcinous tissue, adjacent tissue of cancer, and non-carcinous tissue were similar [fig_ref] Table 4: The χ 2 test outcomes of SSTR-2, SSTR-3, and SSTR-5 mRNA expression [/fig_ref].
We concluded that the expression of SSTR-2 was significantly associated with survival rates [fig_ref] Table 1: The basic characteristics of patients in this study [/fig_ref]. The survival rate of patients with a positive expression was higher than those with a negative expression (P < 0.05*). The expression of SSTR-2 mRNA and SSTR-3 mRNA were relevant with gender. The female positive rate was significantly higher than males (P < 0.05**). These three kinds of mRNA's expressions had no obvious relationships with age, TNM stage, and cancer differentiation (P > 0.05).
Survival curves are shown in [fig_ref] Figure 1: Overall survival according to SSTR-2,SSTR-3,SSTR-5 mRNA expression in pancreatic cancer patients [/fig_ref] -c. Patients with positive SSTR-2 mRNA expression in carcinous tissue had a significantly better overall survival rate than those with negative expression (P < 0.05, the log-rank test, χ2 = 4.40). The difference of a patients' survival rate between positive and negative SSTR-3 expression in carcinous tissue had no statistical significance (P > 0.05, the log-rank test, χ2 = 0.003), and the difference of a patients' survival rate between positive and negative SSTR-5 expression in carcinous tissue also had no statistical significance (P > 0.05, the log-rank test, χ2 = 0.391).
# Discussion
Somatostatin can inhibit growth of pancreatic cancer cells even in CPT-insensitive human pancreatic carcinoid BON cells [bib_ref] Somatostatin receptor-targeted anti-cancer therapy, Sun [/bib_ref]. The inhibition mechanism involves the combined interaction of somatostatin and its analogs with SSTR-2 in tumor tissues, mainly through binding to tumor tissue receptors [bib_ref] Vascular targeting to the SST2 receptor improves the therapeutic response to near-IR..., Starkey [/bib_ref] , which directly inhibits tumor cell proliferation. This occurs through the process of either directly inhibiting division and proliferation of tumor cells or inhibiting the activity of various growth factors, such as vascular endothelial growth factor (VEGF), insulin-like growth factor (IGF) and marked stimulation of the reticuloendothelial system [bib_ref] Protective role of somatostatin receptor 2 against retinal degeneration in response to..., Dal Monte [/bib_ref] [bib_ref] Effects of a somatostatin analogue (SMS201-995) on the growth and development of..., Nott [/bib_ref]. Moreover, it counteracts tumorigenesis and tissue proliferation. In addition, it was found that somatostatin and its analogs could arrest the pancreatic cancer cell cycle at S phase and thus induce apoptosis [bib_ref] Inhibitory effect of somatostatin peptide analogues on DNA polymerase activity and human..., Kuriyama [/bib_ref]. In cloned somatostatin receptor subtypes, synthetic long-acting somatostatin analogs 6 or 8, such as octreotide and vapreotide have high affinity to SSTR-2 and SSTR-5 receptor subtypes, low affinity to SSTR-3, and no affinity to other types of receptor subtypes [bib_ref] Somatostatin analogues inhibit cancer cell proliferation in an SSTR2-dependent manner via both..., Zou [/bib_ref]. Therefore, before implementing therapy by long-acting somatostatin analogs such as octreotide in pancreatic cancer patients, to make sure that tumor tissue have high affinity somatostatin receptors, it is important to improve efficacy and reduce unnecessary costs. There is genre and organ specificity in the expression of somatostatin receptors; for example, the normal pancreas mainly expresses SSTR-2 subtype receptors, while only SSTR-3 genes are expressed in rat pancreatic tissue; in the outer periphery of the body organization, only SSTR-5 subtype receptors are found in the hypothalamus and pituitary tissue [bib_ref] The somatostatin receptor family, Patel [/bib_ref]. Therefore, in relation to drug receptor binding affinity and organspecific aspects, we studied SSTR-2, SSTR-3, and SSTR-5 expression in the pancreatic cancer.
There are few reports about research on somatostatin receptor gene expression in pancreatic tumor tissue. Somatostatin receptor expression in pancreatic tissue is detected by using isotopically labeled somatostatin [bib_ref] Expression of the somatostatin receptor subtype-2 gene predicts response of human pancreatic..., Fisher [/bib_ref]. However, because natural or synthetic somatostatin can bind to various somatostatin receptors, we cannot determine which kind of somatostatin subtype receptor binds to isotopically labeled somatostatin. RT-PCR assay has good specificity and high sensitivity for detection of SSTR-2 gene expression in tumor specimens.
From the results of this study, there were 88 cases with positive expression of SSTR-2 mRNA out of 108 pancreatic cancer cases; thus, the rate was 81.5%. This shows that the majority of pancreatic tumor tissues have SSTR-2 receptor subtype expression. Because the current study shows that the ability of somatostatin to inhibit tumor cell growth is mediated by SSTR-2 [bib_ref] Somatostatin receptor SST2 reduces Akt activity and aggravates hypoxic/ischemic death in cerebral..., Stumm [/bib_ref] [bib_ref] Analysis of somatostatin receptors and somatostatin promoter methylation in human gastric cancer, Shi [/bib_ref] , it is very important for patients suffering from pancreatic cancer to receive somatostatin treatment. In experimental pancreatic tumor, the growth of both functioning pancreatic tumors and nonfunctioning tumors with SSTR-2 gene expression is inhibited by somatostatin, while there is no inhibitory effect on non-SSTR-2 gene expression cells. Carcinoid patients had a similar situation [bib_ref] Profiling of somatostatin receptor subtype expression by quantitative PCR and correlation with..., Nakayama [/bib_ref]. Researchers believe that the expression of SSTR-2 in tumor tissue is the premise for patients to accept somatostatin therapy. In addition, SSTR-3 mRNA expression was higher in pancreatic tumor tissue than the adjacent tissue of cancer, and the rate of positive expression decreases with increasing degree of differentiation trend. Significance of pancreatic tissue SSTR-3 mRNA expression is still unclear. Compared with SSTR-2, SSTR-3 has low affinity when compared to six or eight peptide long-acting somatostatin analogs, so further study is needed to determine its role in the treatment of tumors.
There is less difference in the rate of SSTR-5 mRNA expression in pancreatic tumor tissue and tumor adjacent tissue, both were low, which is consistent with previous findings [bib_ref] Characterization of somatostatin receptor expression in human pancreatic cancer using real-time RT-PCR, Li [/bib_ref]. Synthetic six or eight long-acting somatostatin analogs like octreotide and vapreotide have high affinity to the SSTR-5 receptor subtype [bib_ref] Inhibitory effect of somatostatin peptide analogues on DNA polymerase activity and human..., Kuriyama [/bib_ref] ; its specific mechanism in pancreatic cancer remains to be further explored.
# Conclusions
In summary, our findings indicated that the majority of pancreatic cancers have more than one somatostatin receptor subtype gene. According to recent studies, we obtained that somatostatin and its analogs have obvious anti-proliferative effects on various solid tumors which are mediated through somatostatin receptors. We propose that the SSTR-2 could play an important role in clinical implications for patients with pancreatic cancer undergoing somatostatin or its analog therapy.
[fig] Figure 1: Overall survival according to SSTR-2,SSTR-3,SSTR-5 mRNA expression in pancreatic cancer patients. (a) Overall survival according to SSTR-2 mRNA expression in pancreatic cancer patients. (b) Overall survival according to SSTR-3 mRNA expression in pancreatic cancer patients. (c) Overall survival according to SSTR-5 mRNA expression in pancreatic cancer patients. [/fig]
[table] Table 1: The basic characteristics of patients in this study [/table]
[table] Table 4: The χ 2 test outcomes of SSTR-2, SSTR-3, and SSTR-5 mRNA expression [/table]
|
Alterations of RET, Oncogene in Human Adrenal Tumors
Previous studies have revealed specific activations of the RET oncogene in multiple endocrine neoplasia type 2 (MEN 2) and thyroid tumors. To understand the role of the RET proto-oncogene activation in sporadic adrenal tumors, we analyzed the alterations of the RET proto-oncogene in the cysteine-rich extracellular domain (exons 6 and 10), the terminal region of the extracellular domain and transmembrane domain (exon 11) and the tyrosine kinase domain (exons 12-17) in 35 cases of adrenal tumors (including 18 Conn's syndrome, 3 Cushing's syndrome, 2 non-functional adrenocortical tumor and 12 pheochromocytomas by polymerase chain reaction-single strand conformational polymorphism and sequencing methods. One case with pheochromocytoma and one with Conn's syndrome had point mutation. We also detected the rearrangement of the RET gene by reverse transcription-polymerase chain reaction and Southern hybridization. One case with Conn's syndrome and one with Cushing's syndrome were found to harbor RET/PTC1 (RET tyrosine kinase domain rearranged with H4 gene). The above results indicate that RET protooncogene mutations and RET/PTC1 are involved in the pathogenesis of sporadic adrenal tumors. Mutations at codon 634 of the RET gene were also found in adrenal tumors. This suggests that the RET oncogene may also play a role in the tumorigenesis of adrenal tumors, and this possibility requires further investigation.
The RET gene was first identified as a proto-oncogene a decade ago on the basis of its ability to transform NIH 3T3 mouse cells in culture. [bib_ref] Activation of a novel human transforming gene, RET, by DNA rearrangement, Takahashi [/bib_ref] The transforming sequences that were initially recovered actually represented the product of a rearrangement between RET and another gene that had occurred during the transfection assay. Sequence analysis of the RET proto-oncogene proved that it is a member of the tyrosine kinase receptor gene family. [bib_ref] RET transforming gene encodes a fusion protein homologous to tyrosine kinase, Takahashi [/bib_ref] [bib_ref] GDNF-induced activation of the RET protein tyrosine kinase is mediated by GDNFR-α,..., Jing [/bib_ref] The RET proto-oncogene has a calcium-binding cysteinerich extracellular domain along with a cadherin-like ligand binding site, a tyrosine kinase-containing intracellular domain, and a short transmembrane domain. [bib_ref] Structure analysis of the human RET proto-oncogene using exon trapping, Kwok [/bib_ref] [bib_ref] Localization of the gene for multiple endocrine neoplasia type 2A to a..., Mole [/bib_ref] Previous studies have shown that rearrangement of the RET gene in papillary thyroid carcinomas results in a protein with altered or novel tyrosine kinase function. [bib_ref] PTC is a novel rearranged form of the RET proto-oncogene and is..., Grieco [/bib_ref] [bib_ref] cDNA cloning and characterization of RET activated in human papillary thyroid carcinoma..., Ishizaka [/bib_ref] The portion of the RET gene encoding the tyrosine kinase domain is juxtaposed to sequences from one of three other genes through chromosomal rearrangements. [bib_ref] Identification of the product of two oncogenic rearranged forms of the RET..., Lanzi [/bib_ref] [bib_ref] Molecular characterization of a thyroid tumor-specific transforming sequence formed by the fusion..., Bongarzone [/bib_ref] Two of the gene fusions result from rearrangement involving sequences on the same chromosome as RET (the PTC1 and PTC3 chimeras), and the third (PTC2) results from an interchromosomal rearrangement. [bib_ref] At (10, 17) translocation creates the RET/PTC2 chimeric transforming sequence in papillary..., Sozzi [/bib_ref] [bib_ref] Development of a single-step duplex RT-PCR detecting different forms of RET activation,..., Jhiang [/bib_ref] [bib_ref] Frequent activation of RET proto-oncogene by fusion with a new activating gene..., Bongarzone [/bib_ref] The frequency of RET rearrangements differs in different geographical areas, [bib_ref] High frequency of activation of tyrosine kinase oncogenes in human papillary thyroid..., Bongarzone [/bib_ref] [bib_ref] RET oncogene activation in human thyroid neoplasms is restricted to the papillary..., Santoro [/bib_ref] [bib_ref] Detection of RET/PTC transcripts in thyroid adenomas and adenomarous goiter by an..., Ishizaka [/bib_ref] [bib_ref] Low frequency of rearrangement of the ret and trk proto-oncogenes in Japanese..., Wajjwalku [/bib_ref] [bib_ref] Detection of the PTC RET oncogene in human thyroid cancers, Jhiang [/bib_ref] [bib_ref] Low rate of RET protooncogene activation (PTC/retTPC) in papillary thyroid carcinomas from..., Zou [/bib_ref] but whether the differences can be ascribed to environmental factors, or merely to sampling errors or differences in the techniques used, remains to be elucidated. In addi-tion, point mutation of the RET proto-oncogene has so far only been found in multiple endocrine neoplasia type 2A (MEN 2A), MEN 2B, familial medullary thyroid carcinoma (FMTC), sporadic MTC, sporadic pheochromocytomas and Hirschsprung's disease. The point mutation sites are located in codons specifying cysteine residues, tyrosine kinase domains and intracellular domains. [bib_ref] RET proto-oncogene mutations and rearrangements in endocrine diseases, Ricardo [/bib_ref] According to the above studies, the RET proto-oncogene is specifically activated in endocrine tumors. Adrenal tumors are also a kind of endocrine tumor, but no study has yet indicated that RET oncogene activation occurs in adrenal tumors, except for pheochromocytoma. In general, functional adrenal tumors are small and characterized as benign, as well as being active hormone-producing tumors. Malignant functional tumors are relatively rare. [bib_ref] The adrenal cortex, David [/bib_ref] [bib_ref] Catecholamines and the adrenal medulla, Landsberg [/bib_ref] Presentation of typical feature of uncontrollable overproduction of hormones makes for easy clinical diagnosis of functional adrenal tumor with the use of modern biochemical analysis or imaging technology. Because functional tumors can be diagnosed very easily and resected when the tumor is small in size, they can be used for studies. In contrast, non-functional adrenal tumors are hard to obtain. We were able to collect 2 cases of nonfunctional adrenal tumor as controls for our study. In order to clarify the role of the RET gene in the tumorigenesis of human adrenal tumors, we performed molecular studies in 35 adrenal tumor tissues. Our results suggest that the RET gene may be related to the development of these tumors.
# Materials and methods
Patients and tissues Thirty-five adrenal tumors and their paired remnant normal adrenal tissues were obtained from patients who underwent surgery for adrenal lesions [fig_ref] Table I: Alterations of the RET Oncogene in Human Adrenal Tumors [/fig_ref]. These included 18 patients with primary aldosteronism, 3 patients with adrenal Cushing's syndrome, 12 patients with pheochromocytoma, and 2 patients with nonfunctional adrenal tumor. The diagnosis of primary aldosteronism was made on the basis of the clinical features, including hypertension, hypokalemia, suppression of the serum renin and the biochemical examination of increased aldosterone in blood. Further analysis by computer tomography proved the existence of unilateral adenoma. With regard to Cushing's syndrome, the diagnosis was made on the basis of the clinical features including obesity, hypertension, hirsutism, and facial plathora. Biochemical examination showed increased cortical secretion in blood. Computer tomographic scanning and pathologic findings confirmed the existence of adrenocortical tumor. The diagnosis of pheochromocytoma depended on the analysis of computer tomographic scans, pathological findings, and vanillylmandelic acid. All tissue samples were frozen at −80°C until analyzed. DNA extraction Genomic DNA was extracted from tissues by proteinase-K digestion and phenol-chloroform extraction according to Bline and Stanffard. [bib_ref] A general method for isolation of high-molecular-weight DNA from eukaryotes, Blin [/bib_ref] Polymerase chain reaction-single strand conformational polymorphism (PCR-SSCP) analysis To search for mutations of the RET gene using PCR-SSCP analysis, 9 different sets of primers were prepared to amplify regions including all of the coding exons (exons 6, 10-17), as summarized in [fig_ref] Table I: Alterations of the RET Oncogene in Human Adrenal Tumors [/fig_ref]. Oligonucleotides used as primers for PCR were synthesized based on the published RET gene sequence. [bib_ref] GDNF-induced activation of the RET protein tyrosine kinase is mediated by GDNFR-α,..., Jing [/bib_ref] The reaction mixture contained 50 pmol of each primer, 2.5 units of Taq DNA polymerase (Boehringer Mannheim GmbH, Mannheim, Germany), 100 mM of each deoxy-nucleotide 5′-triphosphate (NTP), [α-32 P]deoxy-cytidine 5′-triphosphate (3000 Ci/mmol, 10 mCi/ml, New England Nuclear Research Products, Boston, MA), 1.5 mM MgCl 2 , 50 mM KCl, 10 mM Tris-HCl (pH 8.3), and gelatin at 10 µg/ml. A programmable thermal cycler (PTC-100, MJ Research, Watertown, MA) was used to perform 35 cycles of denaturation for 15 s each at 94°C and annealing for 15 s at 55°C for exons 6, 10, 12, 14-17, or at 60°C for exons 11 and 13, with an extension for an additional 30 s at 72°C. The total final extension time was 5 min at 72°C. The PCR-SSCP reactions were electrophoresed on 6% neutral polyacrylamide gels. Cloning and sequencing Sequence analysis of PCRamplified RET exons 6 and 11 was performed after subcloning the amplified fragment in pCR-Script cloning vector (Stratagene, La Jolla, CA). Dideoxynucleotide sequencing was done with a Sequenase kit (U.S. Bio-chemical, Cleveland, OH), using [α-35 S]dATP (10 µCi/µl, New England Nuclear Research Products). Aliquots of the sequencing reaction mixtures were electrophoresed on 8% denaturing polyacrylamide gels. For accuracy, we collected 10 independent clones and performed bidirectional sequencing with the T3 and T7 sequencing primers. RT (reverse transcription)-PCR The total RNA was isolated using the acid-guanidine isothiocyanate-phenol-chloroform method [bib_ref] Single-step method of RNA isolation by acid guanidinium thiocyanate-phenolchloroform extraction, Chomczynski [/bib_ref] and used as a template for cDNA
[formula] 1 F Conn N - N 2 M Conn N - ND 3 M Conn N - ND 4 M Conn N - N 5 F Conn N - ND 6 M Conn 634/TGC→TGG Cys→Trp N 7 F Conn N - N 8 F Conn N - N 9 M Conn N - ND 10 F Conn N - N 11 F Conn N - N 12 F Conn N - N 13 F Conn N - N 14 M Conn N - N 15 M Conn N - N 16 F Conn N - RET/PTC1 17 M Conn N - ND 18 F Conn N - ND 19 F Cushing N - RET/PTC1 20 F Cushing N - N 21 F Cushing N - ND 22 M NFA N - ND 23 F NFA N - N 24 F pheo N - ND 25 F pheo N - ND 26 M pheo N - ND 27 F pheo N - ND 28 F pheo N - N 29 F pheo 393/TTC→TCC 634/TGC→TGG Phe→Ser Cys→Trp N 30 F pheo N - ND 31 F pheo N - N 32 M pheo N - N 33 M pheo N - ND 34 M pheo N - ND 35 M pheo N - N N, [/formula]
# Results
PCR-SSCP One of 12 (8.3%) pheochromocytomas and 1 of 18 (5.6%) patients with primary aldosteronism showed an apparent electrophoretic mobility shift between the tumor and its paired adjacent normal tissue [fig_ref] Figure 1: PCR-SSCP analysis of RET oncogene mutations in human adrenal tumors [/fig_ref] , implying the existence of a mutation. Such differences were detected in exon 6 (1 case) and exon 11 (2 cases). A mobility shift of the amplified exon 11 could be clearly detected in cases 6 and 29 [fig_ref] Figure 1: PCR-SSCP analysis of RET oncogene mutations in human adrenal tumors [/fig_ref]. In addition, a mobility shift of the amplified exon 6 was detected in case 29 [fig_ref] Figure 1: PCR-SSCP analysis of RET oncogene mutations in human adrenal tumors [/fig_ref]. All cases with mutation still showed a normal allelic band. In total, 2 of 35 (5.7%) cases of adrenal tumors had conformational alteration. Sequence analysis To examine the type of mutation, the exon 6 and 11 region was cloned from tumor specimens of case 6 and case 29. The sequencing data revealed a substitution from phenylalanine (TTC) to serine (TCC) at codon 393 (exon 6) in case 29 and a substitution from cysteine (TGC) to tryptophan (TGG) at codon 634 (exon 11) in cases 6 and 29 [fig_ref] Figure 2: Sequencing analysis of the RET oncogene [/fig_ref]. The results of bidirectional sequencing of 10 independent clones confirmed that these sites were mutated in the adrenal tumor specimens we collected. RT-PCR To investigate whether RET gene rearrangement forms, including RET/PTC1, RET/PTC2 and RET/ PTC3, were also present in adrenal tumors, we performed RT-PCR on 20 adrenal tumor samples. These samples were collected from 12 patients with Conn's syndrome, 2 patients with Cushing's syndrome, 5 patients with pheochromocytoma and 1 patient with non-functional adrenal tumor. In all the cases studied before with RET gene rearrangement, the breakpoint of the RET gene occurred in an intronic sequence between the tyrosine-kinase and transmembrane encoding domains. This rearrangement was . The cDNA of c-raf-1 was also amplified to evaluate the quality of each RNA sample [fig_ref] Figure 3: Representative results of RT-PCR [/fig_ref]. RET/PTC2 and RET/PTC3 were not found in any of the tested samples. Southern blot analysis To confirm the presence of RET/ PTC1 in cases 16 and 19 and to rule out the possibility of errors in PCR, we performed Southern blot hybridization using a 1 kb BglII-BamHI RET specific DNA fragment as the probe. This fragment was able to detect the region within the RET gene where the rearrangement had occurred. This probe detected a 6.3 kb fragment after digestion with EcoRI, a 3.7 kb fragment after digestion with BamHI, and a 9.3 kb fragment after digestion with HindIII in normal human DNA. Rearranged bands were found in cases 16 and 19 [fig_ref] Figure 4: Activation of the RET oncogene [/fig_ref]. A number of studies have been done on the alterations of the RET gene in MEN2A and MEN2B. They have shown that 95% of the RET mutations occur in exons 10 and 11 in MEN2A, whereas more than 90% of them occur in exon 16 (codon 918) in MEN2B. [bib_ref] Genetic basis of endocrine disease, multiple endocrine neoplasia type 2, Mulligan [/bib_ref] Recent studies have revealed that the frequency of RET gene mutation was 97% and the mutation hot spot was at codon 634 (86%) in MEN2A. [bib_ref] Clinical screening as compared with DNA analysis in families with multiple endocrine..., Lips [/bib_ref] [bib_ref] Germ-line mutations of the RET proto-oncogene in multiple endocrine neoplasia type 2A, Mulligan [/bib_ref] This suggests that mutation at codon 634 is associated with the transforming ability of the RET gene. Furthermore, Santoro et al. [bib_ref] Activation of RET as a dominant transforming gene by germ line mutations..., Santoro [/bib_ref] have indicated that the dimerization of the RET oncoprotein caused the activation of the RET gene in MEN2A. For the above reasons, the RET gene plays an important role in MEN2A. To understand the role of the RET gene in adrenal tumors, we examined 18 cases with Conn's syndrome, 3 cases with Cushing's syndrome, and 12 cases with pheochromocytoma. The results showed that 2 patients, one with Conn's syndrome and the other with pheochromocytoma, had RET gene mutations at codon 634, resulting in a cysteineto-tryptophan substitution. Codon 634 is located in the extracellular domain close to the transmembrane domain. Mutations of codon 634 may influence the structure of the RET protein receptor, [bib_ref] Activation of RET as a dominant transforming gene by germ line mutations..., Santoro [/bib_ref] leading to loss of specificity of ligand binding or activation of the RET protein in the absence of ligand binding. Uncontrolled RET protein activation would contribute to cell malignancy. This suggests that the mutations at codon 634 of the RET gene may be involved in the development of adrenal tumors. On the other hand, one of the 2 cases showed a phenylalanine-toserine substitution at codon 393, which was also found in Hirschsprung's disease. In addition, a loss-of-function effect of missense mutation at codon 393 of the RET protein has been demonstrated. [bib_ref] Point mutations affecting the tyrosine kinase domain of the RET protooncogene in..., Romeo [/bib_ref] [bib_ref] Mutations of the RET proto-oncogene in Hirschsprung's disease, Edery [/bib_ref] With regard to rearrangement of the RET oncogene, this was found in thyroid carcinomas, with most of them being RET/PTC1. RET/ PTC1 is the rearrangement between the RET gene and H4 (D10S170) gene. The chimera point is located at 1843 bp of the RET gene, flanking intron 11, which is upstream of the tyrosine kinase domain. RET/PTC1 was found in most thyroid tumors. The prevalence of RET/PTC1 varies in different geographical areas and in different ethnic populations. [bib_ref] PTC is a novel rearranged form of the RET proto-oncogene and is..., Grieco [/bib_ref] [bib_ref] cDNA cloning and characterization of RET activated in human papillary thyroid carcinoma..., Ishizaka [/bib_ref] [bib_ref] Identification of the product of two oncogenic rearranged forms of the RET..., Lanzi [/bib_ref] [bib_ref] Molecular characterization of a thyroid tumor-specific transforming sequence formed by the fusion..., Bongarzone [/bib_ref] [bib_ref] At (10, 17) translocation creates the RET/PTC2 chimeric transforming sequence in papillary..., Sozzi [/bib_ref] [bib_ref] Development of a single-step duplex RT-PCR detecting different forms of RET activation,..., Jhiang [/bib_ref] [bib_ref] Frequent activation of RET proto-oncogene by fusion with a new activating gene..., Bongarzone [/bib_ref] [bib_ref] High frequency of activation of tyrosine kinase oncogenes in human papillary thyroid..., Bongarzone [/bib_ref] [bib_ref] RET oncogene activation in human thyroid neoplasms is restricted to the papillary..., Santoro [/bib_ref] [bib_ref] Detection of RET/PTC transcripts in thyroid adenomas and adenomarous goiter by an..., Ishizaka [/bib_ref] [bib_ref] Low frequency of rearrangement of the ret and trk proto-oncogenes in Japanese..., Wajjwalku [/bib_ref] [bib_ref] Detection of the PTC RET oncogene in human thyroid cancers, Jhiang [/bib_ref] [bib_ref] Low rate of RET protooncogene activation (PTC/retTPC) in papillary thyroid carcinomas from..., Zou [/bib_ref] Previous studies have suggested that RET/ PTC1 rearrangement may induce activation of the RET oncogene in three situations. 9) 1. When the rearrangement occurs at the N-terminus, the ligand binding sites change, so the foreign gene insertion might activate RET tyrosine kinase and cause the RET oncoprotein activation. 2. When the replaced region of the RET gene contains a transmembrane domain, the rearrangement will change the domain to a free protein in the cytoplasm, so the protein can easily be activated by other proteins. 3. When deletion of the RET protein receptor occurs, RET/PTC1 might be autophosphorylated without any ligand stimulation. These hypotheses need to be studied further to determine which is the real mechanism of RET oncogene activation. In this study, we found that RET/PTC1 was also present in adrenal tumors, indicating that rearrangement of the RET oncogene also plays a role in tumorigenesis in adrenal tumors. RET gene mutation and rearrangement were not [fig_ref] Figure 4: Activation of the RET oncogene [/fig_ref] and hybridized to a 1.0 kb BamHI-BglII RET-specific DNA probe. Both RET proto-oncogene and the RET/PTC rearrangement band were seen in tumor tissue of cases 16 and 19, indicating that the tumors are both heterozygotic for the rearrangement.
found in many of the cases we analyzed; however, it is known that the mutation site (codon 634) and rearrangement form (RET/PTC1) are associated with the transforming activity of the RET gene. This indicates that alterations of the RET gene may contribute to adrenal tumor development. Additionally, our results show that the occurrence of the RET gene mutation and rearrangement are not associated with the tumor size, indicating that they may be initial events in adrenal tumor development. Furthermore, numerous studies have shown that the RET proto-oncogene is involved in neurodifferentiation and the nervous system is associated with the secretion of adrenocortical cells. Whether the RET gene also plays a crucial role in hormone overexpression is something we are still investigating. (Received
[fig] Figure 1: PCR-SSCP analysis of RET oncogene mutations in human adrenal tumors. Representative samples are shown for exon 6 (A) and exon 11 (B and C). An electrophoretic mobility shift of the bands between the tumor (T) and its paired normal tissue (N) implies a different conformation of the fragment, suggesting the presence of mutation in these exons. The PCR-SSCP analysis of exons 11 and 6 showed an apparent mobility shift in cases 6 and 29. [/fig]
[fig] Figure 2: Sequencing analysis of the RET oncogene (exons 11 and 6) in human adrenal tumor. TGC→TGG (Cys→Trp) and TTC→TCC (Phe→Ser) changes were found at codon 634 and codon 393. [/fig]
[fig] Figure 3: Representative results of RT-PCR. For evaluating the quality of each RNA, a part of the c-raf-1 gene was amplified. RET/PTC1 was detected in tumor tissue of cases 16 and 19. [/fig]
[fig] Figure 4: Activation of the RET oncogene (RET/PTC1) in adrenal tumors. Genomic DNA extract from normal adrenal gland of case 1 and tumor tissue of cases 16 and 19 were digested with EcoRI (lanes 3,6,9), BamHI (lanes 2, 4, 6), or HindIII (lanes 1, [/fig]
[table] Table I: Alterations of the RET Oncogene in Human Adrenal Tumors [/table]
|
Induced ectopic expression of HigB toxin in Mycobacterium tuberculosis results in growth inhibition, reduced abundance of a subset of mRNAs and cleavage of tmRNA
In Mycobacterium tuberculosis, the genes Rv1954A-Rv1957 form an operon that includes Rv1955 and Rv1956 which encode the HigB toxin and the HigA antitoxin respectively. We are interested in the role and regulation of this operon, since toxin-antitoxin systems have been suggested to play a part in the formation of persister cells in mycobacteria. To investigate the function of the higBA locus, effects of toxin expression on mycobacterial growth and transcript levels were assessed in M. tuberculosis H37Rv wild type and in an operon deletion background. We show that expression of HigB toxin in the absence of HigA antitoxin arrests growth and causes cell death in M. tuberculosis. We demonstrate HigB expression to reduce the abundance of IdeR and Zur regulated mRNAs and to cleave tmRNA in M. tuberculosis, Escherichia coli and Mycobacterium smegmatis. This study provides the first identification of possible target transcripts of HigB in M. tuberculosis.
# Introduction
Tuberculosis (TB) is a major global health threat. According to the WHO, TB causes 1.4 million deaths each year, and one-third of the world's population is believed to be latently infected . Treatment of TB involves a combination of four drugs (isoniazid, rifampicin, ethambutol and pyrazinamide) which are taken for 2 months in an intensive phase, followed by 4 months of isoniazid and rifampicin in a continuation phase . However, the drug treatments currently in use mainly attack actively growing bacteria and it is believed that a subpopulation of bacteria is able to evade drug-mediated killing by entering a state of non-replicating persistence [bib_ref] An in vitro model for sequential study of shiftdown of Mycobacterium tuberculosis..., Wayne [/bib_ref]. This persister population has been suggested to be the cause of relapse following drug treatment or reactivation of disease after years of latency [bib_ref] Clinical development of antituberculosis drugs, Mitchison [/bib_ref] [bib_ref] Tuberculosis chemotherapy: the influence of bacillary stress and damage response pathways on..., Warner [/bib_ref] [bib_ref] Cytological and transcript analyses reveal fat and lazy persister-like bacilli in tuberculous..., Garton [/bib_ref] [bib_ref] Characterization and transcriptome analysis of Mycobacterium tuberculosis persisters, Keren [/bib_ref]. Chromosomal toxin-antitoxin systems (TAS) can contribute to persister-mediated drug tolerance in bacteria as shown in a number of studies recently reviewed by Lewis [bib_ref] Persister cells, Lewis [/bib_ref]. TAS contain a toxin which causes growth arrest by inhibiting crucial cellular processes; for example RelE toxin inhibits translation in Escherichia coli [bib_ref] Rapid induction and reversal of a bacteriostatic condition by controlled expression of..., Pedersen [/bib_ref]. Toxin action is neutralized by a cognate antitoxin , and antitoxin generally acts as a transcriptional repressor of TA loci. Stress conditions such as starvation, DNA damage, heat shock or oxidative stress can activate TAS expression [bib_ref] Escherichia coli mazEF-mediated cell death is triggered by various stressful conditions, Hazan [/bib_ref] [bib_ref] Three new RelE-homologous mRNA interferases of Escherichia coli differentially induced by environmental..., Christensen-Dalsgaard [/bib_ref]. This occurs through antitoxin degradation by cellular proteases such as Lon or Clp [bib_ref] RelE, a global inhibitor of translation, is activated during nutritional stress, Christensen [/bib_ref] [bib_ref] Bacterial persistence by RNA endonucleases, Maisonneuve [/bib_ref] , releasing the biologically active toxin and allowing transcription of the TAS operon. Subsequent toxin-mediated growth arrest is believed to be beneficial to the bacteria, preserving nutrients and energy in an unfavourable environment and allowing resumption of growth when conditions have become favourable again [bib_ref] Toxins-antitoxins: diversity, evolution and function, Hayes [/bib_ref].
Toxin-antitoxin systems are ubiquitous in mycobacteria, with members of the VapBC, MazEF and RelBE and ParDE as well as novel families of TAS found across the genus. Interestingly, there seems to have been an expansion of VapBC, MazEF and RelE TA families in the Mycobacterium tuberculosis complex (MTBC) which includes M. tuberculosis, Mycobacterium bovis, Mycobacterium africanum, Mycobacterium canetti, Mycobacterium microti and M. bovis Bacille-Calmette-Guerin. This is not the case for HigBA, where only one locus is present in each of the members of the MTBC [bib_ref] Toxin-antitoxin loci are highly abundant in free-living but lost from host-associated prokaryotes, Pandey [/bib_ref] [bib_ref] Comprehensive functional analysis of Mycobacterium tuberculosis toxin-antitoxin systems: implications for pathogenesis, stress..., Ramage [/bib_ref]. The HigBA locus was first identified on the Rst1 plasmid [bib_ref] A new plasmid-encoded proteic killer gene system: cloning, sequencing, and analyzing hig..., Tian [/bib_ref] , but homologues have subsequently been found on chromosomes of a range of bacteria including clinical isolates of Pseudomonas aeruginosa and methicillin-resistant Staphylococcus aureus [bib_ref] Toxin-antitoxin loci are highly abundant in free-living but lost from host-associated prokaryotes, Pandey [/bib_ref] [bib_ref] Toxin-antitoxin (TA) systems are prevalent and transcribed in clinical isolates of Pseudomonas..., Williams [/bib_ref]. HigB toxins belong to the RelE family and have been characterized as translation-dependent mRNA-cleaving enzymes in Vibrio cholerae, Proteus vulgaris and E. coli [bib_ref] Two higBA loci in the Vibrio cholerae superintegron encode mRNA cleaving enzymes..., Christensen-Dalsgaard [/bib_ref] [bib_ref] Bacterial toxin HigB associates with ribosomes and mediates translationdependent mRNA cleavage at..., Hurley [/bib_ref] [bib_ref] Three new RelE-homologous mRNA interferases of Escherichia coli differentially induced by environmental..., Christensen-Dalsgaard [/bib_ref]. HigB toxins are able to cleave a wide range of mRNAs [bib_ref] Two higBA loci in the Vibrio cholerae superintegron encode mRNA cleaving enzymes..., Christensen-Dalsgaard [/bib_ref] [bib_ref] Bacterial toxin HigB associates with ribosomes and mediates translationdependent mRNA cleavage at..., Hurley [/bib_ref] [bib_ref] Three new RelE-homologous mRNA interferases of Escherichia coli differentially induced by environmental..., Christensen-Dalsgaard [/bib_ref] and associate with the 50S subunit of the ribosome in P. vulgaris [bib_ref] Bacterial toxin HigB associates with ribosomes and mediates translationdependent mRNA cleavage at..., Hurley [/bib_ref]. Both amino acid starvation and chloramphenicol-mediated inhibition of protein synthesis induce the higBA loci of V. cholerae and E. coli [bib_ref] Two higBA loci in the Vibrio cholerae superintegron encode mRNA cleaving enzymes..., Christensen-Dalsgaard [/bib_ref] [bib_ref] Three new RelE-homologous mRNA interferases of Escherichia coli differentially induced by environmental..., Christensen-Dalsgaard [/bib_ref]. The activation of HigB toxin by HigA degradation is Lon protease-dependent in E. coli [bib_ref] Three new RelE-homologous mRNA interferases of Escherichia coli differentially induced by environmental..., Christensen-Dalsgaard [/bib_ref].
The M. tuberculosis HigBA TAS has previously been shown to be a functional toxin-antitoxin system in E. coli and mycobacteria [bib_ref] Killing activity and rescue function of genome-wide toxin-antitoxin loci of Mycobacterium tuberculosis, Gupta [/bib_ref] [bib_ref] Analyzing the regulatory role of the HigA antitoxin within Mycobacterium tuberculosis, Fivian-Hughes [/bib_ref] [bib_ref] SecB-like chaperone controls a toxin-antitoxin stress-responsive system in Mycobacterium tuberculosis, Bordes [/bib_ref]. The HigBA TAS of M. tuberculosis is unusual in terms of its genomic organization. The HigB toxin (Rv1955) and HigA antitoxin (Rv1956) are in an operon that includes Rv1954A and Rv1957 [bib_ref] Experimental determination of translational start sites resolves uncertainties in genomic open reading..., Smollett [/bib_ref] [bib_ref] Analyzing the regulatory role of the HigA antitoxin within Mycobacterium tuberculosis, Fivian-Hughes [/bib_ref]. Rv1957 has recently been identified as a Sec-B like chaperone required for antitoxin stabilization [bib_ref] SecB-like chaperone controls a toxin-antitoxin stress-responsive system in Mycobacterium tuberculosis, Bordes [/bib_ref]. Rv1954A shows homology with a family of YjzC-like proteins that are widely conserved in bacteria but its function remains unknown. The Rv1954A-1957 operon is regulated by two promoters [fig_ref] Figure 1: HigB expression in M [/fig_ref]. P1, located directly upstream of HigB, is induced by DNA damage and is not regulated by HigA (Fivian-Hughes, 2009), while P2, upstream of Rv1954A, is repressed by binding of HigA to a specific motif [bib_ref] Analyzing the regulatory role of the HigA antitoxin within Mycobacterium tuberculosis, Fivian-Hughes [/bib_ref].
The higBA locus is induced by DNA damage, heat shock, and during hypoxia and growth in activated macrophages, indicating that it might be important for bacterial survival under stress conditions encountered during infection [bib_ref] Dissection of the heat-shock response in Mycobacterium tuberculosis using mutants and microarrays, Stewart [/bib_ref] [bib_ref] Identification of a promoter motif regulating the major DNA damage response mechanism..., Gamulin [/bib_ref] [bib_ref] Comprehensive functional analysis of Mycobacterium tuberculosis toxin-antitoxin systems: implications for pathogenesis, stress..., Ramage [/bib_ref] [bib_ref] Functional genetic diversity among Mycobacterium tuberculosis complex clinical isolates: delineation of conserved..., Homolka [/bib_ref]. We are interested in the function of this operon in M. tuberculosis, and its potential role in mycobacterial persistence [bib_ref] The three RelE homologs of Mycobacterium tuberculosis have individual, drug-specific effects on..., Singh [/bib_ref]. We have analysed the effect of expressing higB under the control of an inducible promoter system, characterized the effect of higB induction on global translation and identified cleavage of tmRNA in response to HigB expression.
# Results
## Expression of higb inhibits mycobacterial growth
Toxin activity of M. tuberculosis HigB has previously been demonstrated in both E. coli and Mycobacterium marinum [bib_ref] Killing activity and rescue function of genome-wide toxin-antitoxin loci of Mycobacterium tuberculosis, Gupta [/bib_ref] [bib_ref] SecB-like chaperone controls a toxin-antitoxin stress-responsive system in Mycobacterium tuberculosis, Bordes [/bib_ref]. In M. tuberculosis itself, Liquid cultures of transformants were grown from a starting OD580 of 0.05 and monitored over time. No or 300 ng ml −1 ATc were added at the start of the growth curve. All results are the mean values and standard deviation of three independent biological replicates. Where indicated, a significant difference (as determined by Student's t-test) between uninduced (− ATc) and induced (+ ATc) conditions is marked by an asterisk (*) for P < 0.05, ** for P < 0.01, *** for P < 0.0001. HigB toxicity has been inferred from an inability to delete HigA antitoxin without simultaneous deletion of HigB toxin [bib_ref] Analyzing the regulatory role of the HigA antitoxin within Mycobacterium tuberculosis, Fivian-Hughes [/bib_ref]. To test directly the effect of HigB on growth of M. tuberculosis we constructed an inducible expression plasmid, in which higB was placed under the control of a tetracycline-inducible promoter. Functionality of this system was initially tested in Mycobacterium smegmatis, which does not contain a HigBA locus [bib_ref] Comprehensive functional analysis of Mycobacterium tuberculosis toxin-antitoxin systems: implications for pathogenesis, stress..., Ramage [/bib_ref].
Mycobacterium smegmatis was electroporated with the HigB-expression plasmid and the vector control. Transformants were grown to mid-exponential phase in liquid media with and without anhydrotetracycline (ATc) inducer. Quantitative RT-PCR was used to measure the level of higB transcripts present in the different strains in the absence and presence of inducer [fig_ref] Figure 1: HigB expression in M [/fig_ref]. As expected, no higB was detected in the vector control strain. In the HigB-expression strain, some transcripts were detected in the absence of ATc, indicating a degree of leakage associated with the construct; expression increased twofold in the presence of inducer [fig_ref] Figure 1: HigB expression in M [/fig_ref]. These results confirm the functionality of the expression system used.
We proceeded to determine the effect of higB expression on growth in liquid media. Bacterial growth was measured by the optical density of the cultures in the presence or absence of inducer [fig_ref] Figure 1: HigB expression in M [/fig_ref]. No difference was observed during the initial 5 h of growth. However, after 7.5 h in the presence of inducer, growth of strains harbouring the HigB-expression plasmid slowed down and was significantly lower than growth of control strains over a 60 h time-course [fig_ref] Figure 1: HigB expression in M [/fig_ref]. These results show that higB expression inhibits growth of M. smegmatis. Furthermore our data suggest a sharp threshold for HigB toxicity, given that a twofold induction was sufficient to cause growth inhibition.
Next, the effects of toxin expression were tested in M. tuberculosis. Plasmids were electroporated into M. tuberculosis H37Rv and transformants were grown to mid-exponential phase with and without ATc. To confirm that addition of ATc resulted in higB induction in M. tuberculosis the number of higB transcripts was measured by qRT-PCR in vector control and conditional-expression strains with and without inducer [fig_ref] Figure 2: HigB expression in M [/fig_ref]. The level of expression of higB relative to a 16S control was significantly higher in the strains containing the expression plasmid than in the vector control, with a 4.6-fold increase in the presence of ATc, confirming the functionality of this expression system in M. tuberculosis [fig_ref] Figure 2: HigB expression in M [/fig_ref].
Next, we monitored growth of liquid cultures in the presence or absence of inducer [fig_ref] Figure 2: HigB expression in M [/fig_ref]. Both the vector control and the HigB expression strains grew at similar rates over a period of 14 days [fig_ref] Figure 2: HigB expression in M [/fig_ref]. Thus toxin expression in M. tuberculosis wild type does not result in a growth defect.
Having validated that ATc addition resulted in increased higB expression, we reasoned that the lack of an effect on growth may be due to neutralizing activity of endogenous antitoxin and chaperone present in M. tuberculosis wild type. We therefore tested the effect of higB expression on growth in a Rv1955-Rv1957 (ΔTAC) deletion mutant (Fivian-Hughes and Davis, 2010). Growth was seen in the vector control strain and in the uninduced expression strain over the course of 14 days [fig_ref] Figure 2: HigB expression in M [/fig_ref]. In contrast no growth was observed when the expression strain was inoculated into medium containing ATc [fig_ref] Figure 2: HigB expression in M [/fig_ref]. This shows that, in the absence of neutralizing antitoxin and chaperone, HigB expression inhibits early exponential growth.
We then investigated the effect of addition of inducer during mid-exponential phase [fig_ref] Figure 2: HigB expression in M [/fig_ref]. As before, no difference in growth was seen between the vector control strain and the uninduced expression strain [fig_ref] Figure 2: HigB expression in M [/fig_ref]. However, 1 day after ATc induction of the expression strain, optical densities ceased to increase at the same rate as in the control strains, and did not go beyond OD 580 1.4 [fig_ref] Figure 2: HigB expression in M [/fig_ref]. Thus HigB expression during midexponential phase resulted in growth inhibition. Associated loss of viability was evident from analysis of colonyforming units (cfu) following ATc addition. Colony-forming units decreased by 3 log10 7 days after ATc addition to the HigB-expression strain [fig_ref] Figure 2: HigB expression in M [/fig_ref]. In contrast, cfu continued to increase in the ATc-treated vector control strain, confirming that the inducer itself had no bactericidal effect [fig_ref] Figure 2: HigB expression in M [/fig_ref]. Subsequent removal of inducer resulted in resumption of growth of the remaining viable population, as seen by 1 log10 increase in cfu 5 days after ATc was removed from the cells by washing [fig_ref] Figure 2: HigB expression in M [/fig_ref]. Together these data demonstrate that higB expression in the absence of antitoxin and Rv1957c chaperone arrests growth and causes substantial cell death in M. tuberculosis.
## Global gene expression profiling identifies putative higb targets
To investigate the effect of HigB expression on global transcriptional profiles we carried out RNAseq analysis. RNA was extracted from HigB expressing and vector control strains 24 h after ATc addition; libraries were prepared, sequenced and analysed. Validation of RNAseq results by means of qRT-PCR is recorded in [fig_ref] Table 1: RNAseq data comparing fold change between the HigB expressing and the vector... [/fig_ref].
The relative abundance of 34 mRNAs changed significantly [fig_ref] Table 1: RNAseq data comparing fold change between the HigB expressing and the vector... [/fig_ref]. Two genes, Rv0197 (a non-essential gene of unknown function) and bfrB (a non-essential IdeR regulated gene involved in iron storage), were upregulated [bib_ref] The Mycobacterium tuberculosis IdeR is a dual functional regulator that controls transcription..., Gold [/bib_ref] [bib_ref] Genes required for mycobacterial growth defined by high density mutagenesis, Sassetti [/bib_ref]. Of the 32 downregulated genes, a large number are part of operons regulated by IdeR (mbtK, irtAB,hisE, PPE37, mbtHFED-CBAI, Rv3403c, Rv3839), Zur (PPE3) or both (eccB3, A. Quantitative RT-PCR of higB expression relative to 16S rRNA. RNA was isolated from M. tuberculosis wild-type exponential-phase cultures. B-D. Growth of M. tuberculosis wild-type (B) and ΔTAC (C and D) strains. Liquid cultures of transformants were grown from a starting OD580 of 0.05 and monitored over time. No or 300 ng ml −1 ATc were added at the start of the growth curve at day 0 (B and C) or during mid-exponential phase at day 3 (D). E. Survival of ΔTAC transformants following addition of ATc during mid-exponential phase (time 0) and removal of inducer by washing at day 7. All results are the mean values and standard deviation of three independent biological replicates. A significant difference (as determined by Student's t-test) between uninduced (− ATc) and induced (+ ATc) conditions is marked by an asterisk (*) for P < 0.05, ** for P < 0.01, *** for P < 0.0001. eccC3, esxG, esxH, espG3, eccD3, mycP3, eccE3 and mmpL5) [bib_ref] The Mycobacterium tuberculosis IdeR is a dual functional regulator that controls transcription..., Gold [/bib_ref] [bib_ref] ) ideR, an essential gene in Mycobacterium tuberculosis: role of IdeR in..., Rodriguez [/bib_ref] [bib_ref] Global analysis of the Mycobacterium tuberculosis zur (furB) regulon, Maciag [/bib_ref]. qRT-PCR confirmed downregulation of espG3 and mbtC in the HigB expression strain, and also highlighted reduced abundance of rpmE, which had not reached statistical significance in the RNAseq analysis [fig_ref] Figure 1: HigB expression in M [/fig_ref]. RpmE has been implicated in zinc homeostasis in Bacillus subtilis [bib_ref] Proteomic study of the Bacillus subtilis ribosome: finding of zinc-dependent replacement for..., Nanamiya [/bib_ref]. In addition to the metal ion regulons, we also observed a significant decrease in abundance of tmRNA transcripts. Parallel microarray analysis generated analogous results with significant fold change observed for a limited subset of genes enriched in metal ion regulons (data not shown).
## M. tuberculosis higb expression causes tmrna degradation
HigB is related to the RelE family of toxins that are known to target mRNA as well as tmRNA, the stable RNA product of the ssrA gene discovered by Ray and Apirion [bib_ref] Characterization of 10S RNA: a new stable RNA molecule from Escherichia coli, Ray [/bib_ref] [bib_ref] Toxin-antitoxin loci are highly abundant in free-living but lost from host-associated prokaryotes, Pandey [/bib_ref]. tmRNA plays an essential role in trans-translation, which is required to rescue ribosomes stalled by RelE-generated nonstop mRNAs . A role for transtranslation in M. tuberculosis persistence has been highlighted by the demonstration that this process is inhibited by the antimycobacterial drug pyrazinamide [bib_ref] Pyrazinamide inhibits trans-translation in Mycobacterium tuberculosis, Shi [/bib_ref]. We were therefore particularly interested to determine whether M. tuberculosis HigB also targets tmRNA.
Total RNA was extracted from M. tuberculosis ΔTAC strains in mid-exponential growth phase at 2, 6 and 24 h after toxin induction and probed by Northern blot probing for the 5′ and 3′ ends of tmRNA. Only the expected full-length tmRNA transcript was seen in the vector control . In contrast, the amount of the full-length transcript decreased following induction of HigB in the expression strain, along with appearance of two cleaved fragments of ∼ 100 bp and ∼ 150 bp when probing for 5′ . HigB expression affects tmRNA in M. tuberculosis ΔTAC. RNA was extracted from mid-exponential cultures treated with ATc for 2 h, 6 h or 24 h and cDNA was prepared for qRT-PCR analysis. A-C. Northern blots probing for tmRNA (A and B) and 5S (C). Transcript sizes are relative to the position of RNA marker and tmRNA cleavage products are indicated with an arrow. D. Quantitative RT-PCR of transcripts of interest. All results are the mean values and standard deviation of three independent biological replicates. A significant difference (as determined by Student's t-test) between vector control and HigB overexpression strain is marked by an asterisk (*) for P < 0.05, (**) for P < 0.01, (***) for P < 0.0001. tmRNA, and one cleaved fragment of ∼ 250 bp when probing for 3′ tmRNA . Furthermore, the amount of cleavage product increased with time after HigB induction . Toxin induction had no effect on the integrity of ribosomal 5S RNA .
To quantify the effect of HigB expression on tmRNA, qRT-PCR was carried out . As expected in the ΔTAC background, significant expression of higB was only detected in the conditional-expression strain . HigB was fully induced by 2 h after ATc addition. Quantification using primers directed to either the 5′ or the 3′ end of tmRNA revealed a 1.5-fold increase in relative expression after addition of ATc to the vector strain middle and bottom), consistent with a previous observation that tetracycline upregulates expression of tmRNA in M. smegmatis [bib_ref] Expression of tmRNA in mycobacteria is increased by antimicrobial agents that target..., Andini [/bib_ref]. In contrast to this, the relative number of transcripts detected using the 5′ and 3′ probes in the expression strain dropped sharply over the 24 h incubation period . From these results, we conclude that the initial cleavage fragments detected by Northern blot are progressively degraded over time.
tmRNA shares high sequence similarity between M. tuberculosis and E. coli [fig_ref] Figure 4: HigB expression affects tmRNA in M [/fig_ref] , and could be a conserved target of HigB. We tested if HigB also causes tmRNA cleavage in M. smegmatis and E. coli. RNA was prepared from mid-log phase bacteria harbouring HigBexpression plasmid grown in the presence or absence of inducer. Northern blots probing for the 5′ end of tmRNA were carried out [fig_ref] Figure 4: HigB expression affects tmRNA in M [/fig_ref]. In M. smegmatis [fig_ref] Figure 4: HigB expression affects tmRNA in M [/fig_ref] , full-length tmRNA transcript was detected in the HigB expression strain under both conditions tested; two distinct cleavage products of smaller size (> 100 bp) were seen when HigB expression is induced. Similarly in E. coli, full-length transcript was present along with two cleavage products when HigB was expressed [fig_ref] Figure 4: HigB expression affects tmRNA in M [/fig_ref]. Expression of M. tuberculosis HigB in M. smegmatis as well as E. coli results in tmRNA cleavage, identifying this as a conserved target of the M. tuberculosis toxin.
## Tmrna is cleaved in the mrna-like region
HigB toxin expression gives rise to distinct tmRNA cleavage products, indicating that the toxin might directly cleave specific sites within the tmRNA molecule. To investigate this further by mapping the 3′ ends resulting from cleavage, we performed 3′ RACE as described previously [bib_ref] Identification of small RNAs in Mycobacterium tuberculosis, Arnvig [/bib_ref].
RNA was prepared from M. tuberculosis HigBexpressing and vector control strains treated with inducer for 24 h. A poly-A tail was added and cDNA prepared. This was used as a template for PCR using a gene-specific forward primer binding tmRNA just upstream of the 5′ probe used for Northern blots [fig_ref] Figure 5: HigB cleavage sites identified by 3′ RACE [/fig_ref] and a poly-A linker-specific primer. PCR product corresponding to nearly fulllength tmRNA (the gene-specific primer binds 45 bp after the start of ssrA) was obtained from the vector control strain. In the HigB-expression strain, an additional PCR product matching the size of the cleavage products detected by Northern blot (∼ 100-150 bp) was also obtained. PCR products were cloned into a plasmid and sequenced. The 3′ ends of transcripts were identified as the junction with the poly-A tail. All of the 3′ ends identified from the large PCR products corresponded to full-length ssrA [fig_ref] Figure 5: HigB cleavage sites identified by 3′ RACE [/fig_ref]. Sequencing of 25 clones generated from the short PCR products (i.e. the cleaved tmRNA) identified seven 3′ ends located within the mRNA-like coding region of the tmRNA [fig_ref] Figure 5: HigB cleavage sites identified by 3′ RACE [/fig_ref]. This pattern could be generated by multiple cleavage sites, or by a single cleavage followed by 5′-to-3′ exonuclease digestion.
# Discussion
Since their discovery in the 1980s, TA systems have been found in nearly all prokaryotes and archaea [bib_ref] Toxin-antitoxin loci are highly abundant in free-living but lost from host-associated prokaryotes, Pandey [/bib_ref] [bib_ref] Genomes of the most dangerous epidemic bacteria have a virulence repertoire characterized..., Georgiades [/bib_ref] [bib_ref] Toxinantitoxin systems in bacteria and archaea, Yamaguchi [/bib_ref]. Due to their high abundance in M. tuberculosis, there has been a series of research efforts investigating mycobacterial TA systems [bib_ref] The PINdomain toxin-antitoxin array in mycobacteria, Arcus [/bib_ref] [bib_ref] Characterization of mRNA interferases from Mycobacterium tuberculosis, Zhu [/bib_ref] [bib_ref] Persister cells, Lewis [/bib_ref] [bib_ref] Expression of Mycobacterium tuberculosis Rv1991c using an arabinose-inducible promoter demonstrates its role..., Carroll [/bib_ref] [bib_ref] Biochemical characterization of a chromosomal toxin-antitoxin system in Mycobacterium tuberculosis, Zhao [/bib_ref] [bib_ref] Killing activity and rescue function of genome-wide toxin-antitoxin loci of Mycobacterium tuberculosis, Gupta [/bib_ref] [bib_ref] Three Mycobacterium tuberculosis Rel toxin-antitoxin modules inhibit mycobacterial growth and are expressed..., Korch [/bib_ref] [bib_ref] Comprehensive functional analysis of Mycobacterium tuberculosis toxin-antitoxin systems: implications for pathogenesis, stress..., Ramage [/bib_ref] [bib_ref] The vapBC operon from Mycobacterium smegmatis is an autoregulated toxin-antitoxin module that..., Robson [/bib_ref] [bib_ref] Characterization of a chromosomal toxin-antitoxin, Rv1102c-Rv1103c system in Mycobacterium tuberculosis, Han [/bib_ref] [bib_ref] Characterization of an interplay between a Mycobacterium tuberculosis MazF homolog, Rv1495 and..., Huang [/bib_ref] [bib_ref] The three RelE homologs of Mycobacterium tuberculosis have individual, drug-specific effects on..., Singh [/bib_ref] [bib_ref] Characterization of the interaction and cross-regulation of three Mycobacterium tuberculosis RelBE modules, Yang [/bib_ref] [bib_ref] VapC toxins from Mycobacterium tuberculosis are ribonucleases that differentially inhibit growth and..., Ahidjo [/bib_ref] [bib_ref] SecB-like chaperone controls a toxin-antitoxin stress-responsive system in Mycobacterium tuberculosis, Bordes [/bib_ref] [bib_ref] Toxin-antitoxin systems of Mycobacterium smegmatis are essential for cell survival, Frampton [/bib_ref] [bib_ref] A VapBC toxin-antitoxin module is a posttranscriptional regulator of metabolic flux in..., Mckenzie [/bib_ref] [bib_ref] Growth and translation inhibition through sequence-specific RNA binding by Mycobacterium tuberculosis VapC..., Sharp [/bib_ref].
Recent studies have elucidated the role of some M. smegmatis and M. tuberculosis TA families. For example, studies on the M. smegmatis VapBC locus have implicated VapC as playing a role in carbon transport and metabolism, since VapC expression affects genes involved in glycerol utilization [bib_ref] A VapBC toxin-antitoxin module is a posttranscriptional regulator of metabolic flux in..., Mckenzie [/bib_ref]. Furthermore, it has been shown that a M. smegmatis strain which had all of its TA systems deleted displayed a growth defect in complex medium [bib_ref] Toxin-antitoxin systems of Mycobacterium smegmatis are essential for cell survival, Frampton [/bib_ref].
Less is known about the function of M. tuberculosis TA loci, although several have been shown to be induced during stress conditions such as hypoxia or growth in macrophages [bib_ref] Three Mycobacterium tuberculosis Rel toxin-antitoxin modules inhibit mycobacterial growth and are expressed..., Korch [/bib_ref] [bib_ref] Comprehensive functional analysis of Mycobacterium tuberculosis toxin-antitoxin systems: implications for pathogenesis, stress..., Ramage [/bib_ref] [bib_ref] Noncognate Mycobacterium tuberculosis toxin-antitoxins can physically and functionally interact, Zhu [/bib_ref]. Singh and colleagues showed that overexpression of RelE toxins in M. tuberculosis leads to increased survival during antibiotic treatment [bib_ref] The three RelE homologs of Mycobacterium tuberculosis have individual, drug-specific effects on..., Singh [/bib_ref] , and multiple TAS were found to be upregulated in M. tuberculosis selected as drug-tolerant persisters [bib_ref] Characterization and transcriptome analysis of Mycobacterium tuberculosis persisters, Keren [/bib_ref]. Although deletion of single RelE genes did not impair survival of the bacteria in a murine infection model [bib_ref] The three RelE homologs of Mycobacterium tuberculosis have individual, drug-specific effects on..., Singh [/bib_ref] , a significant role for TA systems during M. tuberculosis infection cannot be A. DNA alignment of ssrA from M. tuberculosis and E. coli. Conserved residues are highlighted by an asterisk (*). B. Northern blots probing for tmRNA (B) and 5 S (C). RNA was extracted from M. smegmatis (left panel) and E. coli (right panel) HigB expression strains grown with or without inducer. Strains were grown to mid-exponential phase and inducer was added; ATc for M. smegmatis and L-arabinose for E. coli, for 7.5 h and 2 h respectively. ruled out, given the abundance of TA systems and their potential for redundancy.
We were interested in the function of the M. tuberculosis HigBA locus, which is one of the few TAS without additional homologues in M. tuberculosis [bib_ref] Comprehensive functional analysis of Mycobacterium tuberculosis toxin-antitoxin systems: implications for pathogenesis, stress..., Ramage [/bib_ref]. To elucidate the role of HigB toxin, we constructed a tetracycline-inducible expression plasmid and showed that expression of HigB inhibited growth of M. tuberculosis. This is in accord with previous studies where expression of M. tuberculosis HigB inhibited growth in M. marinum and M. bovis BCG [bib_ref] SecB-like chaperone controls a toxin-antitoxin stress-responsive system in Mycobacterium tuberculosis, Bordes [/bib_ref]. We also observed a significant loss of viability following induction of HigB, leaving only a subpopulation with the potential to acquire a persister phenotype. Growth arrest and loss of viability were contingent on deletion of the endogenous copy of the Rv1955-Rv1957 operon, indicating that -as in E. coli, V. cholerae and P. vulgaris [bib_ref] Characterization of a higBA toxin-antitoxin locus in Vibrio cholerae, Budde [/bib_ref] [bib_ref] Killing activity and rescue function of genome-wide toxin-antitoxin loci of Mycobacterium tuberculosis, Gupta [/bib_ref] [bib_ref] Three new RelE-homologous mRNA interferases of Escherichia coli differentially induced by environmental..., Christensen-Dalsgaard [/bib_ref] -HigA antitoxin is able to counteract HigB toxicity in M. tuberculosis.
HigB toxins from other bacteria are characterized by endonuclease activity against mRNAs [bib_ref] Two higBA loci in the Vibrio cholerae superintegron encode mRNA cleaving enzymes..., Christensen-Dalsgaard [/bib_ref] [bib_ref] Bacterial toxin HigB associates with ribosomes and mediates translationdependent mRNA cleavage at..., Hurley [/bib_ref] [bib_ref] Three new RelE-homologous mRNA interferases of Escherichia coli differentially induced by environmental..., Christensen-Dalsgaard [/bib_ref]. In spite of the marked growth arrest and loss of viability phenotypes, transcriptional profiling 24 h after induction of HigB identified significant changes in the abundance of only a restricted subset of genes. We observed a clear downregulation of genes regulated by the IdeR and Zur repressors involved in regulation of iron and zinc homeostasis in mycobacteria [bib_ref] ) ideR, an essential gene in Mycobacterium tuberculosis: role of IdeR in..., Rodriguez [/bib_ref] [bib_ref] Global analysis of the Mycobacterium tuberculosis zur (furB) regulon, Maciag [/bib_ref]. This could reflect degradation of these mRNAs by HigB endonuclease activity, or a downstream regulatory consequence of toxin-induced increase in the intracellular availability of iron and zinc. Induction of HigB expression in media containing varying metal ion concentrations, by supplementation of media with additional iron or zinc [bib_ref] Characterization of a Mycobacterium tuberculosis ESX-3 conditional mutant: essentiality and rescue by..., Serafini [/bib_ref] , or growth in the presence of the zinc chelator TPEN [bib_ref] Identification and characterization of a major Zn(II) resistance determinant of Mycobacterium smegmatis, Grover [/bib_ref] did not uncover any obvious link between the availability of metal ions and toxin-mediated changes in growth and viability [fig_ref] Figure 2: HigB expression in M [/fig_ref].
Given recent interest in the role of trans-translation in drug sensitivity and persistence of M. tuberculosis [bib_ref] Pyrazinamide inhibits trans-translation in Mycobacterium tuberculosis, Shi [/bib_ref] , we were particularly interested in testing whether M. tuberculosis HigB exhibited activity against tmRNA. Our results show that M. tuberculosis HigB expression leads to tmRNA cleavage, generating 5′ and 3′ fragments that are subsequently degraded. Similarly, expression of M. tuberculosis HigB resulted in cleavage of the closely related tmRNA homologues in M. smegmatis and E. coli. Detailed characterization of the initial fragmentation products shows that tmRNA is cleaved within the mRNA-like 12-codon coding sequence. This is in accordance with the general view that HigB toxins (and also E. coli RelE toxin) are translation-dependent and only cleave RNA transcripts during the process of translation [bib_ref] Two higBA loci in the Vibrio cholerae superintegron encode mRNA cleaving enzymes..., Christensen-Dalsgaard [/bib_ref] [bib_ref] Bacterial toxin HigB associates with ribosomes and mediates translationdependent mRNA cleavage at..., Hurley [/bib_ref] [bib_ref] Three new RelE-homologous mRNA interferases of Escherichia coli differentially induced by environmental..., Christensen-Dalsgaard [/bib_ref].
In summary, inhibition of growth following induction of HigB toxin in the absence of its cognate HigA antitoxin had a significant bactericidal effect on M. tuberculosis. This differs from the predominantly bacteriostatic effects observed with HigB toxins of P. vulgaris and V. cholerae [bib_ref] Two higBA loci in the Vibrio cholerae superintegron encode mRNA cleaving enzymes..., Christensen-Dalsgaard [/bib_ref] [bib_ref] Characterization of a higBA toxin-antitoxin locus in Vibrio cholerae, Budde [/bib_ref]. Induction of HigB resulted in decreased abundance of IdeR and Zur regulated mRNAs together with sitespecific cleavage and subsequent degradation of tmRNA. tmRNA cleavage provides a clear phenotypic marker that may be useful in screening infected tissues for evidence of HigB activation during M. tuberculosis infection.
## Experimental procedures
Bacterial strains and culture conditions Escherichia coli Dh5α was used for plasmid construction and grown at 37°C with shaking at 225 r.p.m. in Luria-Bertani (LB) broth or on Luria-Bertani agar. The E. coli HigB overexpression strain, W3110 transformed with pK6-HigB [bib_ref] SecB-like chaperone controls a toxin-antitoxin stress-responsive system in Mycobacterium tuberculosis, Bordes [/bib_ref] , was a kind gift from Pierre Genevaux. Ampicillin was used at 100 mg l −1 , gentamicin at 20 mg l −1 and kanamycin at 50 mg l −1 . M. smegmatis mc 2 155, M. tuberculosis Putative cleavage sites found in the 300 bp and 100-150 bp PCR products are indicated as highlighted 3′ residues before the A tail. Nucleotides between which the cleavage must have occurred are in bold. The sequence of the gene-specific forward primer used is underlined. The 5′ and 3′ probes used for Northern blots are shaded in grey.
H37Rv (ATCC 25618) and M. tuberculosis [bib_ref] Analyzing the regulatory role of the HigA antitoxin within Mycobacterium tuberculosis, Fivian-Hughes [/bib_ref] were grown at 37°C in modified Dubos medium (Difco) or on Difco Middlebrook 7H11 agar (Becton Dickinson) both supplemented with 4% Dubos medium albumin (Difco) and 0.5% or 0.2% w/v glycerol respectively. M. smegmatis liquid cultures were grown shaking at 100 r.p.m. and M. tuberculosis liquid cultures were grown in a roller incubator at 2 r.p.m. Gentamicin was used at 10 mg l −1 and kanamycin at 20 mg l −1 , when required. All procedures with M. tuberculosis were carried out under Containment level 3 conditions.
## Plasmid construction and oligonucleotides
For tetracycline-inducible gene expression, we constructed a novel vector, pTETR3 (available from Addgene, http://www .addgene.org), which combines the tetracycline-inducible promoter Pmyc1tetO from pSE100 [bib_ref] Controlling gene expression in mycobacteria with anhydrotetracycline and Tet repressor, Ehrt [/bib_ref] , and a codon optimized TetR repressor linked to the Pimyc promoter from pTE-10 M-OX [bib_ref] Improved tetracycline repressors for gene silencing in mycobacteria, Klotzsche [/bib_ref]. The vector was based on pKP186, a kanamycin-resistant pMV306 derivative that does not contain integrase [bib_ref] A member of the cAMP receptor protein family of transcription regulators in..., Rickman [/bib_ref]. For electroporations, integrase was supplied by pBS-Int, an ampicillin-resistant mycobacterial suicide vector containing L5 integrase [bib_ref] Instability and site-specific excision of integration-proficient mycobacteriophage L5 plasmids: development of stably..., Springer [/bib_ref]. To amplify HigB (Rv1955) from M. tuberculosis genomic DNA, primers Rv1955T F and Rv1955T R were used. The PCR product was cloned into pTetR3 as a PacI/EcoRI fragment. The resulting plasmid was named pDS227 and verified by DNA sequencing. Primers and oligonucleotides used in this study are listed in . Plasmids were transformed into mycobacteria by electroporation as described previously [bib_ref] Electroporation of mycobacteria, Goude [/bib_ref].
## Assaying the effect of higb overexpression on growth
To assess the effect of toxin overexpression during aerobic growth, liquid cultures were grown to an OD580 of 0.3-0.6 (unless otherwise stated) before addition of inducer. Growth was monitored by measuring optical densities. For cfu determination, 10-fold serial dilution series were prepared and 30 μl of each dilution were spread on three-sector plates (BD-Falcon). Experiments were performed on three independent biological replicates. To induce higB expression in mycobacteria, 300 ng ml −1 anhydrotetracycline (ATc) was added.
Escherichia coli transformed with pPK6-HigB was grown with 0.4% glucose. Glucose was removed by washing before higB expression was induced by the addition of 0.5% L-arabinose.
## Rna preparation and qrt-pcr analysis
Mycobacterium tuberculosis cultures were grown to midexponential phase (unless otherwise stated), and RNA was prepared using a FastRNA pro blue kit (Qbiogene). Contaminating DNA was removed using a TURBO DNA-free kit (Ambion), and 1 μg of RNA was converted to cDNA using SuperScript III reverse transcriptase (RT) (Invitrogen) with 250 ng of random primers (Invitrogen). Quantitative RT-PCR (qRT-PCR) was carried out on a 7500 fast real-time PCR system (Applied Biosystems) using fast SYBR green master mix (Applied Biosystems). RNA without RT (RT−) was analysed alongside cDNA (RT+). Standard curves were performed for each gene analysed, and the quantities of cDNA within the samples were calculated from cycle threshold values. Data were averaged, adjusted for chromosomal DNA contamination (RT+ minus RT−), and normalized to corresponding sigA or 16S values.
## Northern blots
Northern blots were carried out as described previously [bib_ref] Identification of small RNAs in Mycobacterium tuberculosis, Arnvig [/bib_ref]. Unless otherwise stated 5 μg of total RNA were loaded on a 6% denaturing acrylamide gel. Gels were electrophoresed at 12 W for 2 h and electroblotted onto BrightStar-Plus membrane (Ambion). After air-drying, RNA was cross-linked to the membrane by UV irradiation. Membranes were stained in 0.3 M sodium acetate containing 0.03% methylene blue and incubated overnight with labelled probes in ULTRAhyb (Ambion). After washing, membranes were exposed to phosphorimaging and changes in RNA expression were determined by densitometer-scanning of Northern blots. Transcript sizes were compared with RNA marker low from Abnova (20-500 nucleotides).
## Rnaseq
RNA was isolated as previously described [bib_ref] Sequence-based analysis uncovers an abundance of non-coding RNA in the total transcriptome..., Arnvig [/bib_ref] and treated with Turbo DNase (Ambion) until DNA free. The quality of RNA was assessed using a Nanodrop (ND-1000, Labtech) and Agilent RNA chip (2100 Bioanalyser). Total RNA (2-3 μg) was fragmented (Ambion Cat # AM8740) and strand-specific cDNA libraries were constructed using the Illumina directional mRNA-Seq protocol (Part # 15018460 Rev. A) but with exclusion of poly-A tail and size selection to capture all RNA species. Briefly, this protocol ligates the Illumina v1.5 small RNA 3′ adapter followed by a 5′ adapter to preserve strand specificity. Single-end read sequencing was performed on HiSeq 2000 sequencer. Quality of the Illumina produced fastq files was assessed and good quality reads were mapped to the reference sequence of M. tuberculosis H37Rv (EMBL accession code AL123456) as single end data using BWA [bib_ref] Fast and accurate short read alignment with Burrows-Wheeler transform, Li [/bib_ref]. Genome coverage, defined as number of reads mapped per base of H37Rv genome, was calculated using BEDTools [bib_ref] BEDTools: a flexible suite of utilities for comparing genomic features, Quinlan [/bib_ref]. RPKM values (reads per kilobase per million reads) were calculated using only sequence reads that mapped to annotated features unambiguously and on the correct strand. For whole transcriptome differential expression calling, genome coverage of reads mapping to genes, antisense and ncRNAs were used for statistical testing using DESeq [bib_ref] Differential expression analysis for sequence count data, Anders [/bib_ref]. Differentially expressed genes were considered when fold changes were greater or equal than twofold and the corresponding adjusted P-value (Padj) was less than 0.05. RNA-seq data have been deposited in ArrayExpress under Accession No. E-MTAB-1667.
3′ RACE. To map the 3′ sites of ssrA transcripts, 3′ RACE (rapid amplification of cDNA ends) was performed using a Gene Racer kit (Invitrogen) according to the manufacturer's instruction. Five micrograms of DNA-free RNA was poly-A tailed using E. coli poly-A Polymerase (Ambion). cDNA was synthesized using SuperScript III and oligo d(T) primer. cDNA was used as a template for PCR with Phusion HF MasterMix (Finnzymes) using GR3′ (linker specific) and a gene-specific forward primer (TaqSsrAb F). PCR products were separated on a 2% agarose gel and bands of interest cut out, cloned into pCR-II TOPO (Invitrogen) and sequenced. 3′ ends of transcripts were identified as the junction with the poly-A tail.
[fig] Figure 1: HigB expression in M. smegmatis. A. Genomic organization of the M. tuberculosis HigBA locus. B and C. M. smegmatis strains carrying vector control ('Vector') or the HigB expression plasmid ('HigB') grown in the absence (−) or presence (+) of anhydrotetracycline (ATc). (B) Quantitative RT-PCR of higB expression relative to 16S rRNA. RNA was isolated from exponential-phase cultures. (C) Growth of M. smegmatis strains. [/fig]
[fig] Figure 2: HigB expression in M. tuberculosis. M. tuberculosis strains carrying vector control ('Vector') or the HigB expression plasmid ('HigB') grown in the absence (−) or presence (+) of ATc. [/fig]
[fig] Figure 4: HigB expression affects tmRNA in M. smegmatis and E. coli. [/fig]
[fig] Figure 5: HigB cleavage sites identified by 3′ RACE. A schematic showing the DNA sequence of ssrA. Amino acids encoded in the mRNA like region are shown. [/fig]
[table] Table 1: RNAseq data comparing fold change between the HigB expressing and the vector control strains. [/table]
|
The Utilization of National Tobacco Cessation Services among Female Smokers and the Need for a Gender-Responsive Approach
# Introduction
Worldwide, there have been more countries that have achieved a significantly greater decrease in smoking prevalence among men than women. Although the prevalence of male tobacco use still remains higher than that of females globally, there has only been a minimal decline in female smoking and even an increase in many low-and middle-income countries. In Korea, while the smoking rate has dropped from 66.3% to 36.7% in adult males in the past 20 years, the adult female smoking rate has increased from 6.5% in 1998 to 7.5% in 2018. In particular, the prevalence of smoking among women in their 20s-40s has more than doubled from 4.4% to.9% in the past 20 years in Korea. Considering the cultural background of social disapproval of female smoking in Asian countries, the actual smoking rate among Korean women may be underestimated due to the under-reporting of smoking status.
One of the key strategies to reduce demand for tobacco provided by the World Health Organization (WHO) Framework Convention on Tobacco Control (FCTC) is contained in Article 14: the demand reduction measures concerning tobacco dependence and cessation. In 2019, the WHO discussed "Offer help to quit tobacco use" in their periodic reports on the global tobacco epidemic. As of 2018, 32% of the global population including Koreans was covered by comprehensive national tobacco cessation support. This bestpractice adoption of cessation included three elements: (1) cost coverage of cessation advice in clinical or community settings, (2) cost coverage of nicotine replacement therapy and (3) national toll-free quit lines.
Smoking cessation and smoking cessation treatments are regarded as a cost-effective way to reduce smoking-induced morbidity and mortality. Combining behavioral support for tobacco cessation from an expert and pharmacotherapy increases the likelihood of successful cessation compared with using no cessation support. Previous research has addressed factors associated with the utilization of smoking cessation services including insurance status, receiving advice from healthcare professionals, a diagnosis of mental health and the self-efficacy to quit.
Studies have shown gender differences in smoking cessation and that women find it more difficult to quit tobacco use than men. Compared with men, women show a lower utilization of smoking cessation services. Due to stigma-related barriers, female smokers tend to be reluctant to reveal their smoking status, resulting in less usage of cessation support. Regarding user characteristics of national smoking cessation services in Korea, female smokers reported more usage of web-and telephone-based cessation services, which can provide more privacy than public health center-based smoking cessation clinics.
To prevent an increase in female smoking prevalence, gender-responsive measures for tobacco control are necessary particularly in the changing social context regarding gender. The FCTC recommends incorporating gender into all measures for tobacco control, to increase access to health services for tobacco dependence among marginalized populations including women and to train healthcare providers to consider sex and gender specificities. However, there is insufficient evidence that tobacco control has been sufficiently gender-responsive. Calls have been made to systematically integrate the analysis of sex and gender in program planning.
Thus, the aim of this study is to examine factors associated with the utilization of national smoking cessation services and to explore the gender-specific characteristics of service users to develop effective recruitment strategies for national cessation services.
# Materials and methods
## National tobacco cessation services
In accordance with its guidelines for the implementation of FCTC Article 14, quitting support strategies and tobacco treatment infrastructure, the Korean government launched the first nationwide public health center-based tobacco cessation services in 2005. In 2006, the government started to provide a toll-free quit line offering telephone counseling. Strengthened cessation support was initiated in 2015 including health insurance covering tobacco dependence treatment and outreach services for socially marginalized smokers such as women. These comprehensive nationwide tobacco cessation services enabled a significant decrease in smoking prevalence in men from 66.3% in 1998 to 36.7% in 2018.
## Data collection
From an online panel of a professional research company, namely, the Hyundae Research Institute, 708 Korean female smokers over the age of 19 were sampled. A proportional quota sampling method was used when recruiting the participants to ensure that the respondents demographically represented the general population with quotas based on female smoking rate by age group. Reference data were derived from the 2018 Korea National Health and Nutrition Examination Survey, which was based on the Korean population and housing census. Emails and text messages with a link to a survey questionnaire were sent to random panel members. The research panel included both males and females of all age groups and smokers and those who had never smoked. A nationwide web survey was conducted from 4-9 December 2020. At the beginning of the web survey, Yes or No questions asking the respondents whether they had used any tobacco product in the past five years, whether their age was over 19 and whether they were female were added. With the disqualification logic, the survey only included adult female respondents who had used any tobacco product at least once within the past five years. We used a self-designed questionnaire based on a previously validated nationwide health and tobacco-related survey. Informed consent was obtained from each participant during the survey.
## Measures
## The utilization of national tobacco cessation services
National tobacco cessation services provided by the Ministry of Health and Welfare include smoking cessation clinics at 256 public health centers, national Quitline services, residential five-day programs (regional smoking cessation centers), smoking cessation outreach services (regional smoking cessation centers) and health insurance-covered smoking cessation treatment services. The participants who selected any of these services to the question "Which smoking cessation service have you experienced?" were defined as users of national smoking cessation services.
## General characteristics
Demographic variables such as age, occupation, marital status, living arrangement and history of pregnancy and childbirth were included. We categorized occupations as (1) non-manual workers (general managers, professionals and office workers), (2) sales/service workers, (3) manual workers (agriculture, fishery and forestry workers, machine operation and assembly workers and craft and related trade workers) and (4) others (students, homemakers and unemployed). Marital status was categorized as (1) single and (2) married (including separated/divorced/widowed). Living arrangements were measured by asking whom they were living with and the response options were partners, children (son/daughter), grandchildren, parents, brothers/sisters, son/daughter-in-law, grandparents, relatives, friends and living alone. We then classified living arrangements into two groups: living alone for those who answered "living alone" and those who did not live alone for the rest of the response options. History of pregnancy and childbirth included former experience of (1) pregnancy and childbirth, (2) pregnancy only and (3) no history.
## Mental health
Depression was assessed using the following question: "Have you ever had feelings of sadness or hopelessness every day for two weeks during the last twelve months?". The response options were yes and no. Perceived stress was measured using the following question: "How would you rate your level of usual stress?". The responses included five options, namely, "very high", "high", "moderate", "low" and "none" and were regrouped into three categories: "very high/high" as "high", "moderate" and "low/none" as "low".
# Tobacco-related characteristics
In this study, smokers were defined as individuals who had used any tobacco product once in their life. The survey only included respondents who had used a tobacco product in the past five years among smokers. Tobacco-related characteristics included the age of smoking initiation, current use of tobacco products, current status of poly tobacco users, smoking status of people around the participants and receipt of advice to quit smoking. Current smokers were defined as individuals who had used a tobacco product in the past 30 days. The respondents were asked to report the tobacco products they currently used. We then classified the respondents as a current cigarette, electronic cigarettes (ecigarettes) and heated tobacco product (HTP) user from answers if they reported any current use of cigarettes, e-cigarettes and HTPs. Additionally, the respondents were classified as current multiple tobacco product users if they reported the current use of two or more tobacco products. The smoking status of people around the participants was measured by asking whether their parents, partners and friends were smokers and the response options included yes and no. The participants were asked whether they had someone around them giving them advice to quit smoking. The response options included parents, brothers/sisters, spouse/lover, children/grandchildren, friends/upper and lower classmates, colleagues, teachers, medical personnel, others and none. The responses were regrouped into two categories: "none" as "no" and the rest as "yes".
## Motivation ruler for tobacco cessation
The motivation rulerwas assessed by three 10-point rulers related to the motivation to quit smoking: (1) the importance ruler: "How important is smoking cessation to you?" (0 = not at all important, 10 = 100% important); (2) the readiness ruler: "How ready are you to quit smoking?" (0 = not at all ready, 10 = 100% ready) and (3) the confidence ruler: "How confident are you that you will quit smoking? (0 = not at all confident, 10 = 100% confident)".
## Concerns about weight gain
Post-cessation weight gain concerns were measured using a 10-point ruler as a visual aid: "Are you worried about gaining weight as you quit smoking? (0 = not at all worried, 10 = 100% worried)".
## The need for gender-specific smoking cessation services
Regarding the need for gender-specific smoking cessation services, the survey participants were asked the following questions by using a five-point Likert scale: "Do you think that smoking cessation counselors should provide smoking cessation support that takes into account female characteristics during individual counseling? (1 = strongly disagree, 5 = strongly agree).
## Data imputation
Of the total 708 sampled participants, one participant with an inconsistent response was excluded. Of the 707 participants, the subjects with missing values were replaced using a multiple imputation procedure with the assumption that data are missing at random. A simulation technique such as Markov Chain Monte Carlo, which generates random draws from non-standard distributions via Markov chains, was used for multiple imputation. The following variables of missing data were included: the utilization of national tobacco cessation services, current cigarette use, current e-cigarette use, current HTP use, current multiple tobacco product use, intention to quit, three motivation rulers (importance, readiness, confidence) and needs for gender-specific smoking cessation services. Of the 707 participants, 44 participants who were not aware of national tobacco cessation services were excluded from the analysis. A total of 663 female smokers were included for the final sample in the analysis.
# Statistical analysis
We first conducted a descriptive analysis to examine the general and smoking-related characteristics of the survey participants. A Pearson's chi-squared test and t-tests were used to test the statistical significance of the differences in the use of smoking cessation services. In this step, the properties that were to be included in the multiple regression model were determined. Parameters with a p-value below 0.05 were included in the logistic regression model. We then performed simple and multiple logistic regression analyses to identify statistically significant factors of service utilization with an adjustment for confounding variables. The second model added two interaction terms: one between current HTP use and age and one between current multiple tobacco product use and age. A multinomial regression was used to evaluate the factors that affected the number of utilized service types. The results from the logistic regression analysis were presented as odds ratios (ORs) with 95% confidence intervals (95% CI). The motivation rulers for tobacco cessation, weight gain concerns and the need for a gender-specific smoking cessation program among different age groups were evaluated with a one-way analysis of variance with a post-hoc Scheffé's test for multiple comparisons. Regarding the normality of all of the variables in the analysis, none of the variables were highly skewed or kurtotic. The absolute skewness and kurtosis of the variables larger than 2 and 7 were considered non-normal, respectively. An alpha level of 0.05 was considered as the cutoff for significance. R statistical software version 4.0.4(R Foundation for Statistical Computing, Vienna, Austria) and SPSS version 27.0 (SPSS, Inc., Chicago, IL, USA) were used for all analyses.
# Results
## Demographic and smoking-related characteristics of the survey participants
The basic characteristics of the study participants are shown in. In this study, 89.0% (590 of 663) of the respondents had never used national smoking cessation services and 11.0% (73 of 663) of the respondents had used one. Regarding marital status among smokers, users of smoking cessation services were more likely to be married than single (p = 0.002). Moreover, those who had a history of both pregnancy and childbirth reported a significantly greater use of smoking cessation services (69.9% vs. 46.4%, p < 0.001). Compared with those without the experience of government-supported smoking cessation services, those with experience expressed more depressive moods. We found that the current use of heated tobacco products (HTPs) and multiple tobacco products were more prevalent among national service users (p = 0.003 and p = 0.001). In addition, users of smoking cessation services were more likely to have smoking parent(s) than those who had never used. With regard to smoking cessation advice, service users received more quitting advice than those who had never used (93.2% vs. 83.4%, p = 0.030).
## Logistic regression analysis of factors associated with the utilization of national tobacco cessation services
A simple logistic regression analysis revealed a statistically significant association between those who had ever used national smoking cessation services and a history of pregnancy and childbirth (OR = 2.60, 95% CI: 1.51-4.47). Those who reported depressive moods had more than twice the odds of using smoking cessation services (OR = 1.94, 95% CI: 1. . Smokers currently using HTPs had an OR of 2.10 (95% CI: 1.29-3.42) for using national smoking cessation services. Current multiple tobacco product use was associated with twice higher odds of using national smoking cessation services (OR = 2.21, 95% CI: 1.35-3.63). Compared with smokers with non-smoking parents, the odds of using national smoking cessation services increased among smokers with smoking parents (OR = 1.82, 95% CI: 1.10-3.01). Among smokers who had received advice to quit smoking, the likelihood of using government-supported smoking cessation services was increased (OR = 2.71, 95% CI: 1.06-6.89). Age, marital status, history of pregnancy and childbirth, depression, current HTP use, multiple tobacco product use, parent(s) smoking status and the receipt of advice to quit were included in Model 1 as confounding factors. After adjusting for confounding variables, a statistically significant association between the utilization of smoking cessation services and depression and parental smoking status was revealed. In Model 2, there were no significant interactions between current HTP use and age and current multiple tobacco product use and age.
## Multinomial logistic regression analysis of factors associated with the number of utilized service type
In, a multinomial logistic regression analysis revealed that reporting depressive moods was associated with twice higher odds of using a single type of tobacco cessation support (OR = 2.06, 95% CI: 1.15-3.69). Among smokers whose partner was a smoker, the likelihood of using more than two types of national tobacco cessation service was increased (OR = 3.24, 95% CI: 1.06-9.87).
## The survey participants' motivation ruler for smoking cessation and concern about weight gain
In terms of motivation rulers for tobacco cessation, users of smoking cessation services provided by the government tended to report significantly higher importance than those who had never used. Post-hoc tests revealed that those in their 50s displayed a higher importance than those in their 20s (p = 0.003, Scheffé's test). Users of smoking cessation services reported more weight gain concerns when quitting smoking than those who had never used. Those in their 20s indicated the lowest weight gain concerns when quitting smoking among all age groups . . The survey participants' motivation ruler for smoking cessation and concern about weight gain (n = 579). Results are shown as mean (± standard deviation) or number (%). * Significance at p < 0.05. † Motivation rulers (importance, confidence, readiness) and weight gain concerns were assessed by an 11-point Likert scale (0 = not at all concerned to 10 = very concerned). a,b Values not sharing the same uppercase letter in a row indicate significant statistical differences among groups based on Scheffé's test.
## Variables national tobacco cessation services
## The survey participants' need for gender-specific smoking cessation services
When asked about the need for gender-specific services for smoking cessation, users of government-supported smoking cessation programs reported higher mean scores than those who had never used. Among the different age groups, those in their 20s and 40s reported the lowest and highest mean scores, respectively. Need for gender-specific smoking cessation services † 3.8 (± 0.8) 4.0 (± 0.7) 3.6 (± 0.9) < 0.001 3.5 (± 0.9) a 3.6 (± 0.9)
# Discussion
To our knowledge, this nationwide study is the first to report the factors associated with the use of government-supported smoking cessation services among female smokers in Korea. We observed a few characteristics associated with the utilization of national smoking cessation services. A few of these factors included general characteristics (history of pregnancy and childbirth, depression) and tobacco-related characteristics (current use of HTPs, current use of multiple tobacco products, parental smoking status, receiving advice to quit smoking). This study showed the differences regarding weight gain concerns among users and those who had never used tobacco cessation support. Our study also provided female smokers' opinions regarding the need for the implementation of gender-specific and gender-sensitive interventions.
This study showed that service users of national tobacco cessation support were more likely to have a history of pregnancy and childbirth. Studieshave shown that pregnancy and parenting provide a powerful motivation for many female smokers to quit smoking. Mothers of young children tend to be more responsive to interventions for smoking cessation. In a qualitative studyconducted among Korean female smokers, younger smokers reported a rather opposed attitude to current smoking cessation services to emphasize the harmful effects of smoking related to pregnancy and childbirth. They recognized that women seem to be considered only within the context of reproductive health; only for giving birth, rather than as individuals equal to men. When providing a gender-responsive tobacco cessation program, whether current pregnancy-and childbearing-focused programs could be suitable for all age groups should be considered.
Our findings that those with depressive symptoms were associated with increased odds of using national tobacco cessation support may be explained in line with previous studies. A four-country International Tobacco Control studyobserved that female smokers with depressive symptoms made more attempts to quit but experienced more smoking relapses. A depressive mood has been identified as part of nicotine withdrawal. Nicotine in tobacco products binds to α4β2 nicotinic acetylcholine receptors in the brain and triggers the release of the neurotransmitter dopamine, which helps to create a good feeling. Once a smoker stops smoking and the brain lacks dopamine, a low mood may be experienced. Moreover, as studies have found sex-specific differences in cravings to smoke when experiencing negative mood situations, attention to psychological and emotional factors when counseling women seems to be important. To provide effective gender-responsive tobacco cessation support, different stimulations and concerns regarding smoking need to be considered in both genders.
In our analysis, the participants who used the national tobacco cessation support were associated with the current use of HTPs. A possible explanation for the greater odds of using cessation support in current HTP users may be that these products are considered smoking cessation aids. A recent studyamong Korean adolescents revealed that HTP use among current combustible cigarette smokers was associated with higher odds of having attempted to quit compared with exclusive cigarette users. Additionally, current multiple product users were more likely to report the use of the national cessation support in this study. Previous studies among Koreanand USadults found that past quitting attempts were associated with poly-tobacco use. Given that a considerable proportion of HTP users are dual or multiple users, efforts should be made for this motivation towards smoking cessation to be connected to successful complete cessation.
In this study, users of national smoking cessation services were more likely to have smoking parents. Parental smoking has been found to be a predictor of smoking in children in previous literature. Adolescents who have smoking parents are more likely to initiate smoking at an earlier age, which can lead to problematic smoking trajectoriesrequiring smoking cessation support. There is another possibility that parents who struggle to quit smoking may have negative attitudes toward tobacco smoking, which can be relayed to their children. In addition, as our study population was over the age of 19, the role of parental smoking in adult children may be different from that in adolescence. Therefore, a further analysis is needed regarding the relationship between parental smoking status and adult children's usage of smoking cessation support.
Quitting advice has been identified as an important determinant of the utilization of smoking cessation services. Our study found that smokers who received quitting advice from people in close relationships or healthcare professionals showed an increased use of smoking cessation services. Quitting advice from a healthcare professional has been reported to promote smoking cessation attempts and increase the possibility of successful cessation. In a studyon spouses' support related to tobacco cessation, smokers who maintained successful abstinence for one, three and six months reported a higher frequency and perceived helpfulness from partner interactions. Data from the US National Adult Tobacco Survey demonstrated that healthcare professional advice was correlated with a higher utilization of a Quitline.
Documented studieshave shown that many female smokers have fears related to post-cessation weight gain. Our data showed that smokers with concerns about gaining weight requested more smoking cessation support. A meta-analysisof 62 studies revealed that an average 4-5 kg increase in body weight after 12 months of abstinence was associated with tobacco cessation. Smokers with a higher nicotine dependence are more likely to gain weight. In addition, compared with younger women, older women are known to gain more weight when quitting tobacco products. As weight gain concerns varied among different age groups in this study, a customized and targeted approach would be helpful when counseling for weight management in female smokers.
The participants in this study were asked to evaluate their agreement on the need for individualized gender-specific tobacco cessation support in counseling. Smokers with experience with a national smoking cessation service expressed more agreement on the need for a gender-specific approach in this study. Evidencesuggests that the needs of female smokers for smoking cessation programs include choosing from a diverse range of services and female-specific educational topics. In our data, the participants in their 20s were less in favor of female-focused smoking cessation services than those in their 40s. In a qualitative study, smokers in their 20s expressed resentful concerns about current pregnancy-and childbirth-focused approaches for tobacco cessation in female smokers. Tobacco cessation support developed for women had put specific emphasis on smoking during pregnancy. Although this is crucial for both the woman and the fetus, it became necessary to develop broader programs to support tobacco cessation for women throughout their lives. Tailored smoking cessation interventions that consider individual challenges in a life cycle can be beneficial. Future research is warranted regarding the perception of specific gender-responsive approaches in different age groups.
# Limitations
Our study has several limitations. First, the sample of the present study was obtained from a research panel, thus limiting its generalizability. While efforts were made to ensure that the survey sample represented the population with a proportional quota sampling design, the sample may not be representative of the overall population. Thus, the statistical inference in this study was for the sample not the general population. Given the findings of this study, further research with a nationally representative sample is needed to explore gendered factors associated with the usage of tobacco cessation support. Second, the relatively high prevalence of usage of electronic nicotine delivery systems and HTP can be explained by the inclusion criteria in this study. As this study only included panels who had used a tobacco product in the past five years in order to reflect on newly developed national tobacco cessation services, the usage rates of the new and emerging tobacco products might be higher than that of general population. In addition, the online survey method may be related to a higher prevalence of female smoking than can be found in general by ensuring respondents' anonymity. The study results therefore need to be interpreted with caution. Finally, regarding the nature of the observational study design, recall bias could occur when the participants were asked about their past experiences. Future research with varied study designs will be helpful to understand the characteristics of female tobacco product users with the intention to quit. Despite the study's limitations, using nationwide survey data among female smokers with a quantitative research design to examine the use of national tobacco cessation services was a notable strength.
# Conclusions
This study suggested several factors associated with the use of national health services. The results could be further used to examine what promoted and hindered the use of national tobacco cessation services among female smokers. Future studies on these related factors will be crucial to improve our understanding of the specific needs for developing gender-sensitive tobacco control policies. Informed Consent Statement: Informed consent was obtained from all participants involved in the study.
# Data availability statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available to protect confidentiality of the research participants. |
An Open‐Label Study to Assess the Effect of Itraconazole and Rifampin on Parsaclisib Pharmacokinetics When Administered Orally in Healthy Participants
Parsaclisib, a selective, potent phosphatidylinositol 3-kinase delta inhibitor being developed for the treatment of cancer and autoimmune diseases, is primarily metabolized by cytochrome P450 (CYP) 3A4. This study assessed the pharmacokinetics (PK) and safety of parsaclisib alone or combined with itraconazole (potent CYP3A inhibitor) or rifampin (potent CYP3A4 inducer) in healthy participants. In this open-label, fixed-sequence study, cohort 1 received oral parsaclisib 10 mg once daily on days 1 and 8 and oral itraconazole 200 mg once daily on days 4-11; cohort 2 received oral parsaclisib 20 mg once daily on days 1 and 11 and oral rifampin 600 mg once daily on days 4-12. Parsaclisib plasma concentration was tested and PK parameters calculated by noncompartmental analysis. Geometric mean ratios (GMRs) and 2-sided 90% confidence intervals (CIs) were estimated by 2-factor analysis of variance. Thirty-six healthy participants were enrolled (18 per cohort). Parsaclisib maximum plasma drug concentration (C max ) and area under the concentration-time curve extrapolated to infinity (AUC 0-∞ ) were increased by 21% and 107% with concomitant itraconazole versus parsaclisib alone (GMR, 1.21; 90%CI, 1.14-1.29; and 2.07; 90%CI, 1.97-2.17, respectively). Parsaclisib C max and AUC were reduced by 43% and 77%, respectively, with concomitant rifampin versus parsaclisib alone (GMR, 0.57; 90%CI, and 0.23; 90%CI, respectively). Headache was the most common adverse event, reported by 13.9% of participants (all in cohort 2). Single-dose parsaclisib alone or combined with itraconazole or rifampin appeared safe and well tolerated in healthy participants. Parsaclisib dose adjustment may be necessary with concomitant administration of strong CYP3A4 inhibitors or inducers.
Dysregulation of phosphatidylinositol 3-kinase delta (PI3Kδ) is implicated in malignant B-cell activation, proliferation, and survival. 1,2 Parsaclisib (INCB050465) is a next-generation, potent inhibitor of the human PI3Kδ kinase enzyme (whole-blood half-maximal inhibitory concentration [IC 50 ] = 10 nM), with approximately 20 000-fold selectivity relative to other PI3K isoforms. [bib_ref] Preliminary safety, efficacy, and pharmacodynamics of a highly selective PI3Kδ inhibitor, INCB050465,..., Forero-Torres [/bib_ref] Preclinical models demonstrate the efficacy of parsaclisib,and clinical studies support the benefit of parsaclisib in relapsed or refractory B-cell malignancies. [bib_ref] Preliminary safety, efficacy, and pharmacodynamics of a highly selective PI3Kδ inhibitor, INCB050465,..., Forero-Torres [/bib_ref] [bib_ref] Parsaclisib, a potent and highly selective PI3Kδ inhibitor, in patients with relapsed..., Forero-Torres [/bib_ref] Parsaclisib is currently under investigation as monotherapy or in combination in a number of clinical trials in patients with B-cell malignancies, solid tumors, and autoimmune disorders.
Human pharmacokinetic (PK) studies with parsaclisib showed a time to maximum plasma drug concentration (t max ) of 0.5 to 1 hour and apparent plasma terminal phase disposition half-life ,β ) of 8.6 to 11.5 hours. [bib_ref] Parsaclisib, a potent and highly selective PI3Kδ inhibitor, in patients with relapsed..., Forero-Torres [/bib_ref] Doses of up to 45 mg once daily have been evaluated in cancer patients, with doses > 5 mg once daily achieving exposure levels exceeding 90% of maximal inhibitory concentration throughout the dosing interval. [bib_ref] Parsaclisib, a potent and highly selective PI3Kδ inhibitor, in patients with relapsed..., Forero-Torres [/bib_ref] No time-dependent PK or sex differences in exposure were observed in humans. Although higher exposure was observed in Japanese patients because of low body weight, body weight-normalized clearance was comparable in white and Japanese patients. In addition to the PK profile allowing a favorable dosing regimen, the molecular structure of parsaclisib was designed to limit off-target Cohort 2: with rifampin (n = 18) D4-10 D11* D12 . Study treatment schedule. D, day; QD, once daily. *PK parameter results for 1 participant on day 11 of cohort 2 were excluded from the statistical analysis based on the rifampin analysis results, which indicated the participant had not taken rifampin as scheduled.
hepatotoxicity seen with first-generation PI3Kδ inhibitors (eg, idelalisib and duvelisib). [bib_ref] Management of adverse events associated with idelalisib treatment: expert panel opinion, Coutré [/bib_ref] [bib_ref] Duvelisib, a novel oral dual inhibitor of PI3K-δ,γ , is clinically active..., Flinn [/bib_ref] In vitro experiments demonstrated that parsaclisib is not likely a substrate of human OATP1B and OATP1B3, and at clinically relevant exposure, the potential of drug-drug interaction (DDI) via inhibition of human uptake transporters OATP1B1, OATP1B3, OAT1, OAT3, OCT2, MATE1, or MATE2K is low (data on file, Incyte Corporation). Parsaclisib is a P-glycoprotein (P-gp) substrate (efflux transport saturated at >100 μM) and a P-gp inhibitor, with IC 50 = 18.1 μM (data on file, Incyte Corporation). However, the potential for a DDI is low with P-gp substrates or inhibitors based on the therapeutic clinical dose, the expected plasma concentrations of parsaclisib at that dose, its inhibitory potency for P-gp, and the decision tree in the U.S. Food and Drug Administration (FDA) guidance.In vitro studies with recombinant human cytochrome P450 (CYP) isozymes have determined that parsaclisib is primarily metabolized by CYP3A4, is not an inhibitor of major CYPs tested, and is not a CYP3A4 inducer. In experiments using human liver microsomes and selective CYP chemical inhibitors, ketoconazole (a potent CYP3A4 inhibitor) inhibited parsaclisib metabolism (data on file, Incyte Corporation). In an in vitro study, parsaclisib metabolism was similar across tested species (rat, dog, and human), with no human-specific metabolite identified. Phase 1 metabolites identified from in vitro preparations were primarily oxidative, and no major metabolites were detected in clinical plasma samples (data on file, Incyte Corporation).
Itraconazole is a potent inhibitor of CYP3A, whereas rifampin is a potent inducer of CYP3A4. [bib_ref] Pharmacokinetic interactions with rifampicin: clinical relevance, Niemi [/bib_ref] [bib_ref] Itraconazole and clarithromycin as ketoconazole alternatives for clinical CYP3A inhibition studies, Ke [/bib_ref] Therefore, inhibition of parsaclisib metabolism by itraconazole has the potential to increase exposure of parsaclisib, whereas induction of parsaclisib metabolism by rifampin has the potential for reduced exposure. We conducted a clinical trial in healthy participants to evaluate the effects of CYP3A4 inhibition and induction on parsaclisib metabolism via CYP3A enzymes. The primary objective of this study was to assess the effect of itraconazole and rifampin on parsaclisib PK. The secondary objective was to evaluate the safety and tolerability of parsaclisib when administered alone or in combination with itraconazole or rifampin.
# Methods
## Study design and drug treatment
The protocol for this study was reviewed and approved by a qualified institutional review board (IRB), Chesapeake IRB (now Advarra, Columbia, Maryland). The study was performed in accordance with the International Council for Harmonisation guideline for Good Clinical Practice, including the Declaration of Helsinki and local ethical and legal requirements. Participants signed an institutional review board/independent ethics committee-approved informed consent form prior to study entry. This study was conducted at 1 study center (Celerion, Tempe, Arizona).
This was an open-label, fixed-sequence DDI study to assess the effect of multiple doses of itraconazole or rifampin on the single-dose PK of parsaclisib. The study consisted of a screening period (up to 28 days), a treatment period (11 days for cohort 1, 12 days for cohort 2), and a posttreatment follow-up period (at least 30 [+6] days after the last dose of parsaclisib). During the treatment period , in cohort 1, parsaclisib 10 mg (5 mg × 2) was administered as a single dose in the fasted state (8 hours before and 4 hours after parsaclisib administration) on day 1, followed by itraconazole 200 mg (100 mg × 2) once daily in the fed state on days 4 to 7. On day 8, both parsaclisib and itraconazole were administered in the fasted state, followed by itraconazole in the fed state on days 9 to 11. In cohort 2, parsaclisib 20 mg was administered as a single dose in the fasted state (8 hours before and 4 hours after parsaclisib administration) on day 1, followed by rifampin 600 mg (300 mg × 2) once daily in the fasted state (8 hours before and 1 hour after rifampin administration) on days 4 to 10. On day 11, both parsaclisib and rifampin were administered in the fasted state (according to parsaclisib fasting criteria), followed by rifampin in the fasted state on day 12. All drugs were administered orally with approximately 240 mL of water.
Parsaclisib (INCB050465) 5-and 20-mg tablets were manufactured by Xcelience (Tampa, Florida). Itraconazole 100-mg capsules were manufactured by Mylan Pharmaceuticals Inc. (Canonsburg, Pennsylvania). Rifampin capsules, USP 300 mg, were distributed by Lannett Company, Inc. (Philadelphia, Pennsylvania).
## Participants
Eligible participants were healthy adults between 18 and 55 years of age at the time of screening, with a body mass index between 18 and 32 kg/m 2 inclusive. Exclusion criteria included current or history of a clinically significant disorder, such as renal (estimated glomerular filtration rate ≤ 80 mL/min/1.73 m 2 ), hepatic, or metabolic impairment. Use of medication within 7 days before study entry, including CYP and P-gp inhibitors and inducers, was not allowed. Conditions present at the time of informed consent were recorded on the Medical History Form in the electronic Case Report Form.
# Bioanalytical methods
Blood samples were collected at scheduled times in the treatment period after each parsaclisib administration, with or without itraconazole (cohort 1) or rifampin (cohort 2) coadministration, to determine plasma concentrations of parsaclisib. For cohort 1, PK samples were collected on study visit day 1 (predose, then 0. Plasma samples were shipped to Incyte Corporation (Wilmington, Delaware) for assessment of parsaclisib plasma concentrations using a validated liquid chromatography with tandem mass spectrometry (LC-MS/MS) method. A 50-μL aliquot plasma sample was placed into respective tubes in a 96-well format. After the addition of the internal standard (INCB050904, a structural analogue of parsaclisib [Incyte Research Institute]), an aliquot of 100 μL of 0.1 M NaHCO 3 was added. Then, 800 μL of methyl tert-butyl ether (MTBE) was added, and the samples were vortexed. After centrifugation, 150 μL of MTBE layer was transferred to clean tubes in a 96-well format by a Tomtec liquid handler (Hamden, Connecticut). The samples were then dried under nitrogen and reconstituted with 250 μL of reconstitution solution (acetonitrile:water, 50:50, v/v) and rigorously mixed. The plates were placed in the autosampler tray, and 1 μL of the sample was injected into an LC-MS/MS system, which was composed of binary high-performance liquid chromatography pumps and an autosampler coupled to a SCIEX 6500 QTRAP tandem mass spectrometer (Framingham, Massachusetts). The mass spectrometer was operated in positive electrospray ionization mode, and the multiple reaction monitoring was m/z 433.2 → 150.2 for parsaclisib and m/z 440.3 → 150.2 for the internal standard.
The assay range was 5 to 5000 nM with a 10-fold dilution factor verified. Calibration standards (including 10 concentrations ranging from 5 to 5000 nM) were prepared in control blank human K2-ethylenediaminetetraacetic acid (EDTA) plasma by spiking with stock solutions of parsaclisib (Incyte Research Institute) prepared in acetonitrile:water (50:50, v/v). Quality control samples were also prepared in control blank human K2-EDTA plasma at 15, 250, and 4000 nM. The assay was confirmed for lack of interference from rifampin and itraconazole prior to sample analysis. An exploratory analysis to quantify the other study drugs was performed using residual plasma samples.
## Safety assessments
New or worsening adverse events (AEs) were recorded, and participants were monitored for new AEs during the posttreatment follow-up period. A new or worsening AE occurring after the first dose of study drug was considered a treatment-emergent adverse event (TEAE). AEs were tabulated by the Medical Dictionary for Regulatory Activities preferred term and system organ class. Severity of AEs was described and graded using a FDA toxicity grading scale.
## Study end points
The primary study end points were parsaclisib maximum plasma drug concentration (C max ) and area under the concentration-time curve (AUC) extrapolated to infinity (AUC 0-∞ ). Secondary PK end points were parsaclisib t max , t1 2 ,β , AUC up to the last measurable concentration (AUC 0-t ), oral dose clearance (CL/F), and apparent oral dose volume of distribution (V z /F). Additional secondary end points included monitoring AEs and vital signs, physical examinations, 12-lead electrocardiograms (ECGs), and clinical laboratory blood and urine sample assessments.
## Statistical methods and planned analyses
The sample size of 18 participants per cohort was chosen to enable evaluation of the magnitude of any changes in parsaclisib PK when administered with a potent CYP3A4 inhibitor or inducer. A 0.05 (2-sided) significance level and 90% confidence interval (CI) were used to test all hypotheses for PK. The PK-evaluable population included participants who received at least 1 dose of study drug and provided at least 1 sample for PK analysis. The PK data were described by descriptive statistics for continuous and categorical variables. Parsaclisib plasma concentration data were analyzed using standard noncompartmental PK methods with Phoenix WinNonlin v8.0 software (Certara, Princeton, New Jersey). Comparison of the log-transformed PK parameters among the treatments was performed using a 1-way, 2-factor analysis of variance (ANOVA) with treatment as the fixed factor and participant as the random factor. Parsaclisib geometric mean ratios (GMRs) and 90%CIs of C max , AUC 0-t , and AUC 0-∞ were calculated based on the least-squares means from the ANOVA. The statistical significance of median treatment difference in t max was assessed by the Wilcoxon signed rank test. The statistical analyses were performed using SAS v9.4 software (SAS Institute, Inc, Cary, North Carolina).
Safety data from the safety-evaluable population (participants who received the study drug) were summarized using descriptive statistics (SAS software).
# Results
## Demographics and baseline clinical characteristics
Thirty-six healthy adults were enrolled in the study with 18 participants each in cohorts 1 and 2. Baseline demographics and clinical characteristics were similar in cohorts [fig_ref] Table 1: Summary of Demographic and Baseline Characteristics [/fig_ref]. Mean ± standard deviation (SD) age was 34.8 ± 9.7 years in cohort 1 and 35.9 ± 10.0 years in cohort 2. The mean ± SD body weight was 79.2 ± 12.3 kg in cohort 1 and 82.3 ± 11.6 kg in cohort 2. All 18 participants in cohort 1 and 16 of 18 participants in cohort 2 were men.
## Participant disposition
Study drugs were administered at the assigned times under the supervision of clinic personnel. All participants were included in the PK analysis with the exception of a single participant in cohort 2. A preliminary analysis of parsaclisib PK profiles showed that 1 participant had similar PK profiles on day 1 and day 11. Exploratory measurement of plasma rifampin concentrations for this participant confirmed levels were below the limit of quantitation at all times, suggesting that the participant may not have taken rifampin as documented by the clinical site (data on file, Incyte Corporation). This participant was excluded from all DDI analyses involving rifampin. All participants were included in the safety analysis population.
## Effect of itraconazole on pharmacokinetics of parsaclisib (cohort 1)
Parsaclisib was absorbed quickly following oral administration and achieved peak plasma concentrations within 1 hour, either alone or with concomitant itraconazole [fig_ref] Figure 2: Mean plasma concentrations of parsaclisib following administration of parsaclisib [/fig_ref] ; parsaclisib plasma concentration declined in a biphasic manner. Concomitant administration of itraconazole increased the parsaclisib geometric mean C max by 21% (90%CI, 14%-29%) and increased the parsaclisib geometric mean AUC 0-∞ by 107% (90%CI, 97%-117%); see [fig_ref] Table 2: Comparison of Parsaclisib Pharmacokinetic Parameters Following Administration of Parsaclisib [/fig_ref]. Comparisons of parsaclisib C max , AUC 0-∞ , and AUC 0-t values for individual participants with and without concomitant treatment with itraconazole are presented in. Itraconazole prolonged the parsaclisib mean t 1 2 ,β by 59% (21.1 hours with itraconazole versus 13.2 hours for parsaclisib alone; P ˂ .0001; [fig_ref] Table 2: Comparison of Parsaclisib Pharmacokinetic Parameters Following Administration of Parsaclisib [/fig_ref]. Itraconazole decreased parsaclisib geometric mean CL/F by approximately 51% (1.4 L/h with itraconazole and 2.9 L/h with parsaclisib alone; P ˂ .0001). Coadministration of itraconazole reduced the mean ± SD V z /F by approximately 24% (41.7 ± 9.1 L) compared with parsaclisib alone (54.7 ± 13.9 L); P ˂ .0001).
## Effect of rifampin on pharmacokinetics of parsaclisib (cohort 2)
Parsaclisib was absorbed quickly following oral administration, either alone or with concomitant rifampin, and reached peak plasma concentrations within 1 hour followed by biphasic decline [fig_ref] Figure 4: Mean plasma concentrations of parsaclisib following administration of parsaclisib [/fig_ref]. Concomitant administration of rifampin reduced the parsaclisib C max by 43% (90%CI, 40%-47%; 1650 ± 272 nM for parsaclisib alone versus 934 ± 184 nM with rifampin) and decreased the parsaclisib AUC 0-∞ by 77% (90%CI, 76%-79%; 14 800 ± 2440 nM·h with parsaclisib alone versus 3340 ± 478 nM·h with rifampin); see [fig_ref] Table 3: Comparison of Parsaclisib Pharmacokinetic Parameters Following Administration of Parsaclisib [/fig_ref].
Comparisons of parsaclisib C max , AUC 0-∞ , and AUC 0-t values for individual participants with and without concomitant rifampin administration are presented in. Rifampin shortened the parsaclisib mean t 1 2 ,β by 71% (4.15 hours with rifampin versus 14.7 hours for parsaclisib alone; P ˂ .0001; [fig_ref] Table 3: Comparison of Parsaclisib Pharmacokinetic Parameters Following Administration of Parsaclisib [/fig_ref]. Rifampin increased the parsaclisib geometric mean CL/F by approximately 4.4-fold (12.9 L/h with rifampin and 2.9 L/h for parsaclisib alone; P ˂ .0001). Mean ± SD V z /F was similar for the 2 treatments, 77.6 ± 15.2 L with rifampin and 63.0 ± 20.7 L for parsaclisib alone (P = .0070).
## Safety and tolerability (all participants)
No dose interruptions or reductions, participant discontinuations, serious AEs, or deaths due to TEAEs occurred in the study. Five participants (27.8%) in cohort 1 and 8 participants (44.4%) in cohort 2 experienced TEAEs. Headache was the most common TEAE, reported by 5 participants (13.9%, all in cohort 2), followed by generalized pruritus and maculopapular rash, both reported by 2 participants (5.6%, all in cohort 2).
TEAEs reported by only 1 participant (2.8% overall) were mucosal edema, ear abrasion, arthralgia, somnolence, and oropharyngeal pain in cohort 1 and palpitations, constipation, geographic tongue, feeling hot, noncardiac chest pain, increased blood creatine phosphokinase, myalgia, dizziness, dry throat, dyspnea, and nasal congestion in cohort 2. No TEAE was considered related to itraconazole or parsaclisib. TEAEs considered related to rifampin alone were reported by 2 participants and included generalized pruritus and maculopapular rash. No vital sign or ECG-related TEAEs were reported in this study. No grade 3 TEAE was reported in either cohort. One participant (5.6%) in cohort 2 experienced a grade 4 TEAE of increased creatine phosphokinase (CPK) accompanied by increased aspartate aminotransferase, at the follow-up visit on day 44 from the start of the study (32 days after the last dose of study treatment). By day 56 results for both laboratory parameters had returned to normal. This acute and transient elevation of CPK was likely because of physical activity. The AE of increased CPK was not considered by the investigator to be related to either parsaclisib or rifampin. The participant's CPK values from screening to day 12 (discharge from the unit) was in the normal range.
# Discussion
This study investigated the impact of CYP3A4 inhibition or induction on the PK of parsaclisib, a potent and selective next-generation PI3Kδ inhibitor. Openlabel and fixed-sequence study design was used because the primary end point, PK parameters, is an objective end point that would not be influenced by a sequence effect. Also, a 2-sequence design would have made it difficult to determine the length of washout period needed following both the induction of CYP3A4 and the inhibition of CYP3A, increasing the chance of carryover parsaclisib concentrations affecting the start of the next sequence.
In a recent phase 1/2 study with parsaclisib in relapsed or refractory B-cell malignancies, the doseexposure relationship for parsaclisib at steady state was linear over the dose range from 5 to 45 mg, and 20 mg once daily was chosen as the phase 2 dose. [bib_ref] Parsaclisib, a potent and highly selective PI3Kδ inhibitor, in patients with relapsed..., Forero-Torres [/bib_ref] Therefore, the parsaclisib dose of 20 mg was selected for the CYP3A4 induction study. Because parsaclisib is primarily metabolized by CYP3A4 and coadministration with a potent CYP3A inhibitor such as itraconazole can potentially increase the drug exposure, a lower dose level of 10 mg was selected for the CYP3A4 inhibition study to avoid unexpected safety issues and was therefore considered acceptable. A PK interaction is anticipated with inhibitors or inducers of CYP3A4 based on in vitro data showing that parsaclisib is primarily metabolized by CYP3A4. This clinical study was therefore conducted to confirm the presence and determine the magnitude of a DDI with a potent CYP3A4 inhibitor and inducer and to help guide potential parsaclisib dosage adjustment recommendations.
Following administration of a single 10-mg dose of parsaclisib, concomitant administration with the potent inhibitor of CYP3A4 itraconazole resulted in an approximately 2-fold increase in the exposure of parsaclisib; the GMRs of C max and AUC 0-∞ of parsaclisib were 1.2 (90%CI, 1.1-1.3) and 2.1 (90%CI, 2.0-2.2), respectively. Conversely, after administration of a single 20-mg dose of parsaclisib, concomitant administration with the potent inducer of CYP3A4 rifampin resulted in an approximately 4-fold decrease in the exposure of parsaclisib; the GMR of C max and AUC 0-∞ of parsaclisib were 0.6 (90%CI, 0.5-0.6) and 0.23 (90%CI, 0.21-0.24), respectively. Parsaclisib dose adjustment may therefore be warranted when administered with a strong CYP3A4 inhibitor (eg, boceprevir, indinavir, lopinavir, nelfinavir, ritonavir, saquinavir, clarithromycin, itraconazole, posaconazole, voriconazole),and it may be appropriate to avoid coadministration with strong CYP3A4 inducers (eg, carbamazepine, phenytoin, rifampin),owing to the potential for increased exposure or reduced efficacy of parsaclisib, respectively. First-generation PI3Kδ inhibitors (eg, idelalisib and duvelisib) have been associated with toxicities including transaminitis (increased alanine and aspartate transaminase levels), colitis, diarrhea, and pneumonitis. [bib_ref] Management of adverse events associated with idelalisib treatment: expert panel opinion, Coutré [/bib_ref] [bib_ref] Duvelisib, a novel oral dual inhibitor of PI3K-δ,γ , is clinically active..., Flinn [/bib_ref] [bib_ref] PI3Kδ inhibition by idelalisib in patients with relapsed indolent lymphoma, Gopal [/bib_ref] Parsaclisib demonstrated a favorable safety profile in a phase 1/2 study, with no dose-limiting toxicities identified. [bib_ref] Parsaclisib, a potent and highly selective PI3Kδ inhibitor, in patients with relapsed..., Forero-Torres [/bib_ref] In the current study, headache was the most common TEAE, reported by 13.9% of participants. Generalized pruritus and maculopapular rash (reported by 5.6% of participants) were considered related to rifampin.
# Conclusions
Single doses of parsaclisib when administered to healthy participants alone or in combination with itraconazole or rifampin appeared to be safe and well tolerated in this study of healthy participants. These data will help inform parsaclisib dosage guidelines and DDI risk management.
[fig] Figure 2: Mean plasma concentrations of parsaclisib following administration of parsaclisib (10 mg) with or without concomitant itraconazole (200 mg) in healthy participants. Error bars represent standard deviation. [/fig]
[fig] Figure 3: (A) C max , (B) AUC 0-∞ , and (C) AUC 0-t values for parsaclisib for individual study participants, following administration of parsaclisib (10 mg) with and without concomitant itraconazole (200 mg). AUC, area under the plasma concentration-time curve; C max , [/fig]
[fig] Figure 4: Mean plasma concentrations of parsaclisib following administration of parsaclisib (20 mg) with or without concomitant administration of rifampin (600 mg) in healthy participants. Error bars represent standard deviation. [/fig]
[fig] Figure 5: (A) C max , (B) AUC 0-∞ , and (C) AUC 0-t values for parsaclisib for individual study participants, following administration of parsaclisib (20 mg) with or without concomitant rifampin (600 mg). AUC, area under the plasma concentration-time curve; C max , maximum plasma drug concentration. [/fig]
[table] Table 1: Summary of Demographic and Baseline Characteristics [/table]
[table] Table 2: Comparison of Parsaclisib Pharmacokinetic Parameters Following Administration of Parsaclisib (10 mg) With and Without Concomitant Itraconazole (200 mg) [/table]
[table] Table 3: Comparison of Parsaclisib Pharmacokinetic Parameters Following Administration of Parsaclisib (20 mg) With and Without Concomitant Rifampin (600 mg) Parsaclisib Alone (n = 18) Parsaclisib + Rifampin (n = 17)AUC, area under the plasma concentration-time curve; CI, confidence interval; CL/F, oral dose clearance; C max , maximum plasma drug concentration; PK, pharmacokinetic; SD, standard deviation; t 1 2 ,β , terminal half-life; t max , time to maximum plasma drug concentration; V z /F, apparent oral dose volume of distribution. a t max is reported as median (range). b Reference is parsaclisib alone. [/table]
|
A Review of Biomonitoring of Phthalate Exposures
Phthalates (diesters of phthalic acid) are widely used as plasticizers and additives in many consumer products. Laboratory animal studies have reported the endocrine-disrupting and reproductive effects of phthalates, and human exposure to this class of chemicals is a concern. Several phthalates have been recognized as substances of high concern. Human exposure to phthalates occurs mainly via dietary sources, dermal absorption, and air inhalation. Phthalates are excreted as conjugated monoesters in urine, and some phthalates, such as di-2-ethylhexyl phthalate (DEHP), undergo secondary metabolism, including oxidative transformation, prior to urinary excretion. The occurrence of phthalates and their metabolites in urine, serum, breast milk, and semen has been widely reported. Urine has been the preferred matrix in human biomonitoring studies, and concentrations on the order of several tens to hundreds of nanograms per milliliter have been reported for several phthalate metabolites. Metabolites of diethyl phthalate (DEP), dibutyl-(DBP) and diisobutyl-(DiBP) phthalates, and DEHP were the most abundant compounds measured in urine. Temporal trends in phthalate exposures varied among countries. In the United States (US), DEHP exposure has declined since 2005, whereas DiNP exposure has increased. In China, DEHP exposure has increased since 2000. For many phthalates, exposures in children are higher than those in adults. Human epidemiological studies have shown a significant association between phthalate exposures and adverse reproductive outcomes in women and men, type II diabetes and insulin resistance, overweight/obesity, allergy, and asthma. This review compiles biomonitoring studies of phthalates and exposure doses to assess health risks from phthalate exposures in populations across the globe.
# Introduction
Phthalates are diesters of phthalic acid (1,2-benzenedicarboxylic acid) and are synthetic organic chemicals used in industries as solvents, plasticizers, and additives in polyvinyl chloride (PVC) plastics or personal care products (PCPs) [bib_ref] Monitoring phthalate exposure in humans, Latini [/bib_ref]. More than 25 phthalates are used in commercial applications, with each adding unique qualities to the product into which it is incorporated. Ten commonly used phthalates [fig_ref] Figure 1: Chemical structures of major phthalates and their metabolites studied in the literature [/fig_ref] , [fig_ref] Table 1: Major phthalate diesters and their corresponding metabolites studied in the literature [/fig_ref] are dimethyl phthalate (DMP), diethyl phthalate (DEP), dibutyl phthalate (DBP), diisobutyl phthalate (DiBP), benzylbutyl phthalate (BzBP), dicyclohexyl phthalate (DCHP), di(2-ethylhexyl) phthalate (DEHP), di-n-octyl phthalate (DnOP), di-isononyl phthalate (DiNP), and di-isodecyl phthalate (DiDP). DEHP, one of the major phthalates in commerce, was first synthesized for use as a plasticizer in 1933 [bib_ref] Metabolism of phthalates in humans, Frederiksen [/bib_ref]. The application of DEHP as an additive in polyvinyl chloride to impart the flexibility of plastic has made phthalates popular around the world. The addition of phthalates to PVC makes it not only flexible but also malleable and durable. PVC products may contain up to 50% (by weight) phthalates [bib_ref] Monitoring phthalate exposure in humans, Latini [/bib_ref]. made phthalates popular around the world. The addition of phthalates to PVC makes it not only flexible but also malleable and durable. PVC products may contain up to 50% (by weight) phthalates [bib_ref] Monitoring phthalate exposure in humans, Latini [/bib_ref].
## Biomonitoring of phthalates
Due to the ubiquitous occurrence and widespread exposure of phthalates, their metabolites are one of the most examined environmental chemicals in human biomonitoring studies. The reported half-life of phthalates diesters in blood plasma or urine of humans and rodents was less than 24 h. Several studies have reviewed pharmacokinetics of phthalate esters, and these studies have found rapid hydrolysis of diesters to monoesters in the gastrointestinal tract [bib_ref] Monitoring phthalate exposure in humans, Latini [/bib_ref] [bib_ref] Metabolism of phthalates in humans, Frederiksen [/bib_ref]. Binding of DEHP metabolites to blood plasma proteins, existence of biliary excretion, and enterohepatic circulation in humans have been suggested [bib_ref] Metabolism of phthalates in humans, Frederiksen [/bib_ref]. Nevertheless, urinary excretion has been the major elimination pathway of phthalates [bib_ref] Metabolism of phthalates in humans, Frederiksen [/bib_ref]. Urinary concentrations of phthalate metabolites are generally 5-20 times higher than that in lipid-rich compartments. For example, urinary concentrations of mono-2-ethylhexyl phthalate (MEHP), mono-isobutyl phthalate (MIBP), mono-ethyl phthalate (MEP), and mono-n-butyl phthalate (MBP) were 20-100 times those in blood or milk [bib_ref] Phthalate diesters and their metabolites in human breast milk, blood or serum,..., Högberg [/bib_ref]. Phthalate metabolites have been measured in various body fluids, including urine [bib_ref] Comparative assessment of human exposure to phthalate esters from house dust in..., Guo [/bib_ref] [bib_ref] Phthalate metabolites in urine from China, and implications for human exposures, Guo [/bib_ref] , serum [bib_ref] Serum concentrations of phthalate metabolites are related to abdominal fat distribution two..., Lind [/bib_ref] [bib_ref] Quantitative Detection of Nine Phthalate Metabolites in Human Serum Using Reversed-Phase High-Performance..., Kato [/bib_ref] , semen [bib_ref] Phthalate exposure and human semen parameters, Duty [/bib_ref] [bib_ref] Altered semen quality in relation to urinary concentrations of phthalate monoester and..., Hauser [/bib_ref] , breast milk [bib_ref] Phthalate esters in human milk: Concentration variations over a 6-month postpartum time, Zhu [/bib_ref] [bib_ref] the Nordic Cryptorchidism Study, G. Persistent pesticides in human breast milk and..., Damgaard [/bib_ref] , and saliva [bib_ref] Detection of phthalate metabolites in human saliva, Silva [/bib_ref] [fig_ref] Table 3: Reported concentrations of major phthalate metabolites in human specimens collected from various... [/fig_ref]. Phthalates can cross the placental barrier [bib_ref] Pharmacokinetics of dibutylphthalate in pregnant rats, Fennell [/bib_ref] and have been measured in amniotic fluid in human studies [bib_ref] Detection of phthalate metabolites in human amniotic fluid, Silva [/bib_ref]. To date, studies that report partitioning of phthalates among various tissues and organs in an organism, at state-state exposure conditions, are not available. It is worth noting that a few earlier reviews have described biomonitoring of phthalates in humans [bib_ref] Identification of exposure to environmental chemicals in children and older adults using..., Choi [/bib_ref]. Biomonitoring studies that report concentrations of phthalates metabolites are presented in [fig_ref] Table 3: Reported concentrations of major phthalate metabolites in human specimens collected from various... [/fig_ref]. Phthalate diesters (parent compounds) were measured in blood plasma of women with endometriosis in India, and a significant association was found between phthalate exposure and the risk of developing endometriosis [bib_ref] High plasma concentrations of polychlorinated biphenyls and phthalate esters in women with..., Reddy [/bib_ref]. Similarly, studies have determined phthalates in serum samples of couples from Greenland, Poland, and Ukraine that showed that the DEHP levels were associated with reduced time to achieve pregnancy [bib_ref] Serum phthalate levels and time to pregnancy in couples from Greenland, Poland..., Specht [/bib_ref]. Phthalate diesters and their metabolites also have been measured in breast milk, serum, and urine from Swedish women [bib_ref] Phthalate diesters and their metabolites in human breast milk, blood or serum,..., Högberg [/bib_ref]. In milk and serum samples, the concentrations of phthalate diesters and their metabolites were below the method limit of detection (0.12-3.0 µg/L). Detectable concentrations of phthalate metabolites, however, were found in urine (0.1-1000 µg/L). Measurements of phthalate diesters in breast milk and serum are prone to false positives due to background contamination. Medical devices, including blood collection devices and plastic containers that are used to collect and store samples, can contain phthalate diesters [bib_ref] Medications as a source of human exposure to phthalates, Hauser [/bib_ref]. If the samples were to be analyzed for phthalate diesters, caution should be taken with the screening devices used to collect and store samples. A comprehensive review of challenges associated with low-level phthalate analysis in biological specimens has been published [bib_ref] A survey of phthalates and parabens in personal care products from the..., Guo [/bib_ref].
## Phthalate metabolites in urine
Although microbial degradation of DEHP to MEHP in soils through lipase and esterase enzymes has been shown, environmental degradation/transformation of parent phthalates is slow [bib_ref] Human breast milk contamination with phthalates and alterations of endogenous reproductive hormones..., Main [/bib_ref] [bib_ref] Automated solid phase extraction and quantitative analysis of human milk for 13..., Calafat [/bib_ref]. Because phthalates have a short half-life in human bodies and are excreted quickly in urine as monoester metabolites, the metabolites are suitable biomarkers for human exposure to parent compounds. The half-life of phthalates in human bodies (in plasma and urine) is less than 24 h, and following metabolism, monoesters of phthalates are conjugated with glucuronide or sulfate and excreted in urine [bib_ref] Improved quantitative detection of 11 urinary phthalate metabolites in humans using liquid..., Silva [/bib_ref]. Analysis of metabolites in urine involves enzymatic deconjugation followed by purification. Assessment of human exposure to phthalates is based mainly on the measurement of their urinary monoester metabolites, although several secondary and oxidative metabolites have been reported to occur in human specimens [bib_ref] Improved quantitative detection of 11 urinary phthalate metabolites in humans using liquid..., Silva [/bib_ref]. For instance, DMP, DEP, and DBP undergo degradation/hydrolysis and form their corresponding monoesters, i.e., MMP, MEP and MBP, respectively. Both hydrolysis and oxidation products are formed from the metabolism of DEHP. MEHP, the hydrolysis product of DEHP, is not a major metabolite. The oxidative metabolites, MEOHP, MEHHP, MECPP, and MCMHP, however, are the major metabolites of DEHP and are appropriate biomarkers of exposure to this compound [bib_ref] Quantification of 22 phthalate metabolites in human urine, Silva [/bib_ref]. Some studies suggest, however, that MEHP is more toxic than are other oxidative metabolites. The general metabolic pathways of phthalate esters in humans are shown in [fig_ref] Figure 2: Metabolic pathways of phthalate esters in humans [/fig_ref].
General Population Adults: A large number of studies have reported measurements of phthalate metabolites in human specimens collected from European (Germany, Netherlands, Denmark, Norway, Sweden, Greece, the Czech Republic, Hungary, Slovakia, and Spain) and Asian countries (Japan, China, South Korea, India, Taiwan, Vietnam, Saudi Arabia, Malaysia, and Kuwait) as well as from North American countries. The number of phthalate metabolites measured in urine samples varied considerably; as new analytical standards become made available commercially, more metabolites were added to the list of compounds measured in urine. Although a majority of the recent studies measure close to 20 phthalate metabolites, studies conducted a decade ago measured 10 or fewer metabolites of phthalates.
In general, the concentrations of the sum of 22 phthalate metabolites measured in human urine were on the order of several to hundreds of parts-per-billion (µg/L) [bib_ref] Quantification of 22 phthalate metabolites in human urine, Silva [/bib_ref]. In a majority of the biomonitoring studies, metabolites of DEHP, DEP, and DBP were the major compounds identified in urine, and the profile varied depending on the country. Urine samples collected from 32 men and 53 women (age: 7-64 years) from northern Bavaria (Germany) contained MBP (median: 181 µg/L), MEP (90.2 µg/L), and major DEHP metabolites, such as MEHHP (46.8 µg/L) and MEOHP (36.5 µg/L) [bib_ref] An estimation of the daily intake of di(2-ethylhexyl)phthalate (DEHP) and other phthalates..., Koch [/bib_ref]. The median concentrations of DEHP metabolites, namely, MEHP, MEOHP, and MEHHP, were 4.5, 28.3, and 35.9 µg/L, respectively, and these three metabolites were highly intercorrelated. The concentration ratios, MEHHP/MEHP, and MEOHP/MEHP, were calculated to be 8.2, and 5.9, respectively. These ratios suggest that MEHP is further oxidized to form MEHHP and MEOHP [bib_ref] An estimation of the daily intake of di(2-ethylhexyl)phthalate (DEHP) and other phthalates..., Koch [/bib_ref]. General Population Adults: A large number of studies have reported measurements of phthalate metabolites in human specimens collected from European (Germany, Netherlands, Denmark, Norway, Sweden, Greece, the Czech Republic, Hungary, Slovakia, and Spain) and Asian countries (Japan, China, South Korea, India, Taiwan, Vietnam, Saudi Arabia, Malaysia, and Kuwait) as well as from North American countries. The number of phthalate metabolites measured in urine samples varied considerably; as new analytical standards become made available commercially, more metabolites were added to the list of compounds measured in urine. Although a majority of the recent studies measure close to 20 phthalate metabolites, studies conducted a decade ago measured 10 or fewer metabolites of phthalates.
In general, the concentrations of the sum of 22 phthalate metabolites measured in human urine were on the order of several to hundreds of parts-per-billion (µg/L) [bib_ref] Quantification of 22 phthalate metabolites in human urine, Silva [/bib_ref]. In a majority of the biomonitoring studies, metabolites of DEHP, DEP, and DBP were the major compounds identified in urine, and the profile varied depending on the country. Urine samples collected from 32 men and 53 women (age: 7-64 years) from northern Bavaria (Germany) contained MBP (median: 181 µg/L), MEP (90.2 µg/L), and major DEHP metabolites, such as MEHHP (46.8 µg/L) and MEOHP (36.5 µg/L) [bib_ref] An estimation of the daily intake of di(2-ethylhexyl)phthalate (DEHP) and other phthalates..., Koch [/bib_ref]. The median concentrations of DEHP metabolites, namely, MEHP, MEOHP, and MEHHP, were 4.5, 28.3, and 35.9 µg/L, respectively, and these three metabolites were highly intercorrelated. The concentration ratios, MEHHP/MEHP, and MEOHP/MEHP, were calculated to be 8.2, and 5.9, respectively. These ratios suggest that MEHP is further oxidized to form MEHHP and MEOHP [bib_ref] An estimation of the daily intake of di(2-ethylhexyl)phthalate (DEHP) and other phthalates..., Koch [/bib_ref].
The urinary concentrations of phthalate metabolites in general populations vary among countries. Some of the highest concentrations of total phthalate metabolites were found in urine collected in 2006-2007 from Kuwaitis, with a maximum value of 19,300 µg/L and a median value for of 1050 µg/L [bib_ref] Occurrence of phthalate metabolites in human urine from several Asian countries, Guo [/bib_ref]. The occurrence of phthalate metabolites was investigated in urine from Germans, and MBP was found at high concentrations (median: 49 µg/L) [bib_ref] Occurrence and daily variation of phthalate metabolites in the urine of an..., Fromme [/bib_ref]. The median concentrations of phthalate metabolites in urine samples from Germany decreased significantly from 2002 to 2008 [bib_ref] Trends of the internal phthalate exposure of young adults in Germany-Follow-up of..., Goen [/bib_ref]. Similarly, urinary phthalate metabolites measured in 2015 were significantly lower than those in 2007 in Germany [bib_ref] Phthalate metabolites in 24-h urine samples of the German Environmental Specimen Bank..., Koch [/bib_ref].
Biomonitoring studies in other European countries, including France [bib_ref] Exposure assessment of phthalates in French pregnant women: Results of the ELFE..., Zeman [/bib_ref] , Belgium [bib_ref] Measurement of urinary biomarkers of parabens, benzophenone-3, and phthalates in a Belgian..., Dewalque [/bib_ref] [bib_ref] Simultaneous determination of some phthalate metabolites, parabens and benzophenone-3 in urine by..., Dewalque [/bib_ref] , Slovakia [bib_ref] Relationship between variation of seasonal temperature and extent of occupational exposure to..., Pilka [/bib_ref] , and Norway [bib_ref] Evaluation of exposure to phthalate esters and DINCH in urine and nails..., Giovanoulis [/bib_ref] , report phthalate metabolite concentrations in the range of 1-100 µg/L in urine from adults. MBP and DEHP metabolites were the predominant compounds found in those studies. Further, a comparative analysis of biomonitoring data in Europe suggested a The occurrence of phthalate metabolites was investigated in urine from Germans, and MBP was found at high concentrations (median: 49 µg/L) [bib_ref] Occurrence and daily variation of phthalate metabolites in the urine of an..., Fromme [/bib_ref]. The median concentrations of phthalate metabolites in urine samples from Germany decreased significantly from 2002 to 2008 [bib_ref] Trends of the internal phthalate exposure of young adults in Germany-Follow-up of..., Goen [/bib_ref]. Similarly, urinary phthalate metabolites measured in 2015 were significantly lower than those in 2007 in Germany [bib_ref] Phthalate metabolites in 24-h urine samples of the German Environmental Specimen Bank..., Koch [/bib_ref].
Biomonitoring studies in other European countries, including France [bib_ref] Exposure assessment of phthalates in French pregnant women: Results of the ELFE..., Zeman [/bib_ref] , Belgium [bib_ref] Measurement of urinary biomarkers of parabens, benzophenone-3, and phthalates in a Belgian..., Dewalque [/bib_ref] [bib_ref] Simultaneous determination of some phthalate metabolites, parabens and benzophenone-3 in urine by..., Dewalque [/bib_ref] , Slovakia [bib_ref] Relationship between variation of seasonal temperature and extent of occupational exposure to..., Pilka [/bib_ref] , and Norway [bib_ref] Evaluation of exposure to phthalate esters and DINCH in urine and nails..., Giovanoulis [/bib_ref] , report phthalate metabolite concentrations in the range of 1-100 µg/L in urine from adults. MBP and DEHP metabolites were the predominant compounds found in those studies. Further, a comparative analysis of biomonitoring data in Europe suggested a significant decline in phthalate metabolite concentrations (especially MEP, MBP, MBzP, and DEHP metabolites) from 2011 to 2016 [bib_ref] Temporal trends of urinary phthalate concentrations in two populations: Effects of REACH..., Tranfo [/bib_ref]. Several alternative plasticizers, however, are used as replacements for DEHP in European countries. Common alternatives include Hexamoll DINCH (DINCH), acetyl tributyl citrate (ATBC), dioctyl terephthalate (DOTP), 2,2,4-trimethyl 1,3-pentanediol diisobutyrate (TXIB), trioctyl trimellitate (TOTM), and di-(2-ethylhexyl) adipate (DEHA).
In North America, the distribution of phthalate metabolites in urine has been summarized in nationwide monitoring surveys. For example, the US National Health and Nutrition Examination Survey (NHANES) of the Centers for Disease Control and Prevention (CDC) showed that MEP, MEHP, MEHHP, and MEOHP concentrations in urine from adults >20 years of age were 167, 3.99, 18.8, and 12.6 µg/g creatinine (CR), respectively [bib_ref] Integrating biomonitoring exposure data into the risk assessment process: Phthalates [diethyl phthalate..., Calafat [/bib_ref]. The NHANES program has measured 15 phthalate metabolites in urine. The weighted geometric mean concentration of 15 phthalate metabolites in the US general population was 125 µg/L for the samples collected in the period of 2007-2008. MEP was the major compound found in urine, accounting for >70% of the total concentrations, which was followed by mono- [bib_ref] The internal exposure of Taiwanese to phthalate-An evidence of intensive use of..., Chen [/bib_ref]. In contrast, DEHP metabolites were predominant in urine from Japan, Malaysia, and Vietnam [bib_ref] Occurrence of phthalate metabolites in human urine from several Asian countries, Guo [/bib_ref]. A nationwide survey of urine samples from 6478 adults during the period of 2012-2014 in Korea showed median urinary concentrations of DEHP metabolites (88.2 µg/L) that were twofold higher than that of MBP (44.2 µg/L) [bib_ref] Exposure to environmental chemicals among Korean adults-updates from the second Korean National..., Choi [/bib_ref] [bib_ref] Associations between urinary phthalate metabolites and bisphenol A levels, and serum thyroid..., Park [/bib_ref]. In Israel, phthalate metabolites were found in urine samples collected from 250 adults (ages 20−74), with median concentrations that ranged from 17.1 µg/L (MEOHP) to 37.6 µg/L (MiBP) [bib_ref] Urinary concentrations of environmental contaminants and phytoestrogens in adults in Israel, Berman [/bib_ref]. DEHP exposure in the Chinese population has increased since 2001 [bib_ref] Exposure assessment issues in epidemiology studies of phthalates, Johns [/bib_ref].
The global distribution of major phthalate metabolites measured in urine from general populations is presented in [fig_ref] Figure 3: Urinary concentrations of phthalate metabolites reported in adults [/fig_ref]. . This value is the highest among all countries studied. The profiles of phthalate metabolites varied, with MEP as the predominant metabolite in Indian and Kuwaiti urine samples (49% of the total), which were similar to those found in the US. In China (52%), MBP was the major metabolite found in urine. In Korea (46%), Japan (31%), and Vietnam (52%), DEHP metabolites were the dominant ones. MMP accounted for <8% of the total phthalate metabolite concentrations in all Asian countries, except for Japan, where it was 20%. Overall, MEP and DEHP metabolites were the major phthalate metabolites found in urine from most Asian countries, a pattern similar to that found in the US [bib_ref] Dietary intake Is associated with phthalate body burden in a nationally representative..., Colacino Justin [/bib_ref]. The reported urinary concentrations of phthalate metabolites among several European countries were similar whereas information for African countries and Australia/Oceanian countries is limited.
Pregnant Women: Phthalates have been widely studied for exposure levels in pregnant women. MEP (222 µg/g CR) was the predominant phthalate metabolite found in urine samples of pregnant women from the Netherlands (Generation R study) [bib_ref] Urinary metabolite concentrations of organophosphorous pesticides, bisphenol A, and phthalates among pregnant..., Ye [/bib_ref]. Similar exposure levels were reported for pregnant women from the US [bib_ref] Consumer product exposures associated with urinary phthalate levels in pregnant women, Buckley [/bib_ref] , Canada [bib_ref] Phthalate and bisphenol A exposure among pregnant women in Canada-Results from the..., Arbuckle [/bib_ref] , and Norway [bib_ref] Determination of 12 urinary phthalate metabolites in Norwegian pregnant women by core-shell..., Sabaredzovic [/bib_ref] , with MEP median concentrations exceeding 30 µg/L. In a study of urinary phthalate metabolite concentrations in Spanish pregnant women (n = 391), the median concentration of MEP was reported at 246 µg/g CR [bib_ref] Variability and predictors of urinary phthalate metabolites in Spanish pregnant women, Valvi [/bib_ref].
Several studies have examined phthalate metabolite concentrations in matched urine samples of newborns and mothers. Maternal urinary concentrations of MEHHP and MEOHP in Korea were 17.7 and 14.7 µg/L, respectively, which were two-to threefold higher than those found in newborns (5.79 and 3.27 µg/L) [bib_ref] Association of diethylhexyl phthalate with obesity-related markers and body mass change from..., Kim [/bib_ref]. Another study, however, showed similar urinary concentrations of phthalate metabolites between 120 mother-and-child pairs [bib_ref] Urinary phthalate concentrations in mothers and their children in Ireland: Results of..., Cullen [/bib_ref]. Occurrence of phthalate metabolites in pregnant women suggests potential exposure in the fetus.
countries were similar whereas information for African countries and Australia/Oceanian countries is limited. Pregnant Women: Phthalates have been widely studied for exposure levels in pregnant women. MEP (222 µg/g CR) was the predominant phthalate metabolite found in urine samples of pregnant women from the Netherlands (Generation R study) [bib_ref] Urinary metabolite concentrations of organophosphorous pesticides, bisphenol A, and phthalates among pregnant..., Ye [/bib_ref]. Similar exposure levels were reported for pregnant women from the US [bib_ref] Consumer product exposures associated with urinary phthalate levels in pregnant women, Buckley [/bib_ref] , Canada [bib_ref] Phthalate and bisphenol A exposure among pregnant women in Canada-Results from the..., Arbuckle [/bib_ref] , and Norway [bib_ref] Determination of 12 urinary phthalate metabolites in Norwegian pregnant women by core-shell..., Sabaredzovic [/bib_ref] , with MEP median concentrations exceeding 30 µg/L. In a study of urinary phthalate metabolite concentrations in Spanish pregnant women (n = 391), the median concentration of MEP was reported at 246 µg/g CR [bib_ref] Variability and predictors of urinary phthalate metabolites in Spanish pregnant women, Valvi [/bib_ref].
Several studies have examined phthalate metabolite concentrations in matched urine samples of newborns and mothers. Maternal urinary concentrations of MEHHP and MEOHP in Korea were 17.7 and 14.7 µg/L, respectively, which were two-to threefold higher than those found in newborns (5.79 and 3.27 µg/L) [bib_ref] Association of diethylhexyl phthalate with obesity-related markers and body mass change from..., Kim [/bib_ref]. Another study, however, showed similar urinary concentrations of phthalate metabolites between 120 mother-and-child pairs [bib_ref] Urinary phthalate concentrations in mothers and their children in Ireland: Results of..., Cullen [/bib_ref]. Occurrence of phthalate metabolites in pregnant women suggests potential exposure in the fetus.
Children: The NHANES data showed that the concentrations of urinary phthalate metabolites in children 6-11 years old were higher than those in adolescents and adults . Several studies support the CDC's findings that children have higher urinary concentrations than do adults of DBP, BzBP, and DEHP [bib_ref] Trends of the internal phthalate exposure of young adults in Germany-Follow-up of..., Goen [/bib_ref] [bib_ref] Internal exposure of nursery-school children and their parents and teachers to di(2-ethylhexyl)..., Koch [/bib_ref]. Differences in urinary concentrations of phthalates among infants, children, and adults may reflect different sources and routes of intake. Ingestion is thought to be a primary pathway of exposure to some phthalates, especially those in food packaging [bib_ref] The AAP Committee on Environmental Health. Pediatric exposure and potential toxicity of..., Shea [/bib_ref]. The mouthing behavior of infants and toddlers could potentially increase their exposures to phthalates in toys and other products made with plasticized polymers. The global distribution of reported urinary phthalate metabolite concentrations in children is shown in [fig_ref] Figure 4: Urinary concentrations of phthalate metabolites reported in children [/fig_ref]. Children: The NHANES data showed that the concentrations of urinary phthalate metabolites in children 6-11 years old were higher than those in adolescents and adults . Several studies support the CDC's findings that children have higher urinary concentrations than do adults of DBP, BzBP, and DEHP [bib_ref] Trends of the internal phthalate exposure of young adults in Germany-Follow-up of..., Goen [/bib_ref] [bib_ref] Internal exposure of nursery-school children and their parents and teachers to di(2-ethylhexyl)..., Koch [/bib_ref]. Differences in urinary concentrations of phthalates among infants, children, and adults may reflect different sources and routes of intake. Ingestion is thought to be a primary pathway of exposure to some phthalates, especially those in food packaging [bib_ref] The AAP Committee on Environmental Health. Pediatric exposure and potential toxicity of..., Shea [/bib_ref]. The mouthing behavior of infants and toddlers could potentially increase their exposures to phthalates in toys and other products made with plasticized polymers. The global distribution of reported urinary phthalate metabolite concentrations in children is shown in [fig_ref] Figure 4: Urinary concentrations of phthalate metabolites reported in children [/fig_ref]. MEP, MBP, and DEHP metabolites were the dominant compounds detected in urine from children. Spot urine samples from 5-to 7-year-old German children contained a median phthalate metabolite concentration (sum of 5 metabolites) of 76.9 µg/L, with DEHP metabolites as major compounds [bib_ref] Exposure to phthalates in 5-6 years old primary school starters in Germany-A..., Koch [/bib_ref]. A similar concentration of DEHP metabolites at 75.7 µg/L was found in urine MEP, MBP, and DEHP metabolites were the dominant compounds detected in urine from children. Spot urine samples from 5-to 7-year-old German children contained a median phthalate metabolite concentration (sum of 5 metabolites) of 76.9 µg/L, with DEHP metabolites as major compounds [bib_ref] Exposure to phthalates in 5-6 years old primary school starters in Germany-A..., Koch [/bib_ref]. A similar concentration of DEHP metabolites at 75.7 µg/L was found in urine samples from 8-to 10-year-old German children (n = 465) [bib_ref] Phthalate metabolites and bisphenol A in urines from German school-aged children: Results..., Kasper-Sonnenberg [/bib_ref]. Several biomonitoring studies reported comparable concentrations of DEHP metabolites and MBP in urine from children in China [bib_ref] Phthalate metabolites in urine samples from Beijing children and correlations with phthalate..., Gong [/bib_ref] , Korea [bib_ref] Urinary phthalate metabolites over the first 15months of life and risk assessment-CHECK..., Kim [/bib_ref] , Canada [bib_ref] Maternal and early life exposure to phthalates: The plastics and personal-care products..., Arbuckle [/bib_ref] , Brazil [bib_ref] Urinary concentrations of 25 phthalate metabolites in Brazilian children and their association..., Rocha [/bib_ref] and Portugal [bib_ref] Obesity or diet? Levels and determinants of phthalate body burden-A case study..., Correia-Sa [/bib_ref].
In urine samples collected from children in Beijing, China, MBP was the most abundant metabolite (median: 232 µg/L), followed by MiBP (81.3 µg/L), MECPP (79.1 µg/L), and MEP (28.5 µg/L). A significant association between the concentrations of parent phthalate diesters in handwipes and the corresponding monoester metabolites in urine were observed in urine from children, which suggested that dermal absorption is an important exposure pathway for phthalates in children [bib_ref] Phthalate metabolites in urine samples from Beijing children and correlations with phthalate..., Gong [/bib_ref]. Mean urinary concentrations of MBP decreased as the children aged [bib_ref] Phthalate metabolites and bisphenol A in urine of German children, Becker [/bib_ref]. Among children, urinary DBP and DEHP metabolites in boys were higher than those in girls, whereas urinary MEP concentrations were positively correlated with age in both genders [bib_ref] Urinary excretion of phthalate metabolites in 129 healthy Danish children and adolescents:..., Frederiksen [/bib_ref]. Urinary concentrations of MEP in adolescents were higher than those in children, which was associated with high cosmetic usage among teenagers [bib_ref] Urinary excretion of phthalate metabolites in 129 healthy Danish children and adolescents:..., Frederiksen [/bib_ref] [bib_ref] Biomonitoring and subsequent risk assessment of combined exposure to phthalates in Iranian..., Jeddi [/bib_ref].
Urinary [bib_ref] Human biomonitoring pilot study DEMOCOPHES in Germany: Contribution to a harmonized European..., Schwedler [/bib_ref] , which suggested that children in those countries were more highly exposed to several phthalates than were their mothers. Nevertheless, some studies reported higher urinary MEP concentrations in mothers (45.1-72.0 µg/L) than in children µg/L) [bib_ref] In-utero and childhood chemical exposome in six European mother-child cohorts, Haug [/bib_ref] [bib_ref] Hbmnet Phthalate intake by infants calculated from biomonitoring data, Volkel [/bib_ref]. The concentrations of DEHP metabolites were reported to be similar between mothers and children [bib_ref] In-utero and childhood chemical exposome in six European mother-child cohorts, Haug [/bib_ref] [bib_ref] Biomonitoring of urinary di(2-ethylhexyl) phthalate metabolites of mother and child pairs in..., Song [/bib_ref] [bib_ref] Urinary levels of eight phthalate metabolites and bisphenol A in mother-child pairs..., Cutanda [/bib_ref]. A significant positive correlation existed in urinary phthalate metabolite concentrations between children and their parents. MECPP, an oxidative metabolite of DEHP, was predominant in urine from children (92.7%) relative to that found in adults (56.7-57.6%). Studies have found that children possess enhanced oxidative metabolism for DEHP [bib_ref] Phthalate metabolites and bisphenol A in urine of German children, Becker [/bib_ref] [bib_ref] Biomonitoring of urinary di(2-ethylhexyl) phthalate metabolites of mother and child pairs in..., Song [/bib_ref] [bib_ref] Phthalate exposure in pregnant women and their children in central Taiwan, Lin [/bib_ref]. Another study of urinary phthalate metabolites in 104 paired mothers and school-aged children reported higher concentrations of secondary DEHP metabolites in children than in mothers [bib_ref] Levels of phthalate metabolites in urine among mother-child-pairs-Results from the Duisburg birth..., Kasper-Sonnenberg [/bib_ref]. A study from Austria showed higher urinary concentrations of phthalate metabolites in children than adults [bib_ref] Human biomonitoring of phthalate exposure in Austrian children and adults and cumulative..., Hartmann [/bib_ref]. Overall, these studies suggest higher exposures to phthalates in children than adults.
Highly Exposed Populations: Highly exposed individuals have urinary phthalate metabolite concentrations that often exceed those at the 95th percentile of the general population (https://www. ncbi.nlm.nih.gov/books/NBK215044/). Neonates who receive medical treatments such as transfusions are widely recognized as potentially highlexposed [bib_ref] Exposure to bisphenol A and other phenols in neonatal intensive care unit..., Calafat [/bib_ref]. A study from Slovakia showed that the urinary concentrations of DEHP metabolites, MiBP, and MBP in occupationally exposed individuals from plastic industry were 55.9, 39.2, and 110 µg/L, respectively [bib_ref] Occupational exposure to phthalates in relation to gender, consumer practices and body..., Petrovicova [/bib_ref] , which were higher than those in urine from women of no known occupational exposures [bib_ref] automated online SPE-LC-QTRAP-MS/MS method for the simultaneous analysis of 14 phthalate metabolites..., Heffernan [/bib_ref]. The median concentrations of MEP, MBP, MiBP, and DEHP metabolites in urine from hairdressing apprentices who attended vocational training schools in Slovakia were 201, 103, 61.4, and 82.7 µg/L, respectively [bib_ref] Occupational phthalate exposure and health outcomes among hairdressing apprentices, Kolena [/bib_ref]. Some medications contain phthalates in their coatings or delivery systems [bib_ref] Medications as a source of human exposure to phthalates, Hauser [/bib_ref] and may contribute to the high exposures of children, pregnant women, and others who take these medications.
Exposure Assessment: The concentrations of phthalate metabolites measured in urine can be used to assess the amount of parent phthalate to which humans are exposed, when the fraction of the metabolite excreted in urine is known, as presented in the equation below [bib_ref] Phthalate exposure in pregnant women and their children in central Taiwan, Lin [/bib_ref] : The estimated daily intake (EDI) of parent phthalates is then calculated by taking the average weight of an individual with the average urinary excretion rate, as shown in the equation below:
[formula] Estimated [/formula]
Estimated daily intake (EDI)= Estimated parent phthalate concentration ×Daily urine excretion volume Average body weight [bib_ref] Metabolism of phthalates in humans, Frederiksen [/bib_ref] Several studies have estimated exposure doses to phthalates in populations, which allowed for comparison against a reference dose (RfD), the maximum acceptable oral dose of a toxic substance, of the US EPA. The estimated mean daily exposure doses to DEP and DBP in Asian countries and the US were one to two orders of magnitude below the EPA RfD (DEP = 800, DBP = 100, and DEHP = 20 µg/kg body weight (bw)/day). The estimated daily exposure doses to DEHP in Kuwait and India, however, were close to the RfD of the US EPA [bib_ref] Occurrence of phthalate metabolites in human urine from several Asian countries, Guo [/bib_ref]. Similarly, high concentrations of DEHP metabolites (mean concentration = 338 µg/L) were reported in urine from the Saudi population [bib_ref] Urinary biomarkers of exposure to 57 xenobiotics and its association with oxidative..., Asimakopoulos [/bib_ref].
The calculated EDI max values for DEHP and DBP were 8 and 0.08 µg/kg bw/day, respectively, for the population in Taiwan, which were one to two orders of magnitude lower than the tolerable daily intake (TDI) values (the daily intake amount of a chemical that has been assessed to be safe for human being on a long-term basis) suggested for DEHP (50 µg/kg bw) and DBP (10 µg/kg bw) by the European Food Safety Authority (EFSA) [bib_ref] Phthalate exposure in pregnant women and their children in central Taiwan, Lin [/bib_ref].
The 95th percentile for DEHP exposure doses calculated for the general population (n = 85) and children (n = 254) from Germany were 21 and 25 µg/kg bw/day, respectively, which exceeded the RfD (20 µg/kg bw/day) and the TDI (20-48 µg/kg bw/day) [bib_ref] -ethylhexyl)phthalate (DEHP): Human metabolism and internal exposure-An update and latest results, Koch [/bib_ref]. Further, elevated exposure to phthalates, especially DEHP, in neonates admitted to intensive care units was reported (median: 42 µg/kg bw/day; 95th percentile: 1780 µg/kg bw/day) [bib_ref] -ethylhexyl)phthalate (DEHP): Human metabolism and internal exposure-An update and latest results, Koch [/bib_ref] , and the exposure dose was higher than the RfD.
Some studies defined "Biomonitoring Equivalents (BEs)" as the concentration or range of concentrations of a chemical or its metabolite in a biological medium (blood, urine, or other medium) that is consistent with an existing health-based exposure guideline (e.g., RfD and TDI) [bib_ref] Derivation of Biomonitoring Equivalents for di-n-butyl phthalate (DBP), benzylbutyl phthalate (BzBP), and..., Aylward [/bib_ref] [bib_ref] Biomonitoring Equivalents for di-isononyl phthalate (DINP), Hays [/bib_ref]. BE values for MBP, MBzP, and MEP were reported at 18000, 3800 and 2700 µg/L, respectively [bib_ref] Derivation of Biomonitoring Equivalents for di-n-butyl phthalate (DBP), benzylbutyl phthalate (BzBP), and..., Aylward [/bib_ref] , and the BE values range from 1500 to 3600 µg/L for MiNP [bib_ref] Biomonitoring Equivalents for di-isononyl phthalate (DINP), Hays [/bib_ref]. These values may be used as screening tools for evaluation of biomonitoring data for phthalate metabolites in the context of existing risk assessments and for prioritization of the potential need for additional risk assessment efforts for each of these compounds relative to other chemicals [bib_ref] Derivation of Biomonitoring Equivalents for di-n-butyl phthalate (DBP), benzylbutyl phthalate (BzBP), and..., Aylward [/bib_ref] [bib_ref] Biomonitoring Equivalents for di-isononyl phthalate (DINP), Hays [/bib_ref].
Although current exposure doses in the general population are below the tolerance limits reported by environmental agencies, certainly population groups, especially children, are exposed to high levels of phthalates. Studies of the effects of phthalates from early life stage exposures are warranted.
## Phthalate metabolites in serum
The biomonitoring studies of human phthalate exposure have been based on urinary concentrations of phthalate metabolites. However, when only serum was available for analysis, MEP and MiBP representing low molecular weight phthalates, and MECPP and MCiOP representing high molecular weight phthalates, have been used as indicators of phthalate exposure [bib_ref] Correlations between phthalate metabolites in urine, serum, and seminal plasma from young..., Frederiksen [/bib_ref]. A study reported the correlations of phthalate metabolite concentrations among urine, serum, and seminal plasma of young Danish men [bib_ref] Correlations between phthalate metabolites in urine, serum, and seminal plasma from young..., Frederiksen [/bib_ref]. The mean concentrations of MEP, MBP, and DEHP metabolites were one to two orders of magnitude lower in serum (MEP: 4.2, MBP: 0.4, and DEHP: 7.6 µg/L) and seminal plasma (1.0, 0.8, and 0.6 µg/L) than in urine (326, 42.5, and 115 µg/L). Another study, however, showed that the distribution pattern of monoester metabolites in serum was similar to that of urine [bib_ref] Glucuronidation patterns of common urinary and serum monoester phthalate metabolites, Silva [/bib_ref] , especially for MEHP (the metabolite of DEHP) [bib_ref] Glucuronidation patterns of common urinary and serum monoester phthalate metabolites, Silva [/bib_ref]. Nevertheless, MEHHP, MEOHP, MECPP, and MCMHP were found at much higher concentrations in urine than in serum [bib_ref] New metabolites of di(2-ethylhexyl)phthalate (DEHP) in human urine and serum after single..., Koch [/bib_ref]. The presence of MEHP in serum was more likely related to contamination that arises from sampling devices.
Whole blood and cord blood samples from 128 healthy pregnant women and their newborns were analyzed for phthalate metabolites. Median concentrations of MEHHP and MEOHP were 0.31 and <LOD µg/L in maternal blood and 0.32 and <LOD µg/L in cord blood, respectively. MEHHP and MEOHP also were reported to occur in the placenta at concentrations of 0.09 and <LOD ng/g [bib_ref] Association of diethylhexyl phthalate with obesity-related markers and body mass change from..., Kim [/bib_ref]. MBP, MEHP, MEP, and MiBP were detected in blood serum at median concentrations of 0.54 0.49, 0.50, and 0.5 µg/L, respectively [bib_ref] Phthalate diesters and their metabolites in human breast milk, blood or serum,..., Högberg [/bib_ref] , and these concentrations were at least an order of magnitude lower than those measured in urine.
In the serum of patients who were undergoing dialysis [bib_ref] Quantitative Detection of Nine Phthalate Metabolites in Human Serum Using Reversed-Phase High-Performance..., Kato [/bib_ref] [bib_ref] Phthalic acid is the main metabolite of the plasticizer di(2-ethylhexyl) phthalate in..., Mettang [/bib_ref] [bib_ref] Circulating concentrations of di(2-ethylhexyl) phthalate and its de-esterified phthalic acid products following..., Pollack [/bib_ref] [bib_ref] Association between some endocrine-disrupting chemicals and childhood obesity in biological samples of..., Choi [/bib_ref] , phthalate acid (PA) was found as a metabolite of phthalates at remarkably high concentrations of 5.22 ± 3.94 mg/L [bib_ref] Circulating concentrations of di(2-ethylhexyl) phthalate and its de-esterified phthalic acid products following..., Pollack [/bib_ref]. Another study also reported the occurrence of PA in serum (0.205 ± 0.067 mg/L) of patients who were undergoing dialysis [bib_ref] Phthalic acid is the main metabolite of the plasticizer di(2-ethylhexyl) phthalate in..., Mettang [/bib_ref]. Accumulation of PA in patients who are undergoing dialysis has been suggested [bib_ref] Association between some endocrine-disrupting chemicals and childhood obesity in biological samples of..., Choi [/bib_ref]. Serum concentrations of MEHP and DEHP were reported in autistic children [bib_ref] Increased serum phthalates (MEHP, DEHP) and bisphenol A concentrations in children with..., Kardas [/bib_ref].
## Phthalate metabolites in amniotic fluid, breast milk, semen, and saliva
MBP was found in >93% of amniotic fluid samples collected from the US [bib_ref] Detection of phthalate metabolites in human amniotic fluid, Silva [/bib_ref] at concentrations two-to threefold lower than those of serum and four-to sevenfold lower than those of urine [bib_ref] Detection of phthalate metabolites in human amniotic fluid, Silva [/bib_ref]. Studies have reported the occurrence of phthalates in breast milk [bib_ref] Concentrations of phthalate metabolites in breast milk in Korea: Estimating exposure to..., Kim [/bib_ref] ; the reported concentrations in breast milk were much lower than those in urine but similar to those in amniotic fluid. Monoester metabolites of phthalates were measured in breast milk from 33 lactating mothers in North Carolina. MCPP (0.2 µg/L) and MEOHP (0.3 µg/L), MECPP (0.1-0.4 µg/L), and MEHHP (0.2-0.3 µg/L) were detected in some samples. MiNP was the major metabolite found in breast milk collected from mothers in Denmark (101 µg/L) and Finland (89 µg/L) [bib_ref] Human breast milk contamination with phthalates and alterations of endogenous reproductive hormones..., Main [/bib_ref] [bib_ref] Determination of phthalate monoesters in human milk, consumer milk, and infant formula..., Mortensen [/bib_ref]. Median concentrations of MBP, MBzP, and MEHP in breast milk were 0.54, 0.50, and 0.49 µg/L, respectively [bib_ref] Phthalate diesters and their metabolites in human breast milk, blood or serum,..., Högberg [/bib_ref].
Human saliva samples (n = 39) also contained phthalate metabolites [bib_ref] Detection of phthalate metabolites in human saliva, Silva [/bib_ref] [bib_ref] Quantifying phthalate metabolites in human meconium and semen using automated off-line solid-phase..., Kato [/bib_ref]. Salivary concentrations of phthalate metabolites in 39 adult volunteers were in the ranges of <1 to 10.6 µg/L for PA, 91.4 µg/L for MEP, 65.8 µg/L for MBP, and 354 µg/L for MBzP. MBP was the most (85%) frequently detected compound in saliva [bib_ref] Detection of phthalate metabolites in human saliva, Silva [/bib_ref]. Two phthalate metabolites (2.2 µg/L MCPP and 2.3 µg/L MECPP) were detected in a saliva sample from a US woman.
MBP and MBzP were found in semen from US men [bib_ref] Phthalate exposure and human semen parameters, Duty [/bib_ref] [bib_ref] Altered semen quality in relation to urinary concentrations of phthalate monoester and..., Hauser [/bib_ref]. High concentrations of DEHP and its metabolites ( 40.6 µg/L) were found in semen from German men [bib_ref] Urinary di(2-ethylhexyl)phthalate (DEHP)-Metabolites and male human markers of reproductive function, Herr [/bib_ref]. Studies have also indicated that semen quality can be affected by environmentally relevant phthalate exposures [bib_ref] The internal exposure of Taiwanese to phthalate-An evidence of intensive use of..., Chen [/bib_ref]. Further, DEHP (4.20 µg/L) and DBP (2.06 µg/L) were reported at high concentrations in male seminal plasma from men in the US. The metabolites of DEHP ( 0.98 µg/L) and MBP (2.97 µg/L) also were present in considerable concentrations in seminal plasma in the same study [bib_ref] Endocrine disrupting chemicals in seminal plasma and couple fecundity, Louis [/bib_ref]. These results suggested that phthalate metabolites can partition in seminal plasma. Similarly, DEHP (2.09 µg/L) and DBP (1.75 µg/L), as well as their metabolites, were found as the predominant phthalates/phthalate metabolites in seminal plasma from male partners who were planning for pregnancy. This study showed adverse associations between seminal phthalate metabolite concentrations and semen quality [bib_ref] Preconception seminal plasma concentrations of endocrine disrupting chemicals in relation to semen..., Smarr [/bib_ref].
Phthalate metabolites were measured in nail samples from Belgium, and the total concentrations ranged between <12 and 7980 ng/g. It should be noted, however, that some phthalates, especially DBP, are used in nail polishes and that care should be exercised in interpreting such measurements. MEHP, MBP, and MEP were the major metabolites detected in every nail sample, with a median concentration of 138, 74, and 64 ng/g, respectively [bib_ref] Ultrasound assisted extraction combined with dispersive liquid-liquid microextraction (US-DLLME)-A fast new approach..., Alves [/bib_ref]. Another study of nail samples from Oslo, Norway, showed the presence of monoesters, such as MMP (geometric mean 89.7 ng/g), MEP (104.8 ng/g), and MBP (89.3 ng/g) [bib_ref] Evaluation of exposure to phthalate esters and DINCH in urine and nails..., Giovanoulis [/bib_ref]. The utility of other biologic matrices, such as blood, breast milk, semen, and nails, for assessing human exposure to phthalates remains largely unknown due to the limited data.
## Select epidemiological studies linking phthalate exposure and health outcomes
Controlled laboratory animal studies on the toxic effects of phthalates have enabled understanding of biological plausibility and potential mechanisms of actions of this class of chemicals. Thus far, the majority of the laboratory animal exposure/toxicity studies have focused on DEHP and DBP/DiBP, with limited studies examining the toxicities of other phthalates . The reproductive and developmental effects of phthalates are among the most studied and well-described toxic endpoints in those studies. The toxic endpoints determined in animal studies, following phthalate exposure, include retention of nipples, anogenital distance, pathological changes in testes and male reproductive accessory glands, hypospadias, cryptorchidism, and semen parameters. Phthalates have well-documented anti-androgenic activity in rodent studies that result in reduced circulating testosterone. Several reviews have been published on the toxicity of phthalates [bib_ref] Reproductive and developmental effects of phthalate diesters in males, Kay [/bib_ref] [bib_ref] Modes of action and species-specific effects of di-(2-ethylhexyl)phthalate in the liver, Rusyn [/bib_ref] [bib_ref] Sterner-Kock, A.; Chahoud, I. A dose-response study following in utero and lactational..., Andrade [/bib_ref] [bib_ref] In utero exposure to di-(2-ethylhexyl) phthalate exerts both short-term and long-lasting suppressive..., Culty [/bib_ref] [bib_ref] Transgenerational Effects of Di (2-Ethylhexyl) Phthalate in the Male CRL:CD(SD) Rat: Added..., Gray [/bib_ref]. As a class of well-studied endocrine disrupting chemicals, exposure to phthalates has been linked to sex anomalies, endometriosis, altered reproductive development, early puberty and fertility, breast and skin cancer, allergy and asthma, overweight and obesity, insulin resistance, and type II diabetes.
## Diabetes
Diabetes is a metabolic disease that results in elevated blood glucose levels. Epidemiological studies in the US [bib_ref] Association of urinary concentrations of bisphenol A and phthalate metabolites with risk..., Sun [/bib_ref] reported that women with higher urinary concentrations of MBP, MiBP, MBzP, and MCPP and those of DEHP metabolites showed increased risk of diagnosis for diabetes in comparison with those who had lower concentrations of phthalates. Phthalate exposures have been shown to result in insulin resistance [bib_ref] Endocrine disruptors: From endocrine to metabolic disruption, Casals-Casas [/bib_ref] [bib_ref] Urinary phthalates are associated with higher blood pressure in childhood, Trasande [/bib_ref].
## Overweight and obesity
Overweight and obesity can be associated with many chronic diseases, including diabetes. Phthalate exposure was associated with increased body mass and waist circumference [bib_ref] Urinary phthalate metabolites are associated with body mass index and waist circumference..., Wang [/bib_ref]. Some phthalate metabolites (MEP, MBP, and MiBP) were associated with obesity in children, whereas MEHP, MECPP, MEHHP, MEOHP, MBzP, and MCNP were associated with obesity in adults. Further, DEHP metabolites were found to be significantly associated with obesity in adult females and older males [bib_ref] Age and sex differences in childhood and adulthood obesity association with phthalates:..., Buser [/bib_ref]. Urinary concentration of MBP was associated with fat deposition in boys in China [bib_ref] Age and sex-specific relationships between phthalate exposures and obesity in Chinese children..., Zhang [/bib_ref].
Several studies have shown a significant association between obesity and phthalate exposure [bib_ref] Age and sex differences in childhood and adulthood obesity association with phthalates:..., Buser [/bib_ref] [bib_ref] Association of urinary phthalate metabolite concentrations with body mass index and waist..., Hatch [/bib_ref]. MEP, MEHP, MBzP, MEHHP, and MEOHP were associated with obesity in the US population [bib_ref] Association of urinary phthalate metabolite concentrations with body mass index and waist..., Hatch [/bib_ref]. MBzP, MEHHP, MEOHP, and MEP were associated with increased waist circumference and BMIIn contrast, higher concentrations of MEP and DEHP were found in the serum and urine of individuals who were undergoing weight loss [bib_ref] Phthalate metabolites in obese individuals undergoing weight loss: Urinary levels and estimation..., Dirtu [/bib_ref]. Food intake is the main source of phthalate exposure (for high molecular weight phthalates). Therefore, overweight population with high food intake might have high phthalate exposures.
## Allergy and asthma
Exposure to high molecular weight phthalates are is associated with allergies and asthma [bib_ref] Phthalate exposure and children's health, Braun [/bib_ref] [bib_ref] Phthalate exposure and allergy in the U.S. population: Results from NHANES, Hoppin [/bib_ref]. Studies indicated that children are prone to exposure to DEHP, BzBP, DBP, and DEP and that exposure was associated with allergic rhinitis, atopic dermatitis, and conjunctivitis [bib_ref] Exposure to house dust phthalates in relation to asthma and allergies in..., Bamai [/bib_ref]. DEHP and BzBP and their monoesters are regarded as allergens, and exposure to them has been associated with asthma and wheezing in adults [bib_ref] Phthalate exposure and children's health, Braun [/bib_ref] [bib_ref] Phthalate exposure and allergy in the U.S. population: Results from NHANES, Hoppin [/bib_ref]. Exposure of DEHP, BzBP, DBP, and DEP during gestation has been associated with allergic responses in infants and toddlers [bib_ref] Phthalate exposure and children's health, Braun [/bib_ref]. Urinary MEHP concentrations are correlated with asthma in children [bib_ref] Effects ofphthalate exposure on asthma may be mediated through alterations in DNAmethylation, Wang [/bib_ref]. Prenatal exposure to DEHP metabolites and BzBP has been associated with the risk of developing asthma at the age of 7 years and older [bib_ref] Prenatal exposure to bisphenol A and phthalates and childhood respiratory tract infections..., Gascon [/bib_ref].
## Reproductive health
Urinary MEP and MBP and the metabolites of DEHP and DiNP are associated with anomalies in pubertal development in girls [bib_ref] High urinary phthalate concentration associated with delayed pubarche in girls, Frederiksen [/bib_ref]. A significant association between urinary concentrations of MBzP, MEHP, and MEP and increased risk of endometriosis was found in women [bib_ref] Phthalates and risk of endometriosis, Upson [/bib_ref]. Exposure to MEP, MiBP, and MBP pose an increased risk of pregnancy loss in Chinese women.
Poor semen quality was associated with exposure to phthalate metabolites. MBP and MBzP were strongly associated with spermatotoxicity and subfertility in males [bib_ref] Phthalate exposure and human semen parameters, Duty [/bib_ref] [bib_ref] Altered semen quality in relation to urinary concentrations of phthalate monoester and..., Hauser [/bib_ref]. Significantly higher concentrations of DEHP (4.66 µg/mL) and MEHP (3.19 µg/mL) were found in the urine of 40 Turkish boys with gynecomastia as compared to that of control groups [bib_ref] Plasma phthalate levels in pubertal gynecomastia, Durmaz [/bib_ref]. Several reviews have appeared on the reproductive and developmental toxicities of phthalates [bib_ref] Phthalate exposure and male reproductive outcomes: A systematic review of human epidemiological..., Radke [/bib_ref]. Whereas some inconsistencies exist across phthalates for specific health outcomes associated with exposures, moderate to strong evidence of male reproductive effects have been demonstrated in the literature [bib_ref] Phthalate exposure and male reproductive outcomes: A systematic review of human epidemiological..., Radke [/bib_ref]. Because humans are exposed to thousands of harmful chemicals, establishing the link between exposure to a single substance class and adverse health outcomes is fraught with uncertainties.
# Conclusions and perspectives
Human biomonitoring studies are useful in elucidating exposures and body burdens of phthalates at a population level. Although the sources of exposure to phthalates are well described, several questions about cumulative exposures to phthalates throughout the life span, relative contributions of various sources to cumulative exposures, and mixed exposures that may include phthalates or other chemicals that may elicit common adverse outcomes remain unanswered. Biomonitoring studies clearly demonstrate that human exposures are almost ubiquitous, and, in most cases, children have higher exposures than do adults. The existing studies indicate that the observed associations between phthalate exposure and disease outcomes are exploratory and preliminary, the health effects of phthalate exposure warrant further study. Robust analytical methods exist to measure more than 20 phthalate metabolites in urine, a preferred matrix of choice for biomonitoring studies. Although studies have reported the occurrence of phthalate metabolites in other human specimens, including serum, seminal plasma, and amniotic fluid, the relevance of these matrices in understanding toxic effects needs further investigation. Although biomonitoring studies select major biomarkers/metabolites of phthalates, several other intermediate and transformation products of phthalates appear to exist in human specimens. These intermediates may have more pronounced effects on health. Lack of analytical standards hinders the identification of those intermediate biological transformation products of phthalates. Further, the interaction of phthalate metabolites with other contaminants should be considered in future investigations.
There is a lack of biomonitoring data on phthalate exposures in developing countries in Africa and South America. Studies are needed in those regions with regard to exposures and associated health outcomes in populations. Further, epigenetic effects of phthalate exposures warrant further investigation.
Funding: Research reported in this publication was supported, in part, by the National Institute of Environmental Health Sciences of the National Institutes of Health under Award No. U2CES026542-01. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
## Conflicts of interest:
The authors declare no conflict of interest.
[fig] Figure 1: Chemical structures of major phthalates and their metabolites studied in the literature. [/fig]
[fig] Figure 2: Metabolic pathways of phthalate esters in humans. [/fig]
[fig] Figure 3: Urinary concentrations of phthalate metabolites reported in adults (general population) from several countries (MDEHP: Sum of DEHP metabolites; biomonitoring data published after 2000; median concentration is presented). [/fig]
[fig] Figure 4: Urinary concentrations of phthalate metabolites reported in children (general population) from several countries (MDEHP: Sum of DEHP metabolites; biomonitoring data published after 2000, median concentration is presented). [/fig]
[table] Table 1: Major phthalate diesters and their corresponding metabolites studied in the literature.Table 2. Human exposure doses to total phthalates for the US population through various pathways (µg/kg bw/d). [/table]
[table] Table 3: Reported concentrations of major phthalate metabolites in human specimens collected from various countries. [/table]
|
Extracellular Vesicles As Modulators of Tumor Microenvironment and Disease Progression in Glioma
Diffuse gliomas are lethal tumors of the central nervous system (CNS) characterized by infiltrative growth, aggressive nature, and therapeutic resistance. The recent 2016 WHO classification for CNS tumors categorizes diffuse glioma into two major types that include IDH wild-type glioblastoma, which is the predominant type and IDH-mutant glioblastoma, which is less common and displays better prognosis. Recent studies suggest presence of a distinct cell population with stem cell features termed as glioma stem cells (GSCs) to be causal in driving tumor growth in glioblastoma. The presence of a stem and progenitor population possibly makes glioblastoma highly heterogeneous. Significantly, tumor growth is driven by interaction of cells residing within the tumor with the surrounding milieu termed as the tumor microenvironment. It comprises of various cell types such as endothelial cells, secreted factors, and the surrounding extracellular matrix, which altogether help perpetuate the proliferation of GSCs. One of the important mediators critical to the cross talk is extracellular vesicles (EVs). These nano-sized vesicles play important roles in intercellular communication by transporting bioactive molecules into the surrounding milieu, thereby altering cellular functions and/ or reprogramming recipient cells. With the growing information on the contribution of EVs in modulation of the tumor microenvironment, it is important to determine their role in both supporting as well as promoting tumor growth in glioma. In this review, we provide a comprehensive overview of the role of EVs in tumor progression and glioma pathogenesis.
Diffuse gliomas are lethal tumors of the central nervous system (CNS) characterized by infiltrative growth, aggressive nature, and therapeutic resistance. The recent 2016 WHO classification for CNS tumors categorizes diffuse glioma into two major types that include IDH wild-type glioblastoma, which is the predominant type and IDH-mutant glioblastoma, which is less common and displays better prognosis. Recent studies suggest presence of a distinct cell population with stem cell features termed as glioma stem cells (GSCs) to be causal in driving tumor growth in glioblastoma. The presence of a stem and progenitor population possibly makes glioblastoma highly heterogeneous. Significantly, tumor growth is driven by interaction of cells residing within the tumor with the surrounding milieu termed as the tumor microenvironment. It comprises of various cell types such as endothelial cells, secreted factors, and the surrounding extracellular matrix, which altogether help perpetuate the proliferation of GSCs. One of the important mediators critical to the cross talk is extracellular vesicles (EVs). These nano-sized vesicles play important roles in intercellular communication by transporting bioactive molecules into the surrounding milieu, thereby altering cellular functions and/ or reprogramming recipient cells. With the growing information on the contribution of EVs in modulation of the tumor microenvironment, it is important to determine their role in both supporting as well as promoting tumor growth in glioma. In this review, we provide a comprehensive overview of the role of EVs in tumor progression and glioma pathogenesis.
Keywords: glioblastoma, microenvironment, angiogenesis, microRnAs, extracellular vesicles inTRODUCTiOn Adult diffuse gliomas are histopathologically categorized into grades II-IV as oligodendroglioma, oligo-astrocytoma, astrocytoma, and glioblastoma [bib_ref] The 2007 WHO classification of tumours of the central nervous system, Louis [/bib_ref]. Glioblastoma are highly aggressive and angiogenic tumors with median survival of 12-15 months. To ensure appropriate treatment, it is important to correctly grade glial tumors. However, due to the inter-and intraobserver variability encountered in histopathological analyses of gliomas, tumor grading based on gene expression and DNA methylation signatures is gathering importance. The recent glioma classification is based on the status of mutations of isocitrate dehydrogenase genes 1 and 2 (IDH1/2), 1p/19q co-deletion, ATRX alterations, and TERT promoter mutation status [bib_ref] IDH mutation, 1p19q codeletion and ATRX loss in WHO grade II gliomas, Leeper [/bib_ref]. Using TCGA data that consider genomic signatures, survival time, patient age, and treatment responses, glioblastoma is further subclassified into proneural (PN), neural (N), classical (C), and mesenchymal (MES) subtypes. Here, genomic aberrations in expression of EGFR, NF1, and PDGFRA/IDH1 define the classical, MES, and PN subtypes, respectively (3). MES subtypes demonstrate poor survival as compared to PN, thereby necessitating determination of correct molecular signatures in glioblastoma [bib_ref] Integrated genomic analysis identifies clinically relevant subtypes of glioblastoma characterized by abnormalities..., Verhaak [/bib_ref].
A predominant feature of glioblastoma is the high level of inter-and intracellular heterogeneity, due to the presence of cell population within the tumor that shows various stages of differentiation. Glioblastoma is considered to be propagated by a specific subpopulation of glioma stem cells (GSCs) that are responsible for therapeutic resistance and recurrence [bib_ref] Cancer stem cells in glioblastoma, Lathia [/bib_ref] [bib_ref] Identification of human brain tumour initiating cells, Singh [/bib_ref] [bib_ref] Spontaneous transformation of human adult nontumorigenic stem cells to cancer stem cells..., Shiras [/bib_ref]. The GSCs maintain two mutually exclusive molecular identities, i.e., PN or MES. Following therapy, GSCs display phenotypic transition from Proneural to MES subtype leading to tumor progression and increased aggressiveness. Also, the tumor microenvironment contributes toward a MES signature in glioblastoma [bib_ref] Mesenchymal differentiation mediated by NF-kappaB promotes radiation resistance in glioblastoma, Bhat [/bib_ref]. Interestingly, the stroma in which the GSC pool resides is considered to be the GSC niche and is responsible for tumor aggressiveness. The niche can either be a perivascular niche in which GSCs reside in close proximity to the tumor vasculature or a niche invaded by GSCs where cancer cells co-opt normal blood vessels enabling their migration into brain parenchyma or a hypoxic tumor niche [bib_ref] Glioblastoma: defining tumor niches, Hambardzumyan [/bib_ref]. Angiogenesis is induced by the production of high levels of proteins such as VEGF, FGF, and PDGF by glioma cells [bib_ref] Growth factors in glioma angiogenesis: FGFs, PDGF, EGF, and TGFs, Dunn [/bib_ref]. These factors induce proliferation of endothelial cells that not only help to recruit bone marrow-derived endothelial cells and pericyte precursors but also cause cancer cells to transdifferentiate into endothelial cells or pericytes [bib_ref] Tumour vascularization via endothelial differentiation of glioblastoma stemlike cells, Ricci-Vitiani [/bib_ref] [bib_ref] Glioblastoma stem-like cells give rise to tumour endothelium, Wang [/bib_ref]. This may disrupt the blood-brain barrier (BBB) and lead to treatment failure. The GSCs support growth and the infiltrative character of other cancer cells in a paracrine and autocrine manner by secreting angiocrine factors, cytokines, and chemokines [bib_ref] Autocrine endothelin-3/endothelin receptor B signaling maintains cellular and molecular properties of glioblastoma..., Liu [/bib_ref]. In a hypoxic microenvironment, this creates a permissive atmosphere for malignant progression [bib_ref] Autocrine factors that sustain glioma invasion and paracrine biology in the brain..., Hoelzinger [/bib_ref] [bib_ref] Glioblastoma microvesicles transport RNA and proteins that promote tumour growth and provide..., Oka [/bib_ref]. Several mechanisms exist that help mediate cross talk of GSCs and the surrounding tumor microenvironment. Prominent among them is communication of cancer cells with the outside environment (within its microenvironment and even at distant sites) through extracellular vesicle (EV)-mediated transport. In summary, tumor propagation is a cumulative effect of the GSC population and their communication with the microenvironment that includes the tumor vasculature, immune cells, and non-stem cells. This complex biological network arising from intracellular, intercellular, and distant cell interactions supports growth of aggressive and therapy-resistant glioblastoma tumors.
Molecules that are important in reprogramming, metabolism, and angiogenesis are packaged into EVs and transported to proximal or distant cells, affecting proliferation and angiogenesis (16). These EVs serve as carriers of various types of molecules such as lipids, proteins, mRNAs, miRNAs, long non-coding RNAs (lncRNAs), and DNA. EVs can also directly activate cell surface receptors via bioactive ligands and transfer these to neighboring cells, along with transcription factors, oncogenes or infectious particles [bib_ref] Exosome/ microvesicle-mediated epigenetic reprogramming of cells, Camussi [/bib_ref] , and modulate tumor microenvironment [fig_ref] TABLe 1 |: Composition of putative biomolecules in glioblastoma-derived EVs and their respective functions [/fig_ref]. In this review, we elaborate on the role of EVs in glioblastoma pathogenesis.
## Ev structure, biogenesis, and molecular contents
The EVs are phospholipid bilayer-enclosed vesicles secreted by various cell types displaying a size range between 30 and 1,000 nm. They are broadly categorized into microvesicles (MVs, up to 1,000 nm in diameter) and exosomes (30-100 nm) based on their size, intracellular origin, and biogenesis pathway [bib_ref] Extracellular vesicles: exosomes, microvesicles, and friends, Raposo [/bib_ref] [bib_ref] Exosomes and HIV Gag bud from endosome-like domains of the T cell..., Booth [/bib_ref]. Characteristically, the MVs are formed by outward budding and fission of the cell membrane, whereas exosomes are of endosomal origin [bib_ref] Extracellular vesicles: exosomes, microvesicles, and friends, Raposo [/bib_ref]. The multivesicular body (MVB) formation occurs either through the endosomal sorting complex required for transport (ESCRT) machinery or via an ESCRT-independent manner. The ESCRT machinery consists of four complexes of approximately 30 proteins that are responsible for sequestering ubiquitinated transmembrane proteins in the endosomal membrane followed by their excision in the form of sorted cargo by budding [bib_ref] The ESCRT machinery in endosomal sorting of ubiquitylated membrane proteins, Raiborg [/bib_ref]. The ESCRT-independent manner is mediated via tetraspanin CD63 and enzymes sphingomyelinase, and phospholipase D2 [bib_ref] Ceramide triggers budding of exosome vesicles into multivesicular endosomes, Trajkovic [/bib_ref] [bib_ref] Syntenin-ALIX exosome biogenesis and budding into multivesicular bodies are controlled by ARF6..., Ghossoub [/bib_ref]. showed that the heparin sulfate proteoglycan syndecan and its cytoplasmic adaptor syntenin have roles in exosome formation [bib_ref] Syndecan-syntenin-ALIX regulates the biogenesis of exosomes, Baietti [/bib_ref]. Several posttranslational modifications are involved in the sorting of specific proteins into exosomes, like SUMOylation of heterogeneous nuclear ribonucleoproteins A2/ B1 that promotes the sorting of specific microRNAs into exosomes and also regulates sorting of α-synuclein into EVs [bib_ref] Sumoylated hnRNPA2B1 controls the sorting of miRNAs into exosomes through binding to..., Villarroya-Beltri [/bib_ref] [bib_ref] Extracellular vesicle sorting of alpha-synuclein is regulated by sumoylation, Kunadt [/bib_ref].
Interestingly, exosome secretion is mediated through SNARE and Rab proteins (RAB7, RAB11, RAB27, and RAB35) [bib_ref] Biogenesis, secretion, and intercellular interactions of exosomes and other extracellular vesicles, Colombo [/bib_ref]. The release of EVs followed by their uptake in recipient cells and delivery of cargo may occur in various ways. It occurs either by direct fusion of EVs with the plasma membrane of recipient cells or through fusion with the endosomal membrane following acidification [bib_ref] Obstacles and opportunities in the functional analysis of extracellular vesicle RNA -an..., Mateescu [/bib_ref]. Hsu et al. demonstrated that Rab3 helps in exosome secretion by facilitating the docking and tethering of MVBs to the plasma membrane [bib_ref] Regulation of exosome secretion by Rab35 and its GTPase-activating proteins TBC1D10A-C, Hsu [/bib_ref]. Non-canonical Wnt5a-Ca++ signaling was shown to induce release of exosomes into the extracellular environment of melanoma cells [bib_ref] WNT5A induces release of exosomes containing pro-angiogenic and immunosuppressive factors from malignant..., Ekstrom [/bib_ref]. Interestingly, the release of exosomes by tumor suppressor activated pathway 6 (TSAP6) gene occurs in a p53-dependent manner [bib_ref] Exosome secretion, including the DNA damage-induced p53-dependent secretory pathway, is severely compromised..., Lespagnol [/bib_ref]. Another posttranslational modification, ISGylation was shown to be important in the control of exosome production ISGylation of MVB proteins such as TSG101 regulated exosome release by triggering MVB colocalization with lysosomes and promoted degradation of MVB proteins [bib_ref] ISGylation controls exosome secretion by promoting lysosomal degradation of MVB proteins, Villarroya-Beltri [/bib_ref]. Although the formation of MVs is controlled by ADP-ribosylation factor 6 and membrane lipid microdomains (52), mechanisms responsible for sorting of cargo into the lumen of MVBs that form exosomes are not fully understood [bib_ref] Exosomes: from biogenesis and secretion to biological function, Keller [/bib_ref].
## Role of evs in cellular cross talk and glioblastoma progression
Tumor-derived EVs act as a multicomponent delivery vehicle to transfer genetic information as well as signaling proteins to cells in their vicinity as well as at distant sites . Numerous functions are attributed to EVs in cancer that range from their role in antitumor immunity, drug resistance, metastasis, angiogenesis, and intercellular communication to reprogramming [bib_ref] Oncogenic extracellular vesicles in brain tumor progression, D' Asti [/bib_ref]. Reprogramming is a process of conversion of differentiated cells into a dedifferentiated state and can be mediated by MVs in in vivo conditions [bib_ref] Role of extracellular RNA-carrying vesicles in cell differentiation and reprogramming, Quesenberry [/bib_ref]. Glioblastoma-derived MVs are likely to represent one of the mechanisms by which cancer cells change the tumor microenvironment and make it more permissive for growth and invasion [bib_ref] Gliomas and the vascular fragility of the blood brain barrier, Dubois [/bib_ref]. Therefore, it is worth investigating the molecular cargo present in EVs for early glioma detection. The four glioblastoma subtypes activate different pathways of vesicle formation, and each subtype shows significant differences in expression of the EV regulatory and biogenesis markers [bib_ref] Extracellular vesicles in the biology of brain tumour stem cells -implications for..., Nakano [/bib_ref]. The molecules present in EVs of which expression was subtype-specific include CD63, CD81, RAB27A, RAB27B, FLOT1, FLOT2, TSG101, RAB 5A among others [bib_ref] Exosomes: from biogenesis and secretion to biological function, Keller [/bib_ref]. Recently, Kowal et al. proposed subcategorization of EVs based on relative abundance of specific EV protein markers such as CD63, CD9, and CD81 (60). Godlewski et al. showed that different subtypes of GSCs show highly heterogenous profiles of miRNAs. Moreover, EV-mediated transfer and secretion of miRNAs may contribute to glioblastoma heterogeneity [bib_ref] On the origin of cancer cells, Warburg [/bib_ref]. Importantly, the effect of phenotypic transition of GSCs from PN to MES signature is reflected significantly in the release and content of EVs [bib_ref] Qualitative changes in the proteome of extracellular vesicles accompanying cancer cell transition..., Garnier [/bib_ref] [bib_ref] Two distinct populations of exosomes are released from LIM1863 colon carcinoma cell-derived..., Tauro [/bib_ref].
The significant contribution of EVs in key cellular processes related to disease progression in glioma is outlined below.
## Metabolic regulation
Glial tumors show propensity for non-oxidative metabolism of glucose even in the presence of oxygen, a phenomenon known as the Warburg effect [bib_ref] On the origin of cancer cells, Warburg [/bib_ref] [bib_ref] The metabolism of carcinoma cells, Warburg [/bib_ref]. Glioblastoma cells were also found to be highly oxidative indicating that substrate oxidation also occurs along with aerobic glycolysis and lactate release [bib_ref] Fatty acid oxidation is required for the respiration and proliferation of malignant..., Lin [/bib_ref]. The GSCs have other metabolic strategies or substrates as compared to bulk tumor cells. The tumor microenvironment and genetic factors contribute immensely toward metabolic reprogramming in glioblastoma. The hypoxic microenvironment within the tumor results in a shift toward glycolysis and shows angiogenic phenotype whereas tumor cells at the edge are highly invasive and heavily dependent on mitochondrial respiration for energy production (66, 67). Kucharzewska et al. showed that hypoxiadependent intercellular signaling in glioblastoma is mediated through exosomes [bib_ref] Exosomes reflect the hypoxic status of glioma cells and mediate hypoxia-dependent activation..., Kucharzewska [/bib_ref]. They showed that hypoxia was associated with secretion of exosomes enriched in hypoxia-regulated mRNAs and proteins such as matrix metalloproteinases, IL-8, PDGFs, caveolin 1, and lysyl oxidases which performed pivotal roles in cellular metabolism and cell proliferation. In addition, mutations in metabolic genes such as IDH1/2 were important in gliomagenesis and had prognostic importance (68). Khurshed et al. showed that energy metabolism differed between IDH1 wild-type and mutant glioma [bib_ref] In silico gene expression analysis reveals glycolysis and acetate anaplerosis in IDH1..., Khurshed [/bib_ref]. IDH1 mutant glioma cells used oxidative TCA cycle for metabolism whereas IDH wild-type glioma was more dependent on glycolytic and lactate metabolism. Recently, EVs isolated specifically from cerebrospinal fluid (CSF) contained information regarding the mutational profile of IDH1 in brain tumors [bib_ref] Metabolic reprogramming in glioblastoma: the influence of cancer metabolism on epigenetics and..., Agnihotri [/bib_ref]. Interestingly, several metabolic enzymes were overexpressed in brain tumors, suggesting that the cancer cells derived energy and nutrients needed for proliferation by transferring these enzymes to surrounding cells through EVs under hypoxic conditions. In addition, mitochondrial DNA was also detected in MVs of glioblastoma cells but its function is not yet understood (36).
## Immune modulation
Tumor-derived MVs were found to be enriched in CD39 and CD73 in various types of cancers such as pancreatic, bladder, and breast cancers. CD39 and CD73 were also highly expressed in gliomas causing adenosinergic immunosuppression (71) but its status in glioma EVs is not known. Glioma-derived MVs were shown to activate myeloid-derived suppressor cells
## Figure 1 | (i) biogenesis and secretion of extracellular vesicles (evs) such as mvs and exosomes. sorting of cargo molecules in multivesicular bodies (mvbs) occur in an endosomal sorting complex required for transport (escrt)-dependent manner. exosomes are of endosomal origin and their secretion is mediated by rab
GTPase family proteins and the Wnt5a-Ca++ non-canonical pathway. In an alternative pathway, the release of MVs is governed by ADP-ribosylation factor 6 (ARF6) and membrane lipid microdomains. (ii) Uptake of EVs by recipient cells or binding of surface ligands of EVs to recipient cells is followed by downstream molecular cascades resulting in processes like angiogenesis, therapy resistance, immune modulation, and metabolic reprogramming, (A) angiogenesis; tumor-derived EVs modulate the formation of blood vessels, which supports glioma progression, (B) therapy resistance; exosomes or MVs carry cytokines, which may further activate STAT3 protein via cytokine receptors and ultimately leads to proneural-mesenchymal transition (PMT) and a radiation-resistant phenotype of glioma [bib_ref] STAT3 blockade inhibits radiation-induced malignant progression in glioma, Lau [/bib_ref]. Activation of STAT3 signaling also promotes temozolomide resistance of glioma [bib_ref] STAT3 inhibition overcomes temozolomide resistance in glioblastoma by downregulating MGMT expression, Kohsaka [/bib_ref]. (MDSCs) [bib_ref] Membrane-associated Hsp72 from tumor-derived exosomes mediates STAT3-dependent immunosuppressive function of mouse and..., Chalmin [/bib_ref]. MDSCs modulate immune activity by inhibiting T-cell responses [bib_ref] Regulation of suppressive function of myeloid-derived suppressor cells by CD4+ T cells, Nagaraj [/bib_ref]. Moreover, glioblastoma-derived MVs were shown to contain IL-6 that has a role in phosphorylation of STAT-3 on MDSCs, causing immunosuppression. In addition, TGF-β in MVs caused similar effect in gliomas [bib_ref] Proteomic and immunologic analyses of brain tumor exosomes, Graner [/bib_ref]. van der Vos et al. showed that glioma-derived EVs transferred miR-451 and miR-21 to microglia/macrophages leading to downregulation of their targets [bib_ref] Directly visualized glioblastoma-derived extracellular vesicles transfer RNA to microglia/macrophages in the brain, Van Der Vos [/bib_ref] , whereas uptake of GSC exosomes by monocytes caused failure to mount an immune response against glioma cells [bib_ref] TMIC-09 glioblastoma stem cell-derived exsomes promote M2 polarization of human monocytes, Gabrusiewicz [/bib_ref] [bib_ref] Glioblastoma-derived extracellular vesicles modify the phenotype of monocytic cells, De Vrij [/bib_ref]. Furthermore, glioma cell-derived exosomes suppressed T-cell immune responses by acting on monocyte maturation rather than affecting their direct interaction with T cells [bib_ref] Systemic T cells immunosuppression of glioma stem cell-derived exosomes is mediated by..., Domenis [/bib_ref]. Moreover, glioma-derived MVs were restricted in their capacity to directly prime peripheral immunosuppression [bib_ref] The limited capacity of malignant glioma-derived exosomes to suppress peripheral immune effectors, Iorgulescu [/bib_ref]. Hence, the role of MVs in immune suppression needs further investigation.
## Angiogenesis
Proteins that are expressed under hypoxic conditions, such as HIF are responsible for angiogenesis in glioblastoma [bib_ref] Hypoxia and the hypoxia-inducible-factor pathway in glioma growth and angiogenesis, Kaur [/bib_ref].
Glioblastoma-derived MVs contain VEGF, angiogenin, IL-8, PDGF, and miRNA-19b, and it has been shown that VEGF and angiogenin bind to the cognate receptor on the surface of ECs and promote angiogenesis [bib_ref] Exosomes reflect the hypoxic status of glioma cells and mediate hypoxia-dependent activation..., Kucharzewska [/bib_ref]. Instead, miR-19b-mediated angiogenesis by repressing anti-angiogenic proteins such as thrombospondin-1 and connective tissue growth factor within tumors [bib_ref] mir-17-92, a cluster of miRNAs in the midst of the cancer network, Olive [/bib_ref]. Recently, semaphorin3A was found in the exosomes derived from blood or CSF which acted as a pro-permeability factor but with anti-angiogenic function [bib_ref] Extracellular vesicle-transported Semaphorin3A promotes vascular permeability in glioblastoma, Treps [/bib_ref]. Interestingly, angiogenesis was also induced in glioma by exosomes enriched in lncRNA, POU3F3 [bib_ref] Glioma cells promote angiogenesis through the release of exosomes containing long non-coding..., Lang [/bib_ref]. Svensson and Belting demonstrated a significantly increased content of tissue factor (TF) in glioblastoma cell-derived EVs under hypoxic conditions [bib_ref] Role of extracellular membrane vesicles in intercellular communication of the tumour microenvironment, Svensson [/bib_ref]. In addition, EVs were also shown to transfer the oncogenic form of EGFR, EGFRvIII, between glioblastoma cells as well as to ECs, causing phenotypic modulation of recipient cells [bib_ref] Intercellular transfer of the oncogenic receptor EGFRvIII by microvesicles derived from tumour..., Al-Nedawi [/bib_ref]. Moreover, EGFRvIII-transformed cells became hypersensitive to TF/protease activated receptor (TF/PAR-2) signaling. This kind of receptor transfer may cause angiogenic signaling in recipient cells due to regulation of VEGF gene expression by EGFRvIII. This suggests that under hypoxic conditions, even in the absence of external stimuli, tumor angiogenesis is supported through PAR-2 in ECs.
## Evs in tumor growth, invasion, and therapy resistance
There are several tumor cell resistance mechanisms that affect therapy response in glioblastoma such as
- The cross talk of GSCs with the tumor microenvironment.
- Decreased drug uptake, increased drug efflux and intracellular drug inactivation. - Repair of drug-induced damage or defects in DNA damage response pathway.
Earlier studies showed that chemoresistance of the CD133+ GSC population was due to upregulation of miR-9-2 and MDR1. The protein target of miR-9-2 was patch homolog1. It is expressed at low levels in temozolomide (TMZ)-resistant CD133+ cells in which Gli1 expression was higher [bib_ref] High expression of miR-9 in CD133+ glioblastoma cells in chemoresistance to temozolomide, Munoz [/bib_ref]. Also, miR-9 was shown to be high in TMZ-resistant cells, and MVs were strongly involved in functional delivery of anti-miR-9 from mesenchymal stem cells to glioblastoma cells, imparting TMZ sensitivity [bib_ref] Delivery of functional anti-miR-9 by mesenchymal stem cell-derived exosomes to glioblastoma multiforme..., Munoz [/bib_ref]. Most drugs such as TMZ and cisplatin are alkylating agents and cause DNA damage by exerting their cytotoxic or mutagenic effects on cells. Epigenetic silencing of the DNA repair gene MGMT by promoter methylation compromised DNA repair and was associated with longer survival in glioblastoma patients treated with alkylating agents [bib_ref] MGMT gene silencing and benefit from temozolomide in glioblastoma, Hegi [/bib_ref]. In addition, other DNA repair genes such as ERCC1, ERCC2, MUTYH, and PNKP reduced efficacy of alkylating agents, imparting chemoresistance in glioma [bib_ref] Inhibition of DNA-repair genes Ercc1 and Mgmt enhances temozolomide efficacy in gliomas..., Boccard [/bib_ref]. showed that ERCC1 mRNA levels were upregulated in MVs derived from glioblastoma cells. Also, the mRNA levels of microvesicular MGMT, APNG, or both were elevated in resistant glioblastoma cell lines as compared to sensitive cell lines [bib_ref] Chip-based analysis of exosomal mRNA mediating drug resistance in glioblastoma, Shao [/bib_ref]. Hence, increased patient-derived microvesicular MGMT and APNG mRNA levels are indicative of drug resistance or they predict alkylating drug responses in glioblastoma patients. While MGMT promoter methylation is associated with a better prognosis, mutation/amplification of EGFR is associated with poor prognosis [bib_ref] MGMT gene silencing and benefit from temozolomide in glioblastoma, Hegi [/bib_ref] [bib_ref] PTEN mutation, EGFR amplification, and outcome in patients with anaplastic astrocytoma and..., Smith [/bib_ref]. EGFRvIII protein is transferred in glioblastoma cell-derived MVs, signifying its role as a prognostic biomarker [bib_ref] Intercellular transfer of the oncogenic receptor EGFRvIII by microvesicles derived from tumour..., Al-Nedawi [/bib_ref].
Recent studies also highlight the ability of EVs in promoting glioma growth and invasion. TrkB, a member of the neurotrophin tyrosine kinase receptor-1 family was shown to be highly expressed in exosomes of glioblastoma patients and its level correlated with tumor progression and aggressiveness [bib_ref] TrkB-containing exosomes promote the transfer of glioblastoma aggressiveness to YKL-40-inactivated glioblastoma cells, Pinet [/bib_ref]. Also, it was shown that differential neurotrophin receptor expression levels displayed by exosomes depended on the differentiation status of tumors and YKL-40 expression, thereby making exosomal TrkB a novel biomarker for glioblastoma. In addition, Timp1 as one of the NF-κB target genes with a role in tumor growth was significantly upregulated in exosomes [bib_ref] Exosomes isolated from plasma of glioma patients enrolled in a vaccination trial..., Muller [/bib_ref]. Recently, a circulating protein, CLIC1 with growth stimulatory properties both in vitro and in vivo was identified in EVs of GSCs [bib_ref] Extracellular vesicle-mediated transfer of CLIC1 protein is a novel mechanism for the..., Setti [/bib_ref]. Apart from these molecules enclosed in MVs, a tumor suppressor protein such as PTEN is also exported through exosomes to recipient cells where it suppresses cell proliferation by reducing the abundance of pAkt [bib_ref] The tumor suppressor PTEN is exported in exosomes and has phosphatase activity..., Putz [/bib_ref]. Inhibition of pAkt levels diminished tumor growth and invasion.
## Evs in glioma immune therapy
Exosomes also serve as an attractive candidate for immune therapy of brain tumors as they retain their stability during purification as well as under in vivo conditions. Vaccination with dendritic cell-derived exosomes showed good recovery against malignancy with little adverse effects in phase I and II clinical trials [bib_ref] Vaccination of metastatic melanoma patients with autologous dendritic cell (DC) derived-exosomes: results..., Escudier [/bib_ref] [bib_ref] A phase I study of dexosome immunotherapy in patients with advanced non-small..., Morse [/bib_ref]. Muller et al. found a negative correlation between mRNA expression levels of TIMP1, TGF-β, and IL-8 in exosomes and patient's survival after a vaccination trial in glioblastoma patients. Instead, exosomal mRNA levels of cytokines IL-8 and TGF-β, important in glioma growth and metastasis, showed a clear decrease after vaccination [bib_ref] Exosomes isolated from plasma of glioma patients enrolled in a vaccination trial..., Muller [/bib_ref].
## Mirnas in glioblastoma evs
A large number of microRNAs are found to be encapsulated in EVs in serum of glioma patients. While their functions in relation to microenvironment in glioma are still being explored, they certainly pose great hope as circulating biomarkers for early diagnosis, tumor staging and prognostication. miR-21 and miR-221 were shown to be highly enriched in CSF-derived EVs and serum-derived exosomes of glioblastoma patients, respectively, and hence possessed the potential to serve as a relevant biomarker [bib_ref] MiR-21 in the extracellular vesicles (EVs) of cerebrospinal fluid (CSF): a platform..., Akers [/bib_ref] [bib_ref] Exosomal miR-221 targets DNM3 to induce tumor progression and temozolomide resistance in..., Yang [/bib_ref]. More importantly, the level of miR-221 increased with glioma grades in exosomes. In addition, levels of other miRNAs like miR-24, miR-103, and miR-125 along with miR-21 were also high in exosomes derived from CSF of glioblastoma patients [bib_ref] miRNA contents of cerebrospinal fluid extracellular vesicles in glioblastoma patients, Akers [/bib_ref]. Although several other miRNAs were detected in glioma MVs [fig_ref] TABLe 1 |: Composition of putative biomolecules in glioblastoma-derived EVs and their respective functions [/fig_ref] , their mechanism of action in target cell is largely unknown which limits their use in glioblastoma therapy.
## Future prospects
Exosomes play a critical role in mediating intercellular communication. Being nano-sized and lipid bilayered, they can easily cross the BBB under both physiological as well as abnormal conditions. Moreover, the enclosed biomolecules are stably retained in an active state and are functional after uptake by recipient cells. These characteristics make MVs and/or exosomes candidates for therapeutic applications. Exosomes derived from different cell types can be used to selectively deliver therapeutic nucleic acid analogs (tumor suppressor miRNAs/ncRNA mimics or oncogenic miRNA inhibitors) or conventional drugs for applications in tumor therapy [bib_ref] Extracellular vesicles: biology and emerging therapeutic opportunities, Ela [/bib_ref]. Moreover, the study of molecular cargo of EVs is helpful for the identification of novel biomarkers in disease diagnosis and monitoring [bib_ref] Tumor exosomes: cellular postmen of cancer diagnosis and personalized therapy, Sharma [/bib_ref]. The tumor-derived MVs of glioblastoma show upregulation of several signature molecules, which offer rapid discrimination between tumor-derived EVs and normal cell-derived EVs. This simplifies the diagnosis and circumvents the use of invasive methods such as biopsies.
Interestingly, the EVs derived from CSFs contain RNA signatures reflective of the underlying molecular genetic status of glioblastoma in terms of wt EGFR expression and EGFRvIII status.
The EVs being more enriched in CSF than in serum are easier to detect using non-invasive tools such as PCR or droplet digital PCR [bib_ref] BEAMing and droplet digital PCR analysis of mutant IDH1 mRNA in glioma..., Chen [/bib_ref]. With advances in such technologies, it is possible to identify as well as sub-classify glioblastoma tumors from inaccessible locations. Moreover, we need to overcome safety issues when applying MVs and exosomes as modes for drug delivery in cancer. The use of EVs in medicine is still in its infancy, and there are many potholes to cover, but their utility as diagnostic tools or as delivery vehicles in glioblastoma is an unmet challenge.
# Author contributions
AS conceptualized the review and wrote it. AM and DK prepared the draft and figures; contributed equally to this review. SP helped in preparation of the draft.
# Funding
This review was supported through funding support from Department of Biotechnology, Ministry of Science and Technology, Govt of India, New Delhi, India; Award number: BT/ PR 10852 and intra-mural support from National Centre for Cell Science(NCCS); Pune, India.
## References
[table] TABLe 1 |: Composition of putative biomolecules in glioblastoma-derived EVs and their respective functions. [/table]
|
In How Many Ways is the Approximate Number System Associated with Exact Calculation?
The approximate number system (ANS) has been consistently found to be associated with math achievement. However, little is known about the interactions between the different instantiations of the ANS and in how many ways they are related to exact calculation. In a cross-sectional design, we investigated the relationship between three measures of ANS acuity (non-symbolic comparison, non-symbolic estimation and non-symbolic addition), their cross-sectional trajectories and specific contributions to exact calculation. Children with mathematical difficulties (MD) and typically achieving (TA) controls attending the first six years of formal schooling participated in the study. The MD group exhibited impairments in multiple instantiations of the ANS compared to their TA peers. The ANS acuity measured by all three tasks positively correlated with age in TA children, while no correlation was found between non-symbolic comparison and age in the MD group. The measures of ANS acuity significantly correlated with each other, reflecting at least in part a common numerosity code. Crucially, we found that non-symbolic estimation partially and non-symbolic addition fully mediated the effects of non-symbolic comparison in exact calculation.
# Introduction
An extensive literature that comprises psychophysical [bib_ref] Symbols and quantities in parietal cortex: Elements of a mathematical theory of..., Dehaene [/bib_ref] [bib_ref] Large number discrimination in 6-month-old infants, Xu [/bib_ref] , electrophysiological [bib_ref] Coding of cognitive magnitude: compressed scaling of numerical information in the primate..., Nieder [/bib_ref] , and neuroimaging data [bib_ref] Tuning curves for approximate numerosity in the human intraparietal sulcus, Piazza [/bib_ref] has demonstrated that human infants and adults share an approximate number system (ANS), which is dedicated to representing large magnitudes in an analog fashion. Number representation within the ANS is very similar to the intuition that we have for space and time magnitudes [bib_ref] Space, time, and number: a Kantian research program, Dehaene [/bib_ref] and can be well described by Weber-Fechner's law [bib_ref] Coding of cognitive magnitude: compressed scaling of numerical information in the primate..., Nieder [/bib_ref] [bib_ref] Tuning curves for approximate numerosity in the human intraparietal sulcus, Piazza [/bib_ref] [bib_ref] Calibrating the mental number line, Izard [/bib_ref] [bib_ref] Time required for judgements of numerical inequality, Moyer [/bib_ref]. Because the ANS is already present in newborns [bib_ref] Newborn infants perceive abstract numbers, Izard [/bib_ref] and interacts with culturally derived symbolic representations during development [bib_ref] Varieties of numerical abilities, Dehaene [/bib_ref] , it is considered to be an important start-up tool for the acquisition of mathematical knowledge [bib_ref] Neurocognitive start-up tools for symbolic number representations, Piazza [/bib_ref].
Converging evidence from correlational [bib_ref] Individual differences in nonverbal number acuity correlate with maths achievement, Halberda [/bib_ref] , cross-sectional [bib_ref] Impaired acuity of the approximate number system underlies mathematical learning disability (dyscalculia), Mazzocco [/bib_ref] [bib_ref] Developmental trajectory of number acuity reveals a severe impairment in developmental dyscalculia, Piazza [/bib_ref] , longitudinal [bib_ref] The approximate number system and its relation to early math achievement: evidence..., Bonny [/bib_ref] [bib_ref] Non-symbolic arithmetic abilities and mathematics achievement in the first year of formal..., Gilmore [/bib_ref] [bib_ref] Preschool acuity of the approximate number system correlates with school math ability, Libertus [/bib_ref] [bib_ref] Is approximate number precision a stable predictor of math ability?, Libertus [/bib_ref] [bib_ref] Preschoolers' precision of the approximate number system predicts later school mathematics performance, Mazzocco [/bib_ref] and training studies [bib_ref] Training the approximate number system improves math proficiency, Park [/bib_ref] has provided robust support for the link between the ANS and arithmetics. Halberda et al. [bib_ref] Individual differences in nonverbal number acuity correlate with maths achievement, Halberda [/bib_ref] showed that the Weber fraction calculated from a non-symbolic number comparison task in adolescents retroactively correlated with standardized math achievement scores from Kindergarten up to the sixth grade. A series of other studies replicated this finding using not only general standardized math achievement scores [bib_ref] Preschoolers' precision of the approximate number system predicts later school mathematics performance, Mazzocco [/bib_ref] [bib_ref] Can we predict mathematical learning disabilities from symbolic and non-symbolic comparison tasks..., Desoete [/bib_ref] [bib_ref] Non-verbal number acuity correlates with symbolic mathematics achievement: but only in children, Inglis [/bib_ref] [bib_ref] Typical and atypical development of basic numerical skills in elementary school, Landerl [/bib_ref] but also simple arithmetics operations [bib_ref] Developmental trajectory of number acuity reveals a severe impairment in developmental dyscalculia, Piazza [/bib_ref] [bib_ref] Children's mapping between symbolic and nonsymbolic representations of number, Mundy [/bib_ref].
Importantly, a number of other studies failed to find an association between non-symbolic comparison and math achievement, but rather found significant associations between math achievement and the symbolic version of the task (see [bib_ref] How do symbolic and nonsymbolic numerical magnitude processing skills relate to individual..., Smedt [/bib_ref] for a review of the inconsistency of those findings).
However, two recent meta-analyses confirmed the existence of a robust association between non-symbolic comparison and math achievement from childhood to adulthood. Fazio, Bailey, Thompson and Siegler [bib_ref] Relations of different types of numerical magnitude representations to each other and..., Fazio [/bib_ref] analyzed 19 published studies and found that although non-symbolic processing is less strongly correlated with math achievement compared to symbolic processing, there is a robust and specific significant association between non-symbolic comparison and math achievement. Chen and Li [bib_ref] Association between individual differences in non-symbolic number acuity and math performance: A..., Chen [/bib_ref] investigated 36 cross-sectional studies and found that the association between non-symbolic comparison and math achievement is moderate but statistically significant (r = 0.20, 95% CI = [0.14, 0.26]), even after controlling the effect of general cognitive abilities. Importantly, non-symbolic comparison was found to prospectively predict later math performance (r = 0.24, 95% CI = [0.11, 0.37]; 6 samples) and it is also retrospectively correlated to early math achievement (r = 0.17, 95% CI = [0.07, 0.26]; 5 samples). Based on the estimated effect sizes, the authors conducted power analyses and confirmed that many previous studies failed to find a significant association between non-symbolic comparison and math achievement because of insufficient statistical power due to small sample sizes.
Moreover, other measures of ANS acuity, such as number estimation, were also found to correlate with math achievement in children and adolescents [bib_ref] Impaired acuity of the approximate number system underlies mathematical learning disability (dyscalculia), Mazzocco [/bib_ref] [bib_ref] Preschool acuity of the approximate number system correlates with school math ability, Libertus [/bib_ref] [bib_ref] Numerical and nonnumerical estimation in children with and without mathematical learning disabilities, Mejias [/bib_ref].
Noticeably, longitudinal and training studies have provided evidence for a foundational role of the ANS on the development of math abilities. Using the non-symbolic number comparison task, Mazzoco et al. [bib_ref] Preschoolers' precision of the approximate number system predicts later school mathematics performance, Mazzocco [/bib_ref] showed that the ANS acuity measured prior to formal mathematical instruction was selectively predictive of arithmetics achievement in the first grade (see also Libertus et al. [bib_ref] Is approximate number precision a stable predictor of math ability?, Libertus [/bib_ref]. Similarly, Gilmore et al. [bib_ref] Non-symbolic arithmetic abilities and mathematics achievement in the first year of formal..., Gilmore [/bib_ref] found that non-symbolic calculation abilities measured in a group of Kindergarten children were a robust predictor of later math achievement. Complementary, Park and Brannon [bib_ref] Training the approximate number system improves math proficiency, Park [/bib_ref] showed that training adults in a nonsymbolic addition and subtraction task specifically improves exact addition and subtraction. Interestingly, the ANS acuity was also found to improve with mathematical education. Piazza, Pica, Izard, Spelke, and Dehaene [bib_ref] Education enhances the acuity of the nonverbal approximate number system, Piazza [/bib_ref] investigated a group of Mundurukus, an indigenous population in Brazil that does not have a system for representing exact numbers [bib_ref] Exact and approximate arithmetic in an Amazonian indigene group, Pica [/bib_ref] , and found that the ANS acuity, as quantified by a non-symbolic number comparison task, was modulated by the level of formal instruction at the standard Brazilian school system. This result provides support for a bidirectional association between the most basic forms of number processing and math abilities. Importantly, an analogous bidirectionality has long been found in the reading domain, such as the fact that phonological abilities serve as the base for reading competence and are improved by literacy [bib_ref] Categorizing sounds and learning to read -a causal connection, Bradley [/bib_ref] [bib_ref] Is there a causal link from phonological awareness to success in learning..., Castles [/bib_ref].
Finally, group studies demonstrated that children with developmental dyscalculia (DD), a learning disability specific to calculation, have an impaired ANS compared to their typically achieving (TA) peers. Piazza et al. [bib_ref] Developmental trajectory of number acuity reveals a severe impairment in developmental dyscalculia, Piazza [/bib_ref] showed that the ANS acuity in children with DD at 10 years old, as quantified by the internal Weber fraction, was equivalent to the acuity observed in TA Kindergarten children. Similar results were obtained by Mazzoco et al. [bib_ref] Impaired acuity of the approximate number system underlies mathematical learning disability (dyscalculia), Mazzocco [/bib_ref] , who showed not only that adolescents with DD present higher internal Weber fractions than their TA peers but also that they have an impairment in estimating numerical magnitudes.
Other studies that investigated younger children with DD found only an impairment in the symbolic version of the number comparison task (see review by Noël & Rousselle [bib_ref] Developmental changes in the profiles of dyscalculia: An explanation based on a..., Noël [/bib_ref] , which casts doubt on the assumption of a critical role of the ANS in the acquisition of exact number representations. Chu, van Marle, and Geary [bib_ref] Quantitative deficits of preschool children at risk for mathematical learning disability, Chu [/bib_ref] found that the ANS acuity significantly predicted the risk for DD in children, but measures of symbolic number knowledge were more robust predictors. However, those studies used only one measure to assess the ANS acuity: the non-symbolic number comparison task. Based on Gilmore et al. [bib_ref] Non-symbolic arithmetic abilities and mathematics achievement in the first year of formal..., Gilmore [/bib_ref] it could be the case that different forms of approximate manipulation of numerical information, such as calculation, could be additional important predictors of risk for DD.
Although much progress has been achieved in the establishment of an association between the ANS and arithmetics, it remains largely elusive in how many ways the ANS interacts with exact calculation and through which cognitive mechanisms this association could be grounded. The ANS allows for comparing two different magnitudes, to approximately grasp how many objects are present in a scene and to manipulate quantities using simple operations such as addition and subtraction [bib_ref] Origins of mathematical intuitions: the case of arithmetic, Dehaene [/bib_ref]. In this sense, different tasks have been used to measure the ANS acuity, such as comparison [bib_ref] Individual differences in nonverbal number acuity correlate with maths achievement, Halberda [/bib_ref] [bib_ref] Impaired acuity of the approximate number system underlies mathematical learning disability (dyscalculia), Mazzocco [/bib_ref] [bib_ref] Developmental trajectory of number acuity reveals a severe impairment in developmental dyscalculia, Piazza [/bib_ref] [bib_ref] Basic numerical skills in children with mathematics learning disabilities: a comparison of..., Rousselle [/bib_ref] , estimation [bib_ref] Impaired acuity of the approximate number system underlies mathematical learning disability (dyscalculia), Mazzocco [/bib_ref] [bib_ref] Children's mapping between symbolic and nonsymbolic representations of number, Mundy [/bib_ref] [bib_ref] Numerical and nonnumerical estimation in children with and without mathematical learning disabilities, Mejias [/bib_ref] [bib_ref] Developmental and individual differences in pure numerical estimation, Booth [/bib_ref] and approximate calculation [bib_ref] Non-symbolic arithmetic abilities and mathematics achievement in the first year of formal..., Gilmore [/bib_ref] [bib_ref] Nonsymbolic arithmetic in adults and young children, Barth [/bib_ref] [bib_ref] Defective number module or impaired access? Numerical magnitude processing in first graders..., Smedt [/bib_ref]. Importantly, very little attention was given to the fact that these measures are tapping different instantiations of the ANS. Comparison, estimation and approximate calculation, although possibly operating at the same level of representation (the ANS), involve very different computational processes and consequently could have specific contributions to the development of exact number representations and mathematics. Indeed, Mazzocco et al. [bib_ref] Impaired acuity of the approximate number system underlies mathematical learning disability (dyscalculia), Mazzocco [/bib_ref] found that non-symbolic comparison and estimation accounted for unique proportions of the variance when predicting math achievement.
Moreover, given the complexity of arithmetics, the link between basic number processing (e.g., magnitude comparison) and exact calculation is possibly not direct and might involve the recruitment of other cognitive processes. Indeed, the study by Lyons and Beilock [bib_ref] Numerical ordering ability mediates the relation between number-sense and arithmetic competence, Lyons [/bib_ref] found that the ability to identify the order of a series of digits fully mediated the association between the ANS acuity (as measured with the non-symbolic number comparison task) and exact calculation in adults. Importantly, van Marle, Chu, Li and Geary [bib_ref] Acuity of the approximate number system and preschoolers' quantitative development, Van Marle [/bib_ref] provided a conceptual replication of the study of Lyons and Beilock [bib_ref] Numerical ordering ability mediates the relation between number-sense and arithmetic competence, Lyons [/bib_ref] in children, and proposed that the ANS acuity facilitated the early acquisition of symbolic number knowledge and was indirectly associated with math achievement through this knowledge. In line with these findings, it might also be the case that there is a type of hierarchical association between different instantiations of the ANS, from the most elementary abilities to more complex operations and manipulations of magnitude information. That is, non-symbolic estimation and calculation could be intermediate steps between simple number discrimination and exact calculation.
Surprisingly, to date there is only one study in adults and no study with children that directly compared different measures of the ANS. Gilmore, Attridge, and Inglis [bib_ref] Measuring the approximate number system, Gilmore [/bib_ref] measured the ANS with non-symbolic versions of the number comparison and approximate addition tasks and found null correlations between the internal Weber fractions calculated from each task, placing in doubt the assumption of a single underlying ANS. However, this result is very puzzling and deserves further examination, because both tasks used non-symbolic magnitudes and, even though different cognitive mechanisms might be recruited during performance, both tasks should at least partially activate the representation of numbers and its underlying brain circuitry. Indeed, using a conjunction analysis, Park, Park and Polk [bib_ref] Parietal functional connectivity in numerical cognition, Park [/bib_ref] recently showed that non-symbolic comparison and non-symbolic addition activated common brain circuitries in the right parietal cortex.
Therefore, a more comprehensive investigation of the association between different instantiations of the ANS (comparison, estimation and calculation), their cross-sectional trajectories in children with typical and atypical math abilities and how they interact with exact calculation is needed.
## The present study
Measures that are related to the ANS acuity appear to be normally distributed in the population [bib_ref] Individual differences in nonverbal number acuity correlate with maths achievement, Halberda [/bib_ref] and are systematically associated with arithmetics achievement. In this sense, the present study first addressed the hypothesis that children with math difficulties (MD) who were selected according to a relatively liberal criterion (below the 25 th percentile on a standardized math achievement testwould present with lower ANS acuity compared to their TA peers. To this end, we calculated specific psychophysical parameters for each of three different tasks as indices of ANS acuity. The internal Weber fraction (w) [bib_ref] Symbols and quantities in parietal cortex: Elements of a mathematical theory of..., Dehaene [/bib_ref] was calculated for the non-symbolic number comparison task, and the coefficient of variation (cv) was calculated for the non-symbolic estimation and non-symbolic addition tasks. The cv is a normalized measure of dispersion of a probability distribution and it is defined as the ratio of the standard deviation to the mean. Therefore, like the w, the higher the cv, the lower the precision. Based on the previous results obtained by and Piazza et al. [bib_ref] Developmental trajectory of number acuity reveals a severe impairment in developmental dyscalculia, Piazza [/bib_ref] , we expected TA children to have higher ANS acuity (lower values in the psychophysical parameters) compared to children with MD. Second, as noted by Noël and Rousselle [bib_ref] Developmental changes in the profiles of dyscalculia: An explanation based on a..., Noël [/bib_ref] , one should expect to find differences between the TA and MD groups in the cross-sectional trajectories of the ANS.
More specifically, group differences in ANS acuity, at least as measured by non-symbolic number comparison, should have a trend to increase across development. We finally tested the degree of association between the measures of ANS acuity and exact calculation. Because the psychophysical parameters extracted from the tasks that measure the ANS acuity are at least partially related to the degree of noise in the representation of numerosity, they should be positively correlated to one another. Moreover, based on previous studies [bib_ref] Individual differences in nonverbal number acuity correlate with maths achievement, Halberda [/bib_ref] [bib_ref] Preschoolers' precision of the approximate number system predicts later school mathematics performance, Mazzocco [/bib_ref] , it is expected that the ANS acuity will have a specific impact on exact calculation, even after controlling for the effects of general developmental factors and other abilities that are related to mathematics, such as language. Crucially, based on the results by Lyons and Beilock [bib_ref] Numerical ordering ability mediates the relation between number-sense and arithmetic competence, Lyons [/bib_ref] , who showed that number ordering fully mediated the effect of non-symbolic comparison in exact calculation, we further investigated the relationship between the ANS acuity and calculation using mediation models. Six mediation models were estimated with all possible permutations between measures of ANS acuity as predictors or mediators and exact calculation as the outcome.
# Materials and methods
## Participants
This study was approved by the ethics review board of the Federal University of Minas Gerais, Brazil (COEP-UFMG). Informed consent was obtained in written form from the parents and orally from the children. Children from first to sixth grade were recruited from public and private schools in Brazil and were assigned to different groups according to their performance in the Arithmetics and word Spelling subtests of the Brazilian School Achievement Test (Teste de Desempenho Escolar, TDE. The typical achievement group (TA, n = 172) was composed of children who performed above the 25 th percentile in both the Arithmetics and Spelling subtests of TDE. The mathematical difficulties group (MD, n = 45) performed below the 25 th percentile in the Arithmetics and above that in the Spelling subtest of the TDE.
There were no statistically significant differences in age and sex between groups. All of the children had normal intelligence, as measured by Raven's Colored Progressive Matrices (IQ scores above 85).
Children were assessed using an exact calculation task comprising addition, subtraction and multiplication, a simple reaction time task and three tasks that measured the ANS acuity: non-symbolic comparison, non-symbolic estimation and nonsymbolic addition (see the detailed description of the tasks below).
A subgroup of 10 children from the TA and 5 from the MD group were excluded from further analyses, because either they had a poor fit (R 2 ) for estimation of the w on the non-symbolic comparison task (R 2 ,0.2), and/or they showed a w that exceeded the limit of discriminability of the comparison task (w.0.6). The final sample was composed of 162 TA children and 40 children with MD. The subject details are presented in [fig_ref] Table 1: Descriptive data of the individual assessment sample [/fig_ref] (for the descriptive data of the individual assessment samples by grade, see [fig_ref] Table 1: Descriptive data of the individual assessment sample [/fig_ref] in the Supporting Information).
## Tasks
The Brazilian School Achievement Test. The TDEis the most widely used standardized test of school achievement that has norms for the Brazilian population (see also Oliveira-Ferreira et al. [bib_ref] School Achievement Test: normative data for a representative sample of elementary school..., Oliveira-Ferreira [/bib_ref]. We used the Arithmetics and Spelling subtests, which can be applied in groups. Norms are provided for school-aged children between first and sixth grade. The Arithmetics subtest is composed of three simple verbally presented word problems (e.g., ''If you had three candies and received four, how many candies do you have now?'') and 35 written arithmetic calculations of increasing complexity (e.g., very easy: ''4-1''; easy: ''1230+150+ 1620''; intermediate: ''823 * 96''; hard: ''3/4+2/8''). The Spelling subtest constitutes a dictation of 34 words that have increasing syllabic complexity (e.g., ''toca''; ''balanço''; ''cristalização''). The reliability coefficients (Cronbach's a) of the TDE subtests are 0.89 or higher. The children were instructed to work on the problems to the best of their capacity but without time limit.
Raven's Coloured Progressive Matrices. General intelligence was assessed with the Raven's Coloured Progressive Matrices, according to Brazilian norms. Exact Calculation. The task was divided in two sets of items: symbolic and written verbal calculations. The symbolic calculation set was composed of additions (27 items), subtractions and multiplications . Problems that were printed on separate sheets of paper. Children were instructed to answer as fast and as accurately as possible. Arithmetic operations were balanced at two levels of complexity and were presented to children in separate blocks: one block was composed of simple arithmetic table facts and the other block had more complex facts. Simple addition items had results below 10 (e.g., 3+5), while complex addition results ranged between 11 and 17 (e.g., 9+5). Tie problems (e.g., 4+4) were not considered for addition. Simple subtraction was composed of problems in which the operands were below 10 (e.g., , while in complex subtractions, the first operand ranged from 11 to 17 (e.g., . No negative results were included in the subtraction problems. Simple multiplication constituted operations that had results below 25 or that had the number 5 as one of the operands (e.g., 2 * 7, 5 * 6), whereas for the complex multiplication, the result of the operands ranged from 24 to 72 (e.g., 6 * 8). Tie problems were not used for multiplication. The time limit per block was set to 1 minute. The written verbal calculation set was composed of four additions and eight subtractions with single-digit operands (e.g. ''Isabella has 9 cents. She gives 3 to Pedro. How many cents does Isabella have now?''). Problems were presented to children on a sheet of paper and read aloud by the examiner to avoid reading proficiency bias. The child had to solve the problems mentally and write down the answer in Arabic format as fast and as accurately as possible. The time limit per problem was 1 minute. The total score was calculated as a simple sum of all correct answers combining both symbolic and written verbal items (max score = 94). The task was highly reliable (all Cronbach's a.0.90) [bib_ref] School Achievement Test: normative data for a representative sample of elementary school..., Oliveira-Ferreira [/bib_ref] [bib_ref] A hand full of numbers: a role for offloading in arithmetics learning?, Costa [/bib_ref].
Simple Reaction Time. The computerized Reaction Time (RT) task was a simple task in which a picture of a wolf (height = 9.31 cm; length = 11.59 cm) was displayed in the center of a black screen for a maximum time of 3,000 ms [bib_ref] A hand full of numbers: a role for offloading in arithmetics learning?, Costa [/bib_ref]. Upon appearance of the wolf on screen, children were instructed to press the spacebar on the keyboard at the moment they saw the wolf, as fast as possible. Trials terminated with the first key press. The task had 30 trials, with an inter-trial interval of 2,000 ms, 3,500 ms, 5,000 ms, 6,500 ms or 8,000 ms. This task was used to control for possible differences in basic processing speed that were not related to numerical tasks.
Non-symbolic Comparison. In the non-symbolic comparison task, participants were instructed to compare two sets of black dots, which were simultaneously presented in two white circles on the left and on the right side of the screen, and they were instructed to choose the larger numerosity by pressing a key congruent to its side (left or right) (see [bib_ref] School Achievement Test: normative data for a representative sample of elementary school..., Oliveira-Ferreira [/bib_ref] [bib_ref] A hand full of numbers: a role for offloading in arithmetics learning?, Costa [/bib_ref] [bib_ref] Count on dopamine: influences of COMT polymorphisms on numerical cognition, Júlio-Costa [/bib_ref]. Black dots were presented on a white circle against a black background. On each trial, one of the two white circles contained 32 dots (reference numerosity), and the other contained 20, 23, 26, 29, 35, 38, 41, or 44 dots. Each numerosity was presented eight times, and every presentation was arranged in a different configuration. The task comprised 64 testing trials. The maximum stimulus presentation time was 4,000 ms, and the intertrial interval was 700 ms. Between trials, a fixation point appeared on the screen for 500 ms; the fixation point was a cross printed in white and that had 3 cm for each line. To prevent the use of non-numerical cues, the sets of dots which represent the non-symbolic numerosities were designed and generated using a MATLAB scriptsuch that on half of the trials, dot size remained constant, and total dot area covaried positively with the numerosity; on the other half of the trials, total dot area was held constant and dot size covaried negatively with numerosity. The data were trimmed for each child to exclude responses of 63 SD from the individual mean RT. The w was calculated for each child as a measure of ANS acuity, based on the Log-Gaussian model of the number representation [bib_ref] Symbols and quantities in parietal cortex: Elements of a mathematical theory of..., Dehaene [/bib_ref] , with the methods described by Piazza et al. [bib_ref] Tuning curves for approximate numerosity in the human intraparietal sulcus, Piazza [/bib_ref].
Non-symbolic Estimation. In the non-symbolic estimation task, participants were asked to estimate, with a verbal response, the quantity of dots shown on the computer screen [bib_ref] Count on dopamine: influences of COMT polymorphisms on numerical cognition, Júlio-Costa [/bib_ref] (see . Black dots were presented on a white circle against a black background. The numerosities were 10, 16, 24, 32, 48, 56 or 64 dots. Each numerosity was presented 5 times, every time in a different configuration such that the same numerosity never appeared in consecutive trials. The task comprised 35 testing trials. To avoid counting, the maximum stimulus presentation time was set to 1000 ms. As soon as the child responded, the examiner, who was seated next to the child, pressed the spacebar on the keyboard and typed the child's answer. Between individual trials, a fixation point appeared on the screen, which was a cross printed in white, with 3 cm for each line. To prevent the use of non-numerical cues, the sets of dots were generated using MATLAB, in such a way that dot size changed but total dot area in a given set was always fixed across the stimuli. Thus, the total occupied area could not serve as a cue for distinguishing between the different numerosities. As a result of this manipulation, the average item size covaried inversely with numerosity. To avoid memorization effects due to the repetition of a specific stimulus, on each trial, the stimuli were randomly chosen from a set of 10 precomputed images with the given numerosity. The data were trimmed for each subject, to exclude the responses 63 SD from the mean chosen value across all of the trials. As a measure of ANS acuity, we calculated the mean cv of the responses for each child.
Non-symbolic Addition. The non-symbolic addition task was based on Knops, Viarouge, and Dehaene [bib_ref] Dynamic representations underlying symbolic and nonsymbolic calculation: evidence from the operational momentum..., Knops [/bib_ref] (see . Participants were instructed to solve approximate addition problems with operands presented in a non-symbolic notation (dots patterns). To adapt the paradigm for the use of children, the addition task was embedded in a small history of a monkey having a box of balls. Hence, a trial started with the presentation of the monkey's face, which was followed by the appearance of a brown box against a black background and the first set of dots that moved into the box. Next, another set of dots moved into the same box. Afterward, the box disappeared from the screen and was replaced by the top-view of five boxes that contained different numerosities. The boxes were arranged in a circular manner around the middle of the screen. The children were to choose which numerosity was the closest to the correct outcome by clicking with the left mouse button on the respective box. The task comprised 2 learning trials and 32 testing trials. In the training trials, the boxes were framed after each response. In a case in which the response was correct, the frame was green, which indicated that the child had chosen the box with the correct number of balls. If the response was incorrect, then the frame was red, and the children were instructed to choose another box. This procedure was repeated until the child had chosen the correct box. Before starting the testing, the children were asked if they had understood the task, and if not, the training was repeated until they confirmed that they understood the task. In the testing period, the childrens' choices were indicated by a neutral blue frame around the chosen box, regardless of whether the response was correct or not. All of the addition problems added up to four possible results (i.e., 10, 16, 26 and 40), which combined ten different operands . To prevent the subjects from memorizing the problems, the operands were randomly ''jittered" by adding a random value r, with r M J and J = {21,0,1}. For each correct response, 7 response alternatives were generated as round (c x 2.5 i/3 ), where c is the correct result and i ranges from 23 to +3. To discourage the use of non-arithmetic strategies, such as ''Always choose a response alternative in the middle of the presented range'', only five of the seven possible results were presented in a trial, such that, in half of the trials, the presented results were the upper five (high range), and thus, the correct response was the second largest numerosity.
In the other half of the trials, the lower five results were shown (low range), and the correct response was the fourth largest numerosity.
To prevent the use of non-numerical cues, the sets of dots were generated using MATLAB, in such a way that dot size covaried inversely with numerosity. To avoid memorization effects due to the repetition of a specific stimulus, on each trial, the stimuli were randomly chosen from a set of 10 precomputed images with the given numerosity. The data were trimmed for each subject, to exclude the responses 63 SD from the mean chosen value across all of the trials. As a measure of ANS acuity, we calculated the mean cv of the four different results.
## Analyses
Initially, the TA and MD groups were compared with regard to exact calculation and the three measures of ANS acuity. Next, the cross-sectional trajectories of the ANS were investigated by calculating the slopes of the regressions between ANS acuity and age for each group separately. Finally, the association between the measures within the ANS and between the ANS acuity and the exact calculation was investigated in three steps. First, crosscorrelations of the measures of ANS acuity and exact calculation were determined. Second, to estimate the specific contributions of ANS acuity measures to explain exact calculation, multiple regression models were conducted with exact calculation as the dependent variable and the three measures of ANS acuity as the predictor variables, regressing out the effects of age, schooling, general intelligence and spelling abilities. Finally, to investigate more deeply the possible mediation effects between the ANS instantiations and exact calculation, six mediation models were estimated with all of the possible permutations between the measures of ANS acuity as predictors or mediators and exact calculation as the outcome. All of the statistical analyses were performed using R statistical software. Raw data is available in the Supporting Information (Data S1).
# Results
First, we verified whether the children's performances in the measures of ANS acuity followed Weber's law. In the nonsymbolic comparison task, we calculated the R 2 of the fitting procedure to calculate the w for each child. In both the TA and MD groups, the R 2 values were high (TA: mean = 0.883, SD = 0.082; MD: mean = 0.849, SD = 0.108), which indicates that the children's performances were well described by the Log-Gaussian model of number representation [bib_ref] Symbols and quantities in parietal cortex: Elements of a mathematical theory of..., Dehaene [/bib_ref]. For the nonsymbolic estimation task, we calculated the coefficients of the regression between the correct outcomes and the mean cv per child in each presented numerosity. In both the TA and MD Similar b coefficients were obtained in the non-symbolic addition task, (TA: mean = 0.094, SD = 0.531; MD: mean = 0.056, SD = 0.581). In this case, the mean slope was not significantly different from zero in both groups (TA: t(161) = 0.122, p = 0.223; MD: t(39) = 1.119, p = 0.270). Taken together, these results demonstrate that the performance in all of the tasks that measure the ANS acuity from both the TA and MD groups can be well described by Weber's law. Group differences are presented in the next section.
## Differences between the ta and md groups in ans acuity
Although all of the children with MD had normal intelligence (IQ.85) and normal spelling achievement (above the 25 th percentile), they scored significantly lower in these measures when compared to their TA peers [fig_ref] Table 1: Descriptive data of the individual assessment sample [/fig_ref] ; see [fig_ref] Table 1: Descriptive data of the individual assessment sample [/fig_ref] in the Supporting Information for the descriptive data separated by grade). For this reason, intelligence and spelling were included as covariates for group comparisons in exact calculation and ANS acuity (see for statistics). As expected, the TA group showed better performance in exact calculation when compared to the children with MD. No group difference was found in the simple reaction time task. More importantly, the TA group presented higher ANS acuity, with significant lower w the nonsymbolic comparison task. Moreover, TA children had lower cv in the non-symbolic estimation task; however, this difference was only marginally significant. Finally, a significantly lower cv was found in the non-symbolic addition task for the TA compared to the MD group.
## Cross-sectional trajectories of ans acuity
Cross-sectional trajectories of the different measures of ANS acuity were investigated separately for the two groups (see [fig_ref] Figure 2: Cross-sectional trajectories of the measures of ANS acuity for the TA and... [/fig_ref]. The w was found to decrease monotonically with age in the TA children (b = 20.184, p = 0.015), but it remained stable in the children with MD (b = 20.016, p = 0.897). This result suggests that difference between the MD and TA groups in ANS acuity measured by the w increases during development. Second, the cv from non-symbolic estimation was found to monotonically decrease with age more or less to the same extent in both the TA and MD groups (TA: b = 20.338; p,0.001; MD: b = 20.425; p = 0.006). Finally, the cv from non-symbolic addition was also found to decrease with age by the same extent in both groups, but was only marginally significant in the TA group and was nonsignificant for the MD group (TA: b = 20.154, p,0.050; MD: b = 20.149, p = 0.375).
To confirm the results of the cross-sectional trajectory of the number comparison task, given the possible lack of statistical power in the MD group to detect significant coefficients, we ran a bootstrap analysis with the regression coefficients of the three measures of the ANS and age. First, we generated 10,000 samples with N = 40 (N of the MD group), allowing repetitive cases for each group separately. Next, we calculated for each sample one b coefficient for each of the regressions: age and non-symbolic comparison, age and non-symbolic estimation, and age and nonsymbolic addition. Afterward, we calculated the percentage of positive coefficients in each group, which we use as a likelihood index for the true direction of the association in the population. In the non-symbolic comparison task, the coefficients of the TA group were found to be negative in 91.27% of the generated samples. This finding was not the case in the MD group, in which only 53.67% of the samples showed negative coefficients. For the other measures of ANS acuity, TA and MD showed similar patterns (non-symbolic estimation: TA = 97.30%, MD = 99.30%; non-symbolic addition: TA: 80.98%, MD: 82:61%). Considering a confidence interval of 90%, both groups showed developmental changes in non-symbolic estimation, but the results were less robust in non-symbolic addition. Crucially, TA showed significant improvement in non-symbolic comparison in contrast to their MD peers, who definitively did not show any sign of improvement in this task during development. . ANCOVAs comparing the TA and MD groups in mathematics and in the measures of ANS acuity controlling for intelligence and spelling. Relationship between the measures of ANS acuity and exact calculation
As expected, all of the three measures of ANS acuity showed significant positive correlations among themselves, even after controlling for the effects of age, spelling and intelligence, which indicates that they share a common construct. Importantly, all three measures of ANS acuity also correlated with exact calculation .
Following the suggestion of one the reviews based on the inhibitory control account of the relationship between the nonsymbolic comparison task and exact calculation [bib_ref] ANS acuity and mathematics ability in preschoolers from low-income homes: Contributions of..., Fuchs [/bib_ref] [bib_ref] Individual differences in inhibitory control, not non-verbal number acuity, correlate with mathematics..., Gilmore [/bib_ref] , we ran separate correlations between the w calculated from two sets of stimuli (wSize: size control; wArea: area control; see Methods) used in the non-symbolic comparison task and exact calculation. Partial correlations controlling for the effects of age, intelligence and spelling revealed that both wSize and wArea significantly correlated with exact calculation (r = 20.13, p = 0.033 and r = 2 0.175, p = 0.007, respectively). Importantly, Fisher's r-to-z transformation revealed that there was no significant difference between the two correlation coefficients (z = 20.459, p = 0.359). Therefore, we used the w calculated from all trials in further analyses.
Next, the specific contributions of the different instantiations of the ANS in exact calculation were determined by calculating three multiple regression models. In all three models, two blocks of variables were defined. In the first block, the intervening variables age, schooling, general intelligence and spelling abilities were added using the method ''enter''. In the second block, the three measures of ANS acuity were included as single predictors in three separate models. The first block of variables explained 57.8% of the variance in exact calculation (see the coefficients in . Importantly, all three measures of ANS acuity remained significant predictors of exact calculation after removing the effects of the intervening variables (non-symbolic comparison: b = 20.135, p = 0.005; non-symbolic estimation: b = 20.162, p, 0.001; non-symbolic addition: b = 20.167; p,0.001). This finding indicates that all three instantiations of the ANS contribute to explaining exact calculation independently of age, schooling, general intelligence and spelling abilities.
Because all measures of ANS acuity are intercorrelated, the extent to which different instantiations of the ANS present unique contributions to exact calculation was determined. A multiple regression model was calculated with the same structure as before. The first block of variables included the same variables as before, and the second block of variables considered the three measures of ANS acuity simultaneously by using the ''stepwise'' method. The regression model kept non-symbolic estimation and non-symbolic addition but excluded non-symbolic comparison. Non-symbolic addition raised the variance explained to 60.1% and non-symbolic estimation to 61.4% . Therefore, the multiple regression analysis showed that both non-symbolic estimation and nonsymbolic addition have unique contributions to exact calculation. Moreover, the contribution of non-symbolic comparison to explain exact calculation is shared with other instantiations of the ANS.
Together, these results reveal that all of the instantiations of the ANS contribute to explaining exact calculation, but their contributions are not always unique. More specifically, the effects of non-symbolic comparison on exact calculation appear to be fully shared by non-symbolic estimation and non-symbolic addition. In contrast, a portion of the impact of these two variables on exact calculation appears to be unique. That is, the effect of non-symbolic comparison on exact calculation is indirect, because it is common to non-symbolic estimation and nonsymbolic addition. Are these results due to mediation processes that act inside the ANS? To specifically test this hypothesis, one must test whether the effect of one instantiation of the ANS (X) on exact calculation (Y) is significantly absorbed by another instantiation of the ANS (M) [bib_ref] A general approach to causal mediation analysis, Imai [/bib_ref] [bib_ref] Experimental designs for identifying causal mechanisms, Imai [/bib_ref]. Moreover, to increase the confidence in the direction of the mediation effect, it is necessary to determine whether the effect of M in exact calculation is also reduced to the same extent by the inclusion of X as a mediator variable.
To investigate these possible mediation effects, we conducted Causal Mediation Analysis [bib_ref] A general approach to causal mediation analysis, Imai [/bib_ref] [bib_ref] Experimental designs for identifying causal mechanisms, Imai [/bib_ref] , as implemented in the R package mediation (version 4.2.2)'' [bib_ref] mediation: R package for causal mediation analysis, Tingley [/bib_ref]. Six models that analyzed all of the possible combinations of different measures of ANS acuity as both predictors (X) and mediators (M) and exact calculation as the outcome (Y) were calculated. In each model, the total effect of each instantiation of the ANS on exact calculation was decomposed into a mediation and a direct effect. The regression coefficients as well as their confidence intervals and statistical significance are depicted in [fig_ref] Table 5: Mediation models with measures of ANS acuity as either predictors [/fig_ref]. To determine the statistical significance of the coefficient estimates, a nonparametric bootstrap method was employed. To obtain reliable estimates, a total of 10,000 samples for bootstrapping were drawn.
As can be seen in [fig_ref] Table 5: Mediation models with measures of ANS acuity as either predictors [/fig_ref] , only Models 1.1 (X = non-symbolic comparison, M = non-symbolic estimation) and 2.1 (X = nonsymbolic comparison; M = non-symbolic addition) presented directional mediation effects (p = 0.038 and p = 0.026, respectively). These results revealed that both non-symbolic estimation and non-symbolic addition mediate the total effect of non-symbolic comparison on exact calculation. While non-symbolic estimation has a partial mediation effect, because the direct effect between non-symbolic comparison to exact calculation remained significant (p = 0.012), non-symbolic addition has a complete mediation effect, because the direct effect from non-symbolic comparison to exact calculation failed to reach statistical significance (p = 0.064). . Partial correlations between measures of ANS acuity and exact calculation, controlling for age, intelligence and spelling. The direction of the possible causal direction between the measures of ANS acuity and exact calculation was further corroborated by the fact that the alternative models 1.2 and 2.2 with non-symbolic comparison as the mediator variable did not show any significant mediation effect (p = 0.069 and p = 0.094, respectively). Finally, the results revealed no mediation directional effects between non-symbolic estimation and non-symbolic addition to exact calculation (p = 0.055 and p = 0.058, respectively).
# Discussion
The present study investigated the relationship between three measures of ANS acuity (non-symbolic comparison, estimation and addition), their cross-sectional trajectories in children with typical and atypical arithmetic abilities, and their specific contributions to exact calculation. The children with MD were found to have impairments in multiple instantiations of the ANS, more specifically in non-symbolic comparison and non-symbolic addition. Moreover, the TA children were more accurate in mapping between non-symbolic magnitudes and number words compared to the children with MD, although this difference was only marginally significant.
Interestingly, the acuity of the non-symbolic comparison was found to develop normally in TA, but not in the MD group. The children with MD did not show any improvement with age in this task. A bootstrapping analysis confirmed that this difference was not due to a lack of statistical power given the smaller sample size of the MD group. Regarding the acuity of non-symbolic addition, both groups improved with age, but the children with MD were less accurate compared to their TA peers. Finally, both groups also improved to the same extent in the non-symbolic estimation task.
The three measures of ANS acuity significantly correlated with each other, which possibly reflects at least in part a common numerosity code, as proposed by Dehaene [bib_ref] Origins of mathematical intuitions: the case of arithmetic, Dehaene [/bib_ref]. Importantly, all three measures of ANS acuity significantly correlated with exact calculation. However, a multiple regression analysis revealed that only non-symbolic estimation and addition contributed with unique proportions of variance in explaining exact calculation. Mediation analysis showed that the effect of non-symbolic comparison on exact calculation was mediated to different degrees by non-symbolic estimation and non-symbolic addition.
## Differences between the ta and md groups in ans acuity
In line with previous studies that have investigated the cognitive mechanisms that underlie MLD, the ANS acuity as measured by non-symbolic comparison was found to be impaired in children . Stepwise regression with exact calculation as the dependent variable and non-symbolic comparison, non-symbolic estimation and non-symbolic addition as predictors, regressing out the effects of age, schooling, general intelligence and spelling abilities. with MD compared to their TA peers [bib_ref] Impaired acuity of the approximate number system underlies mathematical learning disability (dyscalculia), Mazzocco [/bib_ref] [bib_ref] Developmental trajectory of number acuity reveals a severe impairment in developmental dyscalculia, Piazza [/bib_ref] [bib_ref] Symbolic and nonsymbolic number comparison in children with and without dyscalculia, Mussolin [/bib_ref]. Importantly, we were able to detect deficits in the ANS even in a group of MD children selected with a more liberal criterion, which probably includes children with high cognitive heterogeneity. This finding lends support to the view that the different forms of MD are better described as a continuous spectrum rather than qualitatively different categories. The children with MD were also found to be impaired in the acuity of non-symbolic addition. To our knowledge, this study was the first that demonstrated that non-symbolic addition is impaired in children with low achievement in math. De Smedt and Gilmore [bib_ref] Defective number module or impaired access? Numerical magnitude processing in first graders..., Smedt [/bib_ref] found no impairment in a non-symbolic addition task in children with DD during the first year of formal schooling. However, the authors used a two-alternative forced choice task and analyzed only the mean accuracy of the responses. In the present study, the task used allowed children to compare their internally generated sum with five different options that were presented. Accordingly, the acuity of the internal representation of numbers could be determined by calculating the cv. Therefore, our measure is more sensitive to capturing differences.
Finally, in line with previous studies [bib_ref] Impaired acuity of the approximate number system underlies mathematical learning disability (dyscalculia), Mazzocco [/bib_ref] [bib_ref] Numerical and nonnumerical estimation in children with and without mathematical learning disabilities, Mejias [/bib_ref] , TA children were more accurate in mapping between non-symbolic stimuli and number words compared to children with MD, although in our study this difference was only marginally significant.
Both the MD and TA groups had normal intelligence (IQ.85) and normal spelling achievement (above the 25 th percentile), but group differences in those domains were still observed. For this reason, intelligence and spelling abilities were included as covariates in the group comparison analyses. These results are similar to other studies that also found medium to high effect sizes when comparing language-related abilities [bib_ref] Developmental trajectory of number acuity reveals a severe impairment in developmental dyscalculia, Piazza [/bib_ref] and intelligence [bib_ref] Weak task-related modulation and stimulus representations during arithmetic problem solving in children..., Ashkenazi [/bib_ref] between TA and DD groups. Lower language-related and intelligence performance could reflect the more widespread impairment that is frequently observed in children who have specific developmental disorders [bib_ref] Executive function and developmental disorders: the flip side of the coin, Johnson [/bib_ref].
## Cross-sectional trajectories of ans acuity
More detailed analyses of the cross-sectional trajectories of ANS acuity in typically and atypically developing children revealed several findings that merit discussion. First, while the w for the TA children decreased with age, this relationship was not the case for the children with MD. While longitudinal studies [bib_ref] Preschool acuity of the approximate number system correlates with school math ability, Libertus [/bib_ref] [bib_ref] Preschoolers' precision of the approximate number system predicts later school mathematics performance, Mazzocco [/bib_ref] have found that ANS acuity measured by the non-symbolic comparison task prior to formal schooling is a specific predictor of later mathematics achievement in TA children, group studies with younger children (6 to 9 years old) failed to find differences in this task between TA and children with DD (for a review, see Noël & Rousselle [bib_ref] Developmental changes in the profiles of dyscalculia: An explanation based on a..., Noël [/bib_ref]. However, studies with younger children did not use the w as an index of ANS acuity, which might be a more sensitive parameter to capture group differences compared to the commonly used distance effect (e.g. Oliveira-Ferreira et al. [bib_ref] School Achievement Test: normative data for a representative sample of elementary school..., Oliveira-Ferreira [/bib_ref]. Furthermore, those studies tended to use a more liberal criterion to classify the children with DD, for example, below the 15 th percentile [bib_ref] Basic numerical skills in children with mathematics learning disabilities: a comparison of..., Rousselle [/bib_ref] [bib_ref] Defective number module or impaired access? Numerical magnitude processing in first graders..., Smedt [/bib_ref]. In contrast, studies with older children that used a more stringent criterion to characterize the DD group (below the 10 th /5 th percentile) reported an elevated w for these children compared to their TA peers [bib_ref] Impaired acuity of the approximate number system underlies mathematical learning disability (dyscalculia), Mazzocco [/bib_ref] [bib_ref] Developmental trajectory of number acuity reveals a severe impairment in developmental dyscalculia, Piazza [/bib_ref]. Because no correlation was found between non-symbolic comparison and age in children with MD as opposed to their TA peers, our results suggest that it might be easier to detect group differences in this task in older children, because differences seem to increase over the course of development.
Importantly, we were able to detect group differences in the non-symbolic comparison task in a group of 10-year-olds even when using a very liberal criterion to select children with math difficulty. The acuity of the non-symbolic addition was found to increase more or less to the same degree in both the TA and MD groups; however, TA children were systematically more accurate than the children with MD. This finding suggests that even from the initial years of formal schooling, MD children might already present a detectable impairment in more complex manipulations of numeric information. In line with the present results, nonsymbolic addition measured before formal math instruction was found to be a specific predictor of later math achievement [bib_ref] Non-symbolic arithmetic abilities and mathematics achievement in the first year of formal..., Gilmore [/bib_ref]. A similar pattern compared to non-symbolic addition was observed in the non-symbolic estimation task; however, a group comparison revealed that there was only a marginally significant result.
Taken together, the results suggest that the hypothesis put forward by Noël and Rousselle [bib_ref] Developmental changes in the profiles of dyscalculia: An explanation based on a..., Noël [/bib_ref] , which states that the ANS is not impaired in children with DD, based solely on the results of a single measure of ANS acuity (non-symbolic comparison) might be too simplistic. The present results indicate that compared to TA children, younger children with low achievement in math selected even with a liberal criterion already present a lower acuity in nonsymbolic addition, which is a task that probably calls for more manipulations within the ANS than non-symbolic comparison. The characterization of cross-sectional trajectories can be considered as an important step towards the understanding of the evolution of developmental disorders, but should be confirmed in future longitudinal studies [bib_ref] Using developmental trajectories to understand developmental disorders, Thomas [/bib_ref].
## Relationship between the measures of ans acuity and exact calculation
As expected, significantly positive correlations were found between the three measures of ANS acuity. This finding is consistent with the data from Mazzocco et al. [bib_ref] Impaired acuity of the approximate number system underlies mathematical learning disability (dyscalculia), Mazzocco [/bib_ref] , who reported an association between non-symbolic comparison (w) and nonsymbolic estimation (cv) in 14-year-old adolescents. However, the results are inconsistent with the only study that directly investigated the association between more than one measure of ANS acuity [bib_ref] Measuring the approximate number system, Gilmore [/bib_ref]. These authors found null correlations between the w calculated from non-symbolic versions of the number comparison and approximate addition tasks, suggesting that these tasks are measuring completely different constructs. However, this study has an important limitation, which is that for the non-symbolic comparison task, only three ratios were used to fit the psychometric function. As is known from the psychophysical literature, it is very difficult to have good fits from using only three points in the psychometric curve. Thus, the lack of correlation reported by the authors could simply reflect a poor estimation of the coefficients. In contrast, in the present study, a much larger range of data points (eight) was used to calculate the w in the nonsymbolic comparison task. Therefore, the moderate but significant correlations between non-symbolic comparison, non-symbolic estimation and non-symbolic addition found in the present study are compatible with the existence of different instantiations of the ANS which at least partially activate a common underlying numerosity code. Nevertheless we agree with both Gilmore et al. [bib_ref] Measuring the approximate number system, Gilmore [/bib_ref] and Park and Brannon [bib_ref] Improving arithmetic performance with number sense training: An investigation of underlying mechanism, Park [/bib_ref] that it is very misleading to select only one task involving non-symbolic numerical stimuli and present it as a valid index of a supposedly unique ANS. This is a very frequent practice in the numerical cognition literature that should be avoided in further studies.
The indices w and cv correspond to the degree of noise in the internal representation of numerosity and are mathematically equivalent, in the sense that they are on the same scale (for a comprehensive review on the mathematical basis of the internal Weber fraction, see Dehaene [bib_ref] Symbols and quantities in parietal cortex: Elements of a mathematical theory of..., Dehaene [/bib_ref] ; for an intuitive explanation about the relationship between the Weber fraction and the coefficient of variation, see Halberda. Indeed, the values of the parameters in the three tasks have been revealed to be similar (see . However, one should not expect them to be equal, because the tasks that were used to extract them involve very different cognitive processes: simple discrimination, mapping from non-symbolic magnitudes to number words and more complex manipulations of magnitudes in the context of an arithmetic operation, for non-symbolic comparison, estimation and addition. Interestingly, the coefficient values increased from non-symbolic comparison to non-symbolic addition; thus, it is tempting to speculate that this result possibly reflects the summation of noise during the process of accumulation of evidence in more complex forms of magnitude manipulation. For example, the internally generated sum of two given numerosities is possibly noisier than the representation of the numerosities themselves, because during addition there is another source of noise, which arises from the operation itself. In fact, the two existent mathematical models of approximate calculation [bib_ref] Nonsymbolic arithmetic in adults and young children, Barth [/bib_ref] [bib_ref] Nonverbal arithmetic in humans: light from noise, Cordes [/bib_ref] account for this additional source of variation, by including a scaling factor that corresponds to the amount of noise due to the calculation. The same logic could be applied to non-symbolic estimation. In this task, one needs to not only discriminate the magnitudes but also transform the represented value in a symbolic label, which certainly corresponds to another source of noise [bib_ref] Symbols and quantities in parietal cortex: Elements of a mathematical theory of..., Dehaene [/bib_ref]. The precise mechanisms that are involved in different instantiations of the ANS and how their interaction occurs is a very exciting topic, but scarcely addressed in the literature. Therefore, future studies should investigate empirically the different sources of noise during magnitude manipulations and their spatial-temporal neural dynamics.
Next, we investigated more deeply the association between measures of ANS acuity and exact calculation. Significant correlations between all three measures of ANS acuity and exact calculation were found. Importantly, all three measures of ANS acuity have specific contributions to explain variance in exact calculation, which remain significant after partialling out the more general effects of age, schooling, general intelligence and spelling abilities. Accordingly, the link between the different instantiations of the ANS and exact calculation does not seem to be generated by general cognitive processes, but rather by magnitude processing abilities underlying the tasks. However, our results are not definitive. It is still possible that other general cognitive abilities that we didn't include in this study, such as executive functions and especially inhibitory control [bib_ref] ANS acuity and mathematics ability in preschoolers from low-income homes: Contributions of..., Fuchs [/bib_ref] [bib_ref] Individual differences in inhibitory control, not non-verbal number acuity, correlate with mathematics..., Gilmore [/bib_ref] can account for this link.
Previous studies also found that non-symbolic estimation [bib_ref] Impaired acuity of the approximate number system underlies mathematical learning disability (dyscalculia), Mazzocco [/bib_ref] [bib_ref] Numerical and nonnumerical estimation in children with and without mathematical learning disabilities, Mejias [/bib_ref] [bib_ref] Developmental and individual differences in pure numerical estimation, Booth [/bib_ref] and non-symbolic addition [bib_ref] Non-symbolic arithmetic abilities and mathematics achievement in the first year of formal..., Gilmore [/bib_ref] are significantly correlated with exact calculation. Regarding the non-symbolic comparison, the literature is more inconsistent and contains both positive and negative results (see De Smedt et al. [bib_ref] How do symbolic and nonsymbolic numerical magnitude processing skills relate to individual..., Smedt [/bib_ref] for a comprehensive review). As noted by De Smedt et al. [bib_ref] How do symbolic and nonsymbolic numerical magnitude processing skills relate to individual..., Smedt [/bib_ref] , this inconsistency can probably be attributed to methodological differences in the non-symbolic comparison tasks, in the index calculated from behavior (e.g., RT, accuracy, distance effect, Weber fraction) and also in the math tests used. This inconsistency has led some authors to radically argue that the ANS does not make any contribution to explaining exact calculation [bib_ref] Developmental changes in the profiles of dyscalculia: An explanation based on a..., Noël [/bib_ref]. However, this conclusion may have been premature, since two recent meta-analyses reported an association, although moderate, between non-symbolic comparison and math achievement from childhood to adulthood [bib_ref] Relations of different types of numerical magnitude representations to each other and..., Fazio [/bib_ref] [bib_ref] Association between individual differences in non-symbolic number acuity and math performance: A..., Chen [/bib_ref]. Moreover, the present study also showed that partial correlations between ANS acuity and exact calculation are in the same order of magnitude as that reported by Chen and Li [bib_ref] Association between individual differences in non-symbolic number acuity and math performance: A..., Chen [/bib_ref] , and Fazio et al. [bib_ref] Relations of different types of numerical magnitude representations to each other and..., Fazio [/bib_ref] , which cannot be reduced to more general cognitive processes. After inspecting all these results, one can be confident to assume the existence of a specific link between very basic ANS related processes and exact calculation.
An alternative hypothesis is that the non-symbolic comparison task is not a measure of the precision of numerical representations, but rather a measure of inhibitory control, since it is necessary to inhibit the processing of continuous visual parameters to be able to accurately discriminate between two numerosities [bib_ref] ANS acuity and mathematics ability in preschoolers from low-income homes: Contributions of..., Fuchs [/bib_ref] [bib_ref] Individual differences in inhibitory control, not non-verbal number acuity, correlate with mathematics..., Gilmore [/bib_ref]. To ensure that participants are not using non-numerical variables to judge which collection of dots is the larger, researchers normally use different sets of stimuli varying continuous visual properties, such as dot size and dot total area. For example, Fuchs and McNeil [bib_ref] ANS acuity and mathematics ability in preschoolers from low-income homes: Contributions of..., Fuchs [/bib_ref] used three different sets of stimuli: dot total area was constant and dot size positively covaried with numerosity; mean dot size was constant and dot total area positively covaried with numerosity; dot total area and mean dot size were both inversely covaried with numerosity ('inverse ANS acuity' trials). Interestingly, results demonstrated that only the accuracy in the inverse ANS acuity set was significantly correlated with math achievement. Furthermore, accuracy in this set showed the highest correlation coefficient with a measure of inhibitory control. Similarly, Gilmore et al. [bib_ref] Individual differences in inhibitory control, not non-verbal number acuity, correlate with mathematics..., Gilmore [/bib_ref] used two sets of stimuli: dot total area and dot size positively covaried with numerosity ('congruent' trials); dot total area and dot size negativelly covaried with numerosity ('incongruent' trials). Consistently with Fuchs and McNeil [bib_ref] ANS acuity and mathematics ability in preschoolers from low-income homes: Contributions of..., Fuchs [/bib_ref] , results showed that incongruent trials were significantly correlated with math achievement but congruent trials were not.
As described in the Methods, in the present study we used two different sets of stimuli: dot size was constant and dot total area positively covaried with numerosity (size control); dot total area was constant and dot size negatively covaried with numerosity (area control). Therefore, we didn't have the 'inverse ANS acuity' [bib_ref] ANS acuity and mathematics ability in preschoolers from low-income homes: Contributions of..., Fuchs [/bib_ref] or the 'incongruent' [bib_ref] Individual differences in inhibitory control, not non-verbal number acuity, correlate with mathematics..., Gilmore [/bib_ref] sets of trials. Nevertheless, we calculated the w separately for the size control and area control items. Partial correlations controlling for the effects of age, intelligence and spelling revealed that both wSize and wArea significantly correlated with exact calculation and no significant difference were found between the two coefficients. However, we cannot rule out the possibility that inhibitory skills could account at least in part for the relationship between the non-symbolic comparison task and exact calculation in the present study. It is important to note that a serious limitation of separating the items in two different categories, which the above-mentioned studies didn't take into account, is that the number of observations in each category of stimuli dramatically decreases and consequently compromises the stability of the measure. Therefore, we decided to use the w calculated from all items in all the analyses.
Importantly for our results, higher correlations between nonsymbolic comparison and math abilities were reported in studies that measured math ability with standardized achievement batteries, which normally include items closely associated with the representation and manipulation of numerical quantity without invoking knowledge of arithmetic (e.g. TEMA-3 [bib_ref] Individual differences in nonverbal number acuity correlate with maths achievement, Halberda [/bib_ref] [bib_ref] Preschoolers' precision of the approximate number system predicts later school mathematics performance, Mazzocco [/bib_ref]. In fact, more recently, Libertus, Feigenson and Halberta [bib_ref] Numerical approximation abilities correlate with and predict informal but not formal mathematics..., Libertus [/bib_ref] analyzed the association between the non-symbolic comparison task and the items present in the widely used TEMA-3 separated in two distinct categories: items associated with informal (e.g. enumeration and number comparison) and formal (e.g. transcoding and exact calculation) mathematical abilities. Results demonstrated that the performance in the non-symbolic comparison task was only significantly correlated with informal mathematical abilities. Similarly, Piazza et al. [bib_ref] Developmental trajectory of number acuity reveals a severe impairment in developmental dyscalculia, Piazza [/bib_ref] found the performance in the non-symbolic comparison task significantly correlated only with a subtest of a standardized math achievement battery that required children to compute the proximity relations between different numbers, but not with the other subtests, which measured transcoding and exact calculation abilities. Therefore, it is most likely that non-symbolic comparison, a very basic form of number manipulation within the ANS, has an indirect and consequently moderate effect in exact calculations.
Accordingly, Lyons and Beilock [bib_ref] Numerical ordering ability mediates the relation between number-sense and arithmetic competence, Lyons [/bib_ref] found that the ability to order a series of digits fully mediated the correlation between nonsymbolic comparison and basic symbolic arithmetical operations in adults. In the same line, van Marle et al. [bib_ref] Acuity of the approximate number system and preschoolers' quantitative development, Van Marle [/bib_ref] found that the association between non-symbolic comparison and math achievement in children was also fully mediated by a series of symbolic numerical tasks, mainly the knowledge of cardinal value. Following the same logic, it is also possible that other instantiations of the ANS that require different forms of numerical manipulation also account for the effect of non-symbolic comparison in exact calculation. Indeed, results of our multiple regression model revealed that non-symbolic comparison did not uniquely contribute to explain the variance of exact calculation, when all three measures of ANS acuity were considered simultaneously. To our knowledge, no previous study has systematically investigated the effects of non-symbolic comparison, estimation and addition on exact calculation. In order to do that, we calculated six mediation models with all combinations of measures of ANS acuity as either predictors or mediators and exact calculation as the outcome. The results revealed first that non-symbolic estimation partially mediates the relation between non-symbolic comparison and exact calculation. This finding is in line with Mazzocco et al. [bib_ref] Impaired acuity of the approximate number system underlies mathematical learning disability (dyscalculia), Mazzocco [/bib_ref] , who demonstrated that both non-symbolic comparison and nonsymbolic estimation accounted for unique proportions of variance in a math achievement task. Second, a full mediation effect of nonsymbolic addition was found to be present in the relation between non-symbolic comparison and exact calculation. These results are fully compatible with the ones recently reported by in Park and Brannon, who demonstrated that the training on nonsymbolic addition but not in non-symbolic comparison has a significant transfer effect to exact calculation. Therefore, the authors suggested that the active process of manipulating numerical information is the critical mechanism underlying the association between the basic number processing and exact calculation.
Although significantly correlated, the three ANS related tasks investigated in the present study involve different cognitive processes. The non-symbolic comparison task involves a very basic operation of magnitude discrimination, which is found to be already present in infancy. Differently, the non-symbolic estimation task involves a transcoding process from approximate to exact symbolic representations of numbers. Finally, the non-symbolic addition involves a more complex process of arithmetical transformations. Results of the mediation analyses show the existence of multiple associations between the different measures of ANS acuity and exact calculation and suggest the existence of a hierarchy of complexity between different instantiations of the ANS. These different instantiations seems to be organized from the more basic and less cognitively demanding forms of number processing to more elaborate operations that involve more active manipulation of magnitudes. The crucial evidence supporting this hypothesis is that alternative models with non-symbolic comparison as the mediator variable for the association of non-symbolic estimation and non-symbolic addition with exact calculation showed no significant mediation effects. At the neural level, this hierarchical organization of different processes underlying number representation and manipulation might reflect the increasing functional connectivity between and within the left and right parietal cortices, as observed during the performance of numberrelated tasks with increasing demands on the processing of numerical information [bib_ref] Parietal functional connectivity in numerical cognition, Park [/bib_ref]. Finally, the hierarchical structure of the different instantiations of the ANS can account for the finding that exact calculation is more strongly associated with nonsymbolic estimation and non-symbolic addition, compared to nonsymbolic comparison.
# Conclusions
Benefiting from high statistical power, we showed that children with MD, even when selected with a more liberal criterion, present lower acuity in multiple instantiations of the ANS (non-symbolic comparison and addition), even after controlling for the effects of intelligence and spelling abilities. This finding lends support to the view that the different forms of MD are better described as a continuous spectrum rather than qualitatively different categories. Second, the analyses of the cross-sectional trajectories showed that the ANS acuity measured by all three tasks positively correlated with age in TA children, while no correlation was found between non-symbolic comparison and age in the MD group. A plausible explanation for this result is that number discrimination, as the most basic form of numerical manipulation, is less prone to compensatory strategies that MD children could have developed to solve the other number-related tasks. Third, for the first time, we demonstrated that the three instantiations of the ANS investigated were significantly correlated among each other, reflecting at least in part a common numerosity code. Finally, mediation models revealed that non-symbolic estimation partially and non-symbolic addition fully mediated the effects of nonsymbolic comparison in exact calculation. Therefore, the present study represents an important step towards a deeper understanding of the cognitive mechanisms underlying the relationship between basic number processing and mathematics. Given the highly hierarchical nature of mathematics, further studies should focus on precisely investigating the association between each instantiation of the ANS and different forms of mathematical reasoning. This will certainly help to a better understanding of the typical normal development of mathematical abilities as well as the nature of developmental dyscalculia.
## Supporting information
[fig] Figure 2: Cross-sectional trajectories of the measures of ANS acuity for the TA and MD groups. doi:10.1371/journal.pone.0111155.g002 Approximate Number System and Exact Calculation PLOS ONE | www.plosone.org [/fig]
[table] Table 1: Descriptive data of the individual assessment sample. TA: typically achieving; MD: mathematical difficulties. Both TDE Arithmetics and TDE Spelling scores are in a standardized form with mean = 100 and SD = 15; d = Cohen's d. doi:10.1371/journal.pone.0111155.t001Figure 1. Psychophysical tasks used to measure ANS acuity, with non-symbolic comparison, non-symbolic estimation and nonsymbolic addition. The white arrows are used in the bottom picture to illustrate the movement of the dots into the box. doi:10.1371/journal.pone.0111155.g001 Approximate Number System and Exact Calculation PLOS ONE | www.plosone.org [/table]
[table] Table 5: Mediation models with measures of ANS acuity as either predictors (X) or mediators (M) and exact calculation as the outcome (Y). [/table]
[table] Table S1: Descriptive data of the individual assessment sample by grade. (DOCX) Data S1 Raw data. Data_S1.zip. (sheet 1: data; sheet 2: variables specifications). (ZIP) [/table]
|
Bile Acid Receptor Therapeutics Effects on Chronic Liver Diseases
# Introduction
Focal studies of hepatic secretion led to the critical analysis and understanding of bile and its circulation connecting the liver and intestine (1). Bile acids (BAs) are heterogenous compounds whose chemical and amphipathic properties result from the enzymatic breakdown of insoluble cholesterol (2). Since their isolation from bile, the field of BA chemistry has provided intensive study of their complex chemical nature and the physiological effects of their dynamic composition in circulating bile (2-5).
Hepatic BA build up leads to inflammation, necrosis, and apoptosis of various liver cells which then affects BA synthesis and transport perpetuating BA-induced damage (2, 3). Due to the extensive knowledge and research concerning hepatic BA formation and secretion, it is natural to assume BAs contribute to chronic liver damage. The increased synthesis of BAs in combination with interrupted BA signaling can lead to adverse effects in patients of chronic liver diseases. It is estimated that 1.5 billion people worldwide suffer from chronic liver diseases whose complications, cirrhosis and liver cancer, result in 2 million deaths globally (6). Chronic liver diseases have an increased burden in the health care system due to the frequent hospital readmissions, accompanying hepatic decompensation, and risk for infection as liver damage progresses (6). Many chronic liver diseases, such as primary sclerosing cholangitis (PSC) and non-alcoholic fatty liver disease (NAFLD), have minimal treatment options requiring liver transplantation as the only permanent remedy (6, 7).
Various therapeutics have been FDA-approved for clinical trials aiming to improve liver function and relieve adverse effects of disrupted BA signaling. This article serves as a brief review of BA signaling and function in various chronic liver diseases and their regulation of the gut microbiome.
## Circulation, modification, and function of bile acids bile acid synthesis and transport
The catabolism of cholesterol into BAs in the parenchyma is tightly regulated by over 17 enzymes, preferentially expressed in the liver, that are involved in the synthesis and alteration of BAs into bile salts . Impairment of these mechanisms can result in cholestasis, liver damage, impaired lipid metabolism, and other maladies (3, 5, 8). Modifications and conjugations of BAs affect their solubility, hydrophobicity, and receptor binding affinity (8, 10). BA synthesis and conjugations are summarized in Bile acids are secreted into the bile canaliculi by hepatocytes, draining to the bile ducts located in the portal triad. Cholangiocytes are the epithelial cells lining the bile ducts and assist in BA modification and circulation by cholehepatic shunting, the process in which BAs are reabsorbed from the bile and returned to hepatocytes. The intrahepatic bile duct system allows BAs to flow into the intestinal lumen through the common hepatic duct in response to food ingestion to assist in emulsification, metabolism and absorption of dietary lipids and fat-soluble vitamins (A, D, E, and K). Alternatively, BAs are deposited in the gallbladder for storage and prevention of cholesterol-crystallization and gallstone formation. Approximately 95% of BAs are reabsorbed in the distal ileum by the apical sodium-dependent bile acid transporter (ASBT; alternatively known as ileal bile acid transporter, IBAT) and delivered to the liver through the portal system via enterohepatic or portal circulation. The gut microbiome is capable of deconjugating primary bile acids and converting them to secondary bile acids prior to absorption or fecal excretion which affects gut microbiome community, BA pool, and liver health.
Abbreviations: ALKP/ALP, Alkaline phosphatase; ASBT, apical sodium bile acid transporter; AST, aspartate aminotransferase; Bas, bile acids; FXR, Farnesoid X receptor; CA, cholic acid; CCA, cholangiocarcinoma; CDCA, chenodeoxycholic acid; DCA, deoxycholic acid; HCC, hepatocellular carcinoma; IBABP, ileal bile acid binding protein; IBD, irritable bowel disease; IBAT, ileal bile acid transporter; LCA, lithocholic acid; LDL, low density lipoprotein; MDR2, multidrug resistance cassette 2; MDR3, multidrug resistance cassette 3; NAFLD, non-alcoholic fatty liver disease; NASH, non-alcoholic steatohepatitis; OCA, Obeticholic acid; OSTαβ, organic solute transporter α-β; PBC, primary biliary cholangitis; PSC, primary sclerosing cholangitis; TBA, total bile acid; TGR5, Takeda G protein coupled receptor 5; UDCA, Ursodeoxycholic acid.
Aside from their important roles in digestion, BAs can behave as signaling molecules in carbohydrate and lipid metabolism, energy expenditure, and hepatic disease. BAs and activation of their downstream targets including G-protein-coupled bile acid receptor (TGR5), transforming growth factor-α (TGF-α) and sphingosine-1-phosphate receptor-2 (S1PR2) stimulate cholangiocyte proliferation and contribute to the progression of cholangiocarcinoma [CCA, in vivo and in vitro]. Alternatively, Farnesoid X Receptor (FXR) is down regulated in hepatocellular carcinoma (HCC). It has been shown that FXR via increased CYP450 epoxygenase activity suppress NF-κB signaling thereby reducing hepatic inflammation. Further exploration into the anti-inflammatory role of FXR and assessment of BA direct or indirect targets may provide understanding of chronic cholestatic disease establishment and progression.
## Intrahepatic and extrahepatic bile acid modification
The catabolism of cholesterol results in the formation of the primary BAs, cholic acid (CA) or chenodeoxycholic acid (CDCA), through the major (classical) pathway or the minor (alternative/acidic) pathway, respectively. Cholehepatic shunting alters the BA pool via biliary ASBT transport, multidrug resistance cassette 3 (MDR3, human; multidrug resistance cassette 2, mice), and organic solute transporter α-β (OSTα-β) BA secretion into the peribiliary plexus prior to reaching the hepatic sinusoids. Ileal bile acid binding protein (IBABP) is expressed in large cholangiocytes to sequester BAs preventing biliary cytotoxicity. CA and CDCA/Ursodeoxycholic acid (UDCA) are converted to deoxycholic acid (DCA) and lithocholic acid (LCA), respectively, via 7α/β-dihydroxylation by various species of the commensal gut microbiota in the gastrointestinal tract. Human secondary BAs (DCA and LCA) are capable of being circulated back to the liver via enterohepatic circulation leading to an increased hepatic levels of damaging hydrophobic BAs.
## Dysregulation of bile acids in chronic liver diseases psc and pbc
PSC and Primary Biliary Cholangitis (PBC) are rare cholestatic liver diseases that affect the biliary system. PSC is an idiopathic disease with cholestasis, inflammation and eventual fibrosis resulting from strictures of intra-and extrahepatic bile ducts. PSC is one of the most common causes for liver transplantation (LT). Due to its heterogenous and spontaneous progression, effective medical therapies have not yet been developed. Fat-soluble vitamin deficiency can occur in PSC patients as a result of decreased bile flow and secretion. It has also been shown that PSC has a positive correlation with ulcerative colitis (UC), a form of inflammatory bowel disease (IBD). PSC/IBD patients display altered BA fecal excretion and decreased gut microbiome diversity compared to healthy or IBD patient controls. Patients with PSC have decreased expression of hepatic FXR, TGR5, and S1PR2. The multidrug resistance cassette 2 knock-out mouse (MDR2 −/− ) is a mouse model utilized to mimic the PSC phenotype including increased cholestasis, intrahepatic bile duct mass and hepatic inflammation due to hepatic BA build up. This murine model has been useful for identifying effects of potential therapeutics, such as UDCA. Meng et al. reported that UDCA treatment in Mdr2 −/− mice reduced serum TBA, elevated hepatic expression of BA transporters, and reduced hepatic inflammation and collagen deposition.
PBC is a chronic auto-immune disease, predominantly affecting middle-aged women, that results in biliary ductopenia and cholestasis. Li et al. reported elevated serum levels of total BAs (TBA) and FGF19 in cirrhotic PBC patients compared to healthy controls and non-cirrhotic PBC patients. Similarly, Trottier et al. demonstrated elevated BAs in serum samples from both PBC and PSC compared to healthy controls. Ursodeoxycholic acid (UDCA), an epimer of CDCA, was the first FDA-approved treatment for PBC. Despite increased bile flow, lower liver enzyme levels, and decreased serum BA levels, one in three PBC patients will have a limited or no response to treatment, strengthening the need for effective therapeutic intervention of PBC progression.
## Nafld
NAFL and non-alcoholic steatohepatitis (NASH) are two of the most common hepatic diseases worldwide due to an increase of sedentary lifestyle and consumption of a high-fat/highcholesterol diet. Its prevalence has demonstrated positive correlation with an increasing number of obese and type II diabetic patients. Currently, there are no approved therapies for the treatment of NAFL and NASH aside from a change of diet and exercise for gradual weight loss.
BA signaling is disrupted in NAFL and NASH patients yielding great interest in the search for exogenous methods of BA regulation.
## Hcc and cca
HCC is currently the third leading cause of cancer deaths worldwide. HCC affects parenchymal cells in the liver, which make up to 70% of the liver tissue. HCC develops in chronic liver disease or cirrhotic liver patients and is usually detected through various imaging methods prior to diagnosis. Patients with HCC can be asymptomatic or present with a range of symptoms including cirrhosis-related pain. The underlying chronic liver injury, and difficulty in diagnosis, both contribute to HCCs high mortality. CCA is a rare but devastating cancer with poor prognosis. Patients present with jaundice, pruritus (intense itch) and acholic (pale) stool due to reduced bile and bilirubin excretion. Due to the intimate relationship between hepatocytes, cholangiocytes and BAs it is important to investigate BA signaling in the establishment and progression of HCC and CCA.
Demonstrating a shift in the BA pool during cancer development, Changbumrung et al. reported elevated ratios for trihydroxy to dihydroxy BAs and for glycine-conjugated to taurine-conjugated BAs in patients with HCC and CCA compared to healthy patients. Luo et al. demonstrated a similar trend with elevated glycine-conjugated BAs in hepatic injury patients, ranging from hepatitis B viral infection to cirrhosis, compared to healthy controls suggesting taurineconjugated BAs as a potential sensitive biomarker for liver injury. This increase in BA pool size is due to reduced inhibition of BA synthesis. FXR activation has been indicated to have anti-cancer properties, with decreased expression in progressing human HCC lesions, since its downstream effects include inhibition of BA synthesis and cell proliferation. . This discovery is likely due to cholangiocyte requirement of TGR5 for BA-induced proliferation and anti-apoptosis signaling. Additional BA receptors, such as S1PR2, have been found to have elevated expression in human CCA tumors, as shown by Liu et al.. Taken together these studies indicate that altered BAs may serve to enhance HCC and CCA progression during disease development.
## Current bile acid-related therapies of chronic liver diseases bile acid therapies
The use of synthetic or naturally occurring BAs to reduce gallstone formation, aid in lipid absorption following gastric surgeries, and assist in reduction of cholestasis has increased in recent years. UDCA is a hydrophilic BA conjugated to undergo enterohepatic circulation or deconjugated and converted into LCA by the gut microbiome and excreted into feces. UDCA was the first FDA-approved therapy for PBC patients exhibiting altered serum liver enzyme levels. UDCA has been a staple therapy due to its ability to prevent gallstone formation, increasing bicarbonate secretion to prevent acidification of bile, and positive side effect profile as compared to treatment with CDCA. In many PBC patients, long-term UDCA treatment when administered at early stage of disease increases survival rate, lowers liver enzyme serum levels, and improves liver histology. Regardless of its beneficial effects, 30-40% of PBC patients do not respond to UDCA treatment.
UDCA treatment can improve liver histology and serum ALT/ASP levels in PSC patients, however, data supporting any long-term efficacy or long-term survival are lacking. Moreover, UDCA prescribed at high doses increased medical complications and mortality in PSC patients. PSC patients treated with norUDCA, a side chain shortened homolog of UDCA, exhibited reduced ALP serum levels compared to placebo in a dosedependent manner, however norUDCA's effects on PSC progression, long-term survival, and mortality have not been investigated (65).
## Bile acid receptor/transporter agonists and antagonists
Obeticholic acid (OCA, Ocaliva) is an FDA-approved, synthetic derivative of CDCA and is a high affinity ligand for the nuclear bile acid receptor, FXR. FXR is a nuclear BA receptor that when expressed is capable of reducing BA synthesis, increasing expression of BA transporters and modulating lipoprotein metabolism. OCA has been studied as a potential therapeutic drug in the treatment of various chronic liver diseases. Following Phase II and Phase III clinical trials, it has been suggested that OCA treatment can be delivered in conjunction with UDCA in PBC, or as a monotherapy in patients who do not tolerate UDCA, respectively. Adverse events including pruritus occurred in nearly all patients and was observed to occur in a dose dependent manner. The beneficial effects of OCA as a monotherapy are currently being investigated in various clinical trials (7,. The use of OCA has promising effects on improving liver enzyme levels, lowering plasma bilirubin and IgM levels, and has recently been shown to improve or stabilize liver fibrosis and biliary injury in PBC patients with cirrhosis. Despite promising outcomes in PBC patients, long-term effects of OCA on disease progression and patient survival are still being investigated.
In the Farnesoid X Receptor Ligand Obeticholic Acid in NASH Treatment (FLINT) clinical trial, NASH patients treated with OCA presented with improved fibrosis scores, elevated low-density lipoprotein (LDL) and increased average weight loss compared to placebo control. NASH patients treated with OCA also had increased ALP levels compared to placebo control. The effects of OCA on insulin resistance and the observed improvement of lipid absorption were not sustained following termination of OCA treatment. Further investigation into the short-term and long-term effects of OCA on liver function and injury are warranted and are actively being explored in the Randomized Global Phase III Study to Evaluate the Impact on NASH with Fibrosis of Obeticholic Acid Treatment (REGENERATE) clinical trial assessing FXR activation in the treatment of NASH. Translational Research and Evolving Alcoholic hepatitis Treatment (TREAT) consortium conducted phase II clinical trial utilizing OCA on patients with moderately severe alcoholic hepatitis (AH) that was completed in 2018, however currently no data is available on this study. The Phase 3 Study of Obeticholic Acid in Patients with Primary Biliary Cirrhosis (POISE) double-blind, placebo controlled clinical trial has been one of the few to publish results demonstrating OCA's long-term safety and effectiveness. POISE study data showed that three-year OCA treatment reduced or stabilized hepatic collagen deposition and ductular injury in PBC patients with cirrhosis. The Combination OCA and Statins for Monitoring of Lipids (CONTROL) clinical trial study found increased LDL in NASH patients treated with OCA (5, 10, or 25 mg) indicating altered lipid metabolism. The observed increase in LDL cholesterol was reduced in NASH patients concurrently treated with OCA and atorvastatin, a statin utilized to reduce endogenous cholesterol production. The authors noted that these two drugs were well-tolerated when utilized together, addressing the observed increase in LDL cholesterol following OCA treatment in NASH patients, but remained inconclusive with respect to the combinatorial effect on hepatic injury. Eaton et al. observed that decompensated patients with cirrhotic PSC and PBC, OCA treatment led to the development of jaundice and elevated liver enzyme levels. While the longterm benefits of OCA are still being evaluated in various liver diseases, it is still being explored as a potential.
During normal enterohepatic circulation, IBAT/ASBT is responsible for the reabsorption of BAs in the intestinal tract prior to secretion into the portal blood system. Additionally, ASBT is responsible for cholehepatic shunting of BAs between cholangiocytes and hepatocytes, which ultimately increases hepatic BA pool. A common symptom of chronic cholestatic liver disease is pruritus, intense itching of the dermis due FIGURE 2 | Current bile acid receptor therapeutics and their effects in bile acid signaling. Briefly, Obeticholic acid (OCA) reduces bile acid (BA) synthesis, circulating BA levels, and hepatic inflammation. This treatment may cause adverse effects such as pruritus and jaundice (in decompensated PBC patients). Ursodeoxycholic acid (UDCA) alters BA pool and reduced hepatic inflammation and damage. In some cases, UDCA treatment has led to development of pruritus. Apical sodium bile acid transporter (ASBT) inhibitors affect bile acid synthesis, circulation, secretion, and microbial composition in the intestine.
Frontiers in Medicine | www.frontiersin.org to increased BA deposition. IBAT/ASBT inhibitors are potential therapeutic candidates for this detrimental symptom. Pilot studies conducted with various synthetic IBAT/ASBT inhibitors have demonstrated variable results. The Al-Dury et al. study reported patients who continued through A4250 (synthetic IBAT/ASBT inhibitor) received relief from itching, but pruritus returned during wash-out and BA sequestrant treatment. Despite improvement of pruritus and lowered serum BAs, many patients in the A4250 study dropped out early due to abdominal pain and diarrhea. Similar benefits were identified in the pilot study of GSK2330672 (synthetic IBAT/ASBT inhibitor), PBC patients resulted with lowered serum BAs and increased serum FGF19. The most common adverse events in this treatment cohort were headaches and diarrhea, which provides reason to limit the treatment length in patients with pruritus.
## Ba receptor agonist effects on the gut/liver axis during liver disease
BA species affect and regulate gut microbial species composition. BA species and concentrations in different portions of the gastrointestinal tract can result in increased side effects including increased intestinal permeability and BA-induced diarrhea. Elevated hydrophobic BAs in the colon are capable of inducing inflammation which is reduced following CDCA treatment alleviating the increased toxicity from insoluble BA concentrations and mast cell secretory factors (i.e., histamine or nerve growth factor NGF).
A small cohort of PBC patients exhibited altered gut microbiome composition, compared to healthy controls, which was partially reversed following 6-month UDCA treatment. This study found eight-PBC associated genera and a reduction in normal gut microbiome associated microbial members including Faecalibacterium, Bacteroides, Sutterella, and Oscillospira spp. A study from Selmi et al. implicated increased N. aromaticivorans population is responsible for the induction of PBC (81). Tang et al. reported increased epithelial infiltration of Enterobacteriaceae in PBC patients indicating increased permeability and decreased gastrointestinal immune response in diseased patients.
The gastrointestinal epithelial barrier plays an important role in maintaining BA enterohepatic circulation homeostasis and reduced inflammation from lipopolysaccharide (LPS) leakage. BA circulation is necessary to avoid damaging increases in colonic BAs that result in inflammation and intestinal damage. Song et al. found that CDCA was capable of reducing paracellular permeability and increased cell-to-cell tight junctions via FXR activation in mice. Removal of commensal gut microbiota in Mdr2 −/− mice resulted in increased serum liver enzyme levels, cholangiocyte senescence, and circulating primary BAs. Alternatively, Li et al. discovered a depletion of BA induced damage to the colon and reduced mast cell activation and degranulation in FXR −/− mice or Z-guggulsterone treated mast cells.
The role of the gut microbiome in BA homeostasis is still being investigated. The disruption of the gut-liver axis can allow infiltration of microbes, or their metabolic products, into the enterohepatic circulation to prolong disease. Chronic liver disease patients all maintain their own gut microbiome signature that inherently play a role in disease development. Future studies of BA-associated therapeutics should consider effects on the gut microbiome and their metabolites in relation to changing BA pool/circulation and chronic liver disease.
# Conclusion
BA signaling, and therapeutics that alter them, have effects on the gut microbiome and organs outside of the liver. The use of BA receptor agonists and antagonists and their indirect effects (BA pool, synthesis, circulation, and the gut microbiome composition) is summarized in . The human and mouse gut microbiome fluctuates with the increase or decrease of BAs, which can lead to increased inflammation and intestinal malabsorption. In depth exploration of BA signaling and circulation during chronic liver diseases may provide insight to disease amelioration or treatment strategies. Understanding the role of the gut microbiome on BA modification and circulation may allow for innovative concurrent treatment for the reductions of inopportune side effects of the currently investigated treatments, OCA and UDCA.
# Author contributions
|
VEGFR-3 and CXCR4 as predictive markers for treatment with fluorouracil, leucovorin plus either oxaliplatin or cisplatin in patients with advanced esophagogastric cancer: a comparative study of the Arbeitsgemeinschaft Internistische Onkologie (AIO)
Background: Combination of fluoropyrimidines and a platinum derivative are currently standards for systemic chemotherapy in advanced adenocarcinoma of the stomach and gastroesophageal junction (GEJ). Nevertheless, individual likelihood for response to these therapeutic regimes remains uncertain. Even more, no predictive markers are available to determine which patients may benefit more from oxaliplatin versus cisplatin or vice versa. The new invasion and stem cell markers VEGFR-3 and CXCR4 have been linked prognostically with more aggressive esophagogastric cancer types. Thus, we aimed to assess correlations of VEGFR-3 and CXCR4 expression levels with clinical outcome in a randomized phase III study of patients with oxaliplatin/leucovorin/5-FU (FLO) versus cisplatin/leucovorin/5-FU (FLP).Methods:The patients data examined in this study (n = 72) were from the collective of the FLO vs. FLP phase III AIO trial. Tumour tissues were stained via immunohistochemistry for VEGFR-3 and CXCR4 expression and results were evaluated by two independent, blinded investigators. Outcome parameter: Survival analysis was calculated for patients receiving FLO vs. FLP in relation to VEGFR-3 and CXCR4 expression.Results: 54% and 36% of the examined tumour tissues showed strong positive expression of VEGFR-3 and CXCR4 respectively. No superiority of each regime was detected in terms of overall survival (OS) in the whole population. Patients with strong expression of CXCR4 on their tumour tissues profited more in terms of OS under the treatment of FLP (mOS: 28 vs 15 months, p = 0.05 respectively). Patients with negative VEGFR-3 and CXCR4 expression had a trend to live longer when FLO regime was applied (mOS: 22 vs. 9 months, p = 0.099 and 20 vs. 10 months, p = 0.073 respectively). In an exploratory analysis of patients older than 60 years at diagnosis, we observed a significant benefit in overall survival for VEGFR-3 and CXCR4-positive patients when treated with FLP (p = 0.002, p = 0.021 respectively). Conclusions: CXCR4 positive patients profited in terms of OS from FLP, whereas FLO proved to be more effective in CXCR4 and VEGFR-3 negative patients. Our results suggest, despite the limited size of the study, a predictive value of these biomarkers concerning chemotherapy with FLP or FLO in advanced esophagogastric cancer.
# Background
Adenocarcinoma of the stomach and the gastroesophageal junction (GEJ) is one of the most common and lethal malignancies with approximately 990,000 new cases and 738,000 deaths per year worldwide [bib_ref] Global cancer statistics, Jemal [/bib_ref]. Most gastric cancers are unfortunately diagnosed at an advanced stage, so that even after a potential curative gastrectomy relapse rates remain at levels of between 40% and 60% [bib_ref] Chemotherapy for advanced gastric cancer, Wagner [/bib_ref].
Systemic chemotherapy is nowadays the gold standard for the palliative treatment of patients with advanced or metastatic cancer of the stomach or GEJ. The combination of fluoropyrimidine and platinum is widely considered to be the treatment of choice in advanced gastric cancers having shown a benefit in overall survival (OS) and progression free survival (PFS) in different studies [bib_ref] Bang YJ: A phase III randomized study of 5-fluorouracil and cisplatin versus..., Kim [/bib_ref] [bib_ref] Final results of a randomized phase III trial of sequential high-dose methotrexate,..., Vanhoefer [/bib_ref] [bib_ref] Randomized trial comparing epirubicin, cisplatin, and fluorouracil versus fluorouracil, doxorubicin, and methotrexate..., Webb [/bib_ref]. Nevertheless, the focus of recent and current studies remains the identification of a superior treatment combination while minimizing toxicity.
To the best of our knowledge, there are two phase III studies that deal with the effect and toxicity of oxaliplatin compared with cisplatin in the treatment of metastatic esophagogastric cancer [bib_ref] Capecitabine and oxaliplatin for advanced esophagogastric cancer, Cunningham [/bib_ref] [bib_ref] Phase III trial in metastatic gastroesophageal adenocarcinoma with fluorouracil, leucovorin plus either..., Al-Batran [/bib_ref]. Data from the REAL-2 trial [bib_ref] Capecitabine and oxaliplatin for advanced esophagogastric cancer, Cunningham [/bib_ref] showed no inferiority of oxaliplatin versus cisplatin or of capecitabine versus 5-FU for treatment in this category of patients. Moreover in a post-hoc subgroup analysis oxaliplatin proved to be more effective than cisplatin in patients >65 years [bib_ref] Phase III trial in metastatic gastroesophageal adenocarcinoma with fluorouracil, leucovorin plus either..., Al-Batran [/bib_ref].
In the search of new biomarkers for advanced esophagogastric carcinoma, VEGFR-3 and CXCR4 have recently become the focus of research [bib_ref] Combination of circulating CXCR4 and Bmi-1 mRNA in plasma: a potential novel..., Xu [/bib_ref] [bib_ref] Blocking VEGFR-3 suppresses angiogenic sprouting and vascular network formation, Tammela [/bib_ref] [bib_ref] Vascular endothelial growth factor receptor 3 is involved in tumor angiogenesis and..., Laakkonen [/bib_ref] [bib_ref] The effect of the expression of vascular endothelial growth factor (VEGF)-C and..., Han [/bib_ref] [bib_ref] Correlation of vascular endothelial growth factor-D expression and VEGFR-3-positive vessel density with..., Choi [/bib_ref] [bib_ref] Vascular endothelial growth factor receptor-3 is a favorable prognostic factor in advanced..., Sung [/bib_ref] [bib_ref] Vascular endothelial growth factor-D and its receptor VEGFR-3: two novel independent prognostic..., Juttner [/bib_ref] [bib_ref] CXCR4/SDF-1 axis is involved in lymph node metastasis of gastric carcinoma, Zhao [/bib_ref] [bib_ref] The expression of CXCL12 and CXCR4 in gastric cancer and their correlation..., Ying [/bib_ref] [bib_ref] Expression of chemokine receptor CXCR4 in esophageal squamous cell and adenocarcinoma, Gockel [/bib_ref]. VEGFR-3 has been associated with lymphangiogenesis, invasion and metastasis of gastric cancer [bib_ref] Blocking VEGFR-3 suppresses angiogenic sprouting and vascular network formation, Tammela [/bib_ref] [bib_ref] Vascular endothelial growth factor receptor 3 is involved in tumor angiogenesis and..., Laakkonen [/bib_ref] [bib_ref] Relationship between LYVE-1, VEGFR-3 and CD44 gene expressions and lymphatic metastasis in..., Ozmen [/bib_ref] [bib_ref] Effect of lentivirus-mediated shRNA targeting VEGFR-3 on proliferation, apoptosis and invasion of..., Qin [/bib_ref] [bib_ref] Suppression of VEGFR-3 signaling inhibits lymph node metastasis in gastric cancer, Shimizu [/bib_ref] [bib_ref] Lymphangiogenesis and the vascular endothelial growth factor receptor (VEGFR)-3 in gastric cancer, Yonemura [/bib_ref] whilst CXCR4 is associated with stimulation of angiogenesis, lymph node metastasis and peritoneal carcinomatosis [bib_ref] The expression of CXCL12 and CXCR4 in gastric cancer and their correlation..., Ying [/bib_ref] [bib_ref] Therapeutics target of CXCR4 and its downstream in peritoneal carcinomatosis of gastric..., Koizumi [/bib_ref] [bib_ref] Expression of CXCR4 and its ligand SDF-1 in intestinal-type gastric cancer is..., Iwasa [/bib_ref] [bib_ref] CCR7 and CXCR4 expression predicts lymph node status including micrometastasis in gastric..., Arigami [/bib_ref] [bib_ref] Role of the CXCL12/CXCR4 axis in peritoneal carcinomatosis of gastric cancer, Yasumoto [/bib_ref] [bib_ref] Blocking on the CXCR4/ mTOR signalling pathway induces the anti-metastatic properties and..., Hashimoto [/bib_ref]. Nevertheless, their role as predictive markers or as potential therapeutic targets in advanced esophagogastric cancer remains unclear. Despite the encouraging results of the addition of bevacizumab in phase II trials in metastatic and loco regional esophagogastric cancer [bib_ref] Multicenter phase II study of irinotecan, cisplatin, and bevacizumab in patients with..., Shah [/bib_ref] [bib_ref] Bevacizumab with peri-operative epirubicin, cisplatin and capecitabine (ECX) in localised gastro-oesophageal adenocarcinoma:..., Okines [/bib_ref] , a significant benefit in terms of OS was not observed in the phase III AVAGAST trial [bib_ref] Bevacizumab in combination with chemotherapy as first-line therapy in advanced gastric cancer:..., Ohtsu [/bib_ref]. Furthermore, there are to date no comparative studies that focus on a correlation of VEGFR-3 and CXCR4 with the clinical outcome using different therapeutic regimes in patients with locally advanced or metastatic adenocarcinoma of the stomach or GEJ.
The aim of this study was to investigate whether VEGFR-3 and CXCR4 could serve as molecular patterns for personalisation of standard chemotherapy in patients with advanced esophagogastric cancer. We therefore examined and compared the effect of combined chemotherapy with oxaliplatin/leucovorin/5-FU (FLO) versus cisplatin/leucovorin/5-FU (FLP) in patients with advanced esophagogastric cancer in relation to tumour VEGFR-3 and CXCR4 expression.
# Methods
## Patients
The patient data examined in this study (n = 72) originate from the collective of the FLO vs. FLP Phase III trial of the AIO. A comparison of the main disease characteristics between patients in our study and the overall trial population is shown in Additional file 1. Patients were recruited in 31 German and one Swiss centre in a time period of 3 years. Eligibility criteria were histological confirmation of locally advanced or metastatic adenocarcinoma of the stomach or GEJ. Patients had to be over 18 years old, have not received any prior palliative chemotherapy, have not suffered from another type of cancer in the previous five years and have a creatinine clearance > 50 ml/min and adequate bone marrow function. Patients gave written informed consent according to the Helsinki protocol before entering the study, which was approved by the ethics committees of the participating institutions and the Federal Institute for Drugs and Medical Devices (Ref. Nr: 4020513).
## Treatment
The participants were randomized into two treatment arms. The FLO-group received infusional oxaliplatin 85 mg/m 2 and leucovorin 200 mg/m 2 over 2 hours every 2 weeks, followed by an infusion of 5-FU 2600 mg/m 2 over 24 hours. Patients in the FLP therapy arm received cisplatin 50 mg/m 2 every 2 weeks, combined with the weekly infusion of leucovorin 200 mg/m 2 over 2 hours and FU 2000 mg/m 2 over 24 hours. After 6 weeks, the FLP-treatment was followed by a 2-week rest period.
## Immunohistochemistry
The expression of VEGFR-3 and CXCR4 was analyzed by immunohistochemistry (IHC). Paraffin-embedded tissue samples were obtained from 70 patients for CXCR4 and 69 for VEGFR-3, due to limited availability of material. The nature of the collected material was mainly tumour resections.
Three micrometer thick tissue sections were cut and mounted on super frost slides. These were deparaffinized, rehydrated and peroxidase blocked (3% H 2 O 2 in methanol, 30 min). After blocking of nonspecific protein binding sites by using fresh frozen plasma (30 min) slides were incubated with the respective primary antibody VEGFR-3 (sc 321, Santa Cruz Biotechnology, UK, 1:200, 2 h) and CXCR4 (CIO115, Capralogics, USA, 1:300, 1.5 h) at room temperature, as described before [bib_ref] Cetuximab with irinotecan, folinic acid and 5-fluorouracil as first-line treatment in advanced..., Moehler [/bib_ref] [bib_ref] An open-label, multicentre biomarker-oriented AIO phase II trial of sunitinib for patients..., Moehler [/bib_ref] [bib_ref] Strong expression of chemokine receptor CXCR4 by pancreatic cancer correlates with advanced..., Wehler [/bib_ref]. Incubation with secondary antibody (anti-rabbit-mouse-goat antibody) was followed by incubation with streptavidin-POD (DAKO, Germany, each 15 min). Specific antibody binding was visualized using DAB solution (DAKO, Germany) and the tissues were counterstained by hemalaun solution (DAKO, Germany). Between each step of staining the specimens were washed in distilled water or DPBS.
Evaluation of staining was performed by two independent, blinded pathologists.
# Statistical analysis
The staining was evaluated by intensity (no staining 0, weak 1, moderate 2, and strong 3) and the extent of the stained tumour area (0% 0, <25% 1, 25-50% 2, 50-75% 3, > 75% 4). These two classifications were added together and divided into the categories negative and positive. Up to a total of 5, staining was scored as negative, from 6 or more it was considered clearly positive. All statistical analyses were done by using MedCalc software 2013 in close cooperation with the IMBEI. The survival analysis was performed by using the Kaplan-Meier method and the log rank test. To investigate the association between the results of immunohistochemistry obtained for VEGFR-3 and CXCR4 and clinical-pathological parameters, univariate statistical analysis were performed using Pearson's Chi-2 test or Fisher's exact test. p values < 0.05 were considered to indicate significant differences.
# Results
## Patients characteristics
The average age of patients was 59.8 years and ranged from 33 to 84 years, two thirds of the patients were male. There were no major differences concerning the main disease characteristics in comparison to the overall phase III trial, see Additional file 1. In the statistical analysis of VEGF receptor 3 (VEGFR-3), a total of 69 tumour specimens were included. The cut-off for positive VEGFR-3 expression was drawn at a value of 6, which was true for 39 specimens (54%). 36 patients were treated with FLO, 33 patients received FLP [fig_ref] Table 1: Patient characteristics [/fig_ref]. In the statistical analysis of CXCR4, tissue samples from 70 patients were included. 37 of these patients received the FLO regimen, 33 were treated with FLP. The cut-off for a positive detection was set at 6 as sum of the scores for proportion and intensity of staining . In 26 (36%) of all tissue samples, a strong CXCR4 expression could be detected.
There was an equal distribution of the patients in regards of treatment and positivity of markers [fig_ref] Table 2: Distribution of patients' treatment in relation to expression of VEGFR-3 or CXCR4 [/fig_ref]. Location of the primary and age showed no statistical significant changes within all groups [fig_ref] Table 1: Patient characteristics [/fig_ref].
# General statistical analysis
In terms of efficacy, no difference in the response to the treatment regimens FLO and FLP could be detected (p = 0.839). Furthermore, there was no significant difference in survival between VEGFR-3 positive and negative patients or for CXCR4 expression (data not shown).
## Results for the vegf receptor 3
The response to either therapy regimen strongly depended on the expression status of VEGFR-3. Patients with a negative VEGFR-3 state significantly benefited from treatment with FLO in survival over 18 months (p = 0.026). A strong trend in favour of the FLO-therapy was detected
## Results for the cxc receptor 4
The CXCR4 expression in tumour tissues showed a strong impact on survival in both treatment groups FLO and FLP. While CXCR4 negative patients showed a trend towards treatment benefit with FLO [fig_ref] Figure 3: Survival analysis under FLO/FLP treatment in regards of CXCR4 expression in an... [/fig_ref] , p = 0.073), patients with strong CXCR expression survived longer under FLP treatment [fig_ref] Figure 3: Survival analysis under FLO/FLP treatment in regards of CXCR4 expression in an... [/fig_ref] , p = 0.05). The median survival of patients with negative CXCR4-status was a statistic relevant 10 months longer under FLO than FLP (20 vs. 10 months), while strong CXCR4 expression was associated with a statistically significant median survival benefit of 13 months for FLP compared with FLO (28 vs. 15 months, p = 0.05).
## Patients older than 60 years
At this part of our study we performed an exploratory analysis examining whether VEGFR-3 and CXCR4 show the same predictive value for older patients.
In the post-hoc subgroup of patients older than 60 years at diagnosis, we found a significant benefit in overall survival for VEGFR-3 and CXCR4 positive patients when treated with FLP [fig_ref] Figure 4: Survival analysis under FLP treatment for patients older than 60 years [/fig_ref] and B, p = 0.002, p = 0.021 respectively). No patient of this subgroup with negative VEGFR-3 expression status survived the first year after diagnosis of gastric cancer. However, similar findings were not observed when older patients with weak tumour expression of VEGFR-3 and CXCR4 were treated with FLO (data not shown).
## Combination of the results for vegfr-3 and cxcr4
Patients with a strong expression of VEGFR-3 and CXCR4 benefited in overall survival from the treatment with the cisplatin-containing FLP scheme [fig_ref] Figure 5: Survival analysis under FLO/FLP treatment in an observation period of 5 years [/fig_ref] , p = 0.126). In contrast, patients with weak expression of CXCR4 and VEGFR-3 lived significantly longer with FLO [fig_ref] Figure 5: Survival analysis under FLO/FLP treatment in an observation period of 5 years [/fig_ref] , p = 0.011).
# Discussion
Seeking reliable markers to predict chemotherapeutic efficacy in advanced esophagogastric cancer, tumour tissues from 72 patients from the collective of the FLO vs. FLP phase III study of the AIO were tested for expression of VEGFR-3 and CXCR4. Both of these molecules have recently been associated with aggressive and metastatic esophagogastric cancer [bib_ref] Combination of circulating CXCR4 and Bmi-1 mRNA in plasma: a potential novel..., Xu [/bib_ref] [bib_ref] Vascular endothelial growth factor receptor 3 is involved in tumor angiogenesis and..., Laakkonen [/bib_ref] [bib_ref] Correlation of vascular endothelial growth factor-D expression and VEGFR-3-positive vessel density with..., Choi [/bib_ref] [bib_ref] Vascular endothelial growth factor receptor-3 is a favorable prognostic factor in advanced..., Sung [/bib_ref] [bib_ref] Vascular endothelial growth factor-D and its receptor VEGFR-3: two novel independent prognostic..., Juttner [/bib_ref] [bib_ref] The expression of CXCL12 and CXCR4 in gastric cancer and their correlation..., Ying [/bib_ref] [bib_ref] Suppression of VEGFR-3 signaling inhibits lymph node metastasis in gastric cancer, Shimizu [/bib_ref] [bib_ref] Therapeutics target of CXCR4 and its downstream in peritoneal carcinomatosis of gastric..., Koizumi [/bib_ref] [bib_ref] Role of the CXCL12/CXCR4 axis in peritoneal carcinomatosis of gastric cancer, Yasumoto [/bib_ref] [bib_ref] Blocking on the CXCR4/ mTOR signalling pathway induces the anti-metastatic properties and..., Hashimoto [/bib_ref].
As previously described [bib_ref] Capecitabine and oxaliplatin for advanced esophagogastric cancer, Cunningham [/bib_ref] [bib_ref] Phase III trial in metastatic gastroesophageal adenocarcinoma with fluorouracil, leucovorin plus either..., Al-Batran [/bib_ref] , no significant difference between the therapeutic regimes FLO vs FLP was observed in terms of survival of the overall population over 5 years.
In our collective almost half of the specimens showed a strong positivity for VEGFR-3 (54%), whereas 36% of all tissue samples proved to be strongly positive for CXCR4 as assessed by immunohistochemistry. Similar findings have been reported previously in esophageal adenocarcinoma [bib_ref] Expression of chemokine receptor CXCR4 in esophageal squamous cell and adenocarcinoma, Gockel [/bib_ref] [bib_ref] Co-expression of receptor tyrosine kinases in esophageal adenocarcinoma and squamous cell cancer, Gockel [/bib_ref] as well as for adenocarcinoma of the stomach [bib_ref] The effect of the expression of vascular endothelial growth factor (VEGF)-C and..., Han [/bib_ref] [bib_ref] Coexpression of receptor-tyrosine-kinases in gastric adenocarcinoma-a rationale for a molecular targeting strategy?, Drescher [/bib_ref]. Taking into consideration that the majority of the CXCR4 and VEGFR-3 measurements, according to the literature, have been performed on tumour tissues after curative gastrectomy, we could expect an increase of the expression of these molecules in patients who are treated in a palliative setting.
In the present study there was a clear benefit in survival when the VEGFR-3 negative patients were treated with oxaliplatin instead of cisplatin. The stronger benefit in this category of patients was seen in the first 18 months of treatment. Conversely, patients with strong VEGFR-3 expression responded better under a cisplatin-containing regimen. Similar results have been reported by Ni et al. [bib_ref] Serum VEGFR-3 and survival of advanced gastric cancer patients treated with FOLFOX, Ni [/bib_ref] who showed significantly longer overall survival when Although the correlation of VEGFR-3 expression and response to chemotherapy has been the focus of research in various forms of cancer [bib_ref] Surrogate biomarkers in evaluating response to anti-angiogenic agents: focus on sunitinib, Deprimo [/bib_ref] [bib_ref] Antitumor activity and biomarker analysis of sunitinib in patients with bevacizumab-refractory metastatic..., Rini [/bib_ref] [bib_ref] VEGF-D expression correlates with colorectal cancer aggressiveness and is downregulated by cetuximab, Moehler [/bib_ref] , there are only limited data concerning gastric cancer [bib_ref] Analysis of anti-proliferative and chemosensitizing effects of sunitinib on human esophagogastric cancer..., Lyros [/bib_ref] [bib_ref] A randomized trial of somatostatin to regulate the VEGFs/VEGFRs in patients with..., Zhao [/bib_ref]. We have previously shown the addition of the VEGFR-1-3 inhibitor, sunitinib to enhance the chemosensitivity of gastric cancer cell lines in vitro [bib_ref] Analysis of anti-proliferative and chemosensitizing effects of sunitinib on human esophagogastric cancer..., Lyros [/bib_ref]. Our present results demonstrate a clear benefit of patients with strong VEGFR-3 expression in favour of FLP specifically when compared with patients with a low tumour VEGFR-3 expression [fig_ref] Figure 2: Survival analysis in patients treated with FLO versus FLP in relation to... [/fig_ref]. In those patients older than 60 years the benefit of a cisplatin containing therapy was even greater in the VEGFR-3 positive group, as no patient in the VEGFR-3 negative subgroup receiving FLP survived the first year [fig_ref] Figure 4: Survival analysis under FLP treatment for patients older than 60 years [/fig_ref].
A possible explanation for the enhanced chemosensitivity of VEGFR-3 positive gastric cancer cells to cisplatin may be deduced by inhibition of the Notch pathway. Tamella and colleagues [bib_ref] Blocking VEGFR-3 suppresses angiogenic sprouting and vascular network formation, Tammela [/bib_ref] revealed that VEGFR-3 upregulation in endothelial tip cells, caused by inhibition of the Notch signalling pathway, played a crucial role in sprouting angiogenesis in tumours. Since Notch inhibition has also been connected with the enhanced toxicity of cisplatin in colorectal cancer lines and nasopharyngeal adenocarcinoma [bib_ref] Gamma-secretase inhibition combined with platinum compounds enhances cell death in a large..., Aleksic [/bib_ref] [bib_ref] Inhibition of NOTCH3 signalling significantly enhances sensitivity to cisplatin in EBV-associated nasopharyngeal..., Man [/bib_ref] , our results may indicate a similar pathophysiological mechanism for advanced esophagogastric cancer. The reason why oxaliplatin appeared to be less effective in VEGFR-3 positive patients remains unknown. However our data are consistent with those of Aleksic et al. [bib_ref] Gamma-secretase inhibition combined with platinum compounds enhances cell death in a large..., Aleksic [/bib_ref] , who noticed that several colorectal cancer lines showed no responsiveness to oxaliplatin, compared with cisplatin, when Notch signalling was blocked.
When CXCR4 expression levels on tumour tissues were measured similar results were observed. Patients with weak CXCR4 expression profited more from FLO whereas CXCR4 positive patients had a significantly longer 5 year overall survival under FLP [fig_ref] Figure 3: Survival analysis under FLO/FLP treatment in regards of CXCR4 expression in an... [/fig_ref]. This effect could also be clearly seen in patients older than 60 years with strong CXCR4 positivity as they showed a better response to FLP than the CXCR4 negative ones [fig_ref] Figure 4: Survival analysis under FLP treatment for patients older than 60 years [/fig_ref].
To date, the significance of CXCR4 as a potential predictive marker for chemotherapy in gastric cancer has been reported only in cellular models [bib_ref] CXCR4, a potential predictive marker for docetaxel sensitivity in gastric cancer, Xie [/bib_ref]. Xie et al. showed a correlation of CXCR4 mRNA levels in gastric cancer with docetaxel sensitivity, whereas the blockade of CXCR4 enhanced docetaxel toxicity. Nevertheless, there are no data that provide a possible explanation for the better responsiveness of CXCR4 positive esophagogastric cancer to cisplatin than to other platinum derives.
It is of great interest in relation to the connection of extracellular-signal related kinases (ERKs) with the stromal cell-derived factor 1 (SDF-1) mediated pathway. Similar to VEGFR-3 [bib_ref] Gamma-secretase inhibition combined with platinum compounds enhances cell death in a large..., Aleksic [/bib_ref] [bib_ref] Effect of extracellular signalregulated kinase on p53 accumulation in response to cisplatin, Persons [/bib_ref] [bib_ref] Isolated lymphatic endothelial cells transduce growth, survival and migratory signals via the..., Makinen [/bib_ref] , the activation of CXCR4 by its natural ligand SDF-1 leads to phosphorylation and activation of multiple intracellular domains including ERKs [bib_ref] Role of the intracellular domains of CXCR4 in SDF-1-mediated signaling, Roland [/bib_ref] [bib_ref] The alpha-chemokine, stromal cell-derived factor-1alpha, binds to the transmembrane G-protein-coupled CXCR-4 receptor..., Ganju [/bib_ref] [bib_ref] Stromal cell-derived factor-1alpha stimulates tyrosine phosphorylation of multiple focal adhesion proteins and..., Wang [/bib_ref] [bib_ref] The chemokine SDF-1alpha triggers CXCR4 receptor dimerization and activates the JAK/STAT pathway, Vila-Coro [/bib_ref]. Since cisplatin relies on ERK activation for bioactivity in some cells [bib_ref] Gamma-secretase inhibition combined with platinum compounds enhances cell death in a large..., Aleksic [/bib_ref] [bib_ref] Effect of extracellular signalregulated kinase on p53 accumulation in response to cisplatin, Persons [/bib_ref] , in contrast to oxaliplatin [bib_ref] Additive interaction of oxaliplatin and 17-allylamino-17-demethoxygeldanamycin in colon cancer cell lines results..., Rakitina [/bib_ref] , this might explain the chemosensitivity of VEGFR-3 and CXCR4 positive esophagogastric adenocarcinoma to FLP and not to FLO.
# Conclusions
To the best of our knowledge this is the first comparative study of FLP and FLO in terms of VEGFR-3 and CXCR4 tumour expression. The list below shows the main findings and conclusions of our trial in accordance to the REMARK guidelines. The main limitation of our study is its size (n = 72) and thus, its power is not very high. However, our results suggest a predictive value of these biomarkers concerning chemotherapy with FLP or FLO in advanced esophagogastric cancer. A trend of longer OS was observed when CXCR4 and VEGFR-3 positive patients were treated with FLP, whereas FLO proved to be more effective in eligible. Patients characteristics, inclusion and exclusion criteria have already been described before [bib_ref] Phase III trial in metastatic gastroesophageal adenocarcinoma with fluorouracil, leucovorin plus either..., Al-Batran [/bib_ref]. Specimen characteristics and assay methods 3. Paraffin-embedded tissue samples were obtained from patients with advanced esophagogastric cancer. The expression of VEGFR-3 and CXCR4 was analyzed by immunohistochemistry. Evaluation of staining was performed by two independent, blinded investigators. Study design 4. This study is a retrospective translational analysis of a phase III trial in metastatic gastroesophageal adenocarcinoma. Patients were treated with fluorouracil, leucovorin plus either oxaliplatin or cisplatin as described before [bib_ref] Phase III trial in metastatic gastroesophageal adenocarcinoma with fluorouracil, leucovorin plus either..., Al-Batran [/bib_ref]. Main end point of the study was overall survival in relation to VEGFR-3/CXCR4 under palliative chemotherapy. Statistical analysis methods 5. The staining was evaluated by intensity and the extent of the stained tumour area. These two classifications were added together and divided into the categories negative and positive. All statistical analyses were done by using SPSS statistical analysis software. The survival and univariate analysis were performed by using Kaplan-Meier method, log rank test, Pearson's Chi-2 test or Fisher's exact test. Result Data 6. The primary tumour location as well as the basic demographic characteristics of the patients who participated at the current trial is demonstrated in [fig_ref] Table 1: Patient characteristics [/fig_ref]. There was an equal distribution of the patients in regards of treatment and positivity of markers. Analysis and presentation 7. In the survival analysis, patients with strong expression of CXCR4 on their tumour tissues profited more in terms of overall survival under the treatment of FLP. Patients with negative VEGFR-3 and CXCR4 expression had a trend to live longer when treated with FLO. In an exploratory analysis of patients older than 60 years at diagnosis, there was a significant benefit in overall survival for patients with strong VEGFR-3 and CXCR4 expression when treated with FLP. Discussion 8. CXCR4 positive patients profited in terms of OS from FLP, whereas FLO proved to be more effective in CXCR4 and VEGFR-3 negative patients. The main limitation of our study is its not very high power, due to the size of the examined tissues (n = 72). However, our results suggest a predictive value of these biomarkers concerning chemotherapy with FLP or FLO in advanced esophagogastric cancer. Further studies are required in order to investigate the predictive value of VEGFR-3 and CXCR4 in terms of chemotherapeutic regimes in patients with advanced adenocarcinoma of the stomach and the gastroesophagic junction.
## Additional file
Additional file 1: Comparative patient characteristics between the translational study and the overall phase III trial.
[fig] Figure 2: Survival analysis in patients treated with FLO versus FLP in relation to the expression of VEGFR-3. A. VEGFR-3 negative in observation time of 5 years. B. VEGFR-3 positive in observation time of 60 months. [/fig]
[fig] Figure 3: Survival analysis under FLO/FLP treatment in regards of CXCR4 expression in an observation period of 5 years. A. CXCR4 negative. B. CXCR4 positive. [/fig]
[fig] Figure 4: Survival analysis under FLP treatment for patients older than 60 years. A. VEGF-R3 and B. CXCR4 expression in an observation period of 5 years. [/fig]
[fig] Abbreviations VEGFR- 3: Vascular endothelial growth factor receptor 3; CXCR4: CXC-chemokine receptor type 4; GEJ: Gastroesophageal junction; FLO: Oxaliplatin/leucovorin/ 5-FU; FLP: Cisplatin/leucovorin/5-FU; AIO: Arbeitsgemeinschaft Internistische Onkologie. Competing interests After the last recruited patient M Moehler and S. -E. Al-Batran received honoraria for presentations in satellite symposia by Sanofi-Aventis. The rest of the authors have no competing interests to declare.Authors' contributions TT: acquisition of data, analysis and interpretation of data, drafting of the manuscript, critical revision of the manuscript, statistical analysis, [/fig]
[fig] Figure 5: Survival analysis under FLO/FLP treatment in an observation period of 5 years. A. VEGF-R3 and CXCR4 positive. B. VEGF-R3 and CXCR4 negative. [/fig]
[table] Table 1: Patient characteristics [/table]
[table] Table 2: Distribution of patients' treatment in relation to expression of VEGFR-3 or CXCR4 [/table]
|
Determining the Population Frequency of the CFHR3/CFHR1 Deletion at 1q32
In this study we have used multiplex ligation-dependent probe amplification (MLPA) to measure the copy number of CFHR3 and CFHR1 in DNA samples from 238 individuals from the UK and 439 individuals from the HGDP-CEPH Human Genome Diversity Cell Line Panel. We have then calculated the allele frequency and frequency of homozygosity for the copy number polymorphism represented by the CFHR3/CFHR1 deletion. There was a highly significant difference between geographical locations in both the allele frequency (X 2 = 127.7, DF = 11, P-value = 4.97x10 -22 ) and frequency of homozygosity (X 2 = 142.3, DF = 22, P-value = 1.33x10 -19 ). The highest frequency for the deleted allele (54.7%) was seen in DNA samples from Nigeria and the lowest (0%) in samples from South America and Japan. The observed frequencies in conjunction with the known association of the deletion with AMD, SLE and IgA nephropathy is in keeping with differences in the prevalence of these diseases in African and European Americans. This emphasises the importance of identifying copy number polymorphism in disease. Citation: Holmes LV, Strain L, Staniforth SJ, Moore I, Marchbank K, et al. (2013) Determining the Population Frequency of the CFHR3/CFHR1 Deletion at 1q32. PLoS ONE 8(4): e60352.
# Introduction
Complement genes within the RCA (Regulators of Complement Activation) cluster at chromosome 1q32 are arranged in tandem within two groups [bib_ref] Heine-Suner D (1999) A radiation hybrid map of complement factor H and..., Diaz-Guillen [/bib_ref]. In a centromeric 360 kb segment lie the genes for factor H (CFH) (OMIM 134370) and five factor H-related proteins -CFHR1 (OMIM 134371), CFHR2 (OMIM 600889), CFHR3 (OMIM 605336), CFHR4 (OMIM 605337) and CFHR5 (OMIM 608593). Sequence analysis of this region shows evidence of segmental duplications (SDs) resulting in a high degree of sequence identity between CFH and the genes for the five factor H related proteins [bib_ref] Complement factor H: sequence analysis of 221 kb of human genomic DNA..., Male [/bib_ref] [bib_ref] The factor H protein family, Zipfel [/bib_ref] [bib_ref] Factor H family proteins and human diseases, Jozsi [/bib_ref]. SDs such as those seen in the RCA cluster are frequently associated with genomic rearrangements [bib_ref] Genomic disorders: molecular mechanisms for rearrangements and conveyed phenotypes, Lupski [/bib_ref]. These usually occur as a result of non-allelic homologous recombination (NAHR) between SDs but can also be a result of gene conversion and microhomology mediated end joining (MMEJ) [bib_ref] MMEJ repair of double-strand breaks (director's cut): deleted sequences and alternative endings, Mcvey [/bib_ref]. Genomic disorders at this locus have affected CFH and the CFHRs in a number of ways. Deletions as a result of NAHR lead to the loss of CFHR1, CFHR3 and CFHR4. Deletions within genes, occurring through both NAHR and MMEJ, result in the formation of hybrid genes (CFH/CFHR1, CFHR1/CFH, CFH/ CFHR3, CFHR3/CFHR1) associated with diseases such as atypical haemolytic uraemic syndrome (aHUS) and membranoproliferative glomerulonephritis (MPGN) [bib_ref] Atypical haemolytic uraemic syndrome associated with a hybrid complement gene, Venables [/bib_ref] [bib_ref] A novel hybrid CFH/CFHR3 gene generated by a microhomology-mediated deletion in familial..., Francis [/bib_ref] [bib_ref] A hybrid CFHR3-1 gene causes familial C3 glomerulopathy, Malik [/bib_ref]. Complete deficiency of factor H related proteins 1 and 3 had been found to be occur in ,4% of a European population in protein studies before DNA studies of the region [bib_ref] Polymorphism and deficiency of human factor H-related proteins p39 and p37, Feifel [/bib_ref]. This DNA copy number polymorphism (CNP) has been extensively characterised in health and disease. It has been shown that the deletion is associated with the presence of factor H autoantibodies in aHUS [bib_ref] Deletion of complement factor H-related genes CFHR1 and CFHR3 is associated with..., Zipfel [/bib_ref] [bib_ref] Factor H autoantibodies in atypical hemolytic uremic syndrome correlate with CFHR1/ CFHR3..., Jozsi [/bib_ref] , with an increased risk of SLE [bib_ref] Association of genetic variants in complement factor H and factor H-related genes..., Zhao [/bib_ref] and a decreased risk of age-related macular degeneration [bib_ref] A common CFH haplotype, with deletion of CFHR1 and CFHR3, is associated..., Hughes [/bib_ref] and IgA nephropathy [bib_ref] Genome-wide association study identifies susceptibility loci for IgA nephropathy, Gharavi [/bib_ref] [bib_ref] Geographic Differences in Genetic Susceptibility to IgA Nephropathy: GWAS Replication Study and..., Kiryluk [/bib_ref]. That there might be differences in the population frequency of the CFHR3/1 deletion was suggested from a study published in 2006 which showed that the prevalence of homozygous deletion in African populations was ,16% [bib_ref] Extended haplotypes in the complement factor H (CFH) and CFH-related (CFHR) family..., Hageman [/bib_ref]. Population difference in the deletion have been confirmed in subsequent studies [bib_ref] Association of genetic variants in complement factor H and factor H-related genes..., Zhao [/bib_ref] [bib_ref] Factor H and CFHR1 polymorphisms associated with atypical Haemolytic Uraemic Syndrome (aHUS)..., Leban [/bib_ref] [bib_ref] A 32 kb critical region excluding Y402H in CFH mediates risk for..., Sivakumaran [/bib_ref]. In this study we have measured copy number of CFHR3 and CFHR1 with multiplex ligationdependent probe amplification (MLPA) [bib_ref] Relative quantification of 40 nucleic acid sequences by multiplex ligationdependent probe amplification, Schouten [/bib_ref] in a range of populations derived from the HGDP-CEPH Human Genome Diversity Panel (http://www.cephb.fr/en/hgdp/diversity.php) [bib_ref] A human genome diversity cell line panel, Cann [/bib_ref]. [bib_ref] Association scan of 14,500 nonsynonymous SNPs in four diseases identifies autoimmunity variants, Burton [/bib_ref]. The samples from the Health Protection Agency were originally obtained from a control population of randomly selected, non-related UK Caucasian blood donors. The full collection of samples with the HGDP-CEPH panel consists of 1051 individuals from 51 world populations (http://www.cephb.fr). We selected for analysis 439 samples from 17 different countries comprising 25 different populations [fig_ref] Table 1: HGDP-CEPH samples used for measurement of CFHR3 and CFHR1 copy number [/fig_ref]. We did not include populations for which data is either already available (for example European populations such as France) or where samples numbers were too small to be representative. There were still some populations with a small number of samples, including the sub-Saharan region of Africa. These were combined into 11 geographical locations [fig_ref] Table 2: Allele frequencies and counts of the CFHR3/CFHR1 deletion in UK and HGDP-CEPH... [/fig_ref] for subsequent analysis. In each of these locations the number of samples was greater than 20. In total 133 samples from African populations were analysed, including 83 from sub-Saharan countries. CFHR1 and CFHR3 copy number was measured as described previously [bib_ref] Association of factor H autoantibodies with deletions of CFHR1, CFHR3, CFHR4 and..., Moore [/bib_ref] using multiplex ligation-dependent probe amplification [bib_ref] Relative quantification of 40 nucleic acid sequences by multiplex ligationdependent probe amplification, Schouten [/bib_ref] (MLPA) using a kit from MRC Holland (www. mlpa.com) (SALSA MLPA kit P236-A1 ARMD) and in house probes.
# Methods
# Ethics statement
## Statistics
Chi-square analysis was used to test whether there was deviation from Hardy-Weinberg equilibrium in the geographical locations. A p value of ,0.05 was considered to be not consistent with Hardy-Weinberg equilibrium. Chi-square analysis was undertaken to determine whether there was a significant difference between geographical locations in either the allele frequency of the CFHR3/CFHR1 deletion, the genotype frequencies (del + + /del + 2 /del 2 2 ) or the frequency (del + + ) of a homozygous deletion of CFHR3/CFHR1. Fisher's exact tests were undertaken to determine whether in different geographical locations either the allele frequency of the CFHR3/CFHR1 deletion, the genotype frequencies (del ++ /del +2 /del 22 ) or the frequency of a homozygous deletion of CFHR3/CFHR1 were significantly different to the values for these variables in the UK population, or to their values in all other populations combined.
# Results
The allele frequency of the CFHR3/CFHR1 deletion in the various geographical locations and the individual populations within these locations is shown in [fig_ref] Table 2: Allele frequencies and counts of the CFHR3/CFHR1 deletion in UK and HGDP-CEPH... [/fig_ref]. There was no deviation from Hardy-Weinberg equilibrium in any of the geographical locations. The CFHR3/CFHR1 deletion was not present in either the South American or Japanese locations. The highest allele frequency for the deletion was 54.7% in Nigeria. The deletion was Africa and Russia but was significantly higher in sub-Saharan Africa and Nigeria. The CFHR3/CFHR1 deletion was not found in homozygosity in Mexico, South America, China, Japan, Pakistan or Siberia. The frequency of homozygous deletion was 3.4% in the UK, between 5-10% in Italy, Russia, North Africa and Sub-Saharan Africa, and 33.3% in Nigeria. Differences in genotype frequencies between geographical locations were highly significant (X 2 = 142.3, DF = 22, P-value = 1.33610 219 ). Differences in the frequency of the homozygous del ++ genotype were also highly significant (X 2 = 56.8, DF = 11, P-value = 3.66610 28 ). The level of statistical significance derived using Fisher's exact test for the del ++ /del +2 /
# Discussion
In this study we have used multiplex ligation-dependent probe amplification (MLPA) to determine the copy number of CFHR3 and CFHR1 in a variety of different geographical locations derived from the HGDP-CEPH collection. MLPA has the advantage over other techniques that have been used in that it provides a specific determination of copy number. We measured copy number of both CFHR3 and CFHR1 to determine the deleted allele frequency because measurement of CFHR1 copy number alone is not specific to this allele as it also occurs with the CFHR1/CFHR4 deletion [bib_ref] Association of factor H autoantibodies with deletions of CFHR1, CFHR3, CFHR4 and..., Moore [/bib_ref] [bib_ref] Characterization of complement factor H-related (CFHR) proteins in plasma reveals novel genetic..., Abarrategui-Garrido [/bib_ref]. Using MLPA we have been able to determine both the allele frequency of the deletion and the frequency of a homozygous deletion. For statistical purposes we have set the UK population as our reference. The value of 3.4% for the frequency of a homozygous deletion in the UK population in this study is similar to values that we have obtained in previous studies [bib_ref] Deletion of complement factor H-related genes CFHR1 and CFHR3 is associated with..., Zipfel [/bib_ref] [bib_ref] Association of factor H autoantibodies with deletions of CFHR1, CFHR3, CFHR4 and..., Moore [/bib_ref] [fig_ref] Table 4: Reported population frequencies of the CFHR3/CFHR1 deletion [/fig_ref] and the allele frequency of the deletion is similar to that which we obtained on introduction of the MLPA assay (17.3% in Moore et al [bib_ref] Association of factor H autoantibodies with deletions of CFHR1, CFHR3, CFHR4 and..., Moore [/bib_ref]. The latter value is similar to the frequency of 18.3% that we have obtained in this study.
The values for the allele frequency of the deletion, the genotype frequencies, and the frequency of a homozygous deletion that we obtained for world-wide populations using the HGDP-CEPH collection show marked population differences with the highest frequencies being seen in African populations. The findings in the African groups are consistent with those reported (Hageman et al [bib_ref] Extended haplotypes in the complement factor H (CFH) and CFH-related (CFHR) family..., Hageman [/bib_ref] in African Americans and validate their findings in HGDP-CEPH African samples which were based on a gene specific PCR method that measured frequency of a homozygous deletion Subsequently there have been several other studies documenting the frequency of the CFHR3/CFHR1 deletion in a range of populations. The results from these are shown in [fig_ref] Table 4: Reported population frequencies of the CFHR3/CFHR1 deletion [/fig_ref]. The values in this study for both allele frequency and the frequency of homozygous deletion are consistent with previous studies particularly for the UK, Japanese, Chinese and Nigerian populations. We chose in this study to combine several populations from subsaharan Africa as the numbers for each group were small. However, the study of Sivakumaran et al [bib_ref] A 32 kb critical region excluding Y402H in CFH mediates risk for..., Sivakumaran [/bib_ref] suggests that for this region there are significant differences in the allele frequency of the deleted allele between tribes. For instance, they found an allele frequency of 23.8% in the Maasai tribe of Kenya compared to 42.3% in the Luhya. As can be seen in [fig_ref] Table 2: Allele frequencies and counts of the CFHR3/CFHR1 deletion in UK and HGDP-CEPH... [/fig_ref] we also observed differences in the allele frequencies of the different populations within this geographical location. For instance in the Biaka pygmyies the allele frequency was 8.7% compared to 50% in the Kenyan Bantus and Senegal Mandenka tribes. Recent studies documenting the genetic variation in this region show evidence of at least two different genetic groups derived from the North and South of the Kalahari [bib_ref] The genetic prehistory of southern Africa, Pickrell [/bib_ref] [bib_ref] The genetic structure and history of Africans and African Americans, Tishkoff [/bib_ref]. This may explain the differences in the allele frequency that we have seen in sub-saharan Africa. It is possible that ancestral African populations with a low allele frequency of the deletion were the ones which participated in the ''out of Africa'' dispersal with the associated bottleneck reinforcing the low allele frequency. That generally the current African populations with a low allele frequency of the deletion are Huntergatherers is compatible with this [bib_ref] The great human expansion, Henn [/bib_ref] [bib_ref] Hunter-gatherer genomic diversity suggests a southern African origin for modern humans, Henn [/bib_ref]. The high allele frequency of the deletion in the African-American population is compatible with the allele frequency seen in the Yoruba and Mandenka [bib_ref] The genetic structure and history of Africans and African Americans, Tishkoff [/bib_ref] [bib_ref] Characterizing the admixed African ancestry of African Americans, Zakharia [/bib_ref]. How in evolution has this deletion arisen and how can the population differences be explained? The alternative pathway of complement is thought to be the oldest component of the innate immune system [bib_ref] Genomic view of the evolution of the complement system, Nonaka [/bib_ref]. The earliest components of the alternative complement pathway to have been recognised are activators such as C3 which has been identified in a coral [bib_ref] Characterization of a C3-like cDNA in a coral: phylogenetic implications, Dishaw [/bib_ref] suggesting their presence in the Cnidria. Regulatory components have been first recognised in the Agnatha with for instance identification of a C3 cleaving short consensus repeat protein in lamprey [bib_ref] A short consensus repeat-containing complement regulatory protein of lamprey that participates in..., Kimura [/bib_ref]. A protein (called SBP1) with a high degree of homology to human factor H was first described in the teleost, sand bass [bib_ref] Ancient origin of human complement factor H, Krushkal [/bib_ref]. Factor H has also been identified in the zebrafish [bib_ref] Zebrafish complement factor H and its related genes: identification, evolution, and expression, Sun [/bib_ref]. In the zebrafish, the mouse and humans there are genes encoding SCR proteins with a high degree of homology to factor H in close proximity to the gene encoding factor H. In man there are the five factor H related proteins (CFHR1-5), in the mouse there are three factor H related proteins and in the zebrafish there are 4 factor H like proteins. Sequence analysis of this region in man suggests that these genes have arisen through a series of segmental duplication events [bib_ref] Complement factor H: sequence analysis of 221 kb of human genomic DNA..., Male [/bib_ref]. Analysis of primate genomes undertaken by Sivakumaran et al suggests that chimps have more extensive duplication in this region than humans. The analysis also suggests that the duplications arose in a common ancestor of the chimp and humans after divergence from the orang-utan [bib_ref] A 32 kb critical region excluding Y402H in CFH mediates risk for..., Sivakumaran [/bib_ref]. The duplicated segments predispose to both non-allelic homologous recombination (NAHR) and gene conversion [bib_ref] Structural variation in the human genome, Lupski [/bib_ref]. The available evidence would suggest that the CFHR3/CFHR1 deletion has arisen through NAHR after the initial formation of the SDs. Sivakuram et al used phylogenetic and linkage equilibrium analysis to determine the ancestral orgin of the deletion [bib_ref] A 32 kb critical region excluding Y402H in CFH mediates risk for..., Sivakumaran [/bib_ref] and found a single origin in Caucasians and Asians but a recurrent origin in Africans. We believe that in certain populations that the deletion has resulted in an evolutionary benefit. There is evidence to suggest that polymorphisms in complement proteins are associated with susceptibility to infection [bib_ref] Complement polymorphisms: Geographical distribution and relevance to disease, Ermini [/bib_ref]. For instance mannose-binding lectin (MBL) binds to microbes and activates the lectin pathway. Allelic variants in the gene (MBL2) encoding this protein are associated with differences in both the serum level and function of MBL. The frequency of these allelic variants differs in populations; and the same variants are associated with a differential risk of pneumococcal disease and leprosy. Recently it has been shown that variants in CFH and CFHR3 are associated with susceptibility to meningococcal disease [bib_ref] Genome-wide association study identifies variants in the CFH region associated with host..., Davila [/bib_ref]. These observations taken with the knowledge that complement plays a significant role in the pathogenesis other diseases such as malaria [bib_ref] Complement driven innate immune response to malaria: fuelling severe malarial diseases, Silver [/bib_ref] would suggest that infection has driven the geographical variability seen in complement variants such as the CFHR3/1 deletion.
Since the CFHR3/CFHR1 deletion was first described a number of studies have documented strong linkage disequilibrium of the deletion with common CFH haplotypes [bib_ref] Associations of CFHR1-CFHR3 deletion and a CFH SNP to age-related macular degeneration..., Raychaudhuri [/bib_ref] [bib_ref] Deletion of CFHR3 and CFHR1 genes in age-related macular degeneration, Spencer [/bib_ref]. In some populations the deletion is present on haplotypes H1-5 and absent on H6-7 [bib_ref] Associations of CFHR1-CFHR3 deletion and a CFH SNP to age-related macular degeneration..., Raychaudhuri [/bib_ref]. In other populations the H2 haplotype perfectly tags the deletion [bib_ref] Genome-wide association study identifies susceptibility loci for IgA nephropathy, Gharavi [/bib_ref]. Likewise in some populations individual SNPs have been shown to be in complete LD with the deletion. Zhao et al found that the deletion was in complete LD with rs6677604 in European Americans but not in African Americans (r 2 = 0.60). Whether the deletion confers an independent risk for AMD, SLE and IgA nephropathy or is simply associated with protective/atrisk haplotypes is an area of controversy [bib_ref] A 32 kb critical region excluding Y402H in CFH mediates risk for..., Sivakumaran [/bib_ref] [bib_ref] Associations of CFHR1-CFHR3 deletion and a CFH SNP to age-related macular degeneration..., Raychaudhuri [/bib_ref] [bib_ref] Reply to ''Associations of CFHR1-CFHR3 deletion and a CFH SNP to age-related..., Hughes [/bib_ref]. However, factor H related protein 1 blocks the C5 convertase but binds, in competition with factor H, to host surfaces through its C-terminal regulatory domain [bib_ref] Factor Hrelated protein 1 (CFHR-1) inhibits complement C5 convertase activity and terminal..., Heinen [/bib_ref]. We are, therefore, of the opinion that deletion of CFHR1 has a dual effect with reduced inhibition of terminal complement pathway activity but increased regulation by factor H of the alternative pathway. This may also explain why in some diseases (AMD and IgA nephropathy) the deletion is protective whilst in other others (SLE) it is associated with increased risk.
It is also possible that CFHR3 has functional activity that contributes to the disease association seen with the CFHR3/1 deletion. In African Americans with a higher frequency of the deletion the prevalence of AMD and IgA nephropathy is lower than in European Americans [bib_ref] Low incidence of IgA nephropathy in blacks, Jennette [/bib_ref] [bib_ref] Prevalence of age-related macular degeneration in the US population, Klein [/bib_ref] whereas the prevalence of SLE is higher [bib_ref] Epidemiology of systemic lupus erythematosus: a comparison of worldwide disease burden, Danchenko [/bib_ref]. Thus studying the population frequency of disease associated CNPs such as the CFHR3/CFHR1 deletion provides novel insights into the pathogenesis of such diseases. However, at an individual level we do not think that screening for the deletion in the normal population is currently of any clinical benefit.
[table] Table 1: HGDP-CEPH samples used for measurement of CFHR3 and CFHR1 copy number. [/table]
[table] Table 2: Allele frequencies and counts of the CFHR3/CFHR1 deletion in UK and HGDP-CEPH populations. [/table]
[table] Table 3: Homozygous CFHR3/CFHR1 deletion frequencies in UK and HGDP-CEPH worldwide populations. [/table]
[table] Table 4: Reported population frequencies of the CFHR3/CFHR1 deletion. [/table]
|
Bio-grout based on microbially induced sand solidification by means of asparaginase activity
Bio-grout, a new ground improvement method, has been recently developed to improve the mechanical properties, decrease the permeability of porous materials, reinforce or repair cementitious materials and modify the properties of soil or sand. Bio-grout production depends on microbially induced calcite precipitation (MICP), which is driven mainly by an enzyme, urease. However, urease-based MICP process produces excessive ammonia, in addition to secondary pollution generated by urea that is used as substrate in it. In the present study, we reported asparaginase-based MICP process for sand bio-grout development using Bacillus megaterium, and results were also compared with urease-based bio-grouts. The asparaginase activity led to significantly less ammonia production compared to urease without compromising with desired properties of a novel grout. The UCS of bio-grout was obtained at 980 kPa, while the permeability was decreased substantially. The mineralogical composition of precipitated substance was identified as calcite using XRD and the crystal morphology was observed under SEM. The mass percentage of calcite in bio-grout was calculated by thermogravimetric analysis and XCT verified calcite precipitation in it. The results confirmed that biocalcification by means of bacterial asparaginase is a potential solution for geotechnical problems. The asparaginase-based MICP process could be of wider acceptance in future.The importance of old buildings restoration is growing in the construction sector as problems such as detachments, cracking and crystallization of soluble salts, among others, appear or are accentuated 1 . Such problems can be solved through improving mechanical properties of system that can fill cavities, fissures, voids or cracks by development of grouts. The application of grout materials and grouting technique are widely popular for construction nowadays 2 ; however, environmental safety must be taken in account while selecting grouts due to negative effects of chemical grouts on environment. In addition, though many grouts were reported with high strength, the compressive strength of grout-injected walls was not found to directly proportional to the compressive strength of the injection grouts used for grouting 3 . Moreover, the grout strength depends mainly on the bond between grout and in situ material, which is not necessarily proportional to the compressive or tensile strength of the grout 4 . To achieve this dual nature of novel grout, we should look for such alternative, which is environmentally sound as well as having self-binding property. Researchers are looking for ideal grout that can be used with high injectability and bonding properties. Bio-grout has potential to increase the cohesion with adjacent materials (whether cementitious, sand, stone or soil) that could enhance the mechanical behavior of the whole system. Such grouts can serve as excellent alternative for natural stone also.Recently, bio-grout has been developed based on biological activity as a new ground improvement technology with potential to use with cementitious materials 5 . It could improve the mechanical properties (strength, stiffness, cohesion, friction), decrease the permeability of porous materials, reinforce or
repair cementitious materials and modify the properties of soil or sand [bib_ref] Microbial carbonate precipitation as a soil improvement technique, Whiffin [/bib_ref] [bib_ref] Bio-mediated soil improvement, Dejong [/bib_ref] [bib_ref] Study on microstructure and properties of sandstone cemented by microbe cement, Rong [/bib_ref] [bib_ref] Microbial concrete-A way to enhance the durability of building structures, Achal [/bib_ref]. Bio-grout mainly depends on microbially induced calcite precipitation (MICP), carried out by urease producing bacteria. During this process, bacteria hydrolyze a substrate, urea by producing an enzyme, urease that leads to increment in pH by the production of ammonium ions. Further, CO 2 dissolution takes place and carbonate ions combines with calcium ions either present in the proximal environment or on the cell wall of bacterial cells to precipitate CaCO 3 . The overall reactions may summarize as follow [bib_ref] Biomineralization and evolutionary history, Knoll [/bib_ref] The reactions 6 and 7 may be replaced with following equation.
[formula] + → ( ) + − Ca CO CaCO 8 2 3 2 3 [/formula]
The problem associated with urease driven MICP process is the production of excessive ammonia, which has a negative impact on human health and the environment when the concentration of ammonia and ammonium is higher than safety threshold [bib_ref] The influence of standing time and content of the slurry on bio-sandstone..., Yu [/bib_ref]. Ammonia is poisonous at high concentrations. In addition, as ammonia is lighter than air, it dissipates in the atmosphere easily and its distinctive smell makes unfavorable condition to work in such environment. Moreover, urea is widely used as substrate in the MICP process and because of its agricultural use and being raw material for fertilizer production, economic viability may not be associated with it, in addition to secondary pollutant nature of urea [bib_ref] Microbially mediated sand solidification using calcium phosphate compounds, Akiyama [/bib_ref].
In order to solve such problems, the present study focuses on MICP process driven by asparaginase enzyme with commercially available asparagine as substrate. Further, microbially induced sand solidification was studied based on MICP with potential to use as bio-grout. The research has great importance in reducing effect of ammonia on environment without compromising with mechanical and impermeable properties of bio-grout in sand.
# Results
Asparaginase activity. The present study focused on MICP based on asparaginase, and to compare results with usual enzyme urease for the purpose of biocalcification. At flask level experiments, Nutrient-Asparagine broth (nutrient media supplemented with asparagine) was luxurious for bacterial growth and pH of culture increased up till 9.55. The pH increased continuously up till 144 h, and reduced marginally with value 9.3 at 168 h. The asparaginase activity during a period of 7 days ranged from 25.1 to 40.6 U ml −1 with maximum activity at 144 h and decreased thereafter.
On other hand, urease activity lead to pH increment till 10.8 at 120 h and at the same time maximum urease production (592 U ml −1 ) was recorded. Urease activity decreased after 120 h.
Mechanical properties. The UCS (Unconfined Compressive Strength) was obtained at 980 kPa for asparaginase-based bio-grout, compared to that of 1002 kPa due to urease activity; however, it completely failed for grout samples due to lack of calcite precipitation in the absence of bacterial cells. The permeability of asparaginase-based bio-grout was recorded at 2.3 × 10 −7 m s −1 compared to 2.0 × 10 −7 m s −1 of bio-grout produced from urease activity.
## Sem-eds (scanning electron microscope-energy dispersive x-ray spectrometer) analysis.
Bio-grout samples presented clear crystal morphology of calcium carbonate nature among particles of sand, when observed under SEM, along with rod-shaped cells of Bacillus megaterium [fig_ref] Figure 2: SEM showing imprint of rod-shaped B [/fig_ref].
In bio-grout, the area chiefly housed by bacterial cells showed high amount of calcium, while higher on calcite crystals precipitated, suggesting calcite precipitation as depicted in EDS spectra [fig_ref] Figure 3: EDS showing high amount of calcium due to B [/fig_ref]. On other hand, the grout (without bacterial cells) consisted of only media, showed very small amount of calcium coming from constituent of Nutrient-Asparagine media used (data not shown).
Scientific RepoRts | 5:16128 | DOi: 10.1038/srep16128 XRD (X-ray Diffraction) and XCT (X-ray Computed Tomography). The mineralogical composition of precipitated substance in bio-grout was identified using XRD. The bacterial strain used in the present study was able to produce calcium carbonate based on MICP process. X-ray diffractograph showed the mineralogical composition of the precipitated mineral in bio-grout with only observed mineral as calcite [fig_ref] Figure 4: XRD identifying several peaks of calcite precipitated by B [/fig_ref].
To investigate the inner structure and uniformity of bio-grout coming from distribution of the calcite precipitates in the whole column, XCT analysis was performed. The density distribution of the bio-grout at each scanning steps as a result of the XCT is presented in, while the closer view of the specimen could be seen in. 2D horizontal slice as cross sections from both top and bottom layers of bio-grout specimen is shown in,d. The density of the top cross section was higherthan that of bottom cross section, could be seen clearly due to the presence of air voids. As B. megaterium of the present study is aerobic bacterium, higher microbially induced carbonate is expected to precipitate on the top layer compared to the bottom layer.
Thermogravimetric analysis (TGA). The mass percentage of calcium carbonate in three layers (top, middle and bottom) of bio-grout calculated via TGA was presented in [fig_ref] Figure 6: TGA graphs of top, middle and bottom layers of bio-grout specimen [/fig_ref] in the present study, determined by sharp weight loss due to decarbonation. There were variations in weight losses between different layers of bio-grout as more weight loss was associated with higher calcium carbonate content. The combined large weight loss started at 155 °C and was completed by 1000 °C in top layer of bio-grout; however, started at 405 °C in middle layer of bio-grout [fig_ref] Figure 6: TGA graphs of top, middle and bottom layers of bio-grout specimen [/fig_ref]. The bottom layer of bio-grout behaved differently, probably due to low amount of calcite, where weight loss initiated and completed around 1000 °C [fig_ref] Figure 6: TGA graphs of top, middle and bottom layers of bio-grout specimen [/fig_ref]. The weight loss due to absorbed water was observed between 0 and 100 °C, while dehydrated CaCl 2 attributed to weight loss between 100 and 200 °C. The decarbonation or calcium carbonate decomposition led to weight loss between 600 and 900 °C.
# Discussion
The enzyme activities of asparaginase and urease used in the present study measures ammonia liberation, thus provides a comparative feature on ammonia released into the environment. The enzyme asparaginase hydrolyzes asparagine into aspartate and releases ammonium ions (Eq. 9). The hydrolyzed products equilibrate in water to form alanine and bicarbonate ions (Eq. 10) that give rise to an increase in pH and ultimately shift the bicarbonate equilibrium, resulting in the formation of carbonate ions (Eq. 11). Finally, the increased carbonate concentration will induce an increase in supersaturation level leading to CaCO 3 precipitation around the bacterial cell in the presence of soluble calcium ions (Eqs 12 and 13).
[formula] + → + ( ) − + C H N O H O C+ ↔ + ( ) − − − HCO OH CO H O 11 3 3 2 2 + → − ( ) + + Ca Cell Cell Ca 12 2 2 − + → − ( ) + − − Cell Ca CO Cell CaCO 13 2 3 2 3 [/formula]
The equation 10 is probable reaction that requires further investigation, if any additional enzyme involved in the process.
The asparaginase activity raised the pH up till 9.5 at 144 h; however, on other hand urease-based MICP resulted in pH increase up to around 11 at 120 h and decreased thereafter. The decreased pH might be due to drop in ammonia production as depicted from a decrease in enzyme activity. Unlike urease activity that decreased after 5 days, asparaginase continued increasing up to 6 days. The maximum release of ammonia, as depicted from enzymatic activities, in the present study was 40.6 U ml −1 via asparaginase that was significantly less compared to ammonia production due to urease activity (592 U ml −1 ). In other studies, the maximum urease production by S. pasteurii was 412 U ml −1 14 , while B. megaterium SS3 produced 650 U ml −1 of urease [bib_ref] Bacillus megaterium mediated mineralization of calcium carbonate as biogenic surface treatment of..., Dhami [/bib_ref]. The results of present study suggest that asparaginase activity is preferable over urease in releasing significantly less ammonia into the environment. Further, organic waste could be utilized in future for asparaginase production in the media to make asparaginase-based MICP economic and environment friendly.
To the best of our knowledge, this is the first research where asparaginase rather than urease was explored in detail for MICP. The MICP followed ammonification i.e., the process of amino acid deamination after hydrolysis of protein by heterotrophic bacteria as the pH-increasing reaction [bib_ref] The global nitrogen cycle, Galloway [/bib_ref].
The results of UCS, and permeability test in bio-grout confirmed that biocalcification or carbonate precipitation by means of bacterial asparaginase is a potential solution for geotechnical problems, as it could induce a gain of compressive strength and at the same time could reduce the permeability. The generation of calcium carbonate could improve the surface density and reduce the permeability of bio-grouts. The results were comparative with urease-based bio-grout, where UCS and impermeability was only marginally higher . The MICP process is widely reported to improve the geomechanical properties with a gain of compressive strength [bib_ref] Comparison of the ability of two bacteria to improve the behavior of..., Venda Oliveira [/bib_ref].
The bacterial asparaginase activity led to calcite capitation on sand surface, and between sand particles, as shown in SEM [fig_ref] Figure 2: SEM showing imprint of rod-shaped B [/fig_ref]. It has been reported that formation of calcite promotes solidification and increases strength of sandstones or cementitious materials due to MICP [bib_ref] Strain improvement of Sporosarcina pasteurii for enhanced urease and calcite production, Achal [/bib_ref] [bib_ref] Cementation of sand grains based on carbonate precipitation induced by microorganism, Qian [/bib_ref]. The MICP process provides well-defined phenomenon in filling of pores in sand rich medium with carbonate crystals [bib_ref] Optimization of calcium-based bioclogging and biocementation of sand, Chu [/bib_ref]. Further, EDS indicated the higher amount of Ca (calcium) in bio-grout that indicated the presence of calcite due to MICP [bib_ref] Study on soil solidification based on microbiological precipitation of CaCO3, Lu [/bib_ref].
The XRD analysis demonstrated that calcium carbonate in the sand biogrout was in the form of calcite, as consistent with TGA results. The similar results were obtained in previous studies where calcite was identified as main biomineral 21 ; however, unlike asparaginase based MICP, most of urease driven MICP also resulted in other forms of calcium carbonates such as aragonite and vaterite [bib_ref] Strain improvement of Sporosarcina pasteurii for enhanced urease and calcite production, Achal [/bib_ref] [bib_ref] Corrosion protection of cement-based building materials by surface deposition of CaCO 3..., Qian [/bib_ref]. Further studies are required to explain why only calcite was formed as biomineral due to asparaginase activity in sand bio-grout.
As shown under XCT, the inner structure in biogrout was non-uniform and heterogeneous, also observed in previous study [bib_ref] Study on microstructure and properties of sandstone cemented by microbe cement, Rong [/bib_ref]. This could be attributed by non-uniform distribution of bacteria and precipitated carbonate in the bio-grout. Tomography is highly desired to locate areas inside the bio-grout columns, as that could not be visually apparent by naked eyes. XCT is recently established as technique to evaluate biocementation in a specimen based on the distribution of calcite precipitation [bib_ref] X-ray computed tomography proof of bacterial-based self-healing in concrete, Wang [/bib_ref].
The precipitates in bio-grouts were also confirmed to be calcium carbonate by TGA analysis. The results of TGA clearly differentiated calcite content in different layers of bio-grout with amount of calcite decreasing from top to bottom layer. The results were consistent with previous findings where B. sphaericus LMG 22257 induced calcium carbonate precipitation in sand [bib_ref] Sand bioconsolidation through the precipitation of calcium carbonate by two ureolytic bacteria, Shirakawa [/bib_ref]. The weight loss before 200 °C, especially clearly shown in top layer of bio-grout was caused by the dehydration of the samples. A sharp weight loss occurred in the temperature range of 600-850 °C in all three layers of bio-grout columns. It was attributed due to calcite decarbonation [bib_ref] The use of thermal analysis in assessing the effect of temperature on..., Alarcon-Ruiz [/bib_ref] , while other studies reported weight loss around 650 °C due to decomposition of CaCO 3 [bib_ref] Thermogravimetric analysis of selected group (II) carbonate minerals -Implication for the geosequestration..., Frost [/bib_ref]. It has also been reported that calcite produced by biomineralization get decomposed from 600 to 750 °C [bib_ref] Cementation of sand grains based on carbonate precipitation induced by microorganism, Qian [/bib_ref]. The theoretical decomposition reaction for microbially produced calcite is as follows:
[formula] ( ) → ( ) + ( ) CaCO s CaO s CO g 3 2 [/formula]
TGA demonstrated a substantial increase in calcium carbonate content in bio-grout compared to grout without bacterial cells, further adds a value upon importance of MICP in the process of grouting.
The calcite amount is not reported to continuous or same throughout the microbial sand column in previous studies. As studied previously, slightly more bacteria were trapped within the first few cm (∼ 54% in 0-3.8 cm section) of sandstone compared to (∼ 40% in 3.8-7.5 cm section) due to aerobic nature of S. pasteurii induced mineral precipitation [bib_ref] Transport of Sporosarcina pasteurii in sandstone and its significance for subsurface engineering..., Tobler [/bib_ref]. It is also possible that the transport of bacteria might get minimized or limited after calcite precipitation in top layers of bio-grout, leading to lower calcite precipitation in subsequent layers.
In conclusion, the advantage of using asparaginase-based MICP process over urease one is relatively lower production of ammonia that is better for the environment. Additionally, the process doesn't require urea, which emits secondary pollution. Further research is required to use organic sources of asparagine rather than commercially available to make overall process economic. The asparaginase-based MICP process could reduce huge amount of ammonia released into the environment.
# Materials and methods
Materials. The bacterium, Bacillus megaterium CGMCC 1.1741 was used in this study. The bacterial strain was maintained on Nutrient Agar medium (pH 8.0). For MICP reactions based on asparaginase activity, Nutrient-Asparagine medium was used with following compositions (per L): 5 g tryptone, 5 g yeast extract and 2 g NaCl, amended with filter sterilized 0.2 M asparagine and 40 mM CaCl 2 (pH 7.5).
The quartz sand, locally collected, was used as aggregate in the present study with grain size characteristics d 10 = 150 μ m (10% of the grains were smaller than this diameter); d 90 = 300 μ m.
Asparaginase assay. Asparaginase activity was measured for the confirmation of microbially induced calcite precipitation. The bacterial strain was grown in nutrient broth medium for seed culture development from where 1% of the culture medium was inoculated in Nutrient-Asparagine medium.
Asparaginase assay was carried out as per Imada et al. [bib_ref] Asparaginase and glutaminase activities of micro-organisms, Imada [/bib_ref]. Briefly, 0.5 ml of culture supernatant was collected, added into 0.5 ml of 0.5 M Tris-HCl buffer (pH 8.4) and mixed with 0.5 ml of 0.04 M L-asparagine and milli Q water to a total volume of 2 ml. The mixtures were incubated for 30 min at 37 °C and reaction was stopped by adding 0.5 ml of 1.5 M TCA. To 3.7 ml of milli Q water, 0.1 ml of reaction mixtures was added with 0.2 ml of Nessler's reagent and the color reaction was allowed to proceed for 20 min before recording absorbance at 450 nm. A standard curve was prepared from solutions of ammonium sulfate as the ammonia source.
For comparison, urease activity was determined in nutrient broth containing filter-sterilized 2% urea and 25 mM CaCl 2 , by measuring the amount of ammonia released from urea as described in Achal et al. [bib_ref] Strain improvement of Sporosarcina pasteurii for enhanced urease and calcite production, Achal [/bib_ref] One international unit (U) of L-asparaginase or urease is the amount of enzyme, which liberates 1 μ mole of ammonia per minute.
Bio-grout preparation. Bio-grouts were prepared as a result of microbially induced sand solidification in 60 mL plastic syringes column (internal diameter 3.0 cm, length 11.0 cm). The columns were packed with sand up to 10 cm height, after placing a layer of glass wool at the bottom. The columns filled with sand were saturated twice with Tris-HCl buffer that avoids the inclusion of air pockets between sand grains and positioned vertically on clamp. A pump was connected to an injection point at the bottom of the column to regulate the flow rate, as explained in Rong et al. [bib_ref] Study on microstructure and properties of sandstone cemented by microbe cement, Rong [/bib_ref].
The bacterial cells grown in Nutrient-Asparagine medium with OD 600 1.00 were poured in the sand column at a constant flow rate of 5 mL/min continuously for 10 hours. The columns were left as such and thereafter, Nutrient-Asparagine medium was injected into the column at flow rate of 10 mL/min till no solution leaked out of column. After a week, columns were oven dried at 60 °C for 48 h and bio-grout samples were demoulded for further analyses.
Strength and permeability of Bio-grout. The bio-grout samples as sandstones were tested to record unconfined compressive strength. The falling head permeability test (as per ASTM D2435-04) was carried out to determine the permeability of asparagine-based bio-grouts and results were compared with urease-based bio-grouts Analytical techniques. Calcium carbonate precipitation in samples was visualized by SEM.
Simultaneously, elemental analyses of test piece segments were carried out by using an energy dispersive X-ray spectrometer (EDS) with SEM.
Bio-grout and grout (control) samples were analyzed by XRD to identify cementing components or biominerals. Samples were gently ground in an agate mortar and then classified by size in a 0.037 mm screen aperture with ethyl alcohol [bib_ref] The use of thermal analysis in assessing the effect of temperature on..., Alarcon-Ruiz [/bib_ref]. XRD spectra were obtained using Bruker D8 diffractometer with a Cu anode (40 kV and 30 mA) and scanning from 5° to 80° 2θ . X-ray computed tomography (XCT). XCT, which is a forceful non-intrusive technique, examines the internal structure of object depending on the different X-ray absorption of the non-homogeneity materials. According to Beer-Lambert's law, when monochromatic X-rays are passed through tested material, the final intensities decreases follow this equation: where, I 0 and I are the initial and the final intensities, respectively. u corresponds the linear absorption coefficient, while d is the length of the sample under test to be measured. As the absorption can be easily deduced from equation, the collection of absorption measurements over many views allows 2D cross sectional image (CT image) to be mathematically constructed. In this study, bio-grout samples were scanned using German Compact 225 type industrial X-ray computerized tomography. The 2D-images of cross-section (respectively perpendicular to the x-axis, y-axis) in bio-grout samples were reconstructed by VGS studio MAX 2.0 software.
Scientific RepoRts | 5:16128 | DOi: 10.1038/srep16128 Thermogravimetry analysis (TGA). Thermogravimetry analysis was conducted on powdered grout samples. Prior to testing, the samples were kept in a vacuum desiccator for 24 h. The analysis was carried out in TA Instruments incorporated high-resolution thermogravimetric analyzer (series Q600) in a flowing nitrogen atmosphere (80 cm 3 min −1 ), by increasing the temperature from 40 to 1200 °C with a heating rate of 10 °C/min. It is assumed that all weight loss between 700 and 900 °C in TGA was due to calcite decarbonation [bib_ref] The global nitrogen cycle, Galloway [/bib_ref].
[fig] Figure 1: (a) pH profile and asparaginase activity, and (b) pH profile and urease activity of B. megaterium at different time interval. [/fig]
[fig] Figure 2: SEM showing imprint of rod-shaped B. megaterium precipitating carbonate in bio-grout. Scientific RepoRts | 5:16128 | DOi: 10.1038/srep16128 [/fig]
[fig] Figure 3: EDS showing high amount of calcium due to B. megaterium driven MICP in bio-grout. [/fig]
[fig] Figure 4: XRD identifying several peaks of calcite precipitated by B. megaterium in bio-grout. Scientific RepoRts | 5:16128 | DOi: 10.1038/srep16128 [/fig]
[fig] Figure 5: (a) Bio-grout as seen under XCT, and (b) its closer view from bottom. Cross-section of 2D tomography of bio-grout column from (c) top, and (d) bottom layer. [/fig]
[fig] Figure 6: TGA graphs of top, middle and bottom layers of bio-grout specimen (continuous lines represent Weight %, while dotted lines represent Deriv. Weight %/°C). [/fig]
|
Implementation of a temperature bundle improves admission hypothermia in very-low-birth-weight infants in China: a multicentre study
Background Hypothermia is a common problem that is associated with increased mortality and morbidity among preterm infants, especially in China. The objective of this study was to evaluate the efficacy of a targeted quality improvement (QI) project that applied hypothermia prevention measures for very-low-birth-weight (VLBW) infants in three tertiary neonatal intensive care units (NICUs) in China. Problem Between January 2018 and December 2018, we conducted a prospective analysis and found that the incidence of AH was 88.2% among VLBW infants. Methods The study enrolled preterm infants born at less than 32 weeks' gestation with a VLBW of less than 1500 g who were delivered at three academic tertiarycare hospitals between January 2018 and December 2019. The primary outcome measure was the incidence of hypothermia. The outcomes of the pre-QI group (1 January-31 December 2018) were compared with those of the post-QI group (1 January-31 December 2019). Interventions Based on the literature, our preliminary findings and the needs of each unit, a temperature bundle that included a transport incubator, prewarmed hats, polyethylene wrap, team training and education, and temperature documentation and workflows were implemented in consecutive plan-do-study-act cycles. Results Of the 530 VLBW infants, 235 infants (36.9%) belonged to the pre-QI group, and 295 infants (46.4%) belonged to the post-QI group. The incidence of hypothermia decreased significantly, from 92.3% to 62% (p<0.001), and the mean body temperature on admission to the NICU increased significantly, from 35.5°C to 36°C±0.7°C (p<0.001). There was one case of hyperthermia during the study period. Infants in the post-QI group had a lower mortality rate (16.1% vs 8.8%, p=0.01). Conclusions Targeted interventions can dramatically reduce admission hypothermia and improve the outcome of VLBW infants in China. Trial registration number Chi CTR 1900020861.
# Introduction problem description
On 1 January 2018, our neonatal clinical research team established the China regional neonatal collaboration network known as the Shandong Neonatal Network (SNN). The SNN covers more than 100 million people and prospectively collects neonatal data from participating units. During this time, we found that the incidence of neonatal admission hypothermia (AH) was quite high. Therefore, a retrospective analysis of VLBW infants born between 1 January and 31 December 2017 was conducted to determine key drivers of change. In 2017, the incidence of AH was 87.9% among VLBW infants in the 28 neonatal intensive care units (NICUs). [bib_ref] Hypothermia on admission in both very low and extremely low birth weight..., Wang [/bib_ref] Some projects were carried out in less than 50% of units; interventions included prewarmed hats, places in polyethylene bags (no drying), use of transport incubators and so on. The results showed that there was a negative correlation between the AH rate and the number of quality improvement (QI) measures implemented to prevent hypothermia. Between January 2018 and December 2018, we conducted a prospective analysis and found that the incidence of AH was 88.2% among VLBW infants and that AH was closely related to mortality, especially in cases of moderate-to-severe hypothermia. Preterm infants, especially very-low-birthweight (VLBW) infants, have difficulty maintaining their body temperature after birth due to immature physiological development. The incidence of hypothermia on admission to the NICU among VLBW preterm infants is 31%-78%. With improvements in the economy and intensive care technology, the survival rate of preterm infants is increasing in China, but there is still a large gap compared with high-income countries. Yun et al [bib_ref] Outcomes of very low birth weight infants at discharge: a multicentered cross-sectionaI..., Yun [/bib_ref] conducted a study that included a total of 2956 VLBW infants (gestational age <34 weeks) and reported an estimated overall mortality of 23.9%. Data collected from 25 level III NICUs in 19 provincial-level administrative regions in China showed that Open access compared with Canada, China had higher case fatality rates for each birth weight stratification among VLBW infants. 9
## Available knowledge
International thermoregulation guidelines involve multiple modalities divided into four phases: Predelivery preparation, resuscitation, transfer and NICU admission. [bib_ref] Preventing admission hypothermia in very low birth weight neonates, Fawcett [/bib_ref] [bib_ref] Delivery room management: quality improvement toolkit [monograph on the Internet, Bell [/bib_ref] [bib_ref] Part 13 ：neonatalresuscitation：2015 American heart association GuidelinesUpdate for cardiopulmonary resuscitation and EmergencyCardiovascular..., Wyckoff [/bib_ref] However, the proportion of AH remains high in clinical practice. After the initiation of a QI project, the incidence of AH significantly declines, and an increasing number of evidence-based QI measures have shown success. Caldas et al [bib_ref] Effectiveness of a measure program to prevent admission hypothermia in very low-birth..., Jpdes [/bib_ref] reported that after the initiation of a QI programme, the incidence of AH in VLBW infants decreased from 37.2% to 14.2%. Choi et al [bib_ref] The impact of a quality improvement effort in reducing admission hypothermia in..., Choi [/bib_ref] showed that after QI measures were initiated, the incidence of AH was reduced to 41%. Andrews et al [bib_ref] Quality-Improvement effort to reduce hypothermia among high-risk infants on a mother-infant unit, Andrews [/bib_ref] reported that in a QI study of hypothermia in a mother-infant unit, the incidence of hypothermia decreased by 16.5% (before the QI, it was 29.8%). A 3-year QI study showed that the incidence of AH in VLBW infants decreased from 32.4% before the introduction of QI measures to 9.6% after. [bib_ref] Quality improvement initiative to prevent admission hypothermia in very-low-birth-weight newborns, Frazer [/bib_ref] However, there are few studies on the current situation of hypothermia in preterm infants in China,and multicentre and large sample studies in particular are lacking.
## Rationale
Given these findings and international experience, we hypothesised that the implementation of a targeted QI bundle could reduce the incidence of AH and improve the outcomes of VLBW infants. On 1 January 2019, we established the Shandong Multicentre Study Coordination for Admission Hypothermia in China and initiated a QI project to prevent hypothermia in VLBW infants admitted to all participating NICUs.
## Specific aims
The aims of this study were to evaluate the change of the incidence of AH of VLBW infants born at three homogeneous academic, tertiary-care hospitals and to analyse their mortality and morbidity before and after QI. Our target was to maintain the admission temperature between 36.5°C and 37.5°C to the maximum extent possible. The specific, measurable, attainable, relevant and time-bound goal of the QI project was to decrease the incidence of hypothermia by 10% (from a baseline of approximately 90%-80%) over a period of 12 months.
# Methods
## Context
This prospective, multicentre, time-series cohort study was carried out from 1 January 2018 to 31 December 2019 in three NICUs. The three recruited NICUs belong to academic tertiary-care hospitals, namely, Eyast Branch of Provincial Hospital Affiliated to Shandong University, The First Affiliated Hospital of Shandong First Medical University and Liaocheng People's Hospital Affiliated to Shandong First Medical University; these hospitals have an average of 59 beds in the Neonatology Department and 40 beds in the NICU. All deliveries of VLBW infants are attended by a dedicated resuscitation team that includes a resuscitation nurse, a nurse practitioner, an obstetric physician and a neonatal attending doctor. The roles of the team members are well defined, and their skills and knowledge are validated by regular assessments through simulations and reports. Measures taken to keep infants warm include raising the ambient temperature, preheating the radiant warmers, drying infants immediately after birth and wrapping them in warm blankets, transferring them to a warm bed and placing them in a warm bed for transfer to the NICU.
The study population included all infants born at the hospitals with a birth weight of less than 1500 g and a gestational age less than 32 weeks who were admitted to the NICUs. Infants who were born outside the participating hospitals and had congenital anomalies, or died in the delivery room and those whose mothers had a fever during delivery (temperature ≥38°C) were excluded. These three hospitals are teaching hospitals located in Shandong, China; each hospital has an adjacent level III regional NICU with an average of 40 beds. The NICUs of the three hospitals receive an average of approximately 1623 newborns per year, of which approximately 500 are VLBW infants (30.7%). The average ratios of nurses to beds and physicians to nurses are 1:2.2 and 1:1.9, respectively.
## Interventions
The SNN was formed in January 2018 to collect neonatal data. Data on admission body temperature were prospectively collected. According to a data analysis performed in 2018, the incidence of hypothermia in our participating units was generally high, with an average of 88.2%. [bib_ref] Association between admission hypothermia and outcomes in very low birth weight infants..., Yu [/bib_ref] Our team identified the opportunity to decrease AH. A retrospective analysis of infants born between January 2017 and December 2017 was conducted to determine key drivers of change. Pareto charts were used to identify and prioritise the factors contributing to AH (figure 1). Based on these findings, keeping the infants warm and improving team training and education were identified as key drivers of change. To meet these two objectives, temperature bundle management was designed with 11 key elements. The specific elements are as follows: (1) keep warm: ambient temperature of 25°C 20 ; (2) switch on radiant warmers and set them to maximum heat output;
(3) infant is wrapped in a polyethylene wrap immediately after birth, without drying; (4) infant is quickly weighed after being placed in a prewarmed blanket; (5) a prewarmed hat is placed on the infant's head; (6) thermal mattresses are used; (7) infant is transported in a heated transport incubator; improve team training and education: (8) document temperature (10 min after birth, on arrival at the NICU, every 30 min thereafter); (9) standardise the temperature measurement; (10) provide training on and assessments of temperature measurement for nurses; and (11) monthly chart reporting on hypothermia in preterm infants on admission to the NICU. Our primary goal was to prevent AH in VLBW infants by preventing excessive heat loss in the delivery room and during the transportation process. To achieve this goal, we standardised delivery room management and NICU admission processes for all VLBW infants. Briefly, the components of the bundle were the use of radiant warmers, plastic wrap, thermal mattresses, transport incubators, an increased room temperature, and prewarmed hats and blankets. The bundle also emphasised accurate documentation of temperature at designated time points. The bundles were implemented using the plan-dostudy-act (PDSA) framework.
We set up the Shandong Multicentre Study Coordination for Admission Hypothermia. For convenience of communication, the team members exchanged contact information, such as telephone numbers, QQ numbers, email addresses, and WeChat IDs, and set up a WeChat group. The team was responsible for reviewing recommendations and initiating a preventive hypothermia programme. The team was composed of the department director, medical team leader, neonatologist, unit clinical nurse specialist and charge nurse at each centre. The department director was responsible for the research design, the discussion and development of the diagnostic criteria and the quality control plan. The medical team leader and the neonatologist were responsible for implementing the test plan, collecting data, recording temperatures and verifying the implementation of various measures. The charge nurse was responsible for temperature measurements at all time points and for quality control during the temperature measurement process. We had weekly department staff meetings and monthly online meetings. The medical team leader prepared power point presentations to report to the team on the causes of hypothermia.
## Study of the interventions
Infants born between January 2018 and December 2018 constituted the pre-QI group, and infants born between January 2019 and December 2019 were classified as the post-QI group. All data were obtained from the SNN database. We developed a standard worksheet to document temperature. The worksheet mainly contained information on the time point of the temperature measurement (10 min after birth, on arrival at the NICU or every 30 min until the temperature reached 36.5°C), the value and the measurement time. All measures for preventing hypothermia were available at the centres and did not require additional resources.
The first PDSA bundle occurred in January 2019. The implementation contents are as follows: (1) maintenance of an ambient temperature of 25°C, (2) use of the heating mode of the air conditioner with the temperature set to above 25°C, (3) placement of prewarmed hats (stockinette or woollen) on the infant's head and (4) rapid determination of the infant's weight after placement in a prewarmed blanket. The incidence of AH decreased from 92.3% kept in each NICU/baseline to 70%. The ambient temperature at birth did not reach 25°C, and the temperatures of the hats and blankets were too low in the study period. After discussion, our team decided to increase the ambient temperature and warm up the hat and blanket half an hour before birth.The PDSA cycle was adopted and continued.
The second PDSA bundle was performed between April and June 2019. After repeated training, we moved on to the second stage. All infants were wrapped immediately after birth, without being dried, in a polyethylene wrap measuring 30 cm×40 cm with a T-shirt-type opening and a border that allowed closure at the other end without drying. We chose to use a thermal mattress. Hypothermia decreased from 70.0% to 61.6%. In this research phase, the polyethylene wrap was improperly used. Also, due to material limitations, suitable thermal mattresses were out of stock for a period. We decided to conduct weekly (4) infant immediately wrapped in a polyethylene wrap without being dried; (5) infant wrapped in a prewarmed blanket wrap after the umbilical cord is cut; (6) infant quickly weighed after being placed in a prewarmed blanket; (7) placement of a prewarmed hat on the infant's head; [bib_ref] Outcomes of very low birth weight infants at discharge: a multicentered cross-sectionaI..., Yun [/bib_ref] body temperature measured and recorded 10 min after birth; (9) infant transported in a heated transport incubator; (10) all surfaces that the infant will contact are prewarmed; [bib_ref] Preventing admission hypothermia in very low birth weight neonates, Fawcett [/bib_ref] infant immediately placed in the admission bed (eg, a Giraffe OmniBed); (12) infant's body temperature measured within 1 hour after birth; (13) infant receives hypothermic rewarming;the time at which the temperature reaches ≥36.5°C is recorded; (15) both nursing and medical operations are implemented; (16) monthly chart reporting on hypothermia in preterm infants on admission to the NICU; (17) standardised temperature measurement; (18) training on temperature measurement for nurses and assessments;implementation of quality improvement projects to reduce hypothermia in preterm infants on admission to the NICU. Based on these findings, keeping the infants warm and improving team training and education were identified as key drivers of change. To meet these two objectives, temperature bundle management was designed with 11 key elements. NICU, neonatal intensive care unit.
## Open access
trainings on how to quickly wrap a baby in a polyethylene wrap. The PDSA cycle was adopted and continued.
The third PDSA bundle was performed between July and December 2019. The temperature of the transport incubator was maintained at 35°C~36°C. A debriefing checklist was created for infants with AH, and auditing with weekly feedback was performed. Control charts for hypothermia were reviewed every month. Hypothermia decreased from 61.6% to 58.6%. In this research phase, the temperature of the transport incubator temperature does not meet the standard. our team decided to open the transport incubator 1 hour before the baby was born and to maintain the temperature at 35°C~36°C.
To ensure the implementation of these measures, our team developed a manual list, an implementation list and a verification list. At the monthly online meetings, the department directors followed up on unfinished work so that medical team leaders could be encouraged to complete these measures. In addition, our team leader conducted monthly supervision and site visits to each centre to solve various problems in the QI process in a face-to-face manner. In general, communication among multiple disciplines (obstetrics, operating room/delivery room and neonatal department) was strengthened; regular training for each discipline was organised; and each medical staff member was encouraged to rigorously implement QI measures.
## Measures
The primary outcome was the monthly proportion of newborns with AH and the mean admission temperature. We calculated this proportion by taking the monthly number of VLBW infants who met the inclusion criteria and had a hypothermic event and dividing it by the monthly total number of VLBW infants who met the inclusion criteria. Admission temperatures were measured with a thermometer before the infants were placed in a NICU incubator. AH was defined as a rectal temperature of <36.5°C, measured with a digital thermometer (OMRON, MC-685-HP), when the newborn arrived in the NICU.
Balancing measures were the proportion of newborns with AH (defined as a rectal temperature of >37.5°C). Process measures included maintaining an ambient temperature of 25°C, documenting the infant's temperature (10 min after birth), wrapping the infant in polyethylene wrap without drying them first and warming the transport incubator. The minimum target for compliance with the process measures was ≥80%, with an ideal target of >90%.
Additionally, to ensure that the ambient temperature was maintained above 25°C, we used the same digital laser infrared thermometer for measurements and made adjustments once a month to avoid errors.
The secondary outcomes were the morbidities of severe intraventricular haemorrhage (IVH) (grades III and IV according to the Papile classification), necrotising enterocolitis (NEC) (Bell grades II and III), pulmonary haemorrhage, severe bronchopulmonary dysplasia (BPD) (the need for supplemental oxygen or positive pressure support at 36 weeks of postmenstrual age), retinopathy of prematurity (ROP) (grades≥III) and in-hospital mortality.
# Analysis
The statistical analyses were conducted using SPSS V.25.0. Demographic data are expressed as means (±SDs) or percentages. In the univariate analysis, we used t-tests, χ 2 tests or Fisher's exact tests. We then evaluated the ORs according to admission temperature using a multivariate logistic regression analysis, with adjustments for factors that had a p value of <0.1 in the univariate analysis. A twosided level of significance of 0.05 was used for all analyses.
The monthly rate of AH was tracked by using statistical process control (SPC) methods (QI Macros). Using SPC methods, we created p-charts to evaluate the monthly percentages of hypothermic events over the 24-month study period. When the value was ≥7, subsequent data points were below the one side of the centreline average, indicating a special-cause shift in the signal.Our writing format was based on the SQUIRE V.2.0 guidelines.
The study has been registered at the Chinese Clinical Trial Registry.
# Results
The new thermoregulation protocol was introduced into practice in January 2019. The incidence of AH decreased dramatically in the post-QI group, with a reduction from 92.3% to 62.0% (p<0.001, table 1). The mean admission temperature also significantly increased to 36.0°C±0.7°C from 35.5°C±0.7°C (p<0.001). The rate of AH was tracked with the control chart (figure 2). The control chart revealed that the centreline shifted downward during the post-QI phase. The incidence of AH decreased gradually with each test change and decreased to 58.6% after the third test change [fig_ref] Table 2: Infant characteristics and neonatal outcomes by plan-do-study-act cycle Data are presented as... [/fig_ref].
We collected process measure compliance data throughout the project and shared these data at monthly meetings to understand the key practices that led to improvements. shows a sample chart that shows the ambient temperature in relation to the number of infants per month. The ambient temperature was within the target range 73.2% of the time on average (range: 55.9%-88.6%) throughout the study period. Throughout the study period, the average compliance with documenting the patients' temperatures at 10 min after birth was 63.1% (range: 55.2%-77.8%); compliance with wrapping the infant in a polyethylene wrap without drying them first was 75.6% (range: 57.8%-89.6%); and compliance with the use of a heated transport incubator was 78.4% (range: 56.2%-90.6%).
The baseline characteristics of the infants in the intervention group were similar to those of the historical controls in 2018 [fig_ref] Table 1: Comparison of maternal and infant characteristics [/fig_ref]. There were no differences in the incidence of hypothermia or in neonatal outcomes among the PDSA cycles [fig_ref] Table 2: Infant characteristics and neonatal outcomes by plan-do-study-act cycle Data are presented as... [/fig_ref].
## Open access
There were no significant differences in other secondary outcomes, namely, severe IVH, NEC, pulmonary haemorrhage, severe BPD, LOS or ROP. However, there was a significant reduction in the incidence of in-hospital mortality in the post-QI group compared with the pre-QI group (p=0.01) [fig_ref] Table 3: Outcomes of infants in the QI intervention group compared with historical controls [/fig_ref].
A total of 636 infants with a birth weight <1500 g and a gestational age <32 weeks were enrolled in the study on their day of birth; 58 out-born infants were excluded. Ten infants with maternal fever and 27 infants with missing data were excluded, as were 11 infants whose families did not agree to participate in the study. The remaining 530 infants were included in this analysis.
# Discussion summary
This is a prospective, multicentre cohort study of the implementation of QI programmes to prevent AH in VLBW infants in China. We succeeded in reducing the incidence of AH in VLBW infants using standardised targeted temperature bundle management. The use of an ambient temperature of 25°C, a polyethylene wrap, [bib_ref] Heat loss prevention: a systematic review of occlusive skin wrap for premature..., Cramer [/bib_ref] [bib_ref] Interventions to prevent hypothermia at birth in preterm and/or low birthweight babies, Mccall [/bib_ref] [bib_ref] Vinyl bags prevent hypothermia at birth in preterm infants, Mathew [/bib_ref] [bib_ref] Temperature control of premature infants in the delivery room, Watkinson [/bib_ref] prewarmed hats, 23-25 a heated transport incubator, [bib_ref] Implementation of a multidisciplinary guideline improves preterm infant admission temperatures, Harer [/bib_ref] team training and education, temperature documentation and workflows were all relatively easy changes to implement in practice, but they dramatically improved the admission temperatures of VLBW newborns. The incidence of AH dropped from 92.3% to 62% in the post-QI group, a decrease of 30.3%.
Although warming measures are highly standardised internationally, there are differences in clinical practice and thermoregulation guidelines in China. Therefore, the incidence of hypothermia was very high in this study compared with previous studies. [bib_ref] Preventing admission hypothermia in very low birth weight neonates, Fawcett [/bib_ref] [bib_ref] Delivery room management: quality improvement toolkit [monograph on the Internet, Bell [/bib_ref] [bib_ref] Part 13 ：neonatalresuscitation：2015 American heart association GuidelinesUpdate for cardiopulmonary resuscitation and EmergencyCardiovascular..., Wyckoff [/bib_ref] [bib_ref] Effectiveness of a measure program to prevent admission hypothermia in very low-birth..., Jpdes [/bib_ref] However, the extent to which the incidence of AH decreased as a result of the intervention was consistent with previous studies. Retrospective analysis has shown that body temperature maintenance is still a very weak link in neonatal resuscitation, especially for premature infants, in China. Although the rate of process measure implementation did not meet minimum international standards (≥80%), the study showed an increase in the implementation of process measures from less than 50% 1 to approximately 73.2%. Communication delays, a lack of training and poor coordination among obstetric and paediatric medical staff led to the inadequate implementation of QI measures. In addition, a lack of domestic paediatric medical staff and a lack of attention from senior leaders
## Open access
were responsible for the inadequate implementation of some measures. Finally, the overall duration of our study was 3 years long, and this paper describes only the intervention phase. In the QI maintenance stage, we constantly improved the process based on the problems we encountered during the implementation of various measures to further reduce the incidence of AH. Because hypothermia has been shown to be associated with LOS, IVH and death, 27-31 we monitored survival without serious complications as a secondary outcome.
Concomitant with the near elimination of moderateto-severe hypothermia was the trend towards improved survival. This encouraging result shows that our intervention not only was effective in reducing the incidence of hypothermia but also had a potentially positive impact on the survival rate of VLBW infants.
However, we must also acknowledge undesired outcomes, such as hyperthermia, which occurred in one case after the intervention. One newborn who spent a prolonged time on the chemical mattress had a hyperthermic admission temperature. Such cases were rare; however, when staff obtain rectal temperature measurements every 30 min from the delivery room or operating room to NICU admission, they should use this information to adjust their practices removing the chemical mattress when temperatures exceed 37.0°C.
## Interpretation
Regarding collaborative improvement processes, Paul Batalden popularised the quote, 'Every system is perfectly designed to get the results it gets'; that is, 'if we want different results, we must change (transform) the system'.Therefore, each team must evaluate its own system at the local level to identify its own challenges and the process changes that are necessary to produce improvements. Some of the participating units have delivery rooms or operating rooms that are far from the NICU; for these units, the use of chemically heated mattresses and transfer materials was considered to further enhance the benefits of thermal care. The low incidence of AH demonstrated the success of the QI measures at all process points as a result of communication among multiple disciplines (obstetrics, operating room/delivery room and neonatal Compliance with the process measures of raising the ambient temperature to 25°C. The left y-axis and the dark grey bars represent the number of very-low-birth-weight infants who met the inclusion criteria per month. The right y-axis and the black boxes represent the percentage of those infants who experienced compliance with protocol measure of raising the ambient temperature to 25°C. A sample chart that shows the ambient temperature in relation to the number of infants per month is shown. The ambient temperature was within the target range 73.2% of the time on average (range: 55.9%-88.6%) throughout the study period.
Open access department), regular training for each discipline and the requirement for each medical staff member to rigorously implement the QI measures. To ensure the implementation of QI measures, our team developed a manual list, an implementation list and a verification list. In monthly online meetings, department directors followed up on unfinished work to encourage medical team leaders to complete the implementation of these measures. In addition, our team leader conducted monthly supervision and site visits at each centre to address various problems in the QI process in a face-to-face manner. By refining the temperature bundle management process, we could significantly reduce the incidence of hypothermia among VLBW infants. Although this study examined process indicators, further study is needed; furthermore, the current domestic measures for heat preservation in premature infants are inadequate; senior management's attention to this issue is minimal; and QI research is very limited. Therefore, the QI team from the centres that participated in the project described in this study not only can serve as a resource for other provinces but also can gain the attention of top leaders to formulate incentives to ensure that all QI measures are rigorously implemented by all medical personnel. This simple and practical bundle can be broadly expanded to the increasing number of NICUs affiliated with perinatal medical centres at all levels in China.
# Limitations
Limitations of this study include the following: first, our study was conducted for 1 year, which was followed by a 2-year QI maintenance phase in which a target of ≥80% compliance with process measures was set to promote ongoing reductions in the incidence of AH; second, although we did not identify any demographic shifts between the preintervention and postintervention periods, unmeasured differences, such as seasonal changes in room temperature, may have confounded the results. However, the potential value of promoting an approach in which the NICU leads the way forward will likely greatly outweigh the already minimal cost and difficulty of implementing simple thermoregulation strategies.
# Conclusions
We described a successful QI effort to reduce AH in VLBW infants in China. The results of this QI initiative indicate that relatively simple targeted interventions can dramatically reduce preventable hypothermic events and potentially improve survival without severe morbidity. The overall duration of our study was 3 years long, and this paper describes only the intervention phase. In the QI maintenance stage, we constantly improved the process based on the problems we encountered during the implementation of various measures to further reduce the incidence of AH.
Ethics approval This study involves human participants and was approved by the institutional review board of Shandong Provincial Hospital Affiliated to Shandong University (ethical approval number LCYJ: No. 2019-004). Informed consent was signed by the legal guardian of all participants. The participants gave informed consent to participate in the study before taking part, signed by their legal guardians.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data are available in a public, open access repository. Data are available upon reasonable request. Data may be obtained from a third party and are not publicly available. http://www.ningbx.com/index/login/login.html. Data are presented as the mean or n (%). Although the numbers were small, one infant (0.3%) in the intervention group had admission hypothermia [fig_ref] Table 1: Comparison of maternal and infant characteristics [/fig_ref]. *χ 2 test. BPD, bronchopulmonary dysplasia; IVH, intraventricular haemorrhage; LOS, late-onset neonatal sepsis; NEC, necrotising enterocolitis; NS, not significant; QI, quality improvement; ROP, retinopathy of prematurity.
[fig] Figure 1: Pareto charts were used to identify and prioritise the factors contributing to admission hypothermia. (1) Ambient temperature of 25°C and the resuscitation bed on manual control with maximum heat output; (2) prewarmed hat and blanket prepared; (3) polyethylene wrap prepared; [/fig]
[fig] Figure 2 P: -chart of the hypothermic birth rate with control limits. PDSA1: January 2019-March 2019; PDSA2: April 2019-June 2019; PDSA3: July 2019-December 2019.The incidence of admission hypothermia decreased dramatically in the postquality improvement group, with a reduction from 92.3% to 62%. CL, centreline; LCL, lower control limit; PDSA, plan-do-study-act; UCL, upper control limit. [/fig]
[table] Table 1: Comparison of maternal and infant characteristics [/table]
[table] Table 2: Infant characteristics and neonatal outcomes by plan-do-study-act cycle Data are presented as the mean or n (%). BPD, bronchopulmonary dysplasia; BW, birth weight; GA, gestational age; IVH, intraventricular haemorrhage; LOS, late-onset neonatal sepsis; NEC, necrotising enterocolitis; NS, not significant; ROP, retinopathy of prematurity. [/table]
[table] Table 3: Outcomes of infants in the QI intervention group compared with historical controls [/table]
|
Acute cardiac injury events ≤30 days after laboratory-confirmed influenza virus infection among U.S. veterans, 2010–2012
Background: Cardiac injury is a known potential complication of influenza infection. Because U.S. veterans cared for at the U.S. Department of Veterans Affairs are older and have more cardiovascular disease (CVD) risk factors than the general U.S. population, veterans are at risk for cardiac complications of influenza infection. We investigated biomarkers of cardiac injury characteristics and associated cardiac events among veterans who received cardiac biomarker testing ≤30 days after laboratory-confirmed influenza virus infection. Methods: Laboratory-confirmed influenza cases among veterans cared for at U.S. Department of Veterans Affairs' facilities for October 2010-December 2012 were identified using electronic medical records (EMRs). Influenza confirmation was based on respiratory specimen viral culture or antigen or nucleic acid detection. Acute cardiac injury (ACI) was defined as an elevated cardiac biomarker (troponin I or creatinine kinase isoenzyme MB) >99 % of the upper reference limit occurring ≤30 days after influenza specimen collection. EMRs were reviewed for demographics, CVD history and risk factors, and ACI-associated cardiac events. Results: Among 38,197 patients with influenza testing results, 4,469 (12 %) had a positive result; 600 of those patients had cardiac biomarker testing performed ≤30 days after influenza testing, and 143 (24 %) had one or more elevated cardiac biomarkers. Among these 143, median age was 73 years (range 44-98 years), and 98 (69 %) were non-Hispanic white. All patients had one or more CVD risk factors, and 98 (69 %) had a history of CVD. Eighty-six percent of ACIassociated events occurred within 3 days of influenza specimen collection date. Seventy patients (49 %) had documented or probable acute myocardial infarction, 8 (6 %) acute congestive heart failure, 6 (4 %) myocarditis, and 4 (3 %) atrial fibrillation. Eleven (8 %) had non-cardiac explanations for elevated cardiac biomarkers, and 44 (31 %) had no documented explanation. Sixty-eight (48 %) patients had received influenza vaccination during the related influenza season. Conclusion: Among veterans with laboratory-confirmed influenza infection and cardiac biomarker testing ≤30 days after influenza testing, approximately 25 % had evidence of ACI, the majority within 3 days. Approximately half were myocardial infarctions. Our findings emphasize the importance of considering ACI associated with influenza infection among patients at high risk, including this older population with prevalent CVD risk factors.
# Background
Every year in the United States, influenza results in an estimated 200,000 hospitalizations and 3,300-49,000 deaths [bib_ref] Influenza-associated hospitalizations in the United States, Thompson [/bib_ref] [bib_ref] Mortality associated with influenza and respiratory syncytial virus in the United States, Thompson [/bib_ref]. Although pneumonia is the most common complication of influenza infection, influenza is also associated with exacerbations of underlying medical conditions, including chronic cardiac diseases (e.g., congestive heart failure and atrial fibrillation) [bib_ref] The cardiovascular manifestations of influenza: a systematic review, Estabragh [/bib_ref] [bib_ref] Cardiovascular manifestations associated with influenza virus infection, Mamas [/bib_ref]. In addition, influenza infection can be associated with experiencing cardiac injury, including myocardial infarction (MI), myocarditis, and congestive heart failure [bib_ref] Short-term exposure to environmental parameters and onset of ST elevation myocardial infarction...., Caussin [/bib_ref] [bib_ref] Association of influenza virus infection and inflammatory cytokines with acute myocardial infarction, Guan [/bib_ref] [bib_ref] Influenza infection exerts prominent inflammatory and thrombotic effects on the atherosclerotic plaques..., Naghavi [/bib_ref] [bib_ref] The impact of influenza on the health and health care utilization of..., Fleming [/bib_ref] [bib_ref] The prevalence of myocarditis and skeletal muscle injury during acute viral infection..., Greaves [/bib_ref] [bib_ref] Influenza as a trigger for acute myocardial infarction or death from cardiovascular..., Warren-Gash [/bib_ref] [bib_ref] Influenza epidemics and acute respiratory disease activity are associated with a surge..., Madjid [/bib_ref]. The increased risk for cardiovascular morbidity and mortality, particularly among people aged ≥85 years, has been described in observational studies regarding winter increases in excess deaths associated with influenza epidemics [bib_ref] Excess mortality from causes other than influenza and pneumonia during influenza epidemics, Collins [/bib_ref] [bib_ref] Excess mortality from epidemic influenza, Housworth [/bib_ref] [bib_ref] The contribution of influenza to combined acute respiratory infections, hospital admissions and..., Fleming [/bib_ref] [bib_ref] A study of excess mortality during influenza epidemics in the United States,..., Barker [/bib_ref] [bib_ref] Influenza and the winter increase in mortality in the United States, 1959-1999, Reichert [/bib_ref]. Ecologic studies of the association between the timing of influenza circulation and cardiac injury have also described an increase in incidence of acute MI during influenza epidemic weeks, compared with non-epidemic weeks [bib_ref] Circulating influenza virus, climactic factors, and acute myocardial infarction: a time series..., Warren-Gash [/bib_ref] [bib_ref] The relationship between influenza outbreaks and acute ischemic heart disease in Maryland..., Lichenstein [/bib_ref]. Case-control, retrospective cohort, and randomized controlled trial investigations have also illustrated an increased risk for certain types of cardiac injury, including acute MI or hospital admission for heart failure during influenza seasons; such studies have included documented receipt of influenza vaccine and a decreased risk for vascular events with use of antivirals for influenza infection among patients with history of cardiovascular disease [bib_ref] Influenza vaccination, pneumococcal vaccination and risk of acute myocardial infarction: matched case-control..., Siriwardena [/bib_ref] [bib_ref] Risk of myocardial infarction and stroke after acute infection or vaccination, Smeeth [/bib_ref] [bib_ref] Influenza infection in secondary prevention from coronary ischaemic events in coronary artery..., Ciszewski [/bib_ref] [bib_ref] Influenza vaccination and reduction in hospitalizations for cardiac disease and stroke among..., Nichol [/bib_ref] [bib_ref] Use of oseltamivir after influenza infection is associated with reduced recurrence adverse..., Casscells [/bib_ref].
In fiscal year 2012, approximately 8.8 million U.S. veterans received treatment from the U.S. Department of Veterans Affairs (VA) Veterans Health Administration (VHA) system. VHA is the largest integrated healthcare system in the United States and provides comprehensive medical care to veterans through 151 medical centers and >1,500 associated community-based outpatient clinics, skilled nursing facilities, long-term care facilities, and other facilities. Because U.S. veterans receiving treatment from VHA are older and have more cardiovascular disease risk factors than the general U.S. population, veterans might be particularly vulnerable to cardiac complications of influenza infection. In 2010, the median age of male veterans was 64 years, compared to 36 years in the general U. S. population. Veterans also have a notable burden of cardiovascular disease and cardiovascular risk factors. reported that 80 % of veterans seeking care at a VHA facility had two or more cardiovascular risk factors (defined in that study as male sex, history of coronary heart disease (CHD), family history of CHD, decreased highdensity lipoprotein, smoking, history of cerebrovascular or peripheral vascular disease, hypertension, obesity, or diabetes) [bib_ref] Multiple risk factors may break the bank, Richlie [/bib_ref]. These characteristics of older age and frequent cardiovascular risk factors indicate veterans might be particularly vulnerable to unhealthy outcomes associated with influenza infection, including cardiac complications [bib_ref] Patterns of multimorbidity in elderly veterans, Steinman [/bib_ref] [bib_ref] Are patients at Veterans Affairs Medical Centers sicker?, Agha [/bib_ref].
The VA Office of Public Health conducts routine surveillance to detect and track influenza-like illness and influenza infection among veterans receiving VHA care [bib_ref] Effective detection of the 2009 H1N1 influenza pandemic in the U.S. Veterans..., Schirmer [/bib_ref]. Specifically, the Office of Public Health uses data from electronic medical records and the VA Electronic Surveillance System for the Early Notification of Community-Based Epidemics (ESSENCE), including International Classification of Disease, Clinical Modification, 9 th Revision (ICD-9-CM) codes and laboratory data, to conduct and report weekly influenza surveillance [bib_ref] Centers for Disease Control and Prevention. National Center for Health Statistics. International..., Department Of Health [/bib_ref]. The frequency of cardiac injury associated with influenza infection among veterans, however, is unknown. Using the robust platform of existing routine influenza surveillance within the VHA system's electronic records, we describe the frequency and characteristics of cardiac injury among veterans who had cardiac biomarker testing ≤30 days after laboratory-confirmed influenza diagnosis during the 2010-2012 influenza seasons.
# Methods
We included 154 VHA facilities with laboratory capacity for influenza testing, including reference laboratory testing, and that also had cases of laboratory-confirmed influenza infection reported during October 3, 2010-December 31, 2012, the period for which data were available. Electronic influenza laboratory results for 139 facilities were available through ESSENCE. For 15 facilities where electronic influenza laboratory data were unavailable in ESSENCE, we reviewed hospitalizations and outpatient visits in ESSENCE with an ICD-9-CM diagnosis code of influenza (ICD-9-CM 487 or 488) and performed manual chart reviews of EMRs to determine if influenza testing had been performed. All influenza testing on respiratory samples was included; influenza serum antibody testing was excluded. Laboratory-confirmed influenza infection was defined as occurring in a patient in whom influenza virus was detected in a respiratory sample on the basis of direct fluorescence, enzyme immunoassay, nucleic acid detection, polymerase chain reaction assay, or viral cultures.
To identify cases of cardiac injury, we analyzed electronic laboratory records of patients with laboratoryconfirmed influenza infection to identify veterans who had cardiac biomarker tests performed ≤30 days after a laboratory-confirmed influenza infection specimen collection. The 30-day period was chosen as a reasonable length of time to seek cardiac injury events that might be temporally linked to influenza infection [bib_ref] Influenza as a trigger for acute myocardial infarction or death from cardiovascular..., Warren-Gash [/bib_ref]. The data source for cardiac biomarker results was the VA Corporate Data Warehouse (VA CDW), a national electronic data repository derived from the VA's EMR system and administrative systems; the VA CDW is used for clinical and operational research, quality programs, and surveillance activities. A case of acute cardiac injury (ACI) was defined as an elevated cardiac biomarker (troponin I level or creatinine kinase isoenzyme, CK MB) value of >99 % of the upper reference limit and occurring ≤30 days after the date of laboratory-confirmed influenza specimen collection. This threshold for elevation was chosen as a standard reference for evidence of myocardial injury [bib_ref] Third universal definition of myocardial infarction, Thygesen [/bib_ref].
We categorized ACI-associated events by identification of documented physician diagnoses. Physician diagnoses were abstracted from EMR progress notes associated with the date of positive cardiac biomarker result, including where applicable, cardiology consult notes, and in the case of admissions, inpatient progress notes and discharge summaries associated with dates of admission during which the positive cardiac biomarker results occurred. Physician diagnoses of ACI-associated events were categorized as (1) ST segment elevation MI (STEMI); (2) non-ST segment elevation MI (NSTEMI) or probable NSTEMI (we defined NSTEMI as documentation of "NSTEMI" or "non-ST segment elevation MI" and probable NSTEMI as documentation of "supply demand mismatch" or "demand ischemia" without an explicit diagnosis of NSTEMI); (2) acute congestive heart failure; (3) myocarditis; (4) atrial fibrillation, including recurrence of previously diagnosed atrial fibrillation and excluding unchanged chronic atrial fibrillation; (5) other cardiac event, including hypertensive emergency; (6) non-cardiac diagnoses, including chronic renal failure; and (7) no documented physician explanation for the cardiac biomarker result.
From EMR progress and admission notes, we abstracted the following variables for all patients at the time of laboratory-confirmed influenza specimen collection: (1) demographics; (2) cardiovascular disease risk factors, defined as physician diagnoses of hypertension, diabetes, hyperlipidemia, pharmacologic treatment for hypertension, hyperlipidemia, or diabetes at time of influenza test, any history of tobacco use, or obesity (body mass index ≥30) (male sex was not included as a cardiovascular disease risk factor); (3) history of cardiovascular disease, defined as physician diagnosis of cerebral or peripheral vascular disease, coronary heart disease, MI, atrial fibrillation, or congestive heart failure; (4) acute comorbidities at time of influenza specimen collection, defined as acute non-cardiac presenting diagnoses aside from influenza documented by a physician (e.g., pneumonia, exacerbation of chronic obstructive lung disease, or acute renal failure); (5) severity of influenza-related illness as described by hospital admission, length of stay, and need for intensive care unit admission, including mechanical ventilation; (6) vital status at 30 days after influenza laboratory test, with date of death collected in applicable cases and (7) non-steroidal anti-inflammatory and aspirin use at time of influenza test. Date of influenza symptom onset (non-specific dates (e.g., notations of "a few days" or "several days" were excluded from the analysis), and the longest given day count was used when a range was given (e.g., 4 days would be recorded if "3-4 days" was documented). Cause of death was not routinely available and therefore was not abstracted. Electrocardiogram (EKG) reports associated with the date of positive cardiac biomarker result were reviewed to identify physician-reported changes compared with baseline. These were abstracted from final EKG reports, not from physician initial interpretations documented in progress notes, admission, or discharge notes. We also abstracted influenza vaccination history for the recommended influenza vaccination period corresponding to influenza diagnosis. We defined recommended influenza vaccination period as August 1 of the year of influenza diagnosis (the earliest date at which seasonal influenza vaccines are available within the VHA system) until 2 weeks before influenza test; we excluded vaccination if it had occurred <2 weeks before influenza diagnosis, assuming the patient's immune response to the vaccination was incomplete. Documented receipt of vaccination at outside facilities in the EMR was included in vaccination history. Outside facility records were not available for review. Finally, we abstracted prescription details regarding initiation of antiviral treatment after positive influenza test. Data on antiviral therapy received at outside facilities was not available.
We calculated the time from influenza diagnosis to cardiac injury by using the interval between date of specimen collection of first positive influenza laboratory result and the date of first positive cardiac biomarker result. For patients with multiple elevated cardiac enzyme results or multiple positive influenza results, the first date with a positive cardiac enzyme result or positive influenza result was considered the reference date for time interval calculations. We calculated time from influenza test specimen collection to initiation of antiviral treatment for all patients who received antiviral therapy and time from influenza symptom onset to antiviral treatment for patients with a specified symptom onset date. Patient age was recorded on the basis of the date of influenza specimen collection. This study was reviewed for human subjects protection by the Centers for Disease Control and Prevention and determined to be non-research.
# Results
During October 2010-December 2012 (the period for which data was available), a total of 38,197 patients had influenza laboratory tests performed and captured in ES-SENCE. Among those, 4,278 (11 %) had laboratoryconfirmed influenza virus infection. Among the 15 facilities that do not have electronic laboratory reporting to ESSENCE, an additional 191 patients with laboratoryconfirmed influenza virus infection were identified through manual chart review, for a total of 4,469 (12 %) patients with laboratory-confirmed influenza virus infection. Six hundred (14 %) veterans with laboratoryconfirmed influenza virus infection also had a cardiac biomarker test performed ≤30 days after influenza specimen collection, of whom 143 (24 %) had a troponin or CK-MB result >99 % of the upper reference limit for their test (different facilities used different tests with difference reference ranges). In this descriptive study focusing on patients with both influenza and elevated cardiac biomarkers, we limited our analysis to these 143 patients and did not review data from the 457 patients with laboratory-confirmed influenza diagnosis and normal cardiac biomarker results.
## Demographic characteristics
Among 143 patients with laboratory-confirmed influenza virus infection and a cardiac biomarker result >99 % of the upper reference limit, the median age was 73 years (range 44-98 years); 98 (69 %) were non-Hispanic white; 32 (22 %) were non-Hispanic black and 2 (1 %) were female [fig_ref] Table 1: Clinical and demographic characteristics of U [/fig_ref]. All patients had at least one risk factor for cardiovascular disease. One hundred three (72 %) had two or more risk factors and 28 (20 %) had three or more risk factors. Ninety-eight patients (69 %) had documented history of cardiovascular disease [fig_ref] Table 1: Clinical and demographic characteristics of U [/fig_ref].
## Clinical characteristics
Among the 105 (73 %) patients with influenza symptom onset date recorded, the median time from onset to influenza specimen collection was 2 days (range 0-9 days), with 85 (81 %) patients having influenza specimens collected ≤3 days after symptom onset. Among the 105 patients with symptom onset date recorded, 59 (56 %) had positive cardiac biomarker testing ≤3 days after influenza symptom onset. The median time from influenza symptom onset to ACI was 3 days (range 0-25 days).
Among the total 143 patients with laboratory-confirmed influenza virus infection and ACI, the median time from influenza specimen collection to ACI was 1 day (range 0-24 days), and 123 (86 %) occurred ≤3 days after influenza specimen collection [fig_ref] Figure 1: Number of days from influenza specimen collection to acute cardiac injury among... [/fig_ref]. No STEMI events were documented. Thirty-six (25 %) patients had an NSTEMI documented in the medical chart ≤30 days after influenza specimen collection. In addition, 34 (24 %) were categorized as probable NSTEMI. Eighteen (13 %) patients had documented cardiac diagnoses (e.g., congestive heart failure or new-onset atrial fibrillation) ( Among the total 143 patients with influenza virus infection and ACI, 89 (62 %) had physician documentation of having one or more additional acute non-cardiac presenting comorbidities. Among these 89 patients, the three most common acute non-cardiac presenting comorbid diagnoses were bacterial pneumonia (69 %), chronic obstruction pulmonary disease (COPD) exacerbation (40 %), and acute renal failure (35 %); 35 had two or more of these acute non-cardiac comorbid diagnoses. Among the 70 patients with probable or documented
## Clinical outcomes
Among the total 143 patients with influenza virus infection and ACI, 134 (94 %) were admitted to the hospital ≤30 days after laboratory-confirmed influenza virus infection specimen collection. Among the 134 patients admitted, the median length of stay was 6 days (range 1-106); 44 (33 %) were admitted to an intensive care unit. Among those 44 patients, 25 (57 %) required mechanical ventilation. The median length of intensive care unit (ICU) stay was 6 days (range 2-30). Eighteen (13 %) of 143 patients died of any cause ≤30 days after laboratory-confirmed influenza virus infection specimen collection. Twelve of 18 (67 %) died ≤14 days after laboratory-confirmed influenza virus infection specimen collection [fig_ref] Figure 2: Time to death after positive cardiac biomarker among U [/fig_ref]. Eleven (61 %) of those who died received a diagnosis of NSTEMI or probable NSTEMI ≤30 days after laboratory-confirmed influenza virus specimen collection.
## Vaccination and antiviral treatment
Sixty-eight of 143 (48 %) patients with influenza virus infection and ACI had documented receipt of influenza vaccination during the season of confirmed influenza virus infection and ≥2 weeks before the date of influenza specimen collection. Among 134 patients who were admitted, 71 (53 %) were unvaccinated. Of the 18 who died ≤30 days after laboratory-confirmed influenza infection, 9 (50 %) were unvaccinated. One hundred-eighteen of 143 (83 %) patients with influenza virus infection and ACI received influenza antiviral agents; all received oseltamivir. Among 105 patients with influenza symptom onset data available, median time from influenza symptom onset to initiation of influenza antiviral treatment was 2 days (range 0-18 days). Among 118 who had received influenza antiviral agents, 114 (97 %) received treatment ≤2 days after positive influenza specimen collection. Among 18 patients who died, 17 had received influenza antiviral therapy. Among ten deaths for which exact symptom onset date was available, five had received influenza antiviral treatment ≤2 days after symptom onset.
# Discussion
Overall, among veterans cared for at the US Veterans Health Administration diagnosed with influenza virus infection who also had cardiac biomarker testing performed ≤30 days after laboratory-confirmed influenza infection specimen collection, approximately 25 % had evidence of ACI. Among patients with influenza virus infection and ACI, >80 % of ACI was diagnosed ≤3 days after influenza specimen collection. Approximately half of ACI-associated events were probable or documented Our case-series study is unique in the inclusion of both laboratory-confirmed influenza cases and laboratory-confirmed cardiac injury findings. Prior studies investigating the association between influenza infection and cardiac injury have typically used diagnosis codes or clinical syndrome coding for both definitions [bib_ref] Influenza as a trigger for acute myocardial infarction or death from cardiovascular..., Warren-Gash [/bib_ref] [bib_ref] Risk of myocardial infarction and stroke after acute infection or vaccination, Smeeth [/bib_ref]. Previous studies have confirmed MI clinical diagnosis by using cardiac biomarkers, but none identified cardiac injury of all patients with positive biomarkers. Similarly, case-control studies have compared patients with cardiac injury, most often MI, with population-based control subjects to understand the relative effect of non-specific acute respiratory infection (ARI) on case status [bib_ref] Risk of myocardial infarction and stroke after acute infection or vaccination, Smeeth [/bib_ref] [bib_ref] The possible role of infections in acute myocardial infarction, Meier [/bib_ref] [bib_ref] Recent respiratory infection and risk of cardiovascular disease: case-control study through a..., Clayton [/bib_ref] [bib_ref] Association of acute respiratory symptoms with onset of acute myocardial infarction: prospective..., Spodick [/bib_ref]. These investigations used less-specific exposure definitions based on symptoms and physician diagnosis taken from administrative data (e.g., diagnosis codes), compared with our use of laboratory-confirmed influenza diagnosis, and report the odds ratios of MI associated with ARI ranging from 2.1 (95 % confidence interval (CI) 1.4-3.2) [34] to 3.6 (95 % CI 2.2-5.7) [bib_ref] Risk of myocardial infarction and stroke after acute infection or vaccination, Smeeth [/bib_ref]. Because we identified a wider range of cardiac injury, compared with previous studies investigating the association between influenza infection and cardiac disease using diagnosis codes or clinical syndromes focused primarily on acute MI, our study provides a unique perspective on the role of influenza in cardiovascular disease. Although the diversity of cardiac manifestations associated with influenza infection have been described previously [bib_ref] The cardiovascular manifestations of influenza: a systematic review, Estabragh [/bib_ref] , our review contributes detailed diagnosis, morbidity, and outcome data among a population at risk for both complicated influenza infection and cardiovascular disease.
The fact that approximately one in seven veterans with laboratory-confirmed influenza had cardiac biomarker testing ≤30 days after their infection, and of those tested, approximately 25 % had evidence of ACI, suggests that there may be a link between acute influenza infection and risk for cardiac injury events.
Interestingly, no STEMI events were identified, whereas approximately half were either probable or documented NSTEMI. These NSTEMI events frequently represented MI type 2 presentations, reflecting myocardial injury with necrosis secondary to an imbalance between myocardial supply and demand among critically ill patients [bib_ref] Cardiac troponin T in chest pain unit patients without ischemic electrocardiographic changes:..., De Filippi [/bib_ref]. These patients are still at high risk for unhealthy outcomes because even low-level increases in cardiac biomarkers without EKG changes have documented higher risk for death or MI [bib_ref] Prospective evaluation of the prognostic implications of improved assay performance with a..., Bonaca [/bib_ref] [bib_ref] Cardiac troponin T in chest pain unit patients without ischemic electrocardiographic changes:..., De Filippi [/bib_ref]. Cases of subclinical cardiac injury also might have been missed because of normal cardiac biomarkers. One recent investigation reported that, among 39 patients admitted with severe influenza A (H1N1) who received echocardiography during October 2009-March 2011, 28 (72 %) had abnormal ventricular function on echocardiography; troponin levels were low (mean troponin 0.03 ng/ml (range 0.01-0.3)), with only one markedly elevated in the setting of an arrhythmia, indicating that cardiac dysfunction associated with influenza infection might not always include elevation in cardiac biomarkers [bib_ref] Myocardial dysfunction during H1N1 influenza infection, Fagnoul [/bib_ref]. Approximately half of the patients with probable or documented NSTEMI in our analysis were being treated with aspirin or statins at the time of laboratory-confirmed influenza diagnosis, which might have mitigated their risk for acute thrombotic events [bib_ref] Aspirin in the treatment and prevention of cardiovascular disease: past and current..., Hennekens [/bib_ref] [bib_ref] Effects of statins on inflammation in patients with acute and chronic coronary..., Kinlay [/bib_ref].
We determined that ACI was frequently close to the time of influenza symptom onset and influenza testing. This short time interval between influenza onset and cardiac injury is consistent with the timing of cardiac injury reported in previous investigations that used less-specific markers for influenza infection (e.g., acute respiratory infection ICD-9-CM coding) and looked specifically at MI [bib_ref] Recent respiratory infection and risk of cardiovascular disease: case-control study through a..., Clayton [/bib_ref] [bib_ref] Association of acute respiratory symptoms with onset of acute myocardial infarction: prospective..., Spodick [/bib_ref]. Two self-controlled case-series and one casecontrol study described the highest risk for acute MI during the first 3-5 days immediately after respiratory infection [bib_ref] Recent respiratory infection and risk of cardiovascular disease: case-control study through a..., Clayton [/bib_ref] [bib_ref] Association of acute respiratory symptoms with onset of acute myocardial infarction: prospective..., Spodick [/bib_ref]. These investigations used the date of clinic diagnosis for the exposure variable definition, perhaps a less-specific date than date of symptom onset that we were able to identify from chart review for a subset of patients. Similar to previous investigations, our description of ACI at time of influenza symptom onset and diagnosis indicates that, among older patients with cardiovascular risk factors, clinicians should consider the acute period of influenza symptom onset and diagnosis as a potentially high-risk period for cardiac complications. Additionally, physicians might consider optimizing cardiac risk factor mitigation (e.g., pharmacologic and behavior management) for such patients.
Although the majority of patients with influenza virus infection and ACI, including in this analysis, received influenza antiviral agents, approximately half did not receive them within the recommended 2-day period after influenza symptom onset. We did not have access to date of clinic presentation for this analysis, and possibly, these patients presented for care later than those who were treated ≤2 days after influenza symptom onset and may not have been offered treatment. The Advisory Committee on Immunization Practices (ACIP) recommends that influenza antiviral treatment be offered to any patient with confirmed or suspected influenza who is hospitalized, has severe or complicated illness, or is at higher risk for influenza complications, including adults aged ≥65 years and those with chronic underlying conditions, including cardiovascular disease. ACIP also recommends routine influenza vaccination for all persons aged ≥6 months that do not have contraindications to vaccination. Observational studies have reported that treatment ≤5 days after influenza symptom onset among severely ill populations has clinical benefits that include reduced mortality and duration of symptoms and complications from influenza. A meta-analysis conducted by Hsu et al., which reviewed >1,600 hospitalizations and 150,000 patients, supported early use of antivirals ≤48 h after symptom onset and confirmed mortality, hospitalization length, intensive care unit admission, and respiratory failure reduction with antiviral use ≤5 days among such severely ill hospitalized patients as those identified in our analysis [bib_ref] Antivirals for treatment of influenza, a systemic review and meta-analysis of observational..., Hsu [/bib_ref]. No investigations specifically addressing the role of antivirals and ACI were identified in our literature review.
Although certain vaccinations might have been administered at outside facilities and not recorded in the VA medical record system, given the goal of 90 % influenza vaccination rate for veterans cared for by the VA, our finding of a <50 % vaccination rate among affected patients illustrates a noteworthy opportunity for improvement in preventative services.
A potential limitation of our study is that certain cardiac events might have been missed because of differences in cardiac biomarker testing protocols or atypical clinical presentations where testing was not performed, as well as the possibility that veterans sought care at non-VA hospitals with no corresponding cardiac biomarker results captured in VA data. In addition, cardiacbiomarkers typically reflect ischemic injury and as such, non-ischemic ACI events such as acute congestive heart failure and atrial fibrillation may have been underestimated. By starting with positive cardiac enzymes to identify patients for review, we captured cardiac events that would have been missed by looking at ICD-9 codes alone, as the majority of previously published studies have done. In particular, probable NSTEMI cases, which represent substantial cardiac events, would have been missed if examining diagnosis codes alone because these cases did not have a specific acute cardiac injury type documented in the medical record. Because veterans can receive care at outside facilities and not all this information is recorded in the VA's EMR system, we also were likely missing vaccination events, influenza infection, and ACI events. As we only examined veterans who were tested for influenza, we likely selected for a sicker group who might have other characteristics different from the general population or other veterans. Similarly, because of time and resource limitations in this descriptive study designed for timely feedback for public health use by VHA, we did not review the 457 cases with laboratory-confirmed influenza infection and normal cardiac biomarker testing for comparison. Nevertheless, we believe that our analysis, based on laboratory diagnosis of both influenza and ACI, provides a unique description of ACI-associated events after influenza infection.
# Conclusion
Acute cardiac injury, most commonly occurring during the first week of influenza virus infection and diagnosis, was identified in approximately 25 % of patients with laboratory-confirmed influenza who received cardiac biomarker testing. These events were associated with a high prevalence of cardiac history and risk factors, hospitalization, and a >10 % all-cause mortality rate at 30 days. Influenza antiviral treatment and vaccination were not uniformly received by this population at high risk and are notable missed public health opportunities. Given the high prevalence of cardiac comorbidities and influenza risk among older U.S. veterans, this analysis highlights the importance of (1) considering concurrent cardiac events when diagnosing influenza; (2) encouraging yearly influenza vaccination among older U.S. veterans; and (3) aggressively identifying and modifying possible cardiovascular risk factors when diagnosing influenza, ensuring prompt treatment with influenza antiviral medication.
[fig] Figure 1: Number of days from influenza specimen collection to acute cardiac injury among U.S. veterans, October 2010-December 2012 (n = 143) [/fig]
[fig] Figure 2: Time to death after positive cardiac biomarker among U.S. veterans, October 2010-December 2012 (n = 18) [/fig]
[table] Table 2: Eleven(8 %) were categorized as non-cardiac diagnosis, of which four where attributed to chronic renal disease. Forty-four (31 %) were categorized as no documented physician explanation for the cardiac biomarker result. Of these 44 patients, 22 (50 %) had repeat cardiac biomarkers performed ≤24 h after initial positive cardiac biomarker result. Among those 22, a total of 16 had decreased to a ≤99 % threshold for test positivity and six remained elevated. Among the 127 (89 %) who had finalized EKG interpretation documented at time of positive cardiac biomarker result, 32 (25 %) had any EKG change, compared with prior EKG interpretations. [/table]
[table] Table 1: Clinical and demographic characteristics of U.S. veterans with acute cardiac injury ≤30 days after laboratory-confirmed influenza virus infection, October 2010-December 2012 (n = 143) [/table]
|
Sample-size determination for the Bayesian t test and Welch’s test using the approximate adjusted fractional Bayes factor
When two independent means μ 1 and μ 2 are compared, H 0 : μ 1 = μ 2 , H 1 : μ 1 = μ 2 , and H 2 : μ 1 > μ 2 are the hypotheses of interest. This paper introduces the R package SSDbain, which can be used to determine the sample size needed to evaluate these hypotheses using the approximate adjusted fractional Bayes factor (AAFBF) implemented in the R package bain. Both the Bayesian t test and the Bayesian Welch's test are available in this R package. The sample size required will be calculated such that the probability that the Bayes factor is larger than a threshold value is at least η if either the null or alternative hypothesis is true. Using the R package SSDbain and/or the tables provided in this paper, psychological researchers can easily determine the required sample size for their experiments.
# Introduction
In the Neyman-Pearson approach to hypothesis testinga null and an alternative hypothesis are compared. Suppose the population means of males and females are denoted by μ 1 and μ 2 . Three hypotheses are relevant: the null hypothesis H 0 : μ 1 = μ 2 , the two-sided alternative hypothesis H 1 : μ 1 = μ 2 , and the one-sided alternative hypothesis H 2 : μ 1 > μ 2 . The null hypothesis H 0
The first author is supported by the China Scholarship Council. The second author is supported by a fellowship from the Netherlands Institute for Advanced Study in the Humanities and Social Sciences is rejected if the observed absolute t-statistic falls inside the critical region, where the critical region is a set of values that are equal to or greater than the critical value t 1−α/2,v , where α is the type I error rate, and v is the degree of freedom for a two-sided alternative hypothesis. The null hypothesis H 0 is rejected if the observed t-statistic falls inside the critical region, where the critical region is a set of values that are equal to or greater than the critical value t 1−α,v for a one-sided alternative hypothesis. Statistical power is the probability of finding an effect when it exists in the population, that is, the probability of rejecting the null hypothesis when the alternative is true. Power analysis for Neyman-Pearson hypothesis testing has been studied for more than 50 years.played a pioneering role in the development of effect sizes and power analysis, and he provided mathematical equations for the relation between effect size, sample size, type I error rate and power. For example, if one aims for a power of 80%, the minimum sample size per group should be 394, 64 and 26 for small (d = 0.2), medium (d = 0.5) and large (d = 0.8) effect sizes, respectively for an independent samples twosided t test at type I error rate α = .05, where Cohen's d is the standardized difference between two means. To perform statistical power analyses for various tests, the G * Power program was developed byandand. Despite the availability of G * Power there is still a lot of underpowered research in the behavioral and social sciences, even though criticism with respect to insufficient power is steadily increasing.
Numerous articles have criticized the Neyman-Pearson approach to hypothesis testing in the classical framework (e.g.,,,,, and. As an alternative,andintroduced the Bayes factor (BF). BF quantifies the relative support in the data for one hypothesis against another, and in addition to that, cannot only provide evidence in favor of the alternative hypothesis, but also provides evidence in favor of the null hypothesis. This approach for Bayesian hypothesis evaluation is increasingly receiving attention from psychological researchers, see for exampleand, and. Nevertheless, researchers, especially psychologists, find it difficult to calculate BF and several software packages for Bayesian hypothesis evaluation have been developed. The most important are the R package BayesFactor, that can be found at http://bayesfactorpcl. r-forge.r-project.org/ and the R package bainthat can be found at https://informative-hypotheses. sites.uu.nl/software/bain/. The latter is the successor of the stand-alone software BIEMSthat can be found at https://informative-hypotheses.sites. uu.nl/software/biems/. Both BayesFactor and bain are implemented in JASP (https://jasp-stats.org/). The main difference between approximate adjusted fractional Bayes factor (AAFBF) implemented in bain and the Jeffreys-Zellner-Siow Bayes factor implemented in BayesFactor is the choice of the prior distribution. We focus on the AAFBF (to be elaborated in the next section) in this manuscript because it is available for both the t test and the Welch's test.
When two independent group means are compared, there exist two specific cases in which variances are either equal or unequal for the two groups, which correspond to t test or Welch's test. The t test is well known, while Welch's test is often extremely important and useful as demonstrated byand, and. In the Neyman-Pearson approach to hypothesis testing, the formulae for calculating the sample size are given by an a priori power analysis for t test and Welch's test. There is not yet a solid body of literature regarding sample-size determination (SSD) for Bayesian hypothesis evaluation, butand De Santis (2007) give different sample-size determination approaches for testing one mean of the normal distribution with known variance.anddiscuss parameter estimation and use the posterior distribution as a measure of evidence strength, andandintroduce Bayes factor design analysis applied to fixed-N and sequential designs. This paper will elaborate on these approaches in the following manners. (1) in addition to the Bayesian t test the Bayesian Welch's test also will be considered. In practice, Welch's test is more widely used, which is a necessary improvement in this manuscript; (2) both two-sided and one-sided alternative hypotheses are considered. One-sided alternative hypothesis can effectively reduce the required sample size and it is recommended to be used. This manuscript will provide a comprehensive analysis for both two-sided and one-sided alternative hypotheses; (3) the sample size will be calculated such that the probability that the Bayes factor is larger than a user specified threshold is at least η if either the null hypothesis or the alternative hypothesis is true; (4) we use the dichotomy method to compute the sample size very fast. In the previous publication, the sample size is computed through progressively increase the sample size with one until the threshold value is reached. This method is simple and easily used but with high computation effort, especially for the case when the required sample size is large, e.g., the sample size of 500 will cause several hundreds of iterations, while only 12 iterations are required with our method; (5) the sensitivity of SSD with respect to the specification of the prior will be highlighted. This is very important when Bayes factor is used for the hypothesis testing evaluation, because there exists some uncertainty for the required sample size for different prior distributions.
The outline of this paper is as follows. First, we give a brief introduction of the AAFBF, show how it can be computed, discuss the specification of the prior distribution and sensitivity analyses. Subsequently, sample-size determination is introduced. Thereafter, we will discuss the role of samplesize determination in Bayesian inference. The paper continues with an introduction of the ingredients required for sample-size determination. Then, the algorithm used to determine the sample size will be elaborated. Next, features of SSD are described. Thereafter, three examples are presented that will help psychological researchers to use the R package SSDbain if they plan to compare two independent means using the t test or the Welch's test. The paper ends with a short conclusion.
## Bayes factor
In this paper, the means of two groups, μ 1 and μ 2 , are compared for both Model 1: the within-group variances for group 1 and 2 are equal,
[formula] y p = μ 1 D 1p + μ 2 D 2p + p with p ∼ N(0, σ 2 ),(1) [/formula]
and Model 2: the within-group variances for group 1 and 2 are not equal,
[formula] y p = μ 1 D 1p + μ 2 D 2p + p with p ∼ N(0, D 1p σ 2 1 + D 2p σ 2 2 ),(2) [/formula]
where D 1p = 1 for person p = 1, · · · , N and 0 otherwise, D 2p = 1 for person p = N + 1, · · · , 2N and 0 otherwise, N denotes the common sample size for group 1 and 2, p denotes the error in prediction, σ 2 denotes the common within-group variance for group 1 and 2, and σ 2 1 and σ 2 2 denote the different within-group variances for group 1 and 2, respectively. In this paper, the AAFBFis used to test hypotheses: H 0 : μ 1 = μ 2 against H 1 : μ 1 = μ 2 1 or against H 2 : μ 1 > μ 2 . The Bayes factor (BF) quantifies the relative support in the data for a pair of competing hypotheses. Specifically, if BF 01 = 5, the support in the data is five times stronger for H 0 than for H 1 ; if BF 01 = 0.2, the support in the data is five times stronger for H 1 than for H 0 . As was shown inthe BF in terms of comparing the constrained hypothesis H i (i = 0, 2) with the hypothesis H 1 can be expressed in a simple form:
[formula] BF i1 = f i c i ,(3) [/formula]
where c i denotes the complexity of the hypothesis H i , and f i denotes the fit of the hypothesis H i . The complexity c i (a hypothesis with smaller complexity provides more precise predictions) of H i describes how specific H i is, and the corresponding fit f i (the higher the fit the more a hypothesis is supported by the data) describes how well the data support H i . The formulae of the fit and complexity are:
[formula] f i = μ∈H i g 1 (μ | y, D 1 , D 2 )dμ,(4)c i = μ∈H i h 1 (μ | y, D 1 , D 2 )dμ,(5) [/formula]
where g 1 (μ | y, D 1 , D 2 ) denotes the posterior distribution, and h 1 (μ | y, D 1 , D 2 ) the prior distribution of μ under H 1 . In case of H 2 , f 2 and c 2 are the proportions of the posterior distribution g 1 (·) and prior distribution h 1 (·) in agreement with H 2 , respectively; in case of H 1 Eq. 3 reduces to the Savage-Dickey density ratio. The BF for H 0 against H 2 is:
[formula] BF 02 = BF 01 BF 21 = f 0 c 0 f 2 c 2 . ( 6 ) g 1 (μ | y, D 1 , D 2 ) = N μ 1 μ 2 , σ 2 /N 0 0σ 2 /N ,(7) [/formula]
when Model 1 is considered; and
[formula] g 1 (μ | y, D 1 , D 2 ) = N μ 1 μ 2 , σ 2 1 /N 0 0σ 2 2 /N ,(8) [/formula]
when Model 2 is considered, whereμ 1 andμ 2 denote the maximum likelihood estimates of the means of group 1 and group 2, respectively, andσ 2 ,σ 2 1 andσ 2 2 denote unbiased estimates of the within-group variances. Due to the normal approximation, the general form of the AAFBF can be used to evaluate hypothesis evaluation in a wide range of statistical models such as structural equation modeling, logistic regression, multivariate regression, AN(C)OVA, etc. Therefore, it is currently the most versatile method for Bayesian hypotheses evaluation.
The prior distribution is based on the fractional Bayes factor approach. It is constructed using a fraction of information in the data. As elaborated inandthe prior distribution is given by:
[formula] h 1 (μ | y, D 1 , D 2 ) = N 0 0 , 1 bσ 2 N 0 0 1 bσ 2 N ,(9) [/formula]
where b is the fraction of information in the data used to specify the prior distribution, when Model 1 is considered, and
[formula] h 1 (μ | y, D 1 , D 2 ) = N 0 0 , 1 bσ 2 1 N 0 0 1 bσ 2 2 N ,(10) [/formula]
when Model 2 is considered. The prior distribution is NOT used to represent the prior knowledge about the effect size under H 1 or H 2 . The prior distribution is chosen such that a default Bayesian hypothesis evaluation of H 0 vs H i is obtained, that is, subjective input from the researcher is not needed. This is an advantage of default Bayesian hypothesis evaluation , that is, turn a noninformative prior into a proper prior using a small proportion of the information in the data. For our situation this is equivalent to using one half observation from group 1 and one half observation from group 2 is used, which is in total one observation. This makes sense because the focus is on one contrast, that is, μ 1 − μ 2 , which means that one parameter needs to be estimated. This choice is too some extend arbitrary, for example, we could also use 2b (one person is needed to estimate each mean) or 3b (one person for each mean and the half for the residual variance), which still maintains the spirit of the minimal training sample approach. In summary, the goal is to compare H 0 with H i (i = 1, 2) by means of Bayes factor, but not comparing the prior distribution of H 0 with H i (i = 1, 2) through the Bayes factor. To achieve this, the prior distributions are calibrated such that H 0 and H i can be evaluated without requiring user input. However, there is some uncertainty in the calibrating, hence the AAFBF can be computed using the fractions b, 2b, and 3b, and the required sample sizes can be computed accordingly.
As an illustration, Tables 1 and 2 list the BF for the comparison of H 0 with the two-sided alternative H 1 and the one-sided alternative H 2 , respectively, when equal withingroups variances are considered (Model 1). From, we can see that when H 0 is true (e.g., the entry with b), the support in the observed data is 13 times larger for H 0 than for H 1 ; when H 1 is true, the support in the observed data is 22 (1/0.045) times larger for H 1 than for H 0 .shows that the data were nearly 18 times more likely to support H 0 when H 0 is true; the support in the data is more than 45 (1/0.022) times more likely to support H 2 when H 2 is true. Therefore, for the same sample size per group, it is much easier to get strong evidence for the one-sided than for the two-sided hypothesis (e.g., compare the corresponding shaded areas of the columns BF 01 in, BF 20 =1/BF 02 is larger than BF 10 =1/BF 01 ). The fit is higher for the true hypothesis (e.g., see column f 0 in, f 0 = 2.816 when H 0 is true is larger than f 0 = 0.009 when H 1 is true). As can be seen in Tables 1 and 2 (bottom two panels) the BF is sensitive to the choice of the fraction. The complexity c 0 becomes larger for H 0 if the fraction increases (from 0.209 to 0.295, then to 0.362), while the complexity c 2 is not affected by the fraction for H 2 (0.5 for any value of fraction). This is because the complexity of a hypothesis specified using only inequality constraints is independent of the fraction, seefor a proof. The corresponding BF for H 0 becomes smaller (e.g., in the column BF 01 , BF decreases from 13.49 to 9.54, then to 7.79), and the BF for H 2 does not change.
## Criteria for sample-size determination
For the Neyman-Pearson approach to hypothesis testing power analysis renders an indication of the sample sizes needed to reject the null-hypothesis with a pre-specified probability if it is not true. If the sample sizes are sufficiently large, under-powered studies can be avoided. A power analysis is conducted prior to a research study, and can be executed if three ingredients, type I error rate, type II error rate, and effect size are given. The main difficulty is getting an a priori educated guess of the true effect size. In practice, often one of two approaches to choose the effect size is used: use an estimate of the effect size based on similar studies in the literature, experts' opinion or a pilot study; or, use the smallest effect size that is considered to be relevantly different from zero for the study at hand. If the chosen effect size is smaller than the unknown true effect size, the sample sizes will be larger than necessary, which can be costly or unethical, and if the chosen effect size is larger than the unknown true effect size, the sample sizes will be too small and the resulting study will be underpowered. y 1 andȳ 2 are the sample means of the two groups, s 2 is the sample variance of the two groups, N is the sample size per group When the Bayes factor is used for hypothesis testing, sample-size determination instead of power analysis is used although the goals are similar. The main ingredients for SSD in a Bayesian framework are explained in . Panel (a) on the left: t test, sample size N = 26 per group, distribution of BF 01 when data are repeatedly sampled from a population in which H 0 : μ 1 = μ 2 is true. Panel (b) on the right: t test, sample size N = 104 per group, distribution of BF 10 when data are repeatedly sampled from a population in which μ 1 = μ 2 , but with the addition that the effect size has to be chosen (here we use effect size d = 0.5 to simulate data). We face the same problem as for power analysis, namely an unknown true effect size, but as will be elaborated in the next section, the combination of SSD and Bayesian updating can be used to address this problem.
Sample size will be determined such that P (BF 01 > BF thresh |H 0 ) ≥ η and P (BF 10 > BF thresh |H 1 ) ≥ η, that is, the probability that BF 01 is larger than a user specified threshold value if H 0 is true should be at least η, and the probability that BF 10 is larger than the threshold value if H 1 is true should be at least η. This is in line with power analysis in Neyman-Pearson approach to hypothesis testing in which the type I error rate α and type II error rate β are given beforehand. In the Bayesian framework, instead of type I error rate and type II error rates, we use the probability that the Bayes factor is larger than BF thresh under the null hypothesis and under the alternative hypothesis. With respect to the choice of BF thresh , two situations can be distinguished. Situation 1: if one wants to explore which hypothesis is more likely to be supported, one can set BF thresh =1. Situation 2: if one wants to find compelling evidence to support the true hypothesis, one can set BF thresh equal to 3, 5, or 10, depending on the strength of the evidence that is required. With respect to the choice of η it should be noted that 1 − η are, for the null and alternative hypotheses, the Bayesian counterparts of the type I and the type II error rates. In high-stakes research, the probability of an erroneous decision should be small, therefore a larger value of η such as 0.90 should be used. In low-stakes or more exploratory research erroneous decisions may be less costly and smaller values like η = 0.80 could be used.
## The role of sample-size determination in bayesian inference
In the Bayesian framework, updatingcan be seen as an alternative for sample-size determination that does not require specification of the effect size under the alternative hypothesis. Bayesian updating proceeds along the following steps: (i) specify an initial sample size per group and the required support in terms of BF; (ii) collect data with the initial sample size; (iii) compute the BF; (iv) if the support in favor of either H 0 or H 1 is large enough the study is finished; if the support is not large enough, increase the sample size and return to (iii). Because in the Bayesian framework the goal is not to control the Type I and Type II error rates (the goal is to quantify the support in the data for the hypotheses under consideration) this is a valid procedure.
With the availability of Bayesian updating and samplesize determination, two strategies can be used to obtain sufficient support for the hypotheses under consideration, which will be described in the next two sub-sections: (i) sample-size determination as a pre-experimental phase in case updating is not an option; and, (ii) sample-size determination followed by updating.
## Sample-size determination as a pre-experimental phase
If updating can be used, it is an approach that avoids prespecification of the effect size under the alternative hypothesis and is a worthwhile option to pursue. However, updating cannot always be used or sample-size determination is a The sampling distribution of BF 01 under H 0 and BF 10 under H 1 . The vertical dashed line denotes the BF thresh = 3. The grey area visualizes η = 0.80. Note that, as will be illustrated inlater in this paper, the sample size is the maximum of 26 and 104 required step before updating can be executed. Consider the following situations. Situation 1. The population of interest is small, for instance, persons with a rare disease or cognitive disorder. The control and treatment groups will very likely not be large. Updating is in this situation not an option. However, if, for example, a researcher is interested to detect an effect size of Cohen's d (for the t test) equal to .8 with a probability η = 0.80 that the Bayes factor is at least 5, the sample size required is 67 per group (see, which will be discussed after the next two sections). Since such a large sample size cannot be obtained, it is decided not to execute the experiment in this form. Situation 2. Next month a survey will start in which 150, currently single, men and women will be tracked for 21 years. Updating is not an option in such a longitudinal cohort study, butshows that 104 persons per group are needed to have a probability of at least η = 0.80 to obtain a Bayes factor larger than 3 if the effect size is Cohen's d = .5. Since the effect size is expected to be 0.5, the study can be actually conducted because the sample size is 150 persons per group. Situation 3. The researchers have to submit the research plans to the (medical) ethical committee. They want to use updating, but both the researchers and the committee's members may want an indication of the sample size needed to obtain sufficient support for different effect sizes under the alternative hypothesis. Only with these numbers they can argue that they have sufficient funding and research time to execute the research plan. Sample-size determination can be used to obtain an indication of the sample sizes needed to obtain sufficient support for different effect sizes. These numbers can be included in the researcher's research proposal for the (medical) ethical committee.
## Sample-size determination followed by updating
When sample-size determination is used, however, as will be highlighted using Situations 4 and 5, having to specify the effect size under the alternative hypothesis may have two undesirable consequences. Consider the following situations. Situation 4. If the alternative hypothesis is true, the researchers expect an effect size Cohen's d = .5. They determine the sample sizes such that an effect size of Cohen's d (for the t test) equal to .5 with η = 0.80 that the Bayes factor is at least 3 is detected, that is, 104 persons per group. After collecting data, they obtain BF 01 = 2.5. This is an undesirable result because they did not achieve the desired support. They can remedy this by updating, that is, increasing the sample size until the Bayes factor is at least 3. The latter is only possible if updating is an option. Situations 1 and 2 are examples of cases where this is not an option. Situation 5. Analogous to Situation 4, but now the researchers find BF 01 = 8.3. This is a problem in the sense that they spent more funds and research time than would have been necessary. The researchers plan and are able to collect the data from 104 persons per group. If the research design permits this they can update until they reach the required support (which may be achieved at a sample size smaller than 104 per group), which will save funds and research time. The combination of sample-size determination and updating is the most powerful approach, whenever it is applicable.
## Ingredients for sample-size determination
Sample-size determination for the Bayesian t test and the Bayesian Welch's test is implemented in the function SSDttest of the R package SSDbain available at https://github.com/Qianrao-Fu/SSDbain. In this section, we introduce and discuss the necessary input for samplesize determination with the SSDttest function. In the sections that follow, we will provide the algorithms used for Bayesian SSD, and a discussion of SSD properties using three tables for Cohen's d equal to .2, .5, and .8, respectively. Furthermore, three examples of the application of SSDttest are presented.
After loading the SSDbain library, the following call is used to determine the sample size per group: library(SSDbain) SSDttest(type='equal',Population_mean= c(0.5,0),var=NULL,BFthresh=3,eta=0.80, Hypothesis ='two-sided',T=10000)
The following ingredients are used: Hypothesis='two-sided' when the competing hypotheses are H 0 : μ 1 = μ 2 , H 1 : μ 1 = μ 2 ; Hypothesis='one-sided' when the competing hypotheses are H 0 : μ 1 = μ 2 , H 2 : μ 1 > μ 2 . The default setting is Hypothesis='two-sided'. This argument is used to decide whether a two-sided (labelled H 1 earlier in the paper) or a one-sided (labeled H 2 earlier in the paper) alternative hypothesis is to be used. For example, one may wish to compare a new drug with an existing drug. If the researcher is not certain if the new drug will be more or less effective than the existing drug, a two-sided alternative hypothesis should be chosen. If the researcher has strong reasons to believe the new drug is more effective than the old one, a one-sided alternative hypothesis should be chosen. 7. T, a positive integer that specifies the number of data sets sampled from the null and alternative populations to determine the required sample size. The default setting is T=10,000, and the recommended value is at least 10,000. This argument will be elaborated in the next section.
The output results include the sample size required and the corresponding probability that the Bayes factor is larger than the BF thresh when either the null hypothesis or the alternative hypothesis is true:
Using N=xxx and b P(BF0i>BFthresh|H0)=xxx P(BFi0>BFthres}|Hi)=xxx Using N=xxx and 2b P(BF0i>BFthresh|H0)=xxx P(BFi0>BFthresh|Hi)=xxx Using N=xxx and 3b P(BF0i>BFthresh|H0)=xxx P(BFi0>BFthresh|Hi)=xxx where xxx will be illustrated in the examples that will be given after the next section.presents Algorithm 1, which is the basic algorithm used to determine the sample size. The ingredients in the first four Steps have been discussed in the previous section. In
## Algorithm used in bayesian sample-size determination
Step 5, T = 10, 000 data sets are sampled from each of the populations of interest (e.g., H 0 vs. H 1 ), starting with a sample size N = 10 per group. In Step 6 the Bayes factor for each data set sampled from each hypothesis is computed. In Step 7, the probabilities P (BF 0i > BF thresh |H 0 ) and P (BF i0 > BF thresh |H i ) are computed. If both are larger than η specified in Step 4, the output presented in the previous section is provided. If one or both are smaller than η, N is increased by 1 per group and the algorithm restarts in Step 5. To be able to account for the sensitivity of the Bayes factor to the specification of the prior distribution, this algorithm is executed using fractions equal to b, 2b, and 3b. The Appendix presents a refinement of Algorithm 1 that reduces the number of iterations in Algorithm 1 to maximally 12.
## Features of ssd
In this section, features of SSD will be discussed. This will be done usingand 5, which were constructed using SSDttest. The tables differ in effect size:is for effect size d = 0.2,is for effect size d = 0.5, andis for effect size d = 0.8. The following features 1.00 1.00 1.00 1.00 1 the means μ 1 = 0.8, μ 2 = 0 and the variance σ 2 = 1 2 the means μ 1 = 0.8, μ 2 = 0 and the variances σ 2 1 = 1.33, σ 2 2 = 0.67 will be discussed: difference between the Bayesian t test and Bayesian Welch's test, effect of the effect sizes, effect of the fraction b used to construct the prior distribution, and comparison of the two-sided and one-sided alternative hypothesis.
There seems to be little difference between the t test and Welch's test with respect to the sample size required and the corresponding probability that the Bayes factor is larger than BF thresh if either the null or the alternative hypothesis is true. For example, for BF thresh =3, two-sided testing, effect size d = 0.5, and η = 0.80 (see, the sample size is 104 per group, and the probability that the Bayes factor is larger than 3 if H 0 is true is 0.92, and the probability that the Bayes factor is larger than 3 if H 1 is true is 0.80 for the t test. The sample size is 104 per group, and the probability that the Bayes factor is larger than 3 if H 0 is true is 0.92, and the probability that the Bayes factor is larger than 3 if H 1 is true is 0.80 for Welch's test.
As expected, the sample size required decreases as the effect size under H i increases. For example, for the twosided t test, BF thresh =3 and η = 0.80, the sample sizes required for effect sizes 0.2, 0.5, and 0.8 are 769, 104, and 36 per group, respectively. This is because an increase of the effect size makes the alternative more distinguishable from the null hypothesis. However, for some special cases, the sample size required for effect size 0.5 and 0.8 are the same, for example for the two-sided t test, BF thresh =5 and η = 0.80 if the fraction 2b is used for the prior distribution. The reason is that the sample size required is the maximum of the sample size required if the null hypothesis is true and the sample size required if the alternative hypothesis is true. In cases like the examples given, the maximum sample size is determined by the null hypothesis, which is the same for effect size 0.5 and 0.8.
The sample size required increases with the fraction going from b to 2b, and then to 3b if the null hypothesis is true, while the opposite relation is found if the alternative hypothesis is true. This feature can be explained as follows: according to Equations (9) and (10), as the fraction gets larger, the prior variance decreases, the relative complexity c 0 gets larger, thus the Bayes factor under H 0 gets smaller. Consequently, the sample size required increases. Conversely, the sample size required when the alternative hypothesis is true decreases. This feature highlights that a sensitivity analysis is important: results depend on the fraction of information used to specify the prior distribution.
As can be seen in, the required sample sizes for one-sided testing are always smaller than or about equal to the sample sizes required for two-sided testing. Therefore, if a directional hypothesis can be formulated, a one-sided testing is preferred over a two-sided testing.
## Practical examples of ssd
In this section, three examples of SSD will be given. The examples use the function SSDttest because it allows researchers to choose Cohen's d, BF thresh , and η as they desire. As an alternative, researchers can also consult Tables 3, 4 and 5, although there sample sizes are only given for a limited number of values for Cohen's d, BF thresh and η.
Example 1 Researchers want to conduct an experiment to investigate whether there is a difference in pain intensity as experienced by users of two types of local anesthesia. The researchers would like to detect a medium effect size d = 0.5 with a two-sided t test, when either H 0 or H 1 with d = 0.5 is true, such that they have a probability of 0.80 that the resulting Bayes factor is larger than 3. The researchers choose BF thresh = 3 because they want to get a compelling evidence for the high-stakes experiment that one of the two types of anesthesia is better able to reduce the pain intensity for users. As elaborated below, the researchers can combine SSD with Bayesian updating to (i) stop sampling before a sample size of N = 104 per group if the true effect size is larger than d = 0.5 used for SSD, or (ii) to continue sampling beyond N = 104 per group if the true effect size is smaller than 0.50. The sample size required to detect d = 0.5 is obtained using the following call to SSDttest: SSDttest(type='equal',Population_mean= c(0.5,0),var=c(1,1), The results are as follows:
Using N=104 and b P(BF01>3|H0)=0.92 P(BF10>3|H1)=0.80
The following can be learned from these results: The researchers need to collect 104 cases per type of local anesthesia to get a probability of 0.92 that the resulting Bayes factor is larger than 3 when H 0 is true, and to get a probability of 0.80 that the resulting Bayes factor is larger than 3 when H 1 is true and d = 0.5.
The researchers will execute the Bayesian updating as follows. First, the researchers will start with 25% of the sample size per group, that is, 26 cases per group. If the resulting BF 01 or BF 10 is larger than 3, the desired support is achieved and updating can be stopped. Otherwise, the researchers can add 26 cases per group and recompute and re-evaluate the Bayes factors. Once the threshold of 3 has been achieved, this process can be stopped, otherwise it can be repeated, also beyond a sample size of 26 cases per group. The SSD executed before these researchers started collecting data is useful because it gives an indication of the sample size that are required to evaluated H 0 and H 1 . Updating ensures that the researchers use their resources optimally.
Example 2 Researchers want to carry out a test to explore whether there is a difference between the yield obtained with a new corn fertilizer and with a current fertilizer. They expect the new fertilizer is more effective than the current one. The researchers want to determine the number of field plots used in a study of the test to detect an effect size d = 0.2 with a one-sided t test. When either H 0 or H 2 with d = 0.2 is true they want to have a probability of 0.90 that the resulting Bayes factor is larger than 1. The researchers used BF thresh = 1 and η = 0.90 because they want to get a Bayes factor to point to the true hypothesis with a high probability. They are not necessarily interested in strong evidence for the true hypothesis. The sample size required is obtained using the following call to SSDttest: SSDttest(type='equal',Population_mean=c (0.2,0),var=c(1,1),BFthresh=1,eta=0.90, Hypothesis ='one-sided',T=10000)
The results are as follows:
Using N=676 and b P(BF02>1|H0)=0.99 P(BF20>1|H2)=0.90
The following can be learned from the output: The researchers need to collect 676 field plots per fertilizer to get a probability of 0.99 that the resulting Bayes factor is larger than 1 if H 0 is true, and a probability of 0.90.16 that the resulting Bayes factor is larger than 1 if H 2 is true.
Example 3 Researchers wish to compare two weight loss regimens to determine whether there is a difference in the mean weight loss. Past experiments have shown that the standard deviations are different for these two regimens. Researchers want to determine the sample size required to detect the effect size d = 0.5 with a two-sided Welch's test. When either H 0 or H 1 is true they want to have a probability of 0.80 that the resulting Bayes factor is larger than 3. They also want to execute a sensitivity analysis and therefore look at the sample sizes required for b, 2b, and 3b. The required sample size is obtained using the following call to SSDttest: ,var=c(1.33,0.67),BFthresh=3,eta= 0.80, Hypothesis='two-sided',T=10000)
The results are as follows:
Using N=104 and b P(BF01>3|H0)=0.92 P(BF10>3|H1)=0.80
Using N=96 and 2b P(BF01>3|H0)=0.87 P(BF10>3|H1)=0.80
Using N=91 and 3b P(BF01>3|H0)=0.83 P(BF10>3|H1)=0.80
From the results the following can be learned: The output from SSDttest can be used to perform a sensitivity analysis. As can be seen the required sample sizes for b, 2b and 3b are 104, 96, and 91 per group, respectively. This implies that if the researchers plan to execute a sensitivity analysis they should aim for a sample size of at least 104 per group. The probabilities of supporting H 0 and H 1 when they are true become more similar with bigger fractions of information. If this is a desirable feature for the researchers, they can use 3b which renders a required sample size of N = 91 per group and η is about equal to 0.80 both when H 0 and H 1 are true.
# Conclusions
The function SSDttest implemented in the R package SSDbain (https://github.com/Qianrao-Fu/SSDbain) has been developed for sample-size determination for twosided and one-sided hypotheses under a Bayesian t test or Bayesian Welch's test using the AAFBF as implemented in the R package bain. This function was used to construct sample size tables that are counterparts to the frequently used tables in. If the tables are not applicable to the situation considered by researchers, the SSDbain package can be used.
With the growing popularity of Bayesian statistics, it is important tools for samplesize determination in the Bayesian framework become available. In this manuscript, we developed software to calculate sample sizes within the framework of Bayesian t test and Bayesian Welch's test hypotheses using timeefficient algorithms. However, the SSDbain package also has its limitation: we focused on the AAFBF, but as was shortly highlighted in the introduction to this paper, there are other Bayes factors researchers may use. Furthermore, we focused on the Bayesian t test and Welch's test, but in our future research we will extend to other statistical models, such as Bayesian ANOVA, ANCOVA, linear regression, and normal linear multivariate models.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creativecommonshorg/licenses/by/4.0/. |
Fluoride contributes to the shaping of microbial community in high fluoride groundwater in Qiji County, Yuncheng City, China
As a toxic element, excessive amounts of fluoride in environment can be harmful because of its antimicrobial activity, however little is known about the relationship between fluoride and the bacterial community in groundwater systems. Here, we use samples from a typical fluorosis area to test the hypothesis that fluoride concentration is a fundamental structuring factor for bacterial communities in groundwater. Thirteen groundwater samples were collected; high-throughput 16S rRNA gene sequencing and statistical analysis were conducted to compare the bacterial community composition in individual wells. The results showed that Proteobacteria, with most relative abundance in groundwater, decreased along the groundwater fluoride concentration. Additionally, relative abundances of 12 families were also statistically correlated with fluoride concentration. The bacterial community was significantly explained by TOC (P = 0.045) and fluoride concentration (P = 0.007) of groundwater. This suggests that fluoride and TOC likely plays an important role in shaping the microbial community structure in these groundwater systems. Our research suggest that fluoride concentration should be taken into consideration in future when evaluating microbial response to environmental conditions in groundwater system, especially for fluoride rich groundwater.
Based on the previous studies, the antimicrobial activity of fluoride is mediated via three major effects of fluoride: (i) enzyme inhibition. The direct inhibition of the F-ATPase enzyme by fluoride was found to be dependent on cations [bib_ref] Metal Fluoride Inhibition of a P-type H + Pump: stabilization of the..., Pedersen [/bib_ref] [bib_ref] Inhibition of proton-translocating ATPases of Streptococcus mutans and Lactobacillus casei by fluoride..., Sturr [/bib_ref] , and fluoride is a potent inhibitor of enolase, catalase, phosphatases and other enzymes of many organisms [bib_ref] Enolases from fluoride-sensitive and fluoride-resistant streptococci, Bunick [/bib_ref] [bib_ref] Sodium fluoride induces nephrotoxicity via oxidative stress-regulated mitochondrial SIRT3 signaling pathway, Song [/bib_ref] [bib_ref] pH-dependent fluoride inhibition of catalase activity, Thibodeau [/bib_ref] ; (ii) fluoride was found to alter the proton movement through cell membranes and inhibit intracellular acid production [bib_ref] Simultaneous assessment of acidogenesis-mitigation and specific bacterial growth-inhibition by dentifrices, Forbes [/bib_ref] [bib_ref] Effect of low levels of fluoride on proton excretion and intracellular pH..., Guha-Chowdhury [/bib_ref] ; (iii) fluoride can inhibit certain microbes in the ingestion, transformation and utilization of certain nutrients, affecting the synthesis of extracellular polysaccharides and the storage of intracellular polysaccharides [bib_ref] Effect of fluoride on extracellular polysaccharide production by streptococcus mutans, Bowen [/bib_ref] [bib_ref] Effects of fluoride on bacterial growth and its gene/protein expression, Ma [/bib_ref] [bib_ref] Effects of fluoride, lithium, and strontium on intracellular polysaccharide accumulation in S...., Wegman [/bib_ref].
As a toxic element, fluoride influences the metabolic behavior of microbes in natural environment. According to during the dissolution of rock phosphate by Aspergillus niger, fluoride was capable of decreasing fungal growth, citric acid production, and acidification [bib_ref] Inhibition of Aspergillus niger phosphate solubilization by fluoride released from rock phosphate, Mendes [/bib_ref]. In an indoor culture, found that a continuous supply of 1 mM NaF reduced the viable counts of microflora in solution at pH 7 [bib_ref] Prevention of population shifts in oral microbial communities in vitro by low..., Bradshaw [/bib_ref]. It was reported that the per liter concentration of fluoride ranges from several micrograms to dozens of milligrams to even one thousand milligrams in groundwater [bib_ref] Fluoride, nitrate and water hardness in groundwater supplied to the rural communities..., Daesslé [/bib_ref] [bib_ref] Controls on the genesis of some high-fluoride groundwaters in India, Jacks [/bib_ref] [bib_ref] A review of fluoride in african groundwater and local remediation methods, Kut [/bib_ref] [bib_ref] Assessment of groundwater quality in a region of endemic fluorosis in the..., Souza [/bib_ref]. While groundwater systems (groundwater-sediment) contain large numbers of microorganisms [bib_ref] Microbial biodiversity in groundwater ecosystems, Griebler [/bib_ref]. However, how microbial communities in groundwater respond to different fluoride levels is not yet well understood. We hypothesize that fluoride influences microbial community diversity and structure by its toxic effects on microbes with various [F] in groundwater.
A typical area with endemic fluorosis, Qiji County of Yuncheng City, China [bib_ref] Enrichment of fluoride in groundwater under the impact of saline water intrusion..., Gao [/bib_ref] [bib_ref] Fluoride and arsenic hydrogeochemistry of groundwater at Yuncheng Basin, Northern China, Khair [/bib_ref] [bib_ref] Hydrogeochemistry of high-fluoride groundwater at Yuncheng Basin, northern China, Li [/bib_ref] [bib_ref] Arsenic, fluoride and iodine in groundwater of China, Wen [/bib_ref] , was chosen as a model system in which to test our hypothesis. According to , the fluoride concentration ranges from approximately 1.4 mg/L to 14.2 mg/L in Qiji County [bib_ref] Hydrogeochemistry of high-fluoride groundwater at Yuncheng Basin, northern China, Li [/bib_ref] , exceeding the World Health Organization (WHO) limit of 1.5 mg/L in drinking water. And over 50% were meeting or exceeding the levels of fluoride that were reported to be needed to kill bacteria, approximating 0.16 to 0.30 mmol/L (3.04 to 5.70 mg/L) [bib_ref] Effects of fluoride on the microbial ecology of dental plaque, Bowden [/bib_ref]. To obtain samples with various fluoride concentrations, shallow groundwater samples were collected from thirteen different wells for a coupled study of both fluoride and microbial community. The aim of this study is (i) to investigate the microbial community in groundwater of the area; (ii) to discern which environmental parameters significantly influence the microbial community at the phylum, family and OTU levels; and (iii) to discuss the potential effects of F/[F] on the groundwater microbial community.
# Materials and methods
Site description. Qiji County is located in the southwestern part of Yuncheng City, Shanxi Province, China [bib_ref] Diverse mechanisms drive fluoride enrichment in groundwater in two neighboring sites in..., Li [/bib_ref] and covers an area of 77 km 2 (between 34°40′ and 35°02′N and 110°29′ and 110°30′E). The region has a semiarid climate with most of the rainfall (>65%) occurring between June and October. The Quaternary sediment is composed of primarily aeolian loess, lacustrine clays, fluvial sands and gravels in the area. The major shallow aquifer consists of an admixture of fine, to medium grained and coarse sands with depth between 35 to 65 meters in the area. The regional shallow groundwater flows from the northwest to the southeast, toward the salt lake [bib_ref] Enrichment of fluoride in groundwater under the impact of saline water intrusion..., Gao [/bib_ref] [bib_ref] Hydrogeochemistry of high-fluoride groundwater at Yuncheng Basin, northern China, Li [/bib_ref].
Historically, approximately 50% of the people living in the area have been affected by fluorosis in the past several decades. Long-term intake of high-fluoride ground water is the major reason for endemic fluorosis in Yuncheng Basin [bib_ref] Enrichment of fluoride in groundwater under the impact of saline water intrusion..., Gao [/bib_ref] [bib_ref] Trace elements and environmental isotopes as tracers of surface water-groundwater interaction: a..., Gao [/bib_ref] [bib_ref] Coexistence of high fluoride fresh and saline groundwaters in the Yuncheng Basin,..., Gao [/bib_ref] and other sites in the world 2 . The Yuncheng basin has a semi-arid climate with a strong evaporation. Groundwater is the major source of water drinking causing the lack of surface water. The fluoride rich groundwater normally located in the low land areas in the basin. The chemistry of the groundwater with high fluoride concentration is characterized as: Na-rich and Ca-poor, with high pH and HCO 3 − values and low TDS values [bib_ref] Enrichment of fluoride in groundwater under the impact of saline water intrusion..., Gao [/bib_ref] [bib_ref] Hydrogeochemistry of high-fluoride groundwater at Yuncheng Basin, northern China, Li [/bib_ref] [bib_ref] Diverse mechanisms drive fluoride enrichment in groundwater in two neighboring sites in..., Li [/bib_ref] [bib_ref] Geochemical processes controlling the groundwater chemistry and fluoride contamination in the Yuncheng..., Luo [/bib_ref]. The dissolution of fluorine-bearing minerals, cation exchange and evaporation were main factors that control the occurrence of high fluoride shallow groundwater.
Sample collection and geochemical analysis. A total of thirteen water samples were collected in August, 2014, from the power -operated wells around the borehole located in Qiji area, Yuncheng City, Shanxi Province . When sampling, water samples were collected only after the in situ physicochemical parameter, including pH, redox potential (Eh), dissolved O 2 (DO), temperature (T), and electrical conductivity (EC), were stable. These four parameters had been measured using portable HACH EC, DO and pH meters. Subsamples for chemical analysis were filtered through 0.45-μm filters and collected in three acid-washed, high-density polyethylene (HDPE) bottles that had been rinsed with the sample water thoroughly (at least three times before sampling). Alkalinity measurements were performed by pH-verified colorimetric titration. The microbial samples were collected by filtration of 5-10 L of water through 0.22-μm filters, and the biomass-containing filters were wrapped and immediately stored at −80 °C for the analysis of biodiversity. The other hydrogeochemical analyses were performed at the State Key Laboratory of Biological and Environmental Geology, China University of Geosciences (Wuhan). The concentrations of Cl − , SO [bib_ref] The effects of fluoride based fire-fighting foams on soil microbiota activity and..., Montagnolli [/bib_ref] 2− , NO 3 − , and F − were determined using ion chromatography (IC) (Dionex 120, Dionex, Sunnyvale, CA, USA). Samples for cation analysis were acidified with trace metal grade HNO 3 to pH less than 2.0. Major cations (K + , Na + , Ca 2+ and Mg 2+ ) were measured using inductively coupled plasma-atomic emission spectroscopy (ICP-AES) (IRIS Intrepid II XSP, Thermo Elemental, Madison, WI, USA). The instrumental detection limit was 0.06 ppm. The standards used for the ICP-AES calibration were within 5% of external standards. Samples for total organic carbon (TOC), total nitrogen (TN) and total phosphorus (TP) analysis were acidified with H 2 SO 4 . Sub-samples for TOC were analyzed with a TOC analyzer (Analytik Jena, Multi N/C, 3100); samples for TN and TP were analyzed with ultraviolet spectrophotometer (HITACHI U-1900).
DNA extraction, PCR amplification, and sequence analysis. Total genomic DNA was extracted from the filtered groundwater samples (0.22 μm, millipore) at the sampling site. DNA extractions were conducted using the PowerWater ® DNA Isolation kit (Anbiosci Tech LTD, 14900-50-NF) according to the manufacturer's manual. DNA yield was quantified with Nanodrop (Thermo Scientific NanoDrop 2000) and then stored at −80 °C until PCR amplification. PCR amplifications were performed with the 515f/806r primer sets (The reverse primer contains a 6-bp error-correcting barcode to distinguish sequences that originated from different samples) that amplifies the V4 region of the 16S rDNA gene, following the protocol described previously [bib_ref] The importance of HCV RNA measurement for tailoring treatment duration, Peiffer [/bib_ref]. Sequencing was conducted with 200 ng of amplified product on an Illumina MiSeq platform at Novogene (Beijing, China).
Sequence analysis was performed with the UPARSE software package using the UPARSE-OTU and UPARSE-OTUref algorithms [bib_ref] UPARSE: highly accurate OTU sequences from microbial amplicon reads, Edgar [/bib_ref]. Sequences were further denoised using conduct filter analysis by applying the microbial community analysis software QIIME [bib_ref] QIIME allows analysis of high-throughput community sequencing data, Caporaso [/bib_ref]. High-quality sequences were obtained and compared with the data library (Gold database, http://drive5.com/uchime/uchime_download.html), and the chimera sequences were measured and removed 52 (UCHIME Algorithm, http://www.drive5.com/usearch/manual/uchime_algo.html). This output was clustered using Uparse 53 software at the 97% sequence identity level, resulting in 6383 operational taxonomic units (OTUs). A representative sequence from each OTU was classified using the Ribosomal Database Project (RDP) classifier 54 (Version 2.2, http://sourceforge.net/projects/rdp-classifier/) and GreenGene data set 55 http://greengenes.lbl.gov/cgi-bin/nph-index.cgi). The OTUs table was formatted to allow comparisons using β diversity metrics.
# Statistical analysis.
Statistical analyses were performed in QIIME and using package "vegan" in R, version 3.2.4 (https://www.r-project.org/) [bib_ref] River organic matter shapes microbial communities in the sediment of the Rhône..., Fagervold [/bib_ref] [bib_ref] Diversity of the sediment microbial community in the Aha watershed (Southwest China)..., Sun [/bib_ref]. Rarefaction curves, species richness estimators and community diversity indices were calculated in QIIME. The Chao1 richness index, Shannon diversity index and coverage rate were calculated to reflect the alpha-diversity of samples [bib_ref] Hydrogeochemistry of high-fluoride groundwater at Yuncheng Basin, northern China, Li [/bib_ref].
A heat map of abundance data at the class level was constructed. Two kinds of analyses were performed to investigate the relationships between the microbial community (at the OTUs level) and the geochemical parameters: (1) Mantel tests were used to examine the relationship between the bacterial community structure and the geochemical parameters [bib_ref] Control of temperature on microbial community structure in hot springs of the..., Wang [/bib_ref]. The significance of each environmental factor was tested by analysis of variance with 999 permutations. Mantel tests were performed with both weighted Uni Frac and Bray Curtis distance matrices for the microbial community data and distance matrices (Euclidian distance) for each environment factor. (2) Redundancy analysis (RDA) with forward selection to identify the factors that could best explain the variation in the microbial community 56 and significance was tested by a Monte Carlo permutation test based on 999 random permutations. Some additional correlation analyses were performed using SPASS software.
# Results
Groundwater geochemistry. The groundwater was slightly alkaline (7.69-8.50) and fresh to slightly saline . Fluoride concentration in groundwater varied from 3.05 to 8.89 mg/L, which was within the range reported by others 41 from the area. The TOC values ranged from 0.12 mg/L to 4.38 mg/L. Na + was the most dominant cation, with concentrations of 244.5~908.1 mg/L, followed by Mg 2+ , Ca 2+ and K + . Bicarbonate and sulfate were major anions, with concentrations of 548.0~1128 mg/L and 27.26~757.4 mg/L, respectively. This results were resistant with that Na-rich, Ca-poor and high bicarbonate concentration were the mainly characteristic of groundwater in the basin [bib_ref] Hydrogeochemistry of high-fluoride groundwater at Yuncheng Basin, northern China, Li [/bib_ref] [bib_ref] Diverse mechanisms drive fluoride enrichment in groundwater in two neighboring sites in..., Li [/bib_ref]. The correlation analysis (Supplementary showed that the fluoride is significantly correlated with pH (R = 0.678, P < 0.05), TP (R = 0.730, P < 0.01) and Ca 2+ (R = −0.720, P < 0.01). Under high pH alkaline conditions, OHcould exchange with F − adsorbed on minerals and soil [bib_ref] F-OH substitution in natural tremolite, talc, and phlogopite, Abercrombie [/bib_ref] [bib_ref] Partitioning of F-Cl-OH between minerals and hydrothermal fluids, Zhu [/bib_ref] and prevent F − from complexing with cations, which results in a high release of F − in groundwater. The positive correlation between fluoride and TP may partly due to the contamination of the phosphatic fertilizer and the competitive adsorption by PO [bib_ref] The effects of fluoride based fire-fighting foams on soil microbiota activity and..., Montagnolli [/bib_ref] 3− which could cause the desorption of fluoride from mineral/organic matter surfaces within the groundwater system 61 . Meanwhile, the negative correlation between [F] and [Ca] is attributed to the dissolution/precipitation equilibrium of major fluoride bearing minerals, especially fluorite (CaF 2 ) [bib_ref] Diverse mechanisms drive fluoride enrichment in groundwater in two neighboring sites in..., Li [/bib_ref]. It is suggested that the concentration of fluoride in groundwater is under the control of natural water-sediments rock interactions (e.g., cation exchange, competitive adsorption, dissolution/precipitation) in natural groundwater systems [bib_ref] Geochemical factors controlling the occurrence of high fluoride groundwater in the Nagar..., Rafique [/bib_ref]. . Obvious change in groundwater community diversity was apparent across the fluoride concentration gradient [fig_ref] Figure 1: The Chao1 and Shannon index along a fluoride concentration gradient [/fig_ref]. The microorganism communities in samples with high fluoride concentration were substantially simpler than those in samples with medium fluoride concentration. . Significant negative correlations were observed between fluoride concentration with Chao1 (R = −0.705, P = 0.007, [fig_ref] Figure 1: The Chao1 and Shannon index along a fluoride concentration gradient [/fig_ref] and Shannon (R = −0.563, P = 0.010, [fig_ref] Figure 1: The Chao1 and Shannon index along a fluoride concentration gradient [/fig_ref] , which may be interpreted by the inhibitive ability of fluoride toward micro-organisms in groundwater in the area.
## Bacterial community composition. the most abundant prokaryotic organisms in groundwater samples
belonged phylogenetically to the Proteobacteria phylum, accounting for 55.41% (in QJ04) to 84.58% (in QJ13) of the total valid reads in all libraries [fig_ref] Figure 1: The Chao1 and Shannon index along a fluoride concentration gradient [/fig_ref]. The second most abundant prokaryotic organisms were Bacteroidetes (1.86% to 37.95% of the total valid reads), followed by Cyanobacteria (0.11% to 8.32%). Crenarchaeota, Nitrospirae, Firmicutes, Acidobacteria, Actinobacteria. Planctomycetes and OP3 were found to be the minor phyla. The microbial community, at the phylum level, was similar among groundwater samples (QJ07, QJ10, QJ08, QJ11 and QJ13) with a medium fluoride concentration (3.05-4.88 mg/L). Significant differences in bacterial community were observed among the rest of samples with a higher fluoride concentration (5.12-8.89 mg/L). Frequently, fluoride-related changes in the relative abundances of bacterial phyla were observed. For example, Proteobacteria occurred at lower abundances in the libraries from groundwater samples with higher fluoride concentration than that from samples with a medium F concentration [fig_ref] Figure 2: Microbial community composition at phylum level in the study area [/fig_ref]. While the abundance of Proteobacteria was positively related with EC, TOC and major ions [fig_ref] Figure 2: Microbial community composition at phylum level in the study area [/fig_ref].
The correlation analysis conducted with heat maps involved a clustering analysis based on the distribution of the predominant families, and deeper colors represent higher richness [fig_ref] Figure 3: Heat map of log relative abundances of the distribution of dominant family... [/fig_ref]. The column cluster illustrated that most of the samples with higher fluoride concentrations (QJ02, QJ04, QJ06) and one sample with medium fluoride concentration (QJ01) were clustered together (Cluster B). Bacterial communities under lower fluoride concentration (QJ07, QJ08, QJ10, QJ11, QJ05, QJ09, QJ12 and QJ13) conditions were clustered into another group (Cluster A). The dominant family was grouped into three groups. Samples in the Cluster B, with a higher fluoride concentration, have a lower relative abundance of families (soft color) in group one and part of the families in group three (from Desulfovibrionaceae to Alcanivoracaceae at the top of the heat map). In contrast, the samples in Cluster A, with a medium fluoride concentration, have a higher relative abundance of families (deep color) in group one and part in group three. Lower relative abundances of the families Xanthomonadaceae, Caulobacteraceae, Chthonomonadaceae, Nocardiaceae, Microbacteriaceae, Bacteriovoracaceae, Rhodasprilliaceae, Hyphmicrobiaceae, Legionellaceae, Rhizobiaceae, Bdellovibrinaceae and Sphingomonadaceae were observed in the high fluoride groundwater, while higher relative abundances of these families were observed in groundwater with medium fluoride concentrations. Obviously, the richness of the predominant families shows a significant diversity in groundwater with different fluoride concentration levels. [fig_ref] Figure 4: The relative abundance of microbial family along fluoride gradients [/fig_ref] shows the relative abundances of bacterial families across the fluoride concentration. The relative abundances of Pseudomonadaceae and Caulobacteraceae were negatively correlated with fluoride concentration at significant level (p < 0.01). The relative abundances of Xanthomonadaceae, Methylophilaceae, Legionellaceae and Chthonomonadaceae were also negatively correlated (p < 0.05) with fluoride concentration. Additionally, the relative abundances of these families, excepting Methylophilaceae and Legionellaceae, did not show any significant correlations with any other parameters . Therefore, it is cautiously inferred that the decrease of the relative abundance of these families may be attributed to the increase of fluoride concentration in groundwater. It was also shown that the relative abundances of Campylobacteraceae, Methylobacteriaceae, Porphyromonadaceae, Catabacteriaceae and two uncultured lineages designated FW and VC21-Bac22 were positively correlated (p < 0.05) with fluoride concentration. This implied that some of the families are resistant to a certain higher concentration of fluoride in groundwater. www.nature.com/scientificreports www.nature.com/scientificreports/ Relationships of bacterial community and hydrogeochemical parameters. The differences in microbial community structure could be influenced by hydrogeochemical parameters. For RDA analysis, we employed forward selection by performing a Mantel test with the number of permutations being 999. Then, we drew the bacterial diversity diagram at OTU level with five selected geochemical parameters (pH, Ca 2+ , F − , TOC and TP) that best explained the variance of the bacterial communities in the thirteen groundwater samples . The variance inflation factors of these five parameters were all less than 20.
The first axis of the RDA explained 13.32% of the variance of bacterial community and was positively correlated with the samples at a medium fluoride concentration except QJ12 and negatively correlated with the samples with a higher fluoride concentration. Interestingly, the samples (except QJ12) with a medium fluoride concentration grouped together in the RDA profile. These five hydrogeochemical parameters explained 36.6% of the total variance. TOC and fluoride concentration explained the variability of bacterial community at the significance level (r 2 = 0.45, p = 0.045 vs. r 2 = 0.59, p = 0.007, Supplementary . While the effects from pH could be present but only at the marginal significance level (r 2 = 0.40, p = 0.067). Ca 2+ and TP were insignificant factors.
# Discussion
Knowledge of microbial communities in relation to hydrogeochemical and environmental conditions is important for understanding biogeochemical processes and the mobilization of pollutants (solutes) in groundwater systems [bib_ref] Analysis of the functional gene structure and metabolic potential of microbial community..., Li [/bib_ref]. Microbial community diversity and structure are affected by many environmental factors, such as pH, and the concentrations of various elements, electron donors and acceptors in soils and groundwater systems [bib_ref] The diversity and biogeography of soil bacterial communities, Fierer [/bib_ref] [bib_ref] Responses of microbial community functional structures to pilot-scale uranium in situ bioremediation, Xu [/bib_ref]. It has been reported that fluoride led to a decrease in microbial biomass in a water treatment system 20 and inhibited the activity of soil dehydrogenase, arylsulfatase, alkaline phosphatase, acid phosphatase, peroxidase and ATPase [bib_ref] Fluoride-induced changes in chemical properties and microbial activity of mull, moder and..., Wilke [/bib_ref] [bib_ref] Responses of plants to fluoride: an overview of oxidative stress and defense..., Yadu [/bib_ref] ; it is likely that fluoride had a negative impact on the function of the microbial community. Therefore, we speculated that F may alter the microbial community in groundwater. In this case study, the Chao1 index and Shannon diversity were significantly and negatively correlated with fluoride concentration in the study wells. This indicated that fluoride may have a strong effect on the diversity of bacterial communities in this study.
As a toxic element, fluoride can inhibit or kill microbes by affecting bacterial metabolism through a set of actions with fundamentally different mechanisms [bib_ref] Metal Fluoride Inhibition of a P-type H + Pump: stabilization of the..., Pedersen [/bib_ref] [bib_ref] Simultaneous assessment of acidogenesis-mitigation and specific bacterial growth-inhibition by dentifrices, Forbes [/bib_ref] [bib_ref] Effect of low levels of fluoride on proton excretion and intracellular pH..., Guha-Chowdhury [/bib_ref] [bib_ref] Effects of fluoride on bacterial growth and its gene/protein expression, Ma [/bib_ref]. For example, Al-F or Mg-F, the major metal-fluoride complexes present in high fluoride groundwater, were responsible for fluoride inhibition of phosphate transfer enzymes under laboratory conditions [bib_ref] Anionic charge is prioritized over geometry in aluminum and magnesium fluoride transition..., Baxter [/bib_ref]. This suggests that fluoride, in the form of metal-fluoride complexes, is a possible enzyme inhibitor in groundwater. Previous studies have shown that most of the fluoride movement across the cell membranes of bacteria would be of the HF form, even at pH 7 [bib_ref] Antimicrobial actions of fluoride for oral bacteria, Marquis [/bib_ref]. In groundwater, an increase www.nature.com/scientificreports www.nature.com/scientificreports/ of fluoride concentration possibly cause more fluoride to move across the cell membrane in the form of HF. Accumulation of fluoride in bacteria is largely dependent on its binding to various cell structures, mainly proteins. In the cytoplasm, the dissociation of HF yields F − and H + , which can act as a metabolic inhibitor and reduce the acid tolerance of the bacteria 67 . Indeed, our data were also in agreement with this mechanism of F action as the relative abundance of Proteobacteria decrease in the high fluoride sample, suggesting the presence of a toxic effect from F.
Bulk geochemical parameters, such as pH, fluoride, major ions, TOC and TP have been shown to explain variance in microbial communities in groundwater. The variance in the microbial community in our study, however, was best explained by TOC and fluoride. Numerous studies have shown that microbial community can be strongly correlated with TOC 52,68,69 . Our study and others are in agreement with the fact that carbon, either organic carbon or inorganic carbon, is a necessary nutrient element for microorganisms and is one of the fundamental factors that structures microbial communities by controlling the growth and distribution of microorganism.
The RDA also indicated that fluoride was the most significant factor influencing the bacterial community in groundwater at the OTU level. In this case study, distinct bacterial communities could be formed between individual wells due to the differences of fluoride concentrations. First, frequent fluoride-related changes in the relative abundance of Proteobacteria were observed in the study wells. In addition, at the family level, fluoride concentration was significantly positively or negatively correlated with the relative abundances of twelve families in groundwater. It was reported that some prokaryotes (bacteria and actinomycetes) were resistant to the high www.nature.com/scientificreports www.nature.com/scientificreports/ concentrations of F compounds [bib_ref] Effects of aluminum plant emissions on soil biota in the Kola Peninsula, Evdokimova [/bib_ref]. Genus Pseudomonas, Bacillus sp., Acinetobacter sp. and Streptococcus sp. were reported to be resistant to fluoride by an ancient system composed by fluoride-specific riboswitches and commonly associated proteins such as CrcB [bib_ref] Widespread genetic switches and toxicity resistance proteins for fluoride, Baker [/bib_ref] [bib_ref] Isolation and characterization of fluoride resistant bacterial strains from fluoride endemic areas..., Banerjee [/bib_ref] [bib_ref] Molecular characterization of fluorine degrading bacteria from soil samples for its industrial..., Praveen [/bib_ref] [bib_ref] Identification of anion channels responsible for fluoride resistance in oral Streptococci, Men [/bib_ref]. Besides, Methylobacterium extorquens DM4, encoding at least 10 fluoride riboswitches in its genome, might consume fluorinated hydrocarbons a food source with a robust fluoride sensor and toxicity mitigation response system 71 . This may partly explain why some families were in positive correlation with fluoride concentration. In groundwater ecosystem, microbes are motile (by itself or flow) and faced dynamic biogeochemical conditions which were possibly challengeable [bib_ref] The selective pressures on the microbial community in a metal-contaminated aquifer, Carlson [/bib_ref]. Stochastic processes frequently influences the assembly of microbial communities, while the selective pressures might drive changes in microbial community composition via deterministic effects [bib_ref] Soil pH mediates the balance between stochastic and deterministic assembly of bacteria, Tripathi [/bib_ref] [bib_ref] Disentangling mechanisms that mediate the balance between stochastic and deterministic processes in..., Dini-Andreote [/bib_ref]. Often selective pressures are inferred based on correlations between the relative abundances of microbial taxa and geochemical parameters (e.g., toxic stressors or nutrients), although correlative analysis does not identify causal relationships, and observed correlations can be misleading. Thus, our results indicated that fluoride might be critical factor which can shape bacterial communities in groundwater. www.nature.com/scientificreports www.nature.com/scientificreports/ conclusion Our study revealed a significant association between fluoride concentration and the composition of bacterial communities in shallow groundwater in the Qiji area, northern China. TOC and fluoride concentration plays an important role in modulating the groundwater bacterial communities. These findings give us new insights into the biogeochemical processes of fluoride and other elements in groundwater, suggesting that fluoride concentration should be considered in future when evaluating microbial response to environmental conditions in groundwater system, especially for fluoride rich groundwater. . RDA biplot of bacterial community composition and the five most significant geochemical parameters. Each dot represents one sample. Arrows indicate the direction and magnitude of geochemical parameters associated with bacterial community structures. Total variance is 0.027, and eigenvalues for constrained axes, RDA1 and RDA2, are 0.0036 and 0.0030, respectively. The significant vectors in the figure were highlighted to avoid misunderstanding. Significance codes: 0 < p < 0.001 "***", 0.001 < p < 0.01 "**", 0.01 < p < 0.05 "*", 0.05 < p < 0.1 "- ".
[fig] Figure 1: The Chao1 and Shannon index along a fluoride concentration gradient. (2019) 9:14488 | https://doi.org/10.1038/s41598-019-50914-6 [/fig]
[fig] Figure 2: Microbial community composition at phylum level in the study area. (a) Taxonomic classification of bacterial reads retrieved from different groundwater samples at the phylum level using the RDP classifier. "others" indicates phyla with relative abundances of less than 1.0%; (b) Heat map of correlation between the ten most abundant bacterial phyla and hydrochemical factors. Significance codes: 0 < p < 0.001 "***", 0.001 < p < 0.01 "**", 0.01 < p < 0.05 "*", 0.05 < p < 0.1 "•". (2019) 9:14488 | https://doi.org/10.1038/s41598-019-50914-6 [/fig]
[fig] Figure 3: Heat map of log relative abundances of the distribution of dominant family in the three samples. The double hierarchical dendrogram shows the microbial distribution of the 13 samples. The log relative abundance for the microbial families are depicted by the color intensity; the color key is at the topright. The microbial families with * at their right are related with fluoride concentration analyzed through SPASS software, denoted **(p < 0.01) and *(p < 0.05).Scientific RepoRtS | (2019) 9:14488 | https://doi.org/10.1038/s41598-019-50914-6 [/fig]
[fig] Figure 4: The relative abundance of microbial family along fluoride gradients. Correlation analyses and significance test for species at family level were conducted using SPASS software.Scientific RepoRtS |(2019) 9:14488 | https://doi.org/10.1038/s41598-019-50914-6 [/fig]
|
Assessing heart rate variability in type 1 diabetes mellitus—Psychosocial stress a possible confounder
## | introduc ti on
Reduced heart rate variability (HRV) as a sign of cardiac autonomic dysfunction has been found to be an independent predictor of death and of cardiovascular events, in individuals with diabetes mellitus, and in individuals with different heart conditions. Conversely, high HRV has been linked to longevity in a healthy population.
Damage to the small nerve fibers of the autonomic nervous system, autonomic neuropathy, occurs in diabetes mellitus as a long-term effect of hyperglycemia. The long fibers of the vagus nerve, including the parasympathetic fibers supplying the heart, might be particularly sensitive to damage and therefore affected early on, impacting HRV.
Traditionally, a combination of autonomic reflex tests has been used to diagnose cardiac autonomic neuropathy (CAN).
In 1984, Ewing et al presented a study which suggested that assessment of cardiac parasympathetic indices using 24-hr electrocardiographic (ECG) recordings might be more sensitive than the described reflex tests in detecting early cardiac parasympathetic damage in individuals with diabetes mellitus.
Since then, several studies have demonstrated a reduced HRV in individuals with type 1 diabetes mellitus (T1DM) compared to healthy individuals, with longer diabetes duration and poor glycemic control as prominent risk factors.
Reduced HRV has been reported in children and adolescents with T1DM with very short duration (<1-2 years) in a few older studies. The more recent SEARCH for Diabetes in Youth Cohort Study found an incidence of CAN in young adults (18 years old) with T1DM of 12%, based on HRV from a 5-min ECG recording.
The analysis and the interpretation of HRV parameters, whether from short-or long-term ECG recordings, pose some challenges.
Although recommendations for standards of measurements for HRV were published in 1996 ("Heart rate variability: standards of measurement, physiological interpretation, and clinical use. Task Force of the European Society of Cardiology and the North American Society of , there is no consensus regarding age-specific normal ranges of HRV indices from longterm ECG, which is a disadvantage when studying young individuals, and essential in developing a method for screening purposes and diagnosing CAN on an individual level.
Another methodological difficulty is the presence of possible confounders for reduced HRV, the most important of which is that of psychosocial stress. In the research fields of psychiatric conditions such as clinical depression, or that of biological psychology and stress research, HRV has emerged as an important tool to measure the physiological response to stress, in adults and in children. Several studies show that clinical depression, but also general stressors in life, reduces HRV in both adults and children.
Children and young individuals with diabetes are exposed to additional stressors in life compared with healthy individuals, in part due to the practical everyday management of the disease, but also due to an additional psychological burden simply because of living with a chronic disease. Therefore, when assessing the possible presence of early CAN with 24-hr ECG and HRV analysis, accounting for psychosocial stress seems to be of importance in order to avoid false-positive results.
No previous studies on HRV in young persons with diabetes have taken psychosocial stress into consideration.
Salivary cortisol measurements are frequently used in scientific studies as a biological marker for psychosocial stress both in adults and in children. Cortisol is the product of the hypothalamic-pituitary-adrenal axis (HPA) which responds to both physiological stress and psychosocial stress . Cortisol levels normally follow a circadian rhythm with a peak at awakening and declining levels during the day. The salivary sampling method is attractive compared with blood sampling, especially in children, since it is non-invasive and pain free, and reliably mirrors cortisol blood levels . Evening or midnight cortisol, cortisol awakening response (CAR), and the diurnal cortisol slope are commonly used measurements.
The HPA axis has been studied in subjects with diabetes with varying results, but salivary cortisol has only been used in a limited number of studies on individuals with T1DM in relation to stress. High midnight salivary cortisol was linked to depression, smoking, physical inactivity, and season in adults with T1DM in one meta-analysis, and flattened cortisol slope was associated with stress and subjective socioeconomic disadvantage in older adolescents with T1DM in another study.
Previously described factors of possible importance for the HRV analysis other than psychosocial stress are as follows: hypoglycemia, physical activity, thyroid diseaseand ovulation in fertile women.
The present cross-sectional methodological work, which is part of a larger prospective study, examines several individual and situational factors during a 24-hr ECG registration that might need to be taken into account when interpreting HRV. It focuses on elucidating the usefulness of salivary cortisol as a biomarker of psychosocial stress in the context of assessing reduced HRV in individuals with T1DM.
## | material and me thods
## | subjects
Children and adolescents 6-18 years old with T1DM from the pediatric departments at three different hospitals in the southeast region of Sweden were enrolled consecutively in the study between May 2015 and June 2017 (n = 68). Approximately 18% of the children attending the diabetes clinics accepted to participate. Adults 19-60 years old with T1DM were enrolled consecutively from the department of internal medicine, diabetes, and endocrinology, Kalmar County Hospital, between May 2016 and February 2019 (n = 45, including 8 cases of late autoimmune diabetes in adults, LADA). Retrieving data from the Swedish National Diabetes Register (NDR), a comparison of HbA1c showed that study participants in the child and adolescent group had a lower mean HbA1c than non-participants (56.4 versus 60.6, p < .05); that is, the study cohort mainly consisted of individuals with good or acceptable metabolic control. HbA1c did not differ significantly between adult participants and non-participants (64.9 versus 67.1, p = .3). Healthy controls consisted of friends or partners of the patients (siblings were not allowed to participate in accordance with the ethics committee's directions). Additional 10-18-year-old healthy controls were recruited from schools in Kalmar, by school nurses in spring 2019.
Exclusion criteria were known heart disease, neurological disease, systemic disease or treatment with systemic steroids for any reason, impaired kidney function, depression, or medically treated sleeping disorders, assessed by the treating physician in inclusion form (subjects with T1DM), or self-reported (denied) in inclusion form (controls). Subjects with well-controlled asthma or allergy (n = 8) and
subjects with stable thyroid substitution (n = 9) were included. Body mass index (BMI) was calculated from weight and height measured at the nearest clinical visit or was self-reported. Adult participants were classified as overweight if BMI ≥25. For children, ISO-BMI 25 was used, according to the Swedish pediatric society for endocrinology and diabetes.
## | ecg recordings and hrv analysis
Twenty-four-hour ECG registrations were performed with the Cardiospy device EC-3H 3-channel Holter ECG system (Labtech Ltd). ECG recordings started between 3 and 6 p.m. and were of 22-to 24-hr duration. Sampling rate was 256 Hz. ECG analysis was made using the Labtech Ltd Cardiospy software version 5.04.01.
Algorithm-based automatic identification of QRS complexes and identification of disturbances and ectopic beats were followed by a manual analysis and strict removal of all interferences by an ex- . All HRV parameters were found to be highly interdependent, and for further correlation analysis, the natural logarithm of the high-frequency power spectral density component during sleep, Ln(HFPower sleep), was chosen to represent HRV. The study did not aim to diagnose CAN on an individual level, but compared HRV between groups.
## | diary
The participants were provided with a standardized but open form of diary, and instructed to enter time and nature of different activities during the registration, in particular physical activities and all forms of stress or other events of emotional impact.
Self-reported stress entries in the diary of short duration (<5 min, e.g., "hurrying to catch the bus") were disregarded. Stress comprised occupational, school, or various performance stress, as well as expressed emotions such as anger, irritation, sadness, worry, or anxiety. The presence of psychosocial stress was used as a dichotomous variable, but duration of stress was also estimated in minutes in order to roughly compare stress burden between groups. Emotions associated with hypoglycemia were not classified into the psychosocial stress category, but hypoglycemia was regarded as a separate entity of physiological stress.
Because the recording device contained an accelerometer, visualization, relative quantification, and exact time recording of physical activity were possible. In this study, being physically active was defined as recorded movement and/or a diary entry by the patient describing physical activity, along with an increase in heart rate of >40% from resting heart rate, and a HR >100 bpm. The sum of physically active time (min) was recorded for each participant. A cutoff at 100 min/registration day was used for the dichotomous variable "physical activity."
Exact time for going to bed and for awakening was obligatory, as well as an estimation of the subjective quality of sleep. "Impaired sleep" (dichotomous variable) was used as a stress category in the analysis.
The participants were additionally instructed to write down the amount of ingested beverages containing caffeine (coffee, tea, coke, or other energizers). The amounts were converted to caffeine equivalent doses. Participants were also asked to note medication taken within 48 hr before start and during the registration day.
Women and girls after the onset of menarche were asked to estimate whether they were within two days of ovulation and to note the date of the first day of their last menstruation.
## | cortisol sampling
Cortisol was collected with the oral swab method (SalivaBio ® , Salimetrics ® LLC). Patients were instructed to abstain from eating, drinking, and toothbrushing one hour before sampling. Samples were kept in a refrigerator until returned to the laboratory where the swabs were centrifuged at 700 g for 10 min to recover saliva. Samples were kept frozen at −20°C until analyzed by competitive enzyme-linked immunoassay from Salimetrics ® LLC, at the Department of Clinical Chemistry, Linköping University . Each participant collected a series of three salivary samples at home: an evening sample, collected within 1 hr before going to bed; a morning sample, collected directly at awakening; and a second morning sample, collected 30 min after the first morning sample in order to calculate the cortisol awakening response (CAR). Cortisol amplitude was calculated as the difference between the highest value of the two morning samples and the evening cortisol. We preferred the term "cortisol amplitude" to "diurnal cortisol slope" which is often used, although under various definitions , because the evening salivary sample preceded the morning samples in our study. Especially the morning samples are known to be sensitive to correct timing of the sampling. Adequate collection timing in this context was assured by (a) a precise entry of time in the diary, (b) pressing the event button on the recording device at the time of collecting the salivary samples, and (c) comparing given times with movement recordings on the ECG registration. Only if there was a congruency between either exact time entries in the diary or event entries in the ECG with the movement pattern and increased heart rate (indicating awakening) were the morning samples included in the analysis. Based on this strict selection, 83% of the patients had acceptable cortisol samples and were included in the analysis (167 out of 201 individuals). Individuals with diabetes had a lower rate of successful sampling than controls (80% versus 88%), mostly due to low glucose levels in the morning that impeded cortisol sampling in some cases. Children had a lower rate of successful sampling than adults (80% versus 91%).
## | glucose monitoring
If a participant used a subcutaneous continuous glucose monitoring device (CGM) as a tool for their everyday management of diabetes, and they agreed to share the data, these data were used. Otherwise, participants were asked to supply all manually acquired glucose values from the memory of their testing device or any other form of glucose recorder, or diary, 7 days back before the start of the ECG registration.
Half of the group consisting of children and adolescents had CGM data, whereas only approximately one fourth of the adults had CGM.
Hypoglycemia was defined as a glucose below 3.5 mmol/L if associated with typical symptoms noted in the diary. If asymptomatic, a cutoff at 3 mmol/L was used, and on CGM, two measurements at least 15 min apart were required. These cutoffs are slightly stricter compared to a clinical setting where a glucose below 3.9 mmol/L is cause to take action, but were based on levels when hypoglycemia might influence HRV.
Mean glucose (mmol/L) preceding the registration was calculated seven days back. CGM rendered between 168 and 2 016 entries/7 days, depending on sampling rate. Mean number of manual entries/7 days was 45 (range 8-121).
## Hba1c values (mmol/mol, international federation for clinical
Chemistry and Laboratory Medicine, IFCC) were derived from the patients' records. The measurement nearest in time to the registration, taken during a routine clinical visit, was chosen.
## | statistics
Mean values (SD) and medians (range) were used to describe the continuous variables for the four groups (consisting of diabetes and controls, subdivided into children and adults). Frequencies and percentages were used to describe the categorical variables. The
Mann-Whitney U test or the Fisher exact test was used for comparisons between groups in Tables 1-3. p < .05 was considered statistically significant, but p-values in descriptive and not inferential purposes. Continuous variables were transformed to natural logarithms when appropriate (high-frequency
[HF] and low-frequency [LF] power and cortisol parameters), allowing analyses of variance (ANOVAs) and analysis of covariance (ANCOVA) to be used. When HRV constituted the dependent variable, all statistical tests were corrected for heart rate (HR)and/ or age when appropriate. When cortisol constituted the dependent variable, all statistical tests were corrected for sex and age. Individuals experiencing hypoglycemia preceding a cortisol sample and women in ovulation phase (including all fertile women who were not able to provide a reliable date for their last menstruation and therefore to be safely said to be non-ovulating) were excluded from other analysis of cortisol (see Sections 2.2.1, 2.2.3, and . Non-parametric tests (Wilcoxon test) were used for a sub-analysis in Section 2.1.3.
All statistical analysis was made using the program Dell Statistica, version 13 (Dell Inc.).
## | re sults
No differences in baseline characteristics were found between participants with diabetes and controls in the children/adolescent group, whereas adult participants with diabetes were slightly older than controls with a tendency of being more physically inactive . Raw data for HRV and cortisol are shown in .
## | heart rate variability
Adults with diabetes had significantly lower time domain and frequency domain HRV parameters than controls . HRV represented by the natural logarithm of the high-frequency band (HF) during sleep, Ln(PowerHF Sleep), was significantly reduced in the diabetes group also when corrected for age, sex, and physical activity (p = .022). There was no significant difference in HRV between groups in children (p = .424).
## | hrv and constitutional factors
Heart rate variability correlated with resting heart rate (HR) at the corresponding night (p < .001). HR, in turn, correlated with anthropomorphic data in children (length). Overweight did not correlate significantly with HRV when corrected for age, since the prevalence of overweight was more than twice as high in 35-to 60-year-old subjects (57%) compared with other age groups. There was no difference between sexes after correction for HR and age. Menstrual cycle as in ovulation did not affect HRV (as opposed to cortisol, see 2.2.1 and . HRV decreased with age, both in healthy controls and in individuals with diabetes (p < .001).
## | hrv and metabolic factors
The duration of disease correlated with decreasing HRV in persons with diabetes, regardless of the effect of aging (p = .033). Individuals with a high mean glucose level at short term (≥10 mmol/L) tended to have lower HRV parameters than those with lower mean glucose levels, but this was only significant in adults (p = .012; .
Children 12 years or younger (n = 25) showed no correlation at all between HRV and HbA1c levels. Adolescents on the one hand and adults on the other hand differed significantly (p = .016) in their respective relationships between HRV and HbA1c, adults with HbA1c >55 mmol/mol displaying lower HRV than those with HbA1c <55mmol/mol, and adolescents showing an opposite trend.
## | hrv and possible confounders
Hypoglycemia did not significantly impair whole night HRV on a group level . Evaluating individuals with hypoglycemia during the night (n = 12) and comparing the lowest hourly measurement of HRV during hypoglycemia with whole night mean HRV, the variability was shown to be significantly reduced during hypoglycemia (p = .002, Wilcoxon test). The heart rate during the hour of hypoglycemia, however, was significantly higher compared with the mean heart rate of the night (p = .002, Wilcoxon test).
Psychosocial stress significantly reduced HRV (p < .05, , in both healthy controls and in the diabetes group. Controls and individuals with diabetes experienced similar amounts of psychosocial stress, according to diary entries and episodes counted in minutes .
Children reported less stress than adults (p < .001), and women had a higher sum of stress minutes than men (p = .028). Subanalysis of different stress categories showed that the presence of occupational or school stress was most common and resulted in significantly lower HRV compared with stress neutral subjects (p = .011). Subjective experience of impaired sleep during the registration night, regardless of objective sleeping hours, also correlated significantly with lower HRV (p = .004), although not in healthy control children.
HRV was significantly higher in children/adolescents who were physically active more than 100 min/registration day than in those who were not (p = .010).
Being on thyroid substitution with stable hormone levels or being on maintenance therapy with SSRI, under the condition that the depression itself was in remission judged by the patient herself, did not affect HRV (studied only in a small subgroup of 36-to 60-year-old women). Minor complaints like taking pain killers for a headache or menstrual pain, medication for allergy or for well-controlled asthma, or having a mild cold/minor infection did not seem to affect HRV (these factors were all grouped together and tested against subjects who reported none of these factors).
Caffeine intake did not influence whole night HRV. Season had no impact on either HRV or cortisol values.
## | cortisol
Individuals with diabetes had significantly higher cortisol levels in the evening than controls (p = .010), but there was no significant difference between groups in CAR levels or in cortisol amplitude.
## | cortisol and gender
Up to the age of 12, there was no difference in cortisol levels between the sexes. Female adolescents and fertile women, however, had higher and more variable levels of cortisol than male adolescents and adult males (p = .010). Evening cortisol , morning cortisol 30 min after awakening, and the cortisol amplitude were all significantly elevated (p < .001 and p = .013) in females who might be ovulating around the time of the registration day (n = 16, 10 T1DM, six controls), as compared to both non-ovulating females and males.
## | cortisol and sleep
Longer sleep (particularly in adolescents) lowered CAR (p = .002). Abbreviations: SD, standard deviation. a Note that "Impaired sleep" partially overlaps with psychosocial stress.
## Ta b l e 2 subject details from diary and registration on registration day
## | cortisol and glucose
A hypoglycemia preceding, but not too close to (>3 hr), the time point of sampling of saliva was found to give significantly higher cortisol levels as compared to euglycemic diabetic individuals .
Thus, evening cortisol levels were affected by hypoglycemia the previous day (p = .001), but not in the same evening, and CAR was affected by early night hypoglycemias (p = .020), but not late-night hypoglycemias.
Mean glucose levels above 10 mmol/L during the preceding 7 days to the registration gave higher basal cortisol levels than lower blood glucose levels (p = .044; . Ovulating women with higher cortisol levels, however, had slightly lower glucose levels than non-ovulating women (8.3 versus 9.8 mmol/L, p = .18, not significant).
## Ta b l e 3 results from ambulatory ecg and salivary cortisol measurements
## | cortisol and psychosocial stress
The cortisol amplitude was significantly lower in adult healthy control subjects who experienced some sort of stress during the registration day as compared to those who had made no note of stress in their diary (p = .013). None of the cortisol markers, however, correlated with stress in the diabetes group. The difference in cortisol amplitude between diabetics and controls was significant (p = .030; .
## | d iscuss i on
To summarize the present work, HRV during sleep was found to decrease with older age, longer diabetes duration, higher mean glucose levels (in adults), physical inactivity ( F I G U R E 2 Effect of psychosocial stress on HRV and cortisol Means and 95% confidence intervals shown. Comparison of (a) HRV (LnHFPower Sleep) and (b) cortisol (LnCortisol evening) between two groups: "Neutral": individuals who reported no stress in their diary and who had acceptable subjective sleep quality and "Stress": individuals who reported some sort of psychosocial stress and/or reported impaired sleep quality. Ovulating women and individuals experiencing a hypoglycemia were excluded from analysis (see . Effects are displayed for children and adults, diabetes, and controls separately. Observations: Total effect of psychosocial stress on HRV was significant (p < .05; upper panel), but only healthy adults differed significantly in cortisol amplitude between individuals experiencing stress and those who did not (p = .013, lower panel) was reduced with the presence of perceived psychosocial stress, but only in adult healthy controls, not in individuals with diabetes.
These results can be useful when endeavoring to standardize a method for diagnosing CAN from HRV, avoiding confounders in the process.
A 24-hr registration of ECG has the advantage over short-term ECG containing extensive and complete data over a long time period.
In analogy with 24-hr blood pressure examinations, the long-term ECG registrations are made in the patients' homes or in their everyday environment. Especially when studying the autonomous nervous system, it is important to avoid the "white coat" effect which might bias the short-term measurements of both adults and children in a hospital setting.
Examination conditions over all are less well controlled, however, in ambulatory examinations. Which exact conditions that must be controlled and what restrictions that are to be imposed on the patient during a 24-hr ECG registration in order to obtain interpretable results of HRV is yet to be specified. The present study addressed some of these issues.
For a diagnostic method to be useful in everyday practice, especially if it were to be chosen as a screening tool, it is essential that it should be simple to apply and that it does not impose too many restrictions on the examined person's everyday life, nor limit too extensively the groups of patients possible to examine. This is the reason patients with stable thyroid substitution therapy were permitted to participate in the study. It is also why we permitted the intake of caffeine during the registration day and why the exclusion criteria for being on SSRI or beta-stimulants were limited to patients with active disease, whereas maintenance therapy was allowed.
Thyroid disease is a common comorbidity of diabetes and has in itself been reported to lower HRV. This renders it as a possible confounder when studying CAN.
In relation to age and gender, stable thyroid substitution therapy did not seem to affect HRV significantly in our material. More data especially in the younger age groups are needed, however.
Psychiatric illness as in depressive and anxiety disorders is on the increase in Sweden.
There are indications that clinical depression is overrepresented in individuals with diabetes as compared to a population without chronic disease. While there is evidence that active depression significantly lowers HRV, making it unsuitable to attempt to diagnose CAN at such a point of time in the patient's life, it is unclear whether this effect remains after remission of the disease. A substantial number of patients will remain on SSRI medication for several years after their actual depressive episode. Are they inept for assessment of HRV? Again, it was only possible to analyze a subgroup of patients, which however did not differ in HRV from their peers without SSRI medication.
A number of factors potentially interfering with the HRV analysis will mainly revert to a strong inverse correlation between HRV and HR. Small body size, being on medication such as beta-stimulators for asthma or stimulants for neuropsychiatric disorders, caffeine intake, hypoglycemia with adrenal counter regulation, infection, pain, physical activity, and stress will all increase HR. Factors like nicotine and alcohol intake which also increase HR were not accounted for in this study. Studying nighttime HRV, which was done in the present work, with a resting HR, probably somewhat attenuates the impact of several of the above factors, yet some kind of correction for HR is highly recommended.
It was relatively reassuring, however, in this study, to observe that caffeine intake did not affect whole night HRV parameters, nor did minor complaints of having a cold, transient pain, allergy problems, or well-controlled asthma.
Hypoglycemia has been shown to lower HRV, at least in the short term. There are some indications that hypoglycemic stress over time in an individual might also reduce HRV in the long term. In our material, even if hypoglycemia affected HRV (mainly by slightly elevated HR) in the short term (within the hour), it did not have a significant effect on whole night HRV. As a precaution, however, HRV should probably not be analyzed in patients experiencing long hypoglycemia before or during the registration. Physical activity did significantly impact HRV during the night in young individuals, which is in concordance with previously published data on the beneficial effect of physical activity on HRV. As a well-documented general recommendation, but also in the context of interpreting HRV, and diagnosing CAN, patients should therefore be encouraged to avoid physical inactivity.
Self-reported psychosocial stress, as well as subjective poor sleep quality, correlated significantly with reduced HRV, even after correction for heart rate. This is in line with previous research on stress and HRV, but has not been demonstrated in individuals with diabetes before. Psychosocial stress should therefore be considered a possible confounder when diagnosing CAN from HRV. To control for psychosocial stress by classifying or grading its effects is difficult even when using questionnaires or structural interviews,
however. An objective biological marker of stress would therefore be an attractive alternative.
While salivary cortisol is an extensively documented method and frequently used in stress research, its pitfalls, especially concerning the cortisol awakening response (CAR), are also well known. Compliance with the timing of sampling is difficult, leading to exclusion of patients from the analysis. Although we took great pains to minimize sampling errors (see Section 1.4), the CAR measurement still had greater variations and was less useful than the evening cortisol and cortisol amplitude measurements. As an example, long sleepers who rose late in the morning (mostly adolescents) had lower CAR than their peers who woke up earlier. This is most probably due to partial awakenings and snoozing prior to collecting the morning salivary samples; that is, the actual morning peak of cortisol occurred before the conscious awakening.
Because ovulation proved to have a pronounced effect on both morning and evening cortisol levels, studying fertile women will add a methodological difficulty to the interpretation of salivary cortisol, in having to account for each woman's stage of the menstrual cycle. Adolescent girls are particularly difficult to assess since they are known to have scarce and irregular ovulations around menarche.
One study indicated that HRV, too, might be affected by ovulation, but nothing supported this theory in the present study.
Another study suggested that midnight cortisol might have a seasonal variation that should be taken into account.
However, we found no correlation in this study between cortisol and the month or season the sampling was made.
In concordance with previous studies, we found the amplitude of cortisol (comparable to the diurnal cortisol slope to correlate with self-reported stress in the healthy adult population, but no such correlation could be seen in the diabetes group.
Evening cortisol was higher in individuals with diabetes than in healthy controls, but psychosocial stress did not seem to be a plausible explanation for this.
## | li m itati o n s
Study participants in the children/adolescent group consisted of individuals with better than the average HbA1c levels (see Section 1.1). This might explain why we found no HRV reduction related to metabolic factors in this group compared with the control group. This is gratifying in itself since it probably translates in a low prevalence of CAN in this group, although in the research setting, it entails that it will take longer to detect early signs of CAN in this cohort. Children and adolescents also reported significantly less stress than adults which weakened the power in the correlation tests between HRV and psychosocial stress in this age group. The method of subjectively reporting stress in a diary has its limitations and might be less suitable for children than adults, even if aided by their parents to fill in the diary. The fact that all four groups (children and adults, T1DM, and controls) showed the same trend and contributed to the significant overall effect of psychosocial stress on HRV (see strengthens the connection, however.
## | con clus ion
Psychosocial stress seems to be a factor to account for in the context of screening for CAN in individuals with T1DM by assessing HRV from 24-hr ECGs. Cortisol levels are, however, of limited use when assessing psychosocial stress in individuals with diabetes, mainly because of an overriding effect of the physiological metabolic stress of T1DM itself over psychosocial stress.
## Ack n owled g m ents
The authors thank Joanna Webb, MD, PhD, for valuable comments on the manuscript.
## Co n fli c t o f i nte r e s t
The authors have no conflicts of interest to declare.
## Auth o r co ntr i b uti
|
Barriers and Facilitators to the International Implementation of Standardized Outcome Measures in Clinical Cleft Practice
Objective: To identify barriers and facilitators to international implementation of a prospective system for standardized outcomes measurement in cleft care.Design: Cleft teams that have implemented the International Consortium for Health Outcomes Measurement Standard Set for cleft care were invited to participate in this 2-part qualitative study: (1) an exploratory survey among clinicians, health information technology professionals, and project coordinators, and (2) semistructured interviews of project leads. Thematic content analysis was performed, with organization of themes according to the dimensions of the reach, effectiveness, adoption, implementation and maintenance (RE-AIM) framework: reach, effectiveness, adoption, implementation, and maintenance.Results: Four cleft teams in Europe and North America participated in this study. Thirteen participants completed exploratory questionnaires and 5 interviewees participated in follow-up interviews. Survey responses and thematic content analysis revealed common facilitators and barriers to implementation at all sites. Teams reach patients either via email or during the clinic visit to capture patient-reported outcomes. Adopting routine data collection is enhanced by aligning priorities at the organizational and cleft team level. Streamlining workflows and developing an efficient data collection platform are necessary early on, followed by pilot testing or stepwise implementation. Regular meetings and financial resources are crucial for implementing, sustaining, analyzing collected data, and providing feedback to health care professionals and patients. Fostering patient-centered care was articulated as a positive outcome, whereas time presented challenges across all RE-AIM dimensions. Conclusions: Identified themes can inform ongoing implementation efforts. Intentionally investing time to lay a sound foundation early on will benefit every phase of implementation and help overcome barriers such as lack of support or motivation.
# Introduction
The use of various disease-specific outcome measures to capture what truly matters to patients is of increasing importance in daily clinical practice. Outcome measures can be used to enhance patient-centered care and evaluate treatment effects. To facilitate the measurement of cleft-specific outcomes in clinical practice, the International Consortium for Health Outcomes Measurement (ICHOM) convened a Working Group of cleft experts including clinicians from various specialties, patients and parents, and academicians to establish international consensus on the outcomes that should be measured routinely as a standard part of cleft care. Emphasis was placed on including clinical indicators for all relevant disciplines and patient-reported outcome measures (PROMs) to incorporate patient and parent perspectives. The result was a holistic, patient-centered Standard Set of measures and guidelines for prospective data collection over the course of care, from birth to young adulthood . Within the Standard Set, satisfaction with appearance, speech function, psychosocial function, oral health, breathing, eating, and drinking are assessed by PROMs (CLEFT-Q scales, Nasal Obstruction Symptom Evaluation (NOSE) questionnaire, and Child Oral Health Impact Profile -Oral Symptoms Scale (COHIP-OSS)). Examples of clinical measures are tone audiometry for the assessment of hearing, Percent Consonants Correct for speech assessment, and screening for velopharyngeal incompetence. Recommended time points for collection of these measures are 5 years (only clinical measures), 8 years, 12 years, and 22 years of age (International Consortium of Health Outcomes Measurement, 2020).
The ICHOM Standard Set for the comprehensive appraisal of cleft care (hereafter, "Standard Set") was designed for broad implementation, internationally and across cultures. Over the past 4 years, 4 cleft teams in North America and Europe have implemented Standard Set collection in their routine clinical practice, and implementation is ongoing at multiple other institutions. Collected outcome data are being used toward quality improvement (QI) efforts, research, and inter-center collaborations to identify and disseminate "best practices". These endeavors are of special importance in cleft care, since research has shown that treatment protocols and quality of care vary widely throughout the world.
The work of these pilot sites is important; however, before meaningful outcome comparisons can be made, widespread adoption and implementation of the Standard Set by more cleft centers is needed. Although many teams are keen to adopt the Standard Set, implementation is not easy. Many cleft teams are cautious about the myriad of challenges and obstacles that they will face. Factors hampering implementation efforts include lacking a defined strategy or a clear understanding of conditions that promote or hinder routine outcome measurement.
Following their collaboration with ICHOM to develop the Standard Set, 4 cleft teams including Boston Children's Hospital, Duke Children's Hospital, Erasmus University Medical Center, and Karolinska University Hospital served as pilots for implementing the Standard Set in clinical practice. Their experiences can help inform ongoing implementation endeavors of other cleft teams. The comprehensive evaluation framework reach, effectiveness, adoption, implementation and maintenance (RE-AIM) is often used to evaluate implementation of an intervention or health care program focusing on 5 dimensions: reach, effectiveness, adoption, implementation, and maintenance. The purpose of this investigation is to use the RE-AIM framework to identify facilitators and barriers for implementing the ICHOM Standard Set for cleft care in routine clinical practice, based on the experiences of 4 pilot centers.
# Methods
This study was conducted in 2 phases, beginning with an exploratory survey followed by in-depth interviews to understand the different centers' experiences implementing the Standard Set. In this study, implementation was defined as "the continuous process of actively measuring, collecting, and analyzing outcomes according to the Standard Set in clinical practice". Participants were recruited through purposive sampling from the authors' personal networks, thus ensuring a diversity of stakeholders who could provide rich context and details regarding the implementation of the Standard Set by their teams. These stakeholders included clinicians, team coordinators, administrative personnel, IT professionals, project coordinators, and managers. Because the aim of this study was to provide an overview of facilitators and barriers from a health care provider's perspective, we decided not to recruit patients and families. The pilot sites invited to participate included Boston Children's Hospital, Duke Children's Hospital, Erasmus University Medical Center, and Karolinska University Hospital. Informed consent prior to the survey or interview was provided by all participants. This qualitative analysis of the facilitators and barriers to implementation was designated by the institutional review board as exempt research (MEC-2020-0343).
## Surveys and interviews
A preliminary exploratory survey was constructed based upon the dimensions present in the RE-AIM framework: reach, effectiveness, adoption, implementation, and maintenance; Online Appendix A). The comprehensive evaluation framework RE-AIM is often used to evaluate implementation of an intervention or health care program. Open-ended questions allowed each participant to expound on the implementation process and corresponding facilitators and barriers. The survey also included questions regarding numerical data such as response rates for the dimensions of reach and adoption. The survey was sent via email to all eligible participants followed by 2 reminders at biweekly intervals. Data collection for the exploratory survey took place between March 2, 2020, and April 6, 2020. Responses were transcoded according to overarching themes and tallied to discover what participants deemed the most important facilitators and barriers. Survey responses were described in frequencies of verbalization (n). Because survey respondents were able to name multiple factors in one answer, the total number of verbalizations could outraise the number of participants. The responses were used to elucidate relevant topics to include in subsequent in-depth interviews.
Following the completion of their exploratory surveys, the cleft team leaders or coordinators from each site were invited for in-depth, semistructured interviews to further explore various dimensions of implementation. Two researchers (I.A. and J.P.R.) conducted the interviews between April 3, 2020, and April 8, 2020. The researchers performing the interviews were not involved in the implementation process.
An interview guide (Online Appendix B) ensured the same questions were asked uniformly of all interviewees, but interviewees were allowed to follow their train of thought and bring up any issues that came to mind. Interviews were conducted in English, and all interviewers and interviewees were fluent in English; however, since the native language of some participants was different from English, they were offered the opportunity to add specific words or sentences in their own language to more accurately express feelings and perspectives. If needed, these parts could be separately translated by 2 additional objective researchers (Dutch and Swedish native speakers) with a good understanding of the English language.
Interviews were audio-recorded and transcribed verbatim in English using NVivo 12 Pro Software for Windows (QSR International, 2020). Thematic content analysis was performed by a main coder (I.A.), then reviewed by a second coder (J.P.R) who checked that transcripts were accurate and appropriately coded, and that no sections were missed during analysis . All coded themes were then grouped according to the RE-AIM dimensions, and appropriately subcoded.
# Results
Twenty participants were invited to complete the exploratory survey; 15 from Erasmus University Medical Center, 1 from Duke Children's Hospital, 1 from Boston Children's Hospital, and 3 from Karolinska University Hospital. Completion rate was 65% (n ¼ 13). Five respondents were eligible for in-depth interviews. Every pilot center was represented by at least one interviewee, and one interviewee provided feedback on behalf of 2 centers, since he has been the implementation lead at both centers at different points in time. Interview duration ranged between 47 and 122 minutes. Survey respondent and interviewee characteristics are described in.
Findings from the survey and in-depth interviews are discussed per RE-AIM dimension below.
## Reach
To engage patients in providing PROMs, 3 different approaches were used. One center started by sending paper questionnaires with appointment letters to patients' homes. Due to the amount of work (mailing questionnaires, sorting them, entering data in a digital system, storing paper forms), they switched to inviting patients to complete questionnaires on Proportion of practices and individuals that adopted the program Facilitators and barriers to reach adoption of the Standard Set among individuals involved in cleft care (eg, clinicians, organization, leadership, patients) Implementation Extent to which the program is implemented as intended Facilitators and barriers in implementing the Standard Set in clinical cleft care as planned Maintenance
Extent to which the program is sustained over time Activities executed to sustain the (implementation of the) Standard Set over time an iPad while waiting for their clinic appointment. Teams using the in-clinic iPad approach reached response rates of 85% to 99%. However, interviewees articulated that it was sometimes noted by clinicians that some patients and parents felt uncomfortable thinking about their appearance while surrounded by others in a waiting room. These concerns made one team change to a third approach of sending questionnaires, including information on how answers will be used for clinical care, by email a few days before the clinic visit so patients could answer in a quiet, private environment. Teams using the latter approach reached 75% to 85% of patients; some could not be reached due to incorrect or missing email addresses, encountered most often for 22-year-olds (as a result of moving from their family home, large time gap since last visit, and switching to their own email address, from that of their parents). The latter was verbalized 8 times by survey respondents as a barrier in reaching patients for PROM collection. Notably, interviewees mentioned that some patients and parents shared negative reactions about unsolicited emails with the team members. At the end of implementation, 2 teams used emailed invitations to complete PROMs at home, and 2 used an iPad to complete PROMs during the clinic visit.
## Effectiveness
Interviewees mentioned the ultimate goal of comparing outcomes is not yet possible as individual centers need to reach more robust levels of data first. However, other effects of implementing the Standard Set in routine clinical practice became visible.
Positive effects. Survey respondents most frequently answered that the ability to quickly assess patient's well-being (n ¼ 8) is the main positive outcome of using the Standard Set.
Interviewees and survey respondents (n ¼ 5) added that using PROMs enables them to plan ahead of the clinic visit (n ¼ 5) and provides a launching point for more focused and intentional discussions (n ¼ 5). As a result, interviewees felt that using PROMs routinely fosters connection between patient and team. Additionally, interviewees mentioned that the use of PROMs gives the parents an opportunity to better prepare for the visit together with their child. Two illustrative quotes about the positive effects of using the Standard Set are below: Interviewee # 2: So, I think it [use of PROMs] is great . . . for making the parents . . . more aware of what the concerns are that the children might have, and it makes our work much easier because we can focus on those [concerns] and not miss out on them.
Interviewee # 3: The psychologists say, 'Why haven't we done this [collecting outcome data] before? This is so useful and we're now reaching families, and parents who are struggling, and kids who are struggling, and we never asked these questions [CLEFT-Q psychosocial scales], they never raised it until it was too late.' So, I think there is a true benefit of just using the set.
Furthermore, interviewees reported that introduction of the Standard Set has helped foster team solidarity by generating a common goal and giving the team an opportunity to selfevaluate.
Negative effects. Survey respondents listed time (n ¼ 7) and extra work (n ¼ 4) as negative effects of using the Standard Set.
Interviewees were more nuanced about these limiting factors:
Interviewee # 4: In the beginning we had some people argue that [collecting outcomes] costs a lot of extra time but once you have everything up and running, and you're used to it, . . . , it really fits. You find a way that it fits in the workflow of your team, we don't experience it as a burden or extra administration or those kinds of things.
## Adoption
Survey respondents emphasized that hospital leadership (n ¼ 3) and cleft team coordinators (n ¼ 3) are crucial stakeholders in successful adoption of the Standard Set. Motivation (n ¼ 4) was most frequently mentioned by survey respondents as a facilitator for adoption, and time as a barrier (n ¼ 4). Three themes were identified.
Theme 1: Creating importance and urgency. The hospital boards of all 4 centers were supportive of the initiative, and interviewees felt that lack of leadership support would hinder widespread implementation. To enhance adoption, interviewees advised teams to get on the hospital board's agenda and explain the value of implementing the Standard Set, for example, to improve quality of care by having your own local outcome registry or positioning cleft teams to benchmark (inter)nationally. An additional advice of interviewees was to use cases from the literature and the experiences of pilot institutions to support this process. Interviewees also advised starting with a simplified implementation collecting only specific parts of the Standard Set, to show the feasibility, benefit, and value in expanding data collection. Demonstrating importance was not only found useful to garner commitment from leadership but also to convince other members of the cleft team, another key stakeholder, to adopt the Standard Set:
Interviewee # 5: I think the main person who we're really talking about is the main team director, but it could also be the chair of a department or something like that. In any case, that person needs to convey to the team that this [measuring outcomes] is important, that this is creating a new sense of normal . . . , a new standard operating procedure. That this is not really voluntary, but this is what we as a team want to do, it fulfills our mission . . . So, you have to create a sense of urgency.
Informing patients and parents about the importance of the project varied by institution. When data collection was wrapped into a broader research program, patients underwent informed consent at the beginning of their clinic visit. If collection of Standard Set data was integrated into routine clinical practice for the purpose of QI, this advancement was announced to patients through newsletters, informational meetings, and on cleft team and/or scientific society websites. An interviewee articulated how prioritizing patient engagement in decision-making increased adoption: Interviewee # 3: I think the fact that we ask questions from them [patients] and that we do something with these questions, increases the connection between the patient and the team, knowing that we look into it, that we care, that we listen to what they're saying, and try to do something with it.
Theme 2: Aligning motivation and priorities through regular meetings. Interviewees reported that due to the multidisciplinary nature of cleft care, it is essential to ensure every specialty buys into measuring outcomes routinely. In addition, interviewees stated that interdisciplinary friction points should be discussed and incentives stated clearly, so the project will not be jeopardized later on because of competing priorities. All 4 participating centers held regular meetings to discuss feelings, visions, thoughts, challenges, and organizational matters regarding implementation of the Standard Set to keep everyone engaged. The most frequently discussed topics were how to organize different data collection workflows in clinical practice, what impact PROM questions might have on the child and how to deal with the answers, and what will ultimately happen with the data. Regular meetings also provided opportunities to build an overarching implementation strategy: Theme 3: Securing resources. Interviewees articulated that teams that want to implement the Standard Set but lack financial resources and time will face barriers implementing health information technology (HIT) solutions and will more likely succeed by starting outcomes collection on paper. Two of the 4 hospitals partially financed their implementation projects through grants. Interviewees mentioned that Duke and Erasmus are now providing open-access platforms in collaborative networks, to decrease the startup time for teams wanting to adopt and implement the Standard Set.
## Implementation
The Standard Set was implemented as planned at all 4 centers, but the initial implementation period was longer than anticipated (ranging between 6 and 24 months). Survey respondents most frequently (n ¼ 7) answered that approximately 10 to 15 people were involved in the implementation team. One respondent reported a number of over 40 people. Interviewees articulated that a small core implementation team was preferred over a larger group because communication problems and staff turnover could disrupt the process. Crucial members of the implementation team included the clinical lead (n ¼ 8), a HIT lead (n ¼ 5), and a clinic coordinator (n ¼ 9) or specialized nurse (n ¼ 5). These members did not differ by teams. The participants felt that the implementation lead could come from any specialty, as long as they are enthusiastic and dedicated, familiar with workflows, and able to build good relationships. Representatives of every specialty could be invited to the team and HIT personnel were mostly included by consultation. Three unique themes were identified.
Theme 1: Reorganizing the clinical workflows. Interviewees frequently mentioned that evaluating and transforming workflows and clinical visits are important aspects of the implementation phase. Teams started by evaluating how data collection would best fit their current workflow, ensuring all outcomes are collected. The 4 centers already worked as a multi-or interdisciplinary team, making it easier for them to streamline workflows of the various specialties involved. Developing flowcharts of treatment protocols including designated Standard Set outcome time points and measurements was very useful to gain insights on how to seamlessly integrate data collection into the existing workflow. Awareness of the extra time needed for speech and language therapists to perform additional testing, and of possible increase in patient volume for the psychologist was necessary. Furthermore, assessing patient's answers, providing feedback to them, and recording clinical outcomes resulted in an additional 5 minutes on average per clinical visit per patient.
Three teams reported that each specialty records their own clinical outcome measures in their HIT system for tracking outcomes, which interfaces directly with the patient's electronic medical record (EMR) in 2 of the 4 teams. One team mentioned having a dedicated person who collects all outcomes from the clinicians in a standardized form, and then registers them in the system. At all 4 teams, after the completion of PROMs by the patient, both at home as in clinic, the answers were directly stored in the HIT system without the intervention of a person. Scoring algorithms for each PROM were programmed within the HIT systems, and access to both PROMs and clinical outcomes was the same.
Interviewee # 1: I don't think one [way of collecting data] is right or wrong, but there are some pros and cons to each. . . . By doing it in a specialty-specific way, you guarantee that the data quality is pretty good . . . The downside is that in many cases you might get incomplete data because people forget or they get busy in clinic, whereas the benefit of having a research person who is . . . always available is always making sure data is collected. . . . The downside is if they don't have a clinical background, you might have some incorrect data in there.
Theme 2: Developing an efficient HIT system. All 5 interviewees and 7 survey respondents agreed that a HIT platform was an important facilitator that will save time and increase ease of data management while reducing risk of data loss as compared to tracking outcomes using pen and paper. The most frequently mentioned system requirements were easy access, allowing concurrent users of the database system, dealing with versioning, and keeping permanent records of changes made, with easy data extraction for use in QI projects. There were no teams that had HIT systems that automatically extracted outcomes from the EMR. Interviewees advised making the HIT system as compatible as possible with other systems to aid in future data exchange. Furthermore, interviewees found it helpful to get advice from someone who has dealt with this process before to prevent mistakes that can later create barriers.
Theme 3: Pilot testing and stepwise implementation. One team started with pilot testing the complete Standard Set for 3 to 4 patients with different ages and cleft diagnoses per clinic day. This enabled them to explore time requirements per visit and gave them the opportunity to adjust workflows accordingly to solve errors early on, before measuring outcomes for all patients. Another team started with the complete set, but scaled implementation up from one patient per week to all patients to ease into it, improving the process of data collection gradually, and working out friction points. The other 2 teams preferred stepwise implementation, starting with implementing PROMs followed by clinical measures. This allowed them to spend more time developing their HIT system. Interviewee # 3: We decided to go for a pilot phase. I know that different hospitals in the world have chosen different routes, so some have said 'Okay, we are just going to do only the 5-yearolds', for example, or 'We are only going to do the cleft lips for a while'. That's one approach, a choice you need to make . . . . We then said, 'We are going to do the whole set, we want to have all the patients from the beginning'. So, the HIT-system was built for all diagnoses and for all aspects of the set.
## Maintenance
Theme 1: Analyzing and utilizing collected data. In order to maintain momentum, most survey respondents and all interviewees felt that it is important to analyze and use locally collected data early in the process (n ¼ 9). For example, QI projects like analyzing data completion or complication rates facilitated opportunities for improvement and sustain motivation. Also, interviewees articulated that research on outcomes data and measurement instruments can provide insights to improve future iterations of the Standard Set. Most importantly, it was found that sharing early wins with the entire team is a good way to maintain engagement, since decreasing commitment levels over time was recognized as a barrier.
# Discussion
This study applied qualitative methods and the RE-AIM framework in the evaluation of facilitators and barriers to implementation of the ICHOM Standard Set for the comprehensive appraisal of cleft care. Major themes identified included creating importance and urgency, aligning motivation and priorities through regular meetings, and securing resources. The dimension of implementation was characterized by reorganizing clinical workflows and developing efficient HIT systems, followed by pilot testing and stepwise implementation. Although implementing the Standard Set requires extra time and effort, especially in the beginning, interviewees experienced advancements in patient-centered care as a positive outcome. Analyzing and utilizing the data collected in practice could help sustain implementation over time.
Three methods were identified to reach patients and collect PROM data: paper surveys mailed to the patient's home; email surveys prior to clinic visit; and real-time data collection using an iPad during clinical visits. Only the 2 electronic approaches are now used by the pilot centers because paper forms were too labor-intensive with higher risk of losing information. Several studies have shown that patients are more receptive toward electronic collection systems compared to pen and paper for the collection of PROMs; however, no clear comparisons have been made between completing surveys at home or during the clinical visit and how this is viewed by the pediatric patient population. Cultural or societal differences might play a role, since the North American institutions chose the in-clinic iPad approach, while the European centers incorporated emailed invites to complete PROMs at home. Other factors that could influence choice of data collection method is the payment model of the health care system and a patient's travel time to the clinic. Patients will not come to clinic when they do not experience problems or concerns if they have to pay extra for each visit or travel long distances. Missing out on collecting data for these patients could potentially jeopardize a center's outcomes. These factors should be taken into consideration when deciding on the best way to reach patients. Including patient advocacy groups in this decision could be valuable.
Unfortunately, investing in electronic systems for data collection might not always be feasible, due to limited financial or technological resources, or differing organizational priorities. Middle-and low-income countries might especially face these challenges. Currently, 2 large initiatives offer support in these circumstances. The European Reference Network for rare and/or complex craniofacial anomalies and ear-nose-throat disorders aims to pool disease-specific expertise, knowledge, and resources from across Europe to improve quality of care. The network is currently developing a registry for the collection of outcome measurement data for cleft care. This registry will be accessible to all participating centers for the primary purpose of quality control, and outcomes research in the future . Similarly, the ACCQUIREnet collaborative, led by Duke University, makes its REDcap-based implementation available to member institutions that join the network.
Creating importance, and aligning motivation and priorities among team members and leadership is a crucial and universal part of implementing an outcomes measurement framework in clinical practice. This is consistent with recent literature on understanding and overcoming barriers to change, which states that it is important for health care professionals to understand the benefits of changing practices (National Centre for Health . Across various health care settings, implementation was boosted when collection of outcome data is supportive of patient-centered care at an individual patient level, instead of at an aggregated level.
The current study identified a common belief among cleft professionals that implementation of the Standard Set had a positive effect on their team and on patient-centered care. Previous literature reported that patient-clinician communication, clinician's awareness of symptoms, and patient satisfaction can be improved by the use of PROMs, and by reviewing the results with the patient. In addition, a recent study showed that over 80% of children completing the CLEFT-Q scales, representing 9 of the 12 PROMs in the Set, liked answering the questions, and felt it made them understand their condition and feelings better. The fact that the children get something in return (insight in their own well-being, more individualized care) could be a reason for obtaining relatively high response rates in contrast to the reported email survey response rates of 20% to 40% among adults in literature.
Implementation efforts were most constrained by time. Time, as part of resources, was articulated to have an overarching and continuing influence on all dimensions of the RE-AIM framework, especially on adoption and implementation. In general, approximately 5 extra minutes per patient were necessary during clinical visits for the discussion of the PROM results with the patient, registration of clinical outcome data, and in some cases extra speech or audiometry screening. For the latter, planning extra time for speech therapists and audiological consultants might be necessary and coordination with the specific departments regarding other obligations is of considerable importance when implementing the Standard Set. When barriers are not properly addressed due to time constraints, teams might struggle with problems later on, experience setbacks or jeopardize the project. Therefore, intentionally investing time to set the parameters for implementation will benefit every phase and help overcome barriers such as lack of support or motivation.
## Limitations and future directions
A strength of this study is inclusion of 4 cleft centers with different implementation methods from various countries, representing unique cultures and societal habits. However, all 4 centers are located in high-income countries, limiting the generalizability of these findings to low-and middle-income countries (The World Bank, 2020). It is likely that factors influencing change management will not differ profoundly, while differences in financial and technological resources will be more prominent.
Another important factor limiting the generalizability and interpretability of our findings is the fact that there was a sizeable disparity in the number of people per cleft team approached for participation in this study, and that all interviewees represented one discipline, instead of a variety in stakeholders. The first disbalance is caused by a high turnover of personnel involved in the clinical implementation of the Standard Set, resulting in a limited number of eligible patients at 3 sites for completing the exploratory survey. The loss of continuity in personnel was mentioned by these sites as a barrier in implementation resulting in slowing down the process. The second disbalance of interviewing only surgeons is caused by the fact that the implementation efforts were all led by surgeons as project coordinators, and because a relatively large proportion of clinicians within a cleft team has a surgical expertise. Also, health care management and coordinating tasks are often employed by clinicians, since they are familiar with the clinical workflows.
Furthermore, centers who are currently implementing or have abandoned implementation due to problems were outside the scope of this study. Anecdotally, some of these centers experienced a lack of institutional and financial support. The findings of this study can help teams experiencing challenges in their implementation efforts to move forward, as well as serve as starting point for future research by centers struggling with implementation, and by centers in low-and middleincome countries.
Using an extensive open-ended survey as well as the fact that experts were recruited through purposive sampling could have influenced answers, because participants could assume specific information or opinions are already common knowledge for the researchers. The use of open-ended questions was chosen to gather as many different opinions and feelings as possible, since a qualitative study toward implementation of such a specific outcomes set has not yet been performed. Therefore, it was deemed a preliminary exploratory survey was necessary to explore the main themes and directions for the interviews. A possible lack of in-depth information on the survey was addressed by the follow-up semistructured interviews with clinical leads and coordinators.
# Conclusion
The themes identified in this qualitative study may be helpful to other cleft teams that are considering adopting and implementing the Standard Set. Specifically, each team should strive to adequately communicate to all stakeholders the reason for adopting the standard set, seek to align motivation and priorities, and provide frequent communication during the initial phases of implementation. At the organizational level, proper attention must be given to setting up the HIT platform, the implementation effects on workflow and provider burden, and securing resources for sustaining the endeavor. Multisite collaboratives may assist in facilitating implementation. |
COVID-19 outbreak in an elderly care home: Very low vaccine effectiveness and late impact of booster vaccination campaign
a b s t r a c tBackground: Elderly people in long-term care facilities (LTCF) are at higher risk for (severe) COVID-19, yet evidence of vaccine effectiveness (VE) in this population is scarce. In November 2021 (Delta period), a COVID-19 outbreak occurred at a LTCF in the Netherlands, continuing despite measures and booster vaccination campaign. We investigated the outbreak to assess VE of primary COVID-19 vaccination against SARS-CoV-2 infection and mortality, and to describe the impact of the booster vaccination. Methods: We calculated attack rate (AR) and case fatality (CF) per vaccination status (unvaccinated, primarily vaccinated and boostered). We calculated VE -at on average 6 months after vaccination -as 1risk ratio (RR) using the crude risk ratio (RR) with 95% confidence intervals (CI) for the association between vaccination status (primary vaccination versus unvaccinated) and outcomes (SARS-CoV-2 infection and mortality < 30 days after testing positive for SARS-CoV-2). Results: The overall AR was 67% (70/105). CF was 33% (2/6) among unvaccinated cases, 12% among primarily vaccinated (7/58) and 0% (0/5) among boostered. The VE of primary vaccination was 17% (95% CI À28%; 46%) against SARS-CoV-2 infection and 70% (95% CI À44%; 96%) against mortality. Among boostered residents (N = 55), there were 25 cases in the first week after receiving the booster dose, declining to 5 in the second and none in the third week. Conclusion: VE of primary vaccination in residents of LTCF was very low against SARS-CoV-2 infection and moderate against mortality. There were few cases at 2 weeks after the booster dose and no deaths, despite the presence of susceptible residents. These data are consistent with the positive impact of the booster vaccination in curbing transmission. Timely booster vaccination in residents of LTCF is therefore important. Ewijk), i.hazelhorst@ ggdtwente.nl (E.I. Hazelhorst), [email protected] (S.J.M. Hahné), mirjam.knol@ rivm.nl (M.J. Knol).
# Background
Elderly people in long-term care facilities (LTCF) are at risk for (severe) COVID-19 due to living in a congregated setting, immunosenescence, and pre-existing medical conditions such as dementia, heart-and lung diseases [bib_ref] Increasing risk of breakthrough COVID-19 in outbreaks with high attack rates in..., Suetens [/bib_ref].
Evidence on COVID-19 vaccine effectiveness in residents of LTCF in the Netherlands is scarce. Residents of LTCF are not often admit-ted to hospitals, not tested for SARS-CoV-2 at community testing centres, and accurate data on vaccination coverage in this population is lacking.
Previous research shows substantial waning of vaccine effectiveness against both infection and mortality in resident of LTCF 12 + weeks after primary vaccination, highlighting the frailty of elderly people and the need for continuous vaccine effectiveness monitoring in order to assess the requirement of booster doses in this population [bib_ref] Effectiveness of Pfizer-BioNTech and Moderna Vaccines in Preventing SARS-CoV-2 Infection Among Nursing..., Nanduri [/bib_ref].
In November 2021, a SARS-CoV-2 outbreak occurred at a LTCF in the Netherlands, affecting both staff and residents, and continuing despite measures and the initiation of the booster COVID-19 vaccination campaign. We conducted a retrospective cohort study to describe the outbreak and to assess vaccine effectiveness of the primary COVID-19 vaccination series among residents living at https://doi.org/10.1016/j.vaccine.2022.09.080 0264-410X/Ó 2022 Elsevier Ltd. All rights reserved. the LTCF during the outbreak against SARS-CoV-2 infection and mortality due to COVID-19. In addition, we describe the impact of the booster vaccination campaign that took place at the end of the outbreak.
## The affected long-term care facility
The affected LTCF contained four separate wards -two psychogeriatric and two somatic wards -each with 16 beds with a total of 63 residents. In addition there were three semi-attached housing estates, that offered assisted living to 42, 30 and 16 residents respectively. Residents from the estates lived independently from the wards, but could choose to share the same day-care activities and restaurant facilities as the residents from the wards of the LTCF. Predominantly the residents from the biggest assisted living estate (42 residents) tended to share facilities with residents from the wards. Staff of wards and the individual living estates were different, whereas residents of the other two estates did not socially mix with residents of the wards.
At the time of the outbreak, the LTCF including the three estates consisted 151 residents and approximately 160 staff (such as nurses, facilitators, paramedics, cleaners, kitchen staff).
## Outbreak detection
On 22 November 2021 the first SARS-CoV-2 infection was diagnosed in a fully vaccinated resident of psycho-geriatric ward 1 after having had close contact with a SARS-CoV-2 positive person [fig_ref] Figure 1: Cases of SARS-CoV-2 infections by date of symptom onset [/fig_ref].
All residents and staff of this ward and other close contacts at the LTCF were tested with a polymerase chain rection-test (PCR) as soon as possible and on day 5 after last exposure. Infected staff stayed at home until at least 7 days after onset symptoms and until they were symptom free for at least 24 h. Negative tested staff worked strictly with personal protective equipment but were allowed to work night shifts on other wards due to understaffing. Visitors were allowed if symptom free and with the wearing of medical facemasks. Initially, cases were put in isolation in their rooms, rather than treating the entire ward as potentially infected. However, as more cases came to light at the psycho-geriatric ward 1, it was decided to put the ward into isolation as a group, including staff, on 29 November 2021. Soon after, several staff members who had previously worked on the affected psycho-geriatric ward, developed symptoms. Some had worked night shifts at other wards during the contagious period. Residents of the potentially newly affected wards were tested and put into isolation as well. Daily activities at the LTCF were cancelled, visiting was restricted, and common areas, such as the restaurants, were closed to residents of affected wards. As previously planned, on 6 December 2021 residents of the wards as well as many of those in the assisting living estates, and who had tested negative so far during the outbreak (n = 55), received a COVID-19 vaccine booster dose. Despite these measures, the SARS-CoV-2 outbreak affecting now all wards of the LTCF, also spreading to the biggest living estate (42 residents) but not to the other two smaller estates since residents had not mixed socially with residents from the wards. As a result, all residents of the LTCF and the one affected estate were tested and put into individual isolation, and common areas were shut down for all residents. The outbreak stopped and the last cases left isolation on 22 December 2021.
# Methods
## Definitions
The cohort of the outbreak investigation was defined as residents living on one of the four wards of the LTCF or the biggest assisted living estate (42 residents) between 20 November 2021 (two days before onset symptom of first case) and 3 January 2022 (14 days after last case).
Case finding was done actively and generally followed national protocol by testing all residents of the separate wards of the LTCF every 5 days once the outbreak was established until no further cases were found and isolation/quarantine could be lifted.
A case was defined as a person living at the LTCF who tested positive for SARS-CoV-2 with a Polymerase-Chain-Reaction test (PCR-test) after 22 November 2021 until 3 January 2022.
Asymptomatic cases were defined as cases who did not develop any COVID-19 related symptoms. A mild COVID-19 case was defined as a person with any non-fatal signs or symptoms of COVID-19 who did not need hospitalisation and who did not die within 30 days of testing positive. A severe COVID-19 case was defined as a person with any signs and symptoms of COVID-19 requiring hospitalisation and/or leading to death within 30 days of the positive test. An unvaccinated person did not receive any COVID-19 vaccination at least 14 days before onset symptoms (or test in case of asymptomatic), or had received only one dose < 14 days before. A partially vaccinated person had received only one dose of COVID-19 vaccination at least 14 days before onset symptoms (or test in case of asymptomatic). A fully vaccinated person had received two COVID-19 vaccines at least 14 days before onset symptoms (or test in case of asymptomatic) as part of a primary series, or had received only one dose after a confirmed SARS-CoV-2 infection prior to this outbreak (n = 2). We assume that fully vaccinated residents of whom vaccination dates were unknown (n = 31) were vaccinated at least 14 days before the start of the outbreak. A boostered person had received a booster dose of COVID-19 vaccine at least 7 days before onset symptoms (or test in case of asymptomatic).
## Data collection
Data were collected as part of the routine (public health) surveillance for COVID-19. In the Netherlands, confirmed SARS-CoV-2 infections are notifiable to the regional Public Health Service (PHS). For each case data were obtained on demographics, date of symptoms onset and test, type of ward, hospital admission due to COVID-19, or death within 30 days of testing positive, previous COVID-19 episodes, and vaccination status (vaccine type, number of doses, dates of administration). LTCF staff filled in a protected online form per case which was subsequently submitted to the regional PHS, according to standard PHS protocol.
Missing data regarding cases and data on non-cases were retrospectively collected where possible and collected by the LTCF managers in collaboration with one of the researchers (E.I.H.) from the regional PHS. For each non-case, data were collected on year of birth, sex, ward, date of last negative test, vaccination status, and any previous confirmed COVID-19 episode.
## Sequencing
Whole genome sequencing was performed for 12 PCR-positive swabs; these samples were from all four wards and throughout the whole outbreak period.
## Analyses
We conducted descriptive statistics and calculated attack rates overall and by ward. In addition, we compared (Kruskal-Wallis rank sum test or Fisher's exact test) characteristics by and calculated case fatality per vaccination status (unvaccinated, fully vaccinated and boostered).
We calculated the crude risk ratio (RR) and 95% confidence intervals (CI) for the association between vaccination status (complete primary vaccination series versus unvaccinated) and outcomes (SARS-CoV-2 infection and mortality within 30 days after testing positive for SARS-CoV-2) using epi.2by2 function from the EpiR package. For calculating the RR, residents who became cases (date of onset symptoms or test if asymptomatic) < 7 days after they received a booster dose were still considered as fully vaccinated. Vaccine effectiveness was calculated as 1-RR. Residents with unknown vaccination status were excluded from analyses.
Sensitivity analyses were performed restricting these analyses to residents without a confirmed SARS-CoV-2 infection prior to this outbreak.
Analyses were conducted using R version 4.0.2.
# Ethical considerations
The PHS has legal permission, provided by the Dutch Public Health Act, to process personal information on notifiable diseases for local and national surveillance. Pseudo-anonymised data on non-cases were provided by the LTCF to the regional PHS as part of the legal duties of a PHS in case of an outbreak investigation. This study received medical ethical clearance (EPI-543) by the Centre for Clinical Expertise (KEC) at the RIVM.
# Results
The total cohort consisted of 105 residents, which included residents of the wards and the biggest living estate but excluded the two small living estates [fig_ref] Table 1: Characteristics of residents of the LTCF, the Netherlands, 20 November 2021 to... [/fig_ref]. The residents had a median age of 85 (min 57; max 106) years and 66% was female. Thirteen residents (14%) had confirmed COVID-19 prior to this outbreak. At the start of the outbreak 8/101 (8%) residents were considered unvaccinated, and 93/101 (92%) were considered fully vaccinated. The vaccination status was unknown for 4 residents. All fully vaccinated residents of whom vaccination dates were known (n = 74), received their last dose between 2 March 2021 and 6 July 2021.
On 6 December 2021, 55 (52%) residents -who had not tested positive for SARS-CoV-2 so far during the outbreak -received their booster vaccination with Comirnaty or Spikevax. Determined at the end of outbreak (for non-cases) or at time of infection (for cases) taking into account time since booster vaccination, 63/101 residents (62%) were considered fully vaccinated and 30/101 (30%) boostered.
## Attack rates
From 20 November 2021 to 3 January 2022, 70 cases among residents of the LTCF were identified with dates of symptom onset or test in case of asymptomatic between 22 November 2021 and 16 December 2021, and a peak on 10 December 2021 [fig_ref] Figure 1: Cases of SARS-CoV-2 infections by date of symptom onset [/fig_ref]. The overall attack rate (AR) among residents of the LTCF was 67% (70/105). The ARs were highest for the two wards where the outbreak started first: psycho-geriatric ward 1 with an AR of 94% (15/16) and somatic ward 1 with an AR of 81% (13/16). The AR for psycho-geriatric ward 2 was 56% (9/16) and 47% (7/15) for somatic ward 2. The AR for the assisted living estate was 62% (26/42).
## Genome sequencing
Sequencing was successful for 8/12 samples, showing pangolineage B.1.617.2, subclade AY.43 in all samples, consistent with Delta variant circulating at that time. [fig_ref] Table 2: Characteristics of residents of the LTCF by vaccination status determined at the... [/fig_ref] shows the characteristics of the residents by vaccination status determined at the end of the outbreak (non-cases) or at time of infection determined by date of onset symptom (or test in case of asymptomatic) (cases).
## Characteristics of residents by vaccination status
Of all boostered residents, 8/23 (38%) had a COVID-19 episode prior to this outbreak compared to 1/7 (14%) unvaccinated and 4/62 (6.5%) fully vaccinated (p = 0.002). These reported prior infections had been between March 2020 and February 2021, and one in September 2021.
## Vaccine effectiveness
The vaccine effectiveness of primary vaccination against SARS-CoV-2 infection was 17% (95% CI: À28%; 46%). The vaccine effectiveness against mortality within 30 days after testing SARS-CoV-2 positive was 70% (95%CI À44%; 96%) . Excluding residents with confirmed previous SARS-CoV-2 infections prior to this outbreak did not change the vaccine effectiveness estimates against infection (16%; 95%CI À16; 40%) or mortality (70%; 95% CI À45%; 96%).
The case fatality among unvaccinated cases was 33% (2/6), 12% among fully vaccinated (7/58) and 0% (0/5) among boostered residents [fig_ref] Table 2: Characteristics of residents of the LTCF by vaccination status determined at the... [/fig_ref].
In the first 6 days after introduction of the booster, and when residents were still considered fully vaccinated and included in the calculations of vaccine effectiveness estimates of primary vaccination, there were 25 cases out of the 55 residents that had received their booster dose on 6 December 2021. However, there were relatively few cases in the second week (5 of 30 remaining boostered residents) and zero in the third week after introduction of the booster vaccination [fig_ref] Figure 1: Cases of SARS-CoV-2 infections by date of symptom onset [/fig_ref].
# Discussion
Our outbreak investigation found a very high overall SARS-CoV-2 infection attack rate (67%) in a LTCF despite a high COVID-19 vaccination coverage among the residents (92%) and nonpharmaceutical interventions (NPI). Attack rates varied across wards from 47% to 94%. Data on vaccine effectiveness in residents of LTCF is scarce. Through this outbreak investigation we were able to estimate vaccine effectiveness for primary vaccination in a LTCF population that on average completed their primary vaccination series more than six months prior to this outbreak.
Vaccine effectiveness against SARS-CoV-2 infection of complete primary vaccination was low (17%, 95% CI À28%; 46%), but higher against mortality (70%, 95%CI À44%; 96%).
High attack rates despite high vaccination coverage in a LTCF have been reported previously [bib_ref] Increasing risk of breakthrough COVID-19 in outbreaks with high attack rates in..., Suetens [/bib_ref]. Possible explanations are a high force of infection (through highly infectious case(s) or persistent exposure through a large outbreak), suboptimal adherence to NPI, or an outbreak due to a new SARS-CoV-2 variant associated with decreased vaccine effectiveness. The omicron variant that was discovered in South-Africa end of November 2021 and that is associated with a decrease in vaccine effectiveness, was not yet widely spread in the Netherlands at the time of this outbreak. Whole genome sequencing from infections in cases from different time points in the outbreak and different wards confirmed SARS-CoV-2 Delta variant.
Previous results from household studies -associated with more intense exposure compared to community level -show lower COVID-19 vaccine effectiveness compared to community vaccine effectiveness estimates [bib_ref] Vaccine effectiveness against SARS-CoV-2 transmission to household contacts during dominance of Delta..., De Gier [/bib_ref]. Wards of LTCF could be considered a type of household with more intense exposure, possibly contributing to the observed lower vaccine effectiveness compared to the same age group in the community in the Netherlands: 73% (95%CI 71%; 74%) for age group 70+ years at 6 months after complete primary series with mRNA vaccines.. Research shows vaccination protects against onward transmission of SARS-CoV-2 [bib_ref] Vaccine effectiveness against SARS-CoV-2 transmission to household contacts during dominance of Delta..., De Gier [/bib_ref]. Although the national COVID-19 vaccination turnout among personnel caring for elderly people, including LTCF, was high (80%), we do not have data on vaccination coverage among staff of the LTCF where the outbreak occurred. Unvaccinated and infected staff might have contributed to higher force of infection during the outbreak leading to lower vaccine effectiveness.
Additionally, studies show COVID-19 vaccine effectiveness wanes over time, especially in elderly people [bib_ref] Effectiveness of Pfizer-BioNTech and Moderna Vaccines in Preventing SARS-CoV-2 Infection Among Nursing..., Nanduri [/bib_ref]. Adults living in LTCF were prioritised in the Netherlands for vaccination during the Dutch COVID-19 vaccination campaign that started in January 2021. In our study, the majority (86% of those with known vaccination dates) received their last dose of the primary vaccination series before end of April 2021.
The low vaccine effectiveness of primary vaccination series against infection at (on average) six months after vaccination with a case fatality of 12% (7/58) within 30 days of testing positive, highlight the vulnerability of adults living in LTCF and the need of timely booster vaccinations to increase protection [bib_ref] SARS-CoV-2 Neutralization with BNT162b2 Vaccine Dose 3, Falsey [/bib_ref] [bib_ref] Effectiveness of COVID-19 booster vaccines against covid-19 related symptoms, hospitalisation and death..., Andrews [/bib_ref].
The booster vaccination campaign was rolled-out during this outbreak on 6 December 2021. Research suggests that the effect of booster vaccination in adults 50 + years can be seen 7 days after the booster dose [bib_ref] SARS-CoV-2 Neutralization with BNT162b2 Vaccine Dose 3, Falsey [/bib_ref] [bib_ref] Effectiveness of COVID-19 booster vaccines against covid-19 related symptoms, hospitalisation and death..., Andrews [/bib_ref]. In this outbreak there were still 25 cases among the 55 boostered residents in the first week after their booster dose, with a peak on day 4 with 14 cases. There were relatively few cases in the second week (5 of 30 remaining negatively tested boostered residents) and zero in the third week after introduction of the booster vaccination, despite the fact there were still susceptible residents. These data seem consistent with the positive impact of the booster dose in curbing transmission. Yet optimal protection of a booster vaccination might take longer to develop in residents of LTCF compared to a younger population.
Among the residents receiving a booster vaccination there were more individuals with a confirmed SARS-CoV-2 infection prior to this outbreak. These residents had already gained protection through vaccination as well infection, potentially giving them more protection during this outbreak compared to those without a prior infection. This could lead to a biased selection of residents receiving a booster vaccination -since residents who tested positive during this outbreak were not given a booster dose -and thus overestimation of the role of the booster vaccination in curbing the transmission during this outbreak. Nonetheless, vaccinating residents of LTCF, when vaccination coverage is low and/or time since last vaccination is more than three months, as intervention in case of a SARS-CoV-2 outbreak to curb transmission and prevent morbidity, might be considered. More research is needed, however, to assess the true effect of such an intervention.
# Limitations
There are several limitations to our study. Although the SARS-CoV-2 infection outbreak was large and greatly impacted staff and residents of the LTCF, the sample size for analyses remained small.
We were therefore unable to adjust for potential confounders when estimating vaccine effectiveness for primary vaccination series. Although the cohort can be considered relatively homogenous regarding age, sex, and time since vaccination, there is likely to be residual confounding. For example, we were unable to adjust for or stratify by type of ward as proxy for differences in exposure and unmeasured confounding per ward.
We were unable to estimate vaccine effectiveness for the booster vaccination either against infection or mortality due to small sample size, and the booster being administered during the outbreak leading to survivor bias. Nonetheless, we observed few cases in the second and none in the third week after introduction of the booster despite the presence of susceptible residents, and a low The vaccine effectiveness of complete primary vaccination series against SARS-CoV-2 infection and mortality in a LTCF, the Netherlands, 20 November 2021 to 3 January 2022 (N = 101) a . case fatality (0%). We therefore believe that the booster vaccination may have contributed to stopping this SARS-CoV-2 outbreak in the LTCF.
# Conclusion
During this outbreak in a LTCF, the COVID-19 vaccine effectiveness of primary vaccination at more than 6 months after vaccination was very low against SARS-CoV-2 infection and moderate against mortality. These results highlight the vulnerability of adults in LTCF. There were few cases in the second and none in the third week after booster vaccination, despite the presence of susceptible residents. These data are consistent with the positive impact of the booster vaccination. Importantly, optimal protection of a booster vaccination might take longer to develop in residents of LTCF compared to younger population. Timely booster vaccination in residents of LTCF is therefore important.
## Data availability
Data will be made available on request.
## Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
[fig] Figure 1: Cases of SARS-CoV-2 infections by date of symptom onset (or test) in a LTCF, the Netherlands, 20 November 2021 to 3 January 2022 (N = 70). [/fig]
[table] Table 1: Characteristics of residents of the LTCF, the Netherlands, 20 November 2021 to 3 January 2022 (N = 105). [/table]
[table] Table 2: Characteristics of residents of the LTCF by vaccination status determined at the end of the outbreak or by date of infection, the Netherlands, 20 November 2021 to 3 January 2022 (N = 101 a ). [/table]
|
No Association of Hair Zinc Concentration with Coronary Artery Disease Severity and No Relation with Acute Coronary Syndromes
Citation: Dziedzic, E.A.; Gąsior, J.S.; Tuzimek, A.; Paleczny, J.; Kwaśny, M.; Dąbrowski, M.; Jankowski, P. No Association of Hair Zinc Concentration with Coronary Artery Disease Severity and No Relation with Acute Coronary Syndromes.
# Introduction
Despite great efforts to advance prophylaxis and treatment, cardiovascular diseases (CVDs) are still the leading cause of death in the world. In 2019, CVDs were responsible for 17.9 million deaths worldwide, which is 85% of deaths from myocardial infarctions or strokes. These numbers are growing, as experts predict that in 2030 CVDs-related deaths could rise to 24 million annually. Despite Zinc (Zn) being one of the key microelements of the human body, its role in CVDs pathogenesis has not yet been firmly established. The available data generally show unfavourable levels of this metal in cardiology patients as low concentrations were found particularly in patients with CAD [bib_ref] Systematic Review of Zinc Biochemical Indicators and Risk of Coronary Heart Disease, Hashemian [/bib_ref] , heart failure [bib_ref] Clinical and Echocardiographic Correlates of Serum Copper and Zinc in Acute and..., Alexanian [/bib_ref] [bib_ref] The Relationship between Serum Zinc Level and Heart Failure: A Meta-Analysis, Yu [/bib_ref] [bib_ref] Long-Term Zinc Deprivation Accelerates Rat Vascular Smooth Muscle Cell Proliferation Involving the..., Alcantara [/bib_ref] and atrial fibrillation [bib_ref] Association of Serum Zinc Level with Prognosis in Patients with Heart Failure, Yoshihisa [/bib_ref]. Moreover, patients with left ventricular hypertrophy and atrial fibrillation had an inverse correlation of Zn concentration with heart muscle thickness [bib_ref] Zinc Levels in Left Ventricular Hypertrophy, Huang [/bib_ref]. However, data on the relationship between Zn concentration and coronary artery disease (CAD) are limited [bib_ref] Is Zinc Deficiency a Risk Factor for Atherosclerosis?, Beattie [/bib_ref] [bib_ref] The Relationship between Serum Levels of Zn and Cu and Severity of..., Islamoglu [/bib_ref] [bib_ref] Serum Zinc Measurement, Total Antioxidant Capacity, and Lipid Peroxide Among Acute Coronary..., El-Mahdy [/bib_ref] [bib_ref] Low Zinc Levels Is Associated with Increased Inflammatory Activity but Not with..., De Paula [/bib_ref] [bib_ref] Reduced Serum Zinc Ion Concentration Is Associated with Coronary Heart Disease, Meng [/bib_ref] [bib_ref] Serum Zinc-A2-Glycoprotein Levels Were Decreased in Patients with Premature Coronary Artery Disease, Liu [/bib_ref] and are mainly based on serum Zn concentration. Analytical Biomolecules 2022, 12, 862 2 of 12 methods commonly used to assess Zn levels in blood and urine do not show the full picture of microelement supply, as Zn content in serum is easily influenced by the time of the day, the type of meal eaten before obtaining the sample, the serum protein concentration, as well as the natural homeostasis mechanisms controlling the Zn concentration utilizing tissue storage [bib_ref] The Time of Day and the Interval since Previous Meal Are Associated..., Arsenault [/bib_ref]. Hair sample analysis has several advantages over serum samples. The concentration of Zn in a hair sample is about 100 times higher than in serum, and it is not as labile as serum concentration, which makes it perfect for long-term nutrition assessment [bib_ref] Hair Analysis: Applications in the, Katz [/bib_ref]. In addition, the hair sample better reflects the recent excessive exposure to metals, as cations are quickly transferred from blood to tissue storage [bib_ref] Elemental Hair Analysis: A Review of Procedures and Applications, Pozebon [/bib_ref] [bib_ref] Elemental Determination for Clinical Diagnosis and Prognosis: Challenges and Trends in Sample..., Mesko [/bib_ref]. Hair samples are considered a good retrospective marker of microelement nutrition in the previous 6-8 weeks. As the concentration of Zn in hair is reported to reflect its content in other tissues [bib_ref] A Comparison of Levels of Select Minerals in Scalp Hair Samples with..., Suliburska [/bib_ref] [bib_ref] Hair as a Biopsy Material: Trace Element Data on One Man over..., Klevay [/bib_ref] [bib_ref] Trace Element Profiles in Single Strands of Human Hair Determined by HR-ICP-MS, Gellein [/bib_ref] , we decided to explore this approach.
This research aims to determine whether the Zn content in hair samples measured by inductively coupled plasma optical emission spectrometry (ICP-OES) correlates with the progression of CAD and acute coronary syndrome (ACS). In ICP-OES, elements are excited using heat from an argon plasma. During de-excitation, the atoms emit light with a spectrum consisting of lines specific to a particular element, allowing for the determination of the elemental composition of the sample [bib_ref] Inductively Coupled Plasma Optical Emission Spectrometry (ICP-OES): A Powerful Analytical Technique for..., Khan [/bib_ref]. This method has found an application in physiological samples due to its remarkable sensitivity and versatility. As obtaining a hair sample is simpler and less invasive than phlebotomy, this method could become an appropriate screening tool for patients at risk of CAD.
# Materials and methods
## Study population
This study is based on data obtained from 133 patients (37 women and 96 men) who underwent coronary angiography to assess the extent of CAD between 2013 and 2017 in the Department of Cardiology of Bielanski Hospital, Warsaw, Poland and agreed to participate in the study in writing. The analysis included patients whose hair was not dyed or permanently waved in segments of at least 3 cm, measuring from the scalp. Patients with an active neoplastic disease, significantly increased inflammatory markers, chronic kidney disease above stage III, taking medications or dietary supplements containing zinc, or using shampoos with an increased content of the bio-element were excluded from the study. Patients with a history of previous MI who were treated with coronary angioplasty were included in this study; however, those with thrombosis or restenosis were excluded. All patients lived in Warsaw and they had no history of occupational exposure to chemical elements. The study was carried out according to the principles of the Declaration of Helsinki and was approved by the bioethics committee of the Medical University of Warsaw.
## Coronary angiography
Coronary angiography is the default method for assessing stenosis in CAD [bib_ref] ACC/AHA/SCAI Guideline for Coronary Artery Revascularization: Executive Summary: A Report of the..., Lawton [/bib_ref]. The examination was performed by radial or femoral artery access. The severity of CAD was classified by three independent cardiologists using the Coronary Artery Surgery Study Score (CASSS). This scale reflects the stenosis of one, two or three arteries through a sum of points (0-3), which were assigned as follows: 1 point for stenosis of the main coronary artery (right coronary artery, circumflex branch, or anterior descending branch) exceeding 70% and 2 points for stenosis of the left main coronary artery greater than 50%. Inconclusive findings between moderate or severe stenosis were decided using fractional flow reserve measurement. The diagnosis of acute coronary syndrome (ACS) was based on criteria from the European Society of Cardiology guidelines, including the increased concentration of markers of myocardial injury with the coexistence of at least one of the following: symptoms of stenocardia, changes in the ECG suggestive of ischemia, results of imaging tests showing myocardial necrosis or coronary artery thrombus identified in coronary angiography [bib_ref] ESC Guidelines for the Management of Acute Coronary Syndromes in Patients Presenting..., Collet [/bib_ref].
## Laboratory tests
Non-permed and non-dyed hair samples, weighing between 0.2 and 0.3 g, were obtained from a few separate scalp sites at the back of the head, close to the skin. In preparation for inductively coupled plasma optical emission spectrometry (ICP-OES) the samples were washed using non-ionic detergent (Triton X-100, Sigma Aldrich Sp. z.o.o., Poznań, Poland) water solution (1:100) in an ultrasonic bath for 5 min, then rinsed sequentially with high-purity water, acetone and water and then dried to constant mass. The dry samples of hair, 0.15 g each, were dissolved in 4 mL of 65% nitric acid (Merck, Darmastadt, Germany) and 1 mL of 30% hydrogen peroxide (Merck) in a closed polypropylene tube (8 mL), then incubated in 80 - C for 30 min in a microwave station. After cooling to room temperature, the samples were diluted to a final volume of 10 mL with Milli-Q water and then analysed using an ICP-OES spectrometer (iCAP7400, Thermo Scientific, Waltham, MA, USA). The concentration of zinc in the solution, and then the total content in the hair samples, were calculated according to the previously determined standard curve.
# Statistical analysis
The Shapiro-Wilk test was used to assess the distribution of data. The chi-square statistic was used to identify associations between dichotomous and categorical data. The Mann-Whitney test was used to compare the values between two groups of patients. Kruskal-Wallis analysis by rank was used to determine the dependence between more than two groups. The R Spearman correlation test was used to evaluate the relationship between the variables. A two-sided p-value < 0.05 was regarded as statistically significant. Statistical analysis and figures were performed and created with Statistica 13 (StatSoft Inc., Tulsa, OK, USA). GraphPad Prism 5 (GraphPad Software Inc., San Diego, CA, USA, 2005) was used to create figures.
# Results
## Population characteristics
The median age of the study population was 65 years (range: 37-95). The median BMI value was 27.7 kg/m 2 (range: 16.9-54.1). A total of 39 (29.3%) participants had a normal body weight, 53 (39.9%) were overweight and 41 (30.8%) patients were classified as obese. A history of type 2 diabetes mellitus (t2DM) or diagnosis during the current hospitalization was found in 42 (31.6%) patients and pre-diabetes in 7 (5.3%) patients. On the basis of the lipid profile (total cholesterol-TC, LDL and HDL cholesterol, triglycerides-TG), hyperlipidaemia was assessed in 123 patients and diagnosed in over half of them despite statin treatment, i.e., in 55 (41.4%). Hypertension was present in 114 (85.7%) patients. Acute coronary syndrome (ACS) as the cause of hospitalization was diagnosed in 67 (50.4%) patients, while stable CAD was the cause in 66 (49.6%) patients. A history of myocardial infarction (MI) was noticed in 40 (30.1%) patients. Active smoking during the study was declared by 40 (30.1%) patients, and 17 (12.8%) patients had smoked in the past. Insignificant changes in the coronary arteries (CASSS 0) were found in only 22 (16.5%) patients. One-vessel coronary disease (CASSS 1) was found in 34 (25.6%) patients, two-vessel (CASSS 2) in 46 (34.6%) and three-vessel (CASSS 3) in 31 (23.3%) patients. The median Zn concentration was 166 parts per million (ppm) (range: 25-495). [fig_ref] Table 1: Association between selected parameters, including Zn level and CAD stages [/fig_ref] presents results of measurements for the study group according to CASSS level. A significant difference in sex distribution was observed between CASSS groups. There was also a significant difference in distribution of patients with history of previous MI and cause of hospitalization.
## Association between zn level and severity of cad
## Difference in zn level between patients with stable cad and patients with acs
Significant differences were observed between patients with ACS and stable CAD in LDL and triglyceride (TG) levels [fig_ref] Table 2: Differences in selected parameters between patients with ACS and stable CAD [/fig_ref]. A lack of significant association between Zn level and CASSS in groups of patients without (H = 2.076 p = 0.557; [fig_ref] Figure 1: Association between Zn level and CASSS in group of patients without [/fig_ref] and with a history of MI (H = 0.000 p = 1.000; [fig_ref] Figure 1: Association between Zn level and CASSS in group of patients without [/fig_ref] was observed. the selected subgroup; data for three and two groups are presented separately for statistical purposes.
A lack of significant association between Zn level and CASSS in groups of patients without (H = 2.076 p = 0.557; [fig_ref] Figure 1: Association between Zn level and CASSS in group of patients without [/fig_ref] and with a history of MI (H = 0.000 p = 1.000; [fig_ref] Figure 1: Association between Zn level and CASSS in group of patients without [/fig_ref] was observed. There were no significant differences in Zn level between patients with stable CAD and ACS in groups of patients without (p = 0.159; [fig_ref] Figure 2: Differences in Zn level between patients with stable CAD and ACS in... [/fig_ref] and with a history of MI (p = 0.084; [fig_ref] Figure 2: Differences in Zn level between patients with stable CAD and ACS in... [/fig_ref].
## Association between zn level and selected parameters
There was no significant correlation between Zn and age (R = −0.005, p = 0.952; [fig_ref] Figure 3: Association between Zn level and selected parameters [/fig_ref] or BMI (R = −0.15, p = 0.084; [fig_ref] Figure 3: Association between Zn level and selected parameters [/fig_ref]. There were no significant differences in Zn between males and females (p = 0.218; [fig_ref] Figure 3: Association between Zn level and selected parameters [/fig_ref] , patients with different smoking status (H 2, (N = 133) = 2073; p = 0.355; [fig_ref] Figure 3: Association between Zn level and selected parameters [/fig_ref] , patients with and without hypertension (p = 0.384; [fig_ref] Figure 3: Association between Zn level and selected parameters [/fig_ref] and patients with or without hyperlipidaemia (p = 0.335; [fig_ref] Figure 3: Association between Zn level and selected parameters [/fig_ref]. Significant association was observed between Zn level and t2DM (H 2, (N = 133) = 10,952; p = 0.004; [fig_ref] Figure 3: Association between Zn level and selected parameters [/fig_ref]. Patients with t2DM presented significantly lower Zn values than patients without t2DM (p = 0.006). A lack of significant correlation was observed between Zn and TC (R = −0.070, p = 0.440), HDL (R = 0.091, p = 0.320), LDL (R = −0.020, p = 0.829) [fig_ref] Figure 4: Correlation between Zn level and lipid profile [/fig_ref]. A significant correlation was noted between Zn and TG (R = −0.193, p = 0.032) [fig_ref] Figure 4: Correlation between Zn level and lipid profile [/fig_ref]. There were no significant differences in Zn level between patients with stable CAD and ACS in groups of patients without (p = 0.159; [fig_ref] Figure 2: Differences in Zn level between patients with stable CAD and ACS in... [/fig_ref] and with a history of MI (p = 0.084; [fig_ref] Figure 2: Differences in Zn level between patients with stable CAD and ACS in... [/fig_ref]. the selected subgroup; data for three and two groups are presented separately for statistical purposes.
A lack of significant association between Zn level and CASSS in groups of patients without (H = 2.076 p = 0.557; [fig_ref] Figure 1: Association between Zn level and CASSS in group of patients without [/fig_ref] and with a history of MI (H = 0.000 p = 1.000; [fig_ref] Figure 1: Association between Zn level and CASSS in group of patients without [/fig_ref] was observed. There were no significant differences in Zn level between patients with stable CAD and ACS in groups of patients without (p = 0.159; [fig_ref] Figure 2: Differences in Zn level between patients with stable CAD and ACS in... [/fig_ref] and with a history of MI (p = 0.084; [fig_ref] Figure 2: Differences in Zn level between patients with stable CAD and ACS in... [/fig_ref].
## Association between zn level and selected parameters
There was no significant correlation between Zn and age (R = −0.005, p = 0.952; [fig_ref] Figure 3: Association between Zn level and selected parameters [/fig_ref] or BMI (R = −0.15, p = 0.084; [fig_ref] Figure 3: Association between Zn level and selected parameters [/fig_ref]. There were no significant differences in Zn between males and females (p = 0.218; [fig_ref] Figure 3: Association between Zn level and selected parameters [/fig_ref] , patients with different smoking status (H 2, (N = 133) = 2073; p = 0.355; [fig_ref] Figure 3: Association between Zn level and selected parameters [/fig_ref] , patients with and without hypertension (p = 0.384; [fig_ref] Figure 3: Association between Zn level and selected parameters [/fig_ref] and patients with or without hyperlipidaemia (p = 0.335; [fig_ref] Figure 3: Association between Zn level and selected parameters [/fig_ref]. Significant association was observed between Zn level and t2DM (H 2, (N = 133) = 10,952; p = 0.004; [fig_ref] Figure 3: Association between Zn level and selected parameters [/fig_ref]. Patients with t2DM presented significantly lower Zn values than patients without t2DM (p = 0.006). A lack of significant correlation was observed between Zn and TC (R = −0.070, p = 0.440), HDL (R = 0.091, p = 0.320), LDL (R = −0.020, p = 0.829) [fig_ref] Figure 4: Correlation between Zn level and lipid profile [/fig_ref]. A significant correlation was noted between Zn and TG (R = −0.193, p = 0.032) [fig_ref] Figure 4: Correlation between Zn level and lipid profile [/fig_ref].
## Association between zn level and selected parameters
There was no significant correlation between Zn and age (R = −0.005, p = 0.952; [fig_ref] Figure 3: Association between Zn level and selected parameters [/fig_ref] or BMI (R = −0.15, p = 0.084; [fig_ref] Figure 3: Association between Zn level and selected parameters [/fig_ref]. There were no significant differences in Zn between males and females (p = 0.218; [fig_ref] Figure 3: Association between Zn level and selected parameters [/fig_ref] , patients with different smoking status (H 2, (N = 133) = 2073; p = 0.355; [fig_ref] Figure 3: Association between Zn level and selected parameters [/fig_ref] , patients with and without hypertension (p = 0.384; [fig_ref] Figure 3: Association between Zn level and selected parameters [/fig_ref] and patients with or without hyperlipidaemia (p = 0.335; [fig_ref] Figure 3: Association between Zn level and selected parameters [/fig_ref]. Significant association was observed between Zn level and t2DM (H 2, (N = 133) = 10,952; p = 0.004; [fig_ref] Figure 3: Association between Zn level and selected parameters [/fig_ref]. Patients with t2DM presented significantly lower Zn values than patients without t2DM (p = 0.006). A lack of significant correlation was observed between Zn and TC (R = −0.070, p = 0.440), HDL (R = 0.091, p = 0.320), LDL (R = −0.020, p = 0.829) [fig_ref] Figure 4: Correlation between Zn level and lipid profile [/fig_ref]. A significant correlation was noted between Zn and TG (R = −0.193, p = 0.032) [fig_ref] Figure 4: Correlation between Zn level and lipid profile [/fig_ref].
# Discussion
This study analysed the nutritional status of Zn among patients with angiographically confirmed CAD. Intake was assessed in hair samples using ICP-OES. There was no
# Discussion
This study analysed the nutritional status of Zn among patients with angiographically confirmed CAD. Intake was assessed in hair samples using ICP-OES. There was no
# Discussion
This study analysed the nutritional status of Zn among patients with angiographically confirmed CAD. Intake was assessed in hair samples using ICP-OES. There was no association between the bioelement and the advancement of CAD and episodes of myocardial infarction in the analysed cohort of patients. Moreover, patients with t2DM had significantly lower Zn content in comparison to non-diabetic individuals. In this cohort, serum TG concentration was found to negatively correlate with Zn content in hair.
Zinc (Zn) is one of the key microelements of the human body. Of all the transition metals, it is second only to haemoglobin-bound iron in terms of prevalence in humans. It is widespread in all tissues and bodily fluids [bib_ref] Zinc and Human Health: An Update, Chasapis [/bib_ref]. Zn deficiency has been diagnosed in 17% of the world population. This percentage increases to 35% in developing countries [bib_ref] Estimating the Global Prevalence of Zinc Deficiency: Results Based on Zinc Availability..., Wessells [/bib_ref]. As Zn influences the activity of more than 300 enzymes [bib_ref] Zinc and the Immune System, Rink [/bib_ref] , it contributes to the stabilization of many protein structures, including more than 2000 transcription factors, most notably Zn finger proteins [bib_ref] Homeostasis and Cellular Functions of Zinc, Beyersmann [/bib_ref]. In the human heart, more than 24 Zn transporting proteins have been identified, indicating the important role of Zn in the homeostasis of the cardiovascular system [bib_ref] Analysis of the Human Tissue-Specific Expression by Genome-Wide Integration of Transcriptomics and..., Fagerberg [/bib_ref].
The results of previous studies indicate that Zn has a protective role in atherosclerosis, a fundamental process in CVDs. Zn and Zn-transporting proteins are important for the function of the vessel wall and its integrity. Zn is essential for the superoxide dismutase function, as it is involved in the dimerization of endothelial NO synthase and nitric oxide production. NO enables labile Zn supplies to be released from endothelial cells, which dilate the vessels and protect endothelial cells [bib_ref] Roles for Endothelial Zinc Homeostasis in Vascular Physiology and Coronary Artery Disease, Zalewski [/bib_ref]. Zn deficiency allows atherosclerotic plaque to build through the aggravation of oxidative stress; destruction of NO, NF-kB and the endothelium; as well as the production of pro-inflammatory cytokines [bib_ref] Zinc Deficiency and Cellular Oxidative Stress: Prognostic Implications in Cardiovascular Diseases, Choi [/bib_ref]. An inverse correlation was found between Zn blood concentration and the risk of CVDs [bib_ref] Zinc Status and Risk of Cardiovascular Diseases and Type 2 Diabetes Mellitus-A..., Chu [/bib_ref] , atherosclerosis progression [bib_ref] Emerging Risk Factors as Markers for Carotid Intima Media Thickness Scores, Masley [/bib_ref] and CVDs complications, such as acute coronary syndrome and heart failure [bib_ref] Serum Zinc and Copper Levels in Ischemic Cardiomyopathy, Shokrzadeh [/bib_ref] [bib_ref] Zinc and Cardiovascular Disease, Little [/bib_ref]. Zn also plays a key role in the immune response, as its deficiency causes a weak cellular and humoral response [bib_ref] The Role of Nutrition in Enhancing Immunity in Aging, Pae [/bib_ref]. By regulation of the oxidation-reduction balance of cells [bib_ref] Zinc Deficiency and Cellular Oxidative Stress: Prognostic Implications in Cardiovascular Diseases, Choi [/bib_ref] , this micronutrient facilitates the integrity of the endothelial cell membrane [bib_ref] Zinc Dynamics in the Myocardial Redox Signaling Network, Korichneva [/bib_ref] and protects it from oxidation stress [bib_ref] Marginal Dietary Zinc Deficiency In Vivo Induces Vascular Smooth Muscle Cell Apoptosis..., Allen-Redpath [/bib_ref] [bib_ref] Zinc Supplementation Decreases Incidence of Infections in the Elderly: Effect of Zinc..., Prasad [/bib_ref]. In animal models, Zn deficiency resulted in increased levels of reactive oxygen species [bib_ref] The Oxidative Stress of Zinc Deficiency, Eide [/bib_ref] , decreased levels of glutathione [bib_ref] Zinc Protects Endothelial Cells from Hydrogen Peroxide via Nrf2-Dependent Stimulation of Glutathione..., Cortese [/bib_ref] and superoxide dismutase 1 [bib_ref] Roles of Zinc and Copper in Modulating the Oxidative Refolding of Bovine..., Li [/bib_ref]. Supplementation reduced serum lipid peroxidation [bib_ref] Zinc Supplementation Inhibits Lipid Peroxidation and the Development of Atherosclerosis in Rabbits..., Jenner [/bib_ref] and normalized inducible nitric oxide synthase activity [bib_ref] Zinc Regulates INOS-Derived Nitric Oxide Formation in Endothelial Cells, Cortese-Krott [/bib_ref]. In both animal and human models, it also suppressed the expression of pro-inflammatory cytokines regulated by NF-kB [bib_ref] Zinc and Its Role in Age-Related Inflammation and Immune Dysfunction, Wong [/bib_ref] [bib_ref] Suboptimal Dietary Zinc Intake Promotes Vascular Inflammation and Atherogenesis in a Mouse..., Beattie [/bib_ref]. These effects are components of the anti-inflammatory action, which modulates the development of atherosclerotic plaques. In summary, the experimental data indicate that low levels of Zn correlate with endothelium dysfunction [bib_ref] Suboptimal Dietary Zinc Intake Promotes Vascular Inflammation and Atherogenesis in a Mouse..., Beattie [/bib_ref] , high levels of oxidation stress and vessel wall inflammation [bib_ref] Biological Consequences of Zinc Deficiency in the Pathomechanisms of Selected Diseases, Jurowski [/bib_ref] -all of them being well-established risk factors for atherosclerosis. The importance of adequate Zn intake is also supported by data suggesting a relationship between Zn deficiency and subclinical inflammation [bib_ref] Is Zinc Deficiency a Risk Factor for Atherosclerosis?, Beattie [/bib_ref] [bib_ref] Zinc Decreases C-Reactive Protein, Lipid Peroxidation, and Inflammatory Cytokines in Elderly Subjects:..., Bao [/bib_ref].
In this study, we did not observe a statistically significant difference in Zn levels in hair samples between subgroups of patients with different severity of CAD (CASSS 0-3). At present, the relationship between this bioelement and CAD has not yet been determined, similar to the data on CAD risk in patients and their nutritional status of Zn being equivocal. Although Islamoglu et al. suggested a direct relationship between Zn level and CAD diagnosis [bib_ref] The Relationship between Serum Levels of Zn and Cu and Severity of..., Islamoglu [/bib_ref] and El-Mahdy et al. correlated lower concentrations of this microelement with a higher SYNTAX score [bib_ref] Serum Zinc Measurement, Total Antioxidant Capacity, and Lipid Peroxide Among Acute Coronary..., El-Mahdy [/bib_ref] , de Paula et al. suggested that Zn levels have no association with CAD course [bib_ref] Low Zinc Levels Is Associated with Increased Inflammatory Activity but Not with..., De Paula [/bib_ref]. On the other hand, data from epidemiological studies show a relationship of micronutrient levels, including Zn, on the presence and progression of atherosclerosis [bib_ref] Current Zinc Intake and Risk of Diabetes and Coronary Artery Disease and..., Singh [/bib_ref] and CAD [bib_ref] Is Zinc Deficiency a Risk Factor for Atherosclerosis?, Beattie [/bib_ref]. The ratio of Zn excretion in urine to serum concentration is well-correlated with CAD and its severity [bib_ref] Association of Reduced Zinc Status with Angiographically Severe Coronary Atherosclerosis: A Pilot..., Giannoglou [/bib_ref]. Liu et al. found that levels of serum zinc-α2-glycoprotein were decreased in patients with premature CAD [bib_ref] Serum Zinc-A2-Glycoprotein Levels Were Decreased in Patients with Premature Coronary Artery Disease, Liu [/bib_ref] , while Meng et al. observed that Zn levels were higher in ACS groups than in CAD groups [bib_ref] Decreased Iron Ion Concentrations in the Peripheral Blood Correlate with Coronary Atherosclerosis, Meng [/bib_ref]. Furthermore, their analysis revealed that serum Zn level is an independent risk factor for the development of CAD [bib_ref] Reduced Serum Zinc Ion Concentration Is Associated with Coronary Heart Disease, Meng [/bib_ref]. Gao et al. demonstrated an inverse correlation of Zn levels with the progression of atherosclerotic calcification in CAD [bib_ref] Association of Dietary Zinc Intake with Coronary Artery Calcium Progression: The Multi-Ethnic..., Gao [/bib_ref] and the carotid intima-media thickness test as a subclinical marker of atherosclerosis [bib_ref] Emerging Risk Factors as Markers for Carotid Intima Media Thickness Scores, Masley [/bib_ref] [bib_ref] Dietary Zinc Intake Is Inversely Related to Subclinical Atherosclerosis Measured by Carotid..., Yang [/bib_ref]. The aforementioned articles, presenting a significant difference in Zn levels in CAD patients compared to healthy controls, are not in line with our results. This discrepancy may be due to the characteristics of the patient group whose CAD is angiographically confirmed.
In our cohort, in addition to the lack of a relationship between Zn and CAD, we did not find significant differences in the measured concentrations in patients with ACS and stable CAD. Just as the association of Zn levels with CAD is not well-established, the relationship of this micronutrient with ACS is yet to be conclusively settled. The available data show a significant decrease in Zn levels in serum and hair of patients after ACS compared to healthy controls [bib_ref] Serum Zinc Measurement, Total Antioxidant Capacity, and Lipid Peroxide Among Acute Coronary..., El-Mahdy [/bib_ref] [bib_ref] Plasma Zinc in Acute Myocardial Infarction. Diagnostic and Prognostic Implications, Low [/bib_ref] [bib_ref] Deficient Zinc Levels and Myocardial Infarction: Association between Deficient Zinc Levels and..., Liu [/bib_ref]. Lower amounts of this bioelement were also observed in patients with ST-elevation myocardial infarction compared to those with non-ST-elevation myocardial infarction [bib_ref] Serum Zinc Measurement, Total Antioxidant Capacity, and Lipid Peroxide Among Acute Coronary..., El-Mahdy [/bib_ref]. An inverse correlation of serum Zn concentration with the concentration of myocardial necrosis markers and clinical predictors of ACS was found [bib_ref] Plasma Zinc in Acute Myocardial Infarction. Diagnostic and Prognostic Implications, Low [/bib_ref] [bib_ref] The Relationship Between Serum Zinc Levels, Cardiac Markers and the Risk of..., Huang [/bib_ref]. Furthermore, the higher the concentration of Zn, the less frequently ACS was observed in a group of patients [bib_ref] The Relationship Between Serum Zinc Levels, Cardiac Markers and the Risk of..., Huang [/bib_ref]. In addition to the suggested use of Zn as the prognostic marker for ACS, Lal et al. focused on the cardioprotective role of this microelement after myocardial infarction [bib_ref] Effect of Zinc Sulphate on Infarct Size in Experimental Myocardial Infarction in..., Lal [/bib_ref]. In animal models, Zn administration caused a halving of the area of the induced infarction. In addition, a reduction in the frequency of arrythmia was observed [bib_ref] Protective Role of Intracellular Zinc in Myocardial Ischemia/Reperfusion Is Associated with Preservation..., Karagulova [/bib_ref] [bib_ref] Zinc Plays a Critical Role in the Cardioprotective Effect of Postconditioning by..., Xu [/bib_ref]. Taking into account the discrepancies in the results of a few randomized trials [bib_ref] Iron, Zinc, and Alcohol Consumption and Mortality from Cardiovascular Diseases: The Iowa..., Lee [/bib_ref] , some cohort research articles [bib_ref] Zinc Status and Risk of Cardiovascular Diseases and Type 2 Diabetes Mellitus-A..., Chu [/bib_ref] and our results showing the lack of association of Zn in acute CAD complications, we suggest further research on this topic. A meta-analysis of prospective cohort studies by Chu et al. suggests the possibility of a correlation of higher Zn concentrations with lower CVDs risk observed in three of the five included papers. The effect of this micronutrient was more prominent in patients with t2DM who underwent angiography due to chest pain compared to the healthy population [bib_ref] Zinc Status and Risk of Cardiovascular Diseases and Type 2 Diabetes Mellitus-A..., Chu [/bib_ref]. In addition, the Iowa Women's Health Study observed an inverse correlation of dietary supplementation of Zn with CVD mortality (CAD and stroke included) in more than 30,000 postmenopausal women observed over more than ten years [bib_ref] Iron, Zinc, and Alcohol Consumption and Mortality from Cardiovascular Diseases: The Iowa..., Lee [/bib_ref]. This result was, however, not confirmed by the recent metanalysis (49 studies with more than 300,000 participants) by Schwingshackl et al. [bib_ref] Dietary Supplements and Risk of Cause-Specific Death, Cardiovascular Disease, and Cancer: A..., Schwingshackl [/bib_ref]. It is worth noting that both studies used a multinutrient supplement formula with quite a small dose of Zn (≤20 mg/d), and most of the participants were healthy.
The significantly lower hair Zn content observed in patients with t2DM may indicate a correlation of this element with glucose metabolism. Experimental studies have shown that Zn plays a role in the regulation of synthesis, storage and insulin excretion from β-cells in the pancreas [bib_ref] Nutritional Effects of Zinc on Metabolic Syndrome and Type 2 Diabetes: Mechanisms..., Ruz [/bib_ref] , as well as improving tissue insulin sensitivity [bib_ref] Trace Elements in Diabetes Mellitus, Sujatha [/bib_ref] and regulating the activity of gluconeogenetic enzymes [bib_ref] Zinc and Diabetes, Chabosseau [/bib_ref]. In diabetic patients, both hypozincaemia and hyperzincuria are frequent findings [bib_ref] Hypozincemia in Diabetes Mellitus, Garg [/bib_ref] [bib_ref] Hyperzincuria of Diabetes Mellitus and Possible Genetical Implications of This Observation, Pidduck [/bib_ref]. Supplementation with Zn reduces the risk of t2DM by up to 40% [bib_ref] Zinc Intake and Status and Risk of Type 2 Diabetes Mellitus: A..., Fernández-Cao [/bib_ref] and improves glucose control [bib_ref] A Meta-Analysis of Randomised Placebo Controlled Supplementation Trials in Humans, Capdor [/bib_ref].
Our data, collected from patients with CAD treated with comparable statin doses, reveal a negative correlation between Zn content and serum TG concentration. A similar trend is observed without statistical significance for total cholesterol and LDL. These findings seem to be consistent with other studies since Zn deficiency is found to have several effects on lipid metabolism. Firstly, it decreases the TG absorption [bib_ref] Effect of Nutrient and Micronutrient Intake on Chylomicron Production and Postprandial Lipemia, Desmarchelier [/bib_ref]. Secondly, it causes a decrease in levels of Zn-α2-glycoprotein, which in turn increases lipogenesis [bib_ref] Zinc Status Is Associated with Inflammation, Oxidative Stress, Lipid, and Glucose Metabolism, Olechnowicz [/bib_ref]. Supplementation with Zn decreases total cholesterol levels and increases high-density lipoprotein levels [bib_ref] Effects of Zinc Supplementation on Serum Lipids: A Systematic Review and Meta-Analysis, Ranasinghe [/bib_ref]. Compared to the healthy control group, lower levels of this micronutrient were found in obese patients [bib_ref] Is There a Link between Zinc Intake and Status with Plasma Fatty..., Knez [/bib_ref]. Zn supplementation also decreases insulin resistance and inflammation marker concentrations [bib_ref] Effect of Zinc Supplementation on Markers of Insulin Resistance, Oxidative Stress, and..., Kelishadi [/bib_ref].
Since almost 86% of our patients were normotensive, we were unable to prove or disprove the existence of a correlation between the nutritional status of Zn and hypertension. However, previous studies on this matter showed contradictory results [bib_ref] Dietary Zinc Intake Is Inversely Related to Subclinical Atherosclerosis Measured by Carotid..., Yang [/bib_ref] [bib_ref] The Relationship between Established Coronary Risk Factors and Serum Copper and Zinc..., Ghayour-Mobarhan [/bib_ref]. Similarly, we did not find any correlation between hair Zn levels and BMI or smoking; this subject needs further in-depth research, as previous research was mainly conducted on healthy individuals [bib_ref] Is There a Link between Zinc Intake and Status with Plasma Fatty..., Knez [/bib_ref] [bib_ref] The Influence of Food Consumption and Socio-Economic Factors on the Relationship between..., Knez [/bib_ref].
There are a few notable limitations to this study. In particular, we did not take into consideration factors interfering with Zn levels, e.g., the potential interactions of nutritional ingredients and other microelements, which can modulate the biological effects of Zn. We omitted the differentiation of the source of Zn, which could have an influence on metabolic syndrome and the risk of CVDs [bib_ref] Dietary Intakes of Zinc and Heme Iron from Red Meat, but Not..., De Oliveira Otto [/bib_ref]. We neglected the effect of drugs commonly used in CVDs, such as beta-blockers, angiotensin receptor blockers, angiotensin-converting enzyme inhibitors and diuretics. Our analysis was based on a single method of quantifying the Zn level, namely hair concentration. This omits the fact that Zn can be in different forms and its concentration changes are dynamic. The use of multiple measurement methods (hair, serum, erythrocytes, urine concentration) may result in a better understanding of the relationship between Zn and CVDs. The degree of progression of CAD was determined on the basis of the results of coronary angiography using the CASSS; perhaps using the SYNTAX or Gensini score would be more optimal.
We did not confirm an association of Zn nutritional status with CAD severity and ACS in the cohort of patients with CAD. We found a statistically significant, negative, correlation between Zn levels in hair and serum TG concentration. Furthermore, patients with t2DM had a lower Zn hair content compared to non-diabetic individuals. Evaluation of the impact of Zn on cardiovascular health requires more well-designed randomized studies to specify the advantages, dangers and contraindications for different levels of Zn dietary intake and supplementation. It is necessary to define and implement global dietary recommendations and food fortification strategies, particularly in developing countries, to achieve optimal Zn intake and reduce CVD risk.
# Conclusions
In patients with angiographically confirmed CAD, there are no significant statistical differences in Zn levels between groups with different severity of CAD (CASSS 0-3). Differences in the concentration of the analysed element in hair samples were not statistically significant in patients with ACS compared to those with stable CAD. Significantly lower levels of Zn were found in the hair of patients with t2DM. A negative correlation was identified between Zn content in the hair and serum TG concentration. The role of Zn in the pathogenesis of CAD and its complications requires further well-designed research.
[fig] Figure 1: Association between Zn level and CASSS in group of patients without (A) and with history of MI (B). [/fig]
[fig] Figure 2: Differences in Zn level between patients with stable CAD and ACS in group of patients without (A) and with history of MI (B). [/fig]
[fig] Figure 3: Association between Zn level and selected parameters: (A) sex, (B) hypertension, (C) hyperlipidaemia, (D) smoking status, (E) t2DM, (F) age, (G) BMI. [/fig]
[fig] Figure 4: Correlation between Zn level and lipid profile: (A) TC, (B) HDL, (C) LDL, (D) TG. [/fig]
[fig] Author: Contributions: Conceptualization, E.A.D.; methodology, E.A.D. and M.K.; statistical analysis, J.S.G.; investigation, E.A.D.; data curation, E.A.D.; writing-original draft preparation, E.A.D., J.S.G., A.T. and J.P.; writing-review and editing, all authors; funding acquisition, E.A.D. and M.D. All authors have read and agreed to the published version of the manuscript. Funding: This research was partly supported by a statutory grant to the Cardiology Clinic of Physiotherapy Division from the 2nd Faculty of Medicine, Medical University of Warsaw, Poland (grant number: 2F5/PM2/16). The APC was funded by the Lazarski University in Warsaw. Institutional Review Board Statement: The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Institutional Review Board (or Ethics Committee) of the Medical University of Warsaw (KB/124/2014). [/fig]
[table] Table 1: Association between selected parameters, including Zn level and CAD stages. *-assessed in 123 patients; **-three subgroups of patients due to the low number of patients in the selected subgroup; data for three and two groups are presented separately for statistical purposes. [/table]
[table] Table 2: Differences in selected parameters between patients with ACS and stable CAD. *-assessed in 123 patients; **-three subgroups of patients due to the low number of patients in the selected subgroup; data for three and two groups are presented separately for statistical purposes. [/table]
|
Chemical Characteristics and Antioxidant Properties of Crude Water Soluble Polysaccharides from Four Common Edible Mushrooms
Four crude water soluble polysaccharides, CABP, CAAP, CFVP and CLDP, were isolated from common edible mushrooms, including Agaricus bisporus, Auricularia auricula, Flammulina velutipes and Lentinus edodes, and their chemical characteristics and antioxidant properties were determined. Fourier Transform-infrared analysis showed that the four crude polysaccharides were all composed of β-glycoside linkages. The major monosaccharide compositions were D-galactose, D-glucose and D-mannose for CABP, CAAP and CLDP, while CFVP was found to consist of L-arabinose, D-galactose, D-glucose and D-mannose. The main molecular weight distributions of CABP and the other three polysaccharides were <5.1 × 10 4 Da and >66.0 × 10 4 Da, respectively. Antioxidant properties of the four polysaccharides were evaluated in in vitro systems and CABP showed the best antioxidant properties. The studied mushroom species could potentially be used in part of well-balanced diets and as a source of antioxidant compounds.
(C=O) showed two bands: an asymmetrical stretching band around 1,650 cm −1 (CABP: 1,629.6 cm −1 , CAAP: 1,627.1 cm −1 , CFVP: 1,635.4 cm −1 , CLDP: 1,642.1 cm −1 ) and a weaker symmetric stretching band near 1,400 cm −1 (CABP: 1,403.6 cm −1 , CAAP: 1,424.2 cm −1 , CFVP: 1,402.8 cm −1 , CLDP: 1,400.5 cm −1 ). The characteristic band nearby 1,080 cm −1 (CABP: 1,080.9 cm −1 , CAAP: 1,071.3 cm −1 , CFVP: 1,080.6 cm −1 , CLDP: 1,077.1 cm −1 ) suggested a pyranose form of the glucosyl residue. The characteristic peak at around 880 cm −1 (CABP: 880.9 cm −1 , CAAP: 891.8 cm −1 , CFVP: 875.1 cm −1 , CLDP: 879.4 cm −1 ) was typical for β-configuration of the sugar units. On the other hand, CAAP had more FT-IR features compared with the other three polysaccharides, CLDP, CABP and CFVP. The relatively strong absorption peak at 1,733.0 cm −1 corresponded to a carboxylic ester band, and the characteristic absorption at 1,376.5 cm −1 was assigned to C-H in-plane bending vibration. Moreover, the signal at 1,250.3 cm −1 was due to C-C band stretching vibrations from a ketone sugar, which were only present in CAAP.
## Monosaccharide composition
The monosaccharide composition results of the polysaccharides from four edible mushroom species are summarized in . From GC, the three polysaccharides, CABP, CAAP and CLDP, were found to mainly contain D-mannose, D-galactose and D-glucose. The monosaccharide contents were different for the three polysaccharides. For CABP, the contents of D-mannose, D-galactose and D-glucose were 36.68%, 20.70% and 34.85%, respectively. CAAP contained, in addition to D-glucose (64.40%), considerable proportions of D-mannose (18.23%) and D-galactose (7.84%). The most important monosaccharides in CLDP were D-glucose (59.90%), D-galactose (17.80%) and D-mannose (15.34%). CFVP was composed of four main monosaccharides. L-Arabinose was observed in CFVP and the content of L-arabinose (36.77%) was higher than the other three monosaccharides. The other main monosaccharides contained in CFVP were also D-glucose (34.18%), D-mannose (14.54%) and D-galactose (9.43%). . Monosaccharide composition of the polysaccharides isolated from four species of edible mushrooms.
## Polysaccharides monosaccharide content (%) l-ara ery l-fuc d-gal d-glu d-man l-rha d-rib d-
## Molecular weight distribution
Gel permeation chromatography (GPC) is an effective method for polysaccharide molecular weight determination. The calibration curve for molecular weight determination was made using a series of β-glucan standards: lgM w = −0.376V t + 11.5 (R 2 = 0.995), where M w is molecular weight; V t is retention time. Based on the calibration curve made with a set of β-glucan series standards, Empower software was used for the calculation of molecular weights [bib_ref] Some properties of an acidic protein-bound polysaccharide from the fruit of pumpkin, Caili [/bib_ref]. The main molecular weight distributions of four polysaccharides were shown in [fig_ref] Table 2: Molecular weight distribution of the polysaccharides isolated from four species of edible... [/fig_ref]. CLDP showed a single molecular weight distribution, and its molecular weight distribution was 78.0 × 10 4 Da. The other three polysaccharides showed two or more molecular weight distributions. CABP and CFVP had two different molecular weight distributions. For CABP, the main molecular weight distributions were in 5.09 × 10 4 Da and minor in 2.17 × 10 4 Da. For CFVP, the main molecular weight distributions were in 68.1 × 10 4 Da and 65.5 × 10 4 Da. CAAP had three different molecular weight distributions, primarily in 66.3 × 10 4 Da, and secondarily in 1.32 × 10 4 Da and 0.92 × 10 4 Da, which was consistent with previous report that CAAP was a heteropolysaccharide [bib_ref] FT-IR analysis of black fungus polysaccharides and its inhibition against skin aging..., Peng [/bib_ref].
## Antioxidant properties
For measuring antioxidant activities, different methods have been used corresponding to different levels of antioxidant action. In this experiment, the in vitro antioxidant capacities of four polysaccharides were evaluated using different biochemical methods of hydroxyl, superoxide anion and DPPH radical scavenging assay, and reducing power analysis.
## Scavenging ability on hydroxyl radicals
Removing hydroxyl radicals is important for the protection of living systems because hydroxyl radicals are considered to be a highly potent oxidant, which can react with most biomacromolecules functioning in living cells and induce severe damage to the adjacent biomolecules. Scavenging abilities of the four polysaccharides isolated from four species of edible mushrooms on hydroxyl radicals were depicted in [fig_ref] Figure 2: Scavenging ability of the polysaccharides isolated from four species of edible mushrooms... [/fig_ref]. In this work, the scavenging abilities of CABP and CAAP on hydroxyl radicals were concentration-dependent, however, the scavenging abilities on hydroxyl radicals of CFVP and CLDP were not very concentration dependent, at the tested concentration range of 0.125-2.0 mg/mL. At 1.0 mg/mL, the scavenging abilities of CABP, CAAP, CFVP and CLDP on hydroxyl radicals were 37.63%, 16.68%, 16.74% and 30.89%, respectively. At 2.0 mg/mL, the scavenging abilities of CABP, CAAP, CFVP and CLDP were 70.68%, 31.84%, 20.84% and 51.16%, respectively. Moreover, at 2.0 mg/mL, CABP was found to have a higher scavenging activity on hydroxyl radicals than CAAP, CFVP and CLDP (p < 0.05). Effectiveness in antioxidant properties is inversely correlated with EC 50 values. It is important to note the effectiveness of polysaccharides in reacting with free radicals under different conditions as a lower EC 50 value corresponds to higher antioxidant activity of the edible mushroom's polysaccharides. The EC 50 values of scavenging ability on hydroxyl radicals for CABP and CLDP were 1.05 mg/mL and 1.90 mg/mL, respectively [fig_ref] Table 3: EC50 values of the polysaccharides isolated from four species of edible mushrooms [/fig_ref]. However, the EC 50 values of scavenging ability on hydroxyl radicals were both >2 mg/mL for CAAP and CFVP [fig_ref] Table 3: EC50 values of the polysaccharides isolated from four species of edible mushrooms [/fig_ref]. The present results proved that CABP was a good scavenger for hydroxyl radicals. The polysaccharides minimized the concentration of Fe 2+ in the Fenton reaction, and the scavenging abilities of the polysaccharides might be due to the active hydrogen donating ability of hydroxyl substitutions of polysaccharides. Therefore, CABP had a better potency to donate hydrogen to reactive hydroxyl radicals than CAAP, CFVP and CLDP. However, the scavenging abilities of the four polysaccharides on hydroxyl radicals were all relatively lower than that of Vc at the same concentrations. The EC 50 values for hydroxyl radical-scavenging ability of CABP and CLDP were lower than that of the polysaccharide from another common edible mushroom, Auricularia polytricha, which was about 2.4 mg/mL [bib_ref] Structural characterization and hydroxyl radicals scavenging capacity of a polysaccharide from the..., Sun [/bib_ref]. However, the value of EC 50 of the intracellular polysaccharide from one precious edible mushroom, Pholiota adipose, was 0.042 mg/mL [bib_ref] Extraction and in vitro antioxidant activity of intracellular polysaccharide by Pholiota adiposa..., Deng [/bib_ref] , remarkably much lower than that of the four crude water soluble polysaccharides in this experiment. Superoxide anions are a precursor of active free radicals react with biological macromolecules and induce tissue damage and play important roles in the formation of other reactive oxygen species such as hydrogen peroxide, hydroxyl radicals, and singlet oxygen, which induce oxidative damage in lipids, proteins and DNA [bib_ref] Ageing and the free radical theory, Wickens [/bib_ref]. Therefore, the superoxide radical-scavenging ability is of great importance to its potential antioxidant activity. [fig_ref] Figure 3: Scavenging ability of the polysaccharides isolated from four species of edible mushrooms... [/fig_ref] reveals the scavenging abilities of the four crude polysaccharides on superoxide anion radicals. An increase of the scavenging activity on hydroxyl anion radicals was observed with the increasing tested concentration at the concentration range (0.125-2.0 mg/mL) of the polysaccharides, but didn't exhibit a visible concentration-dependent manner. The scavenging effects of CABP, CAAP, CFVP and CLDP on superoxide anion radicals were 46.84%, 27.29%, 22.2% and 31.77%, respectively, at 1.0 mg/mL. The scavenging effects of CABP, CAAP, CFVP and CLDP on superoxide anion radicals were 65.78%, 45.21%, 27.9% and 43.99% at 2.0 mg/mL, respectively. CABP revealed a better antioxidant activity because the EC 50 values of scavenging ability on superoxide anion radicals for CABP and the other three polysaccharides were 1.17 mg/mL and >2 mg/mL, respectively [fig_ref] Table 3: EC50 values of the polysaccharides isolated from four species of edible mushrooms [/fig_ref]. These results suggested that CABP was an effective scavenger for superoxide anion radicals, and it might be advantageous for preventing injury induced by superoxide radicals in pathological conditions. However, the four studied polysaccharides showed lower scavenging activities on superoxide radicals than Vc. The EC 50 value of scavenging ability on superoxide anion radicals for the intracellular polysaccharide from P. adiposa was 0.0196 mg/mL [bib_ref] Extraction and in vitro antioxidant activity of intracellular polysaccharide by Pholiota adiposa..., Deng [/bib_ref] , indicating much stronger superoxide anion radical-scavenging ability than the four crude water soluble polysaccharides in this experiment.
## Scavenging ability on dpph radicals
The model of scavenging the stable DPPH radicals has been widely accepted as a tool to evaluate the free radical-scavenging activities of materials [bib_ref] Comparative antioxidant activity of individual herbal components used in Ayurvedic medicine, Naik [/bib_ref]. [fig_ref] Figure 4: Scavenging ability of the polysaccharides isolated from four species of edible mushrooms... [/fig_ref] describes the scavenging abilities of different polysaccharides from four species of edible mushrooms on DPPH radicals. At all concentrations tested, CABP, CAAP and CLDP exhibited a dose-dependent DPPH radical-scavenging activity. Nevertheless, for CFVP, the scavenging ability on DPPH radicals slightly increased with increasing sample concentration. In the lower concentration ranged from 0.125 to 1.0 mg/mL, the scavenging abilities on DPPH radicals for CABP and CLDP were increased with increasing sample concentration, which were 64.56% and 67.39%, respectively, at 1.0 mg/mL. However, at 2.0 mg/mL, the scavenging abilities on DPPH radicals for CABP and CLDP were 72.24% and 71.16%, respectively, indicating no significantly increasing ability (p < 0.05). With regard to scavenging ability on DPPH radicals, CABP and CLDP showed very good scavenging abilities as evidenced by their particularly low EC 50 values (≈0.5 mg/mL) [fig_ref] Table 3: EC50 values of the polysaccharides isolated from four species of edible mushrooms [/fig_ref]. For CAAP and CFVP, the EC 50 values of scavenging ability on DPPH radicals were 1.62 and >2.0 mg/mL, respectively [fig_ref] Table 3: EC50 values of the polysaccharides isolated from four species of edible mushrooms [/fig_ref]. These results suggested that CABP and CLDP had higher scavenging abilities than CAAP and CFVP on DPPH radicals. It might be that CABP and CLDP could be better advantageous than CAAP and CFVP for reacting with DPPH radicals to convert them to more stable products and thereby terminate radical chain reactions. However, the scavenging abilities on DPPH radicals of the four polysaccharides were lower than that of Vc. The EC 50 value of the polysaccharide from A. bisporus for DPPH radical-scavenging ability was 2 mg/mL [bib_ref] Antioxidative and immunomodulating activities of polysaccharide extracts of the medicinal mushrooms Agaricus..., Kozarski [/bib_ref] , much higher than that of CABP. Moreover, the EC 50 values of the polysaccharides from wild edible mushrooms of Armillaria mellea, Calocybe gambosa, Clitocybe odora and Coprinus comatus for DPPH radical-scavenging ability were 3.95 mg/mL, 7.08 mg/mL, 3.56 mg/mL and 7.31 mg/mL, respectively [bib_ref] Chemical composition of wild edible mushrooms and antioxidant properties of their water..., Vaz [/bib_ref] , which were remarkably much higher than that of CABP, CAAP and CLDP. It was worthy of noting that the EC 50 value for DPPH radical-scavenging ability of the polysaccharide from another edible mushroom, Agaricus brasiliensis, was 0.27 mg/mL [bib_ref] Antioxidative and immunomodulating activities of polysaccharide extracts of the medicinal mushrooms Agaricus..., Kozarski [/bib_ref] , much lower than that of the four polysaccharides in this experiment.
## Reducing power
The reducing capacity of a compound may serve as a significant indicator of its potential antioxidant activity [bib_ref] Meningitis-associated central nervous system complications are mediated by the activation of poly(ADP-ribose)..., Koedel [/bib_ref]. In this assay, the concentration-dependent profile of reducing power was obvious for all the tested polysaccharides [fig_ref] Figure 5: Reducing power of the polysaccharides isolated from four species of edible mushrooms [/fig_ref]. At 1.0 mg/mL, the reducing powers were 0.388, 0.128, 0.076 and 0.270 for CABP, CAAP, CFVP and CLDP, respectively. At 2.0 mg/mL, the reducing powers were 0.752, 0.218, 0.135 and 0.452 for CABP, CAAP, CFVP and CLDP, respectively. With regard to reducing power, CABP showed the best antioxidant property because of its lowest EC 50 value (1.35 mg/mL), indicating that CABP was good effective as an antioxidant [fig_ref] Table 3: EC50 values of the polysaccharides isolated from four species of edible mushrooms [/fig_ref]. High reducing power of CABP suggested a high potential in hydrogen-donating ability which could react with free radicals to convert them to more stable products and thereby terminate radical chain reactions [bib_ref] Antioxidant properties of polysaccharide fractions with different molecular mass extracted with hot-water..., Zha [/bib_ref]. CABP showed a obviously higher ability than that of the other three polysaccharides, but less than that of Vc. Kozarski et al. [bib_ref] Antioxidative and immunomodulating activities of polysaccharide extracts of the medicinal mushrooms Agaricus..., Kozarski [/bib_ref] and Vaz et al. [bib_ref] Chemical composition of wild edible mushrooms and antioxidant properties of their water..., Vaz [/bib_ref] reported that the EC 50 values for reducing power of the polysaccharides from A. brasiliensis, Calocybe gambosa and Coprinus comatus were 3.13 mg/mL, 2.38 mg/mL and 4.67 mg/mL, respectively, which were higher than that of CABP. Furthermore, for reducing power of the polysaccharide from A. bisporus, EC 50 value was found of 14.83 mg/mL [bib_ref] Antioxidative and immunomodulating activities of polysaccharide extracts of the medicinal mushrooms Agaricus..., Kozarski [/bib_ref] , remarkably higher than that of CABP studied in our experiment. However, the polysaccharides of A. mellea and C. odora showed the lower EC 50 values for reducing power (0.98 mg/mL and 0.94 mg/mL, respectively) than CABP [bib_ref] Chemical composition of wild edible mushrooms and antioxidant properties of their water..., Vaz [/bib_ref]. In general, it is interesting and important to elucidate the relation among chemical structures and chain conformations of polysaccharides and their biological activities [bib_ref] Purification, composition analysis and antioxidant activity of the polysaccharides from Dendrobium nobile..., Luo [/bib_ref]. The antioxidant activities of the polysaccharides might be related to monosaccharide component, molecular size, structure and conformation [bib_ref] Isolation, chemical characteristics and antioxidant properties of the polysaccharides from marine fungus..., Sun [/bib_ref]. Tsiapali et al. [bib_ref] Glucans exhibit weak antioxidant activity, but stimulate macrophage free radical activity. Free..., Tsiapali [/bib_ref] stated that the antioxidant abilities were partially related to monosaccharide composition. Moreover, Lo et al. [bib_ref] Correlation evaluation of antioxidant properties on the monosaccharide components and glycosyl linkages..., Lo [/bib_ref] reported that the monosaccharide composition and the type of glycosyl linkage modulated the antioxidant properties of the polysaccharides. The antioxidant properties of the polysaccharides were dependent on the ratio of different monosaccharides in the composition. Among the monosaccharides, rhamnose was the most significant determinant factor associated with antioxidant properties. The glycosyl linkage of the monosaccharides also affected the antioxidation characteristics of the polysaccharides. Specifically, the arabinose 1→4 and mannose 1→2 linkages of the side-chain were significantly related to the reducing power, whereas the glucose 1→6 linkage and arabinose 1→4 linkage were related to the scavenging on DPPH radicals. In addition, a correlation was found between EC 50 value of the reducing power ability and the amount of total glucans content in polysaccharides [bib_ref] Antioxidative and immunomodulating activities of polysaccharide extracts of the medicinal mushrooms Agaricus..., Kozarski [/bib_ref]. Furthermore, Luo et al. [bib_ref] Purification, composition analysis and antioxidant activity of the polysaccharides from Dendrobium nobile..., Luo [/bib_ref] figured out that the antioxidant activities of polysaccharides were supposed to relate to the structural characteristics of the polysaccharides, including molecular weight, monosaccharide composition and configuration. Polysaccharides owned smaller molecular weight exhibited stronger antioxidant activities. Polysaccharides with β-configuration in pyranose form sugars also exhibited stronger antioxidant activities. Besides, the high content of rhamnose was supposed to relate to the strong antioxidant effects.
In our experiment, the four crude polysaccharides showed different antioxidant activities. However, it is hard to clear the correlation between chemical characteristics and antioxidant properties of crude water soluble polysaccharides because using crude polysaccharides may cause many factors of chemical characteristics to influence their antioxidant properties. The objective of our study was to evaluate and compare the chemical characteristics and antioxidant properties of crude water soluble polysaccharides from four species of common edible mushrooms. Our results can give guidance for well-balanced diets in our daily life and a source of antioxidant compounds. Meanwhile, the correlation between chemical characteristics and antioxidant properties for purified polysaccharide fractions is undergoing in our lab.
## Experimental
# Materials and chemicals
## Isolation of the polysaccharides
Four fruiting bodies of edible mushrooms were dried at 70 C in oven and ground to obtain a fine powder (50 mesh), then weighed (50 g each) and extracted with 20 volumes of distilled water at 100 C for 2 h. The suspension was centrifuged (8,000 × g for 15 min), and the supernatant was concentrated under vacuum. The protein in the product of condensation was deproteinized using the Sevag reagent [bib_ref] Polysaccharides from hot water extracts of roasted Coffea arabica beans: Isolation and..., Navarini [/bib_ref]. After removal of the Sevag reagent, 4 volumes of anhydrous ethanol were added to this concentrate, and the mixture was stirred at room temperature for 20 min and then left at 4 C overnight. The precipitate was collected by centrifugation (8,000 × g for 15 min), subsequently washed three times with anhydrous ethanol, acetone and ether, respectively, then dissolved in distilled water and dialyzed in dialysis bag (molecular weight cut-off, 1.2 × 10 4 Da) against distilled water at room temperature for three successive days. The retained fraction was recovered, concentrated under vacuum and lyophilized to obtain four crude water soluble polysaccharides, which were termed as CABP, CAAP, CFVP and CLDP for A. bisporus, A. auricular, F. velutipes and L. edodes, respectively.
## Fourier transform-infrared (ft-ir) spectroscopy analysis
The FT-IR spectra of polysaccharide extracts were recorded on a Thermo-Nicolet Model 6700 spectrophotometer (Thermo Scientific, Waltham, MA, USA). The spectrophotometer was equipped with DTGS TEC detector and OMNIC 7.3 software. Spectra were recorded in the 400-4,000 cm −1 range using KBr disc technique.
# Monosaccharide composition analysis
For the monosaccharide composition analysis, four polysaccharides (5 mg each) were hydrolyzed with 2 M trifluoroacetic acid (5 mL) at 110 °C for 2 h. The hydrolyzate was repeatedly co-concentrated with methanol to dryness, subsequently reduced with NaBH 4 for 3 h at room temperature and acetylated with acetic anhydride at 100 °C for 2 h, then concentrated with a little methylbenzene under reduced pressure. Finally, the acetylated monosaccharides were extracted by chloroform for gas chromatography (GC) analysis. The monosaccharide standards were acetylated in the same way. GC was performed on Agilent Technologies 7890A gas chromatograph with a DB-1701 capillary column (30 m × 0. 32 mm × 0.25 μm) and equipped with a flame ionization detector (FID). The nitrogen gas was used as the carrier gas at a flow rate of 1 mL/min. The temperatures of injector and detector were set at 230 C and 255 °C, respectively; the initial column temperature was increased from 150 °C at a rate of 10 °C/min to 222 °C and held for 16 min. These experiments were repeated twice.
## Determination of molecular weight distribution
Molecular weight distributions of polysaccharide samples from four edible mushroom species were evaluated and determined by high performance liquid chromatography (HPLC) with a Waters HPLC apparatus (Waters 515, Waters Co. Ltd., Milford, MA, USA) equipped with a 2424 refractive index detector (RID). Ultrahydrogel TM 500 (7.8 × 300 mm, Part NO. WAT11530) and Ultrahydrogel TM 1000 (7.8 × 300 mm, Part No. WAT11535) were used after connection in series and the temperature of the column was kept at 30 °C. The sample concentration was 0.8 mg/mL, and its injection volume was 10 μL. The eluent was 0.01 M NaNO 3 , and the flow rate was 0.8 mL/min. The linear regression was calibrated with a set of β-glucan series standards of known molecular mass (M w : 1.
## Assay of antioxidant properties
## Scavenging ability on hydroxyl radicals
Hydroxyl radical-scavenging activity was determined based on the method described by Smirnoff and Cumbes [bib_ref] Hydroxyl radical scavenging activity of compatible solutes, Smirnoff [/bib_ref] with some modifications. The reaction mixture contained polysaccharide sample with different concentrations (0-2.0 mg/mL, 1 mL) was incubated with a solution containing orthophenanthroline (5 mM, 1 mL), phosphate buffer (7.5 mM, pH 7.4, 0.8 mL) and FeSO 4 (7.5 mM, 0.5 mL). Finally, H 2 O 2 (8.8 mM, 0.5 mL) was added, and the reaction mixture was then incubated at 37 C for 1 h. The absorbance of the resulting solution was measured spectrophotometrically at 532 nm. The scavenging ability of hydroxyl radicals was calculated according to the equation: scavenging ability (%) = (1 − A sample /A control ) × 100, where A control is the absorbance of control without the polysaccharide sample, and A sample is the absorbance in the presence of the polysaccharide sample. The EC 50 value (mg/mL) is the effective concentration at which the hydroxyl radicals are scavenged by 50%. Vc was used for comparison because it is a standard antioxidant.
## Scavenging ability on superoxide anion radicals
The scavenging activity of superoxide anion radicals was assessed referring to the reference [bib_ref] Involvement of the superoxide anion radical in the autoxidation of pyrogallol and..., Marklund [/bib_ref] with several modifications. A tube containing Tris-HCl buffer (50.0 mM, pH 8.2, 3 mL) and polysaccharide sample (0-2.0 mg/mL, 1 mL) was incubated in a water bath at 25 °C for 20 min, then pyrogallic acid (5.0 mM, 0.4 mL) at the same temperature was added and proceed at 25 °C. HCl solution (8.0 M, 0.1 mL) was used to terminate the reaction after 4 min. The absorbance of the mixture was measured at 320 nm. The scavenging ability of superoxide anion radicals was calculated using the following formula: scavenging ability (%) = (1 − A sample /A control ) × 100, where A control is the absorbance of control without the polysaccharide sample, and A sample is the absorbance in the presence of the polysaccharide sample. The EC 50 value (mg/mL) is the effective concentration at which the superoxide anion radicals are scavenged by 50%. For comparison, Vc was used as positive control.
## Scavenging ability on 1,1-diphenyl-2-picrylhydrazyl radicals
The DPPH radical-scavenging activity was measured according to the method of Braca et al. [bib_ref] Antioxidant principles from Bauhinia tarapotensis, Braca [/bib_ref]. Polysaccharide sample with different concentrations (0-2.0 mg/mL, 1 mL) was mixed with methanol solution (2 mL) containing DPPH radicals (0.2 mM). The mixture was shaken vigorously and incubated for 30 min in darkness at room temperature, and then absorbance at 517 nm was measured. The scavenging ability of DPPH radicals was calculated using the following formula: scavenging ability (%) = (1 − A sample /A control ) × 100, where A control is the absorbance of control without the polysaccharide sample, and A sample is the absorbance in the presence of the polysaccharide sample. The EC 50 value (mg/mL) is the effective concentration at which the DPPH radicals are scavenged by 50%. Vc was used as standard antioxidant and positive control.
## Reducing power ability
The reducing power was determined referring to the method described in literature [bib_ref] Fe(III) reductive and free radical-scavenging properties of summer savory (Satureja hortensis L.)..., Dorman [/bib_ref] with slight modifications. Polysaccharide sample with different concentrations (0-2.0 mg/mL, 1 mL) was mixed with phosphate buffer (0.2 M, pH 6.6, 2.5 mL) and K 3 Fe(CN) 6 (1%, w/v, 2.5 mL). The mixture was incubated at 50 °C for 20 min, then trichloroacetic acid (10%, w/v, 2.5 mL) was added to the mixture, subsequently centrifuged at 3,000 × g for 10 min. An aliquot of supernatant (5 mL) was mixed with deionized water (4 mL) and FeCl 3 (0.1%, w/v, 1 mL). After incubating at room temperature for 10 min, the absorbance of the mixture was measured at 700 nm. A higher absorbance in the reaction mixture indicates greater reducing power ability. Vc was used as positive control for comparison.
# Statistical analysis
For each polysaccharide, three samples were prepared for assays of every antioxidant attribute. All bioassay results were expressed as means ± standard deviation (SD) of three replications, and the one-way analysis of variance (ANOVA) was used for the statistical analysis.
# Conclusions
In this study, four crude water soluble polysaccharides were isolated from four species of common edible mushrooms. Chemical characteristics, including polysaccharide profile, monosaccharide composition and molecular weight distribution of the four polysaccharides were determined. FT-IR analysis showed that the four crude polysaccharides were all composed of β-glycoside linkages. Monosaccharide composition results indicated the dominance of D-mannose, D-galactose and D-glucose in CABP, CAAP and CLDP with different contents of monosaccharide. CFVP was mainly composed of four monosaccharides, namely L-arabinose, D-mannose, D-galactose and D-glucose. The main molecular weight distributions of CABP and the other three polysaccharides were <5.1 × 10 4 Da and >66.0 × 10 4 Da, respectively. For antioxidant activities of the four polysaccharides, CABP was the best natural antioxidant. The water soluble polysaccharides of edible mushrooms acted as natural antioxidants could be most good sources of antioxidant foods and the consumption of edible mushrooms might give a certain level of health protection against oxidative damages.
[fig] Figure 1: FT-IR spectra of the polysaccharides isolated from four species of edible mushrooms. [/fig]
[fig] Figure 2: Scavenging ability of the polysaccharides isolated from four species of edible mushrooms on hydroxyl radicals. [/fig]
[fig] Figure 3: Scavenging ability of the polysaccharides isolated from four species of edible mushrooms on superoxide anion radicals. [/fig]
[fig] Figure 4: Scavenging ability of the polysaccharides isolated from four species of edible mushrooms on DPPH radicals. [/fig]
[fig] Figure 5: Reducing power of the polysaccharides isolated from four species of edible mushrooms. [/fig]
[table] Table 2: Molecular weight distribution of the polysaccharides isolated from four species of edible mushrooms. [/table]
[table] Table 3: EC50 values of the polysaccharides isolated from four species of edible mushrooms. [/table]
|
Epidermolysis bullosa acquisita as an adverse effect from rituximab therapy
Rationale: Rituximab is a monoclonal antibody directed against B cells and is a first-line agent for the treatment of B cell lymphoma and a second-line agent for the treatment of idiopathic thrombocytopenic purpura (ITP). It has also been used for the treatment of several other autoimmune diseases. Epidermolysis bullosa acquisita (EBA) has never been reported as an adverse effect resulted from rituximab therapy.Patient concerns: A 54-year-old female presented with relapse of the ITP for around eight months. She was treated with rituximab. Intramuscular chlorpheniramine and intravenous methylprednisolone and cimetidine were used as premedication before rituximab infusion. The infusion was initially started at 50 mg/h for 1 h followed by 100 mg/h till the end of infusion. The day after rituximab infusion, the patient noticed pruritic blisters on both arms and chest skin. The next day, the lesions increased in severity and extent.Diagnosis: The skin biopsy established the diagnosis of EBA. H&E staining revealed subepidermal blisters infiltrated by inflammatory cells, including eosinophils and lymphocytes. Direct immunofluorescence (DIF) showed linear deposition of IgG and C3 at the dermoepidermal junction. Indirect immunofluorescence with the patient's serum on salt-split skin revealed exclusive dermal binding of circulating IgG antibasement membrane antibodies at a titer of 1:160.Interventions: She was treated with intravenous methylprednisolone and was continued on oral prednisolone.Outcomes: The lesions regressed. Six weeks later, she had a recurrence of similar lesions but in milder form. This episode subsided in 4 to 5 days with topical steroid application.Lessons: Physicians should consider this diagnosis when a patient develops bullous skin eruptions while undergoing Rituximab therapy.Abbreviations: DIF = Direct Immunofluorescence, EBA = Epidermolysis Bullosa Acquisita, ITP = Idiopathic Thrombocytopenic Purpura.
# Introduction
Rituximab, a chimeric monoclonal antibody directed against the CD20 antigen, is commonly used for the treatment of B-cell lymphoma.It is a second line regimen for the treatment of idiopathic thrombocytopenic purpura (ITP).Infusion reactions such as flushing, itching, dyspnea, heartache, fever, and nausea are the most common adverse effects of rituximab.Rituximab is also associated with several dermatologic reactions.Furthermore, cytokine release during the intravenous infusion of rituximab can cause urticaria and angioedema.
Epidermolysis bullosa acquisita (EBA) has never been reported in patients undergoing rituximab therapy. We herein report a case of EBA in a patient with ITP. In the present case, the condition developed during the first cycle of rituximab therapy. The patient has given a written informed consent for reporting this case.
## Case report
A 54-year-old female presented with relapse of the ITP for around 8 months. She was diagnosed to have ITP about 10 years earlier. She was treated with prednisolone and cyclosporine. Serum complements were normal, and anti-dsNA antibody was negative. However, despite adequate trough levels of cyclosporine (C0 level maintained >100 ng/mL), there was no response after 3 months of treatment, which manifested by persistent thrombocytopenia. In view of the failed therapy, further treatment options were discussed, and an informed decision was made to use rituximab. Patient was counselled regarding its efficacy and adverse effects. Rituximab monotherapy with 375 mg/m 2 was planned. Intramuscular chlorpheniramine and intravenous methylprednisolone and cimetidine were used as premedication before rituximab infusion. The infusion was initially started at 50 mg/h for 1 h followed by 100 mg/h till the end of infusion.
The day after rituximab infusion, the patient noticed pruritic blisters on both arms and chest skin. On examination, these were lesions with clear fluid predominantly involving the skin of the flexor and extensor aspects of both upper limbs and chest. The next day, the lesions increased in severity and extent.
After a skin biopsy, she was treated with intravenous methylprednisolone (according to prednisone 1 mg/kg·d) and the lesions regressed. The skin biopsy established the diagnosis of EBA. H&E staining revealed subepidermal blisters infiltrated by inflammatory cells, including eosinophils and lymphocytes. Direct immunofluorescence (DIF) showed linear deposition of IgG and C3 at the dermoepidermal junction. Indirect immunofluorescence with the patient's serum on salt-split skin revealed exclusive dermal binding of circulating IgG antibasement membrane antibodies at a titer of 1:160.
She was continued on oral prednisolone (prednisone minus one tablet per week). Six weeks later, she had a recurrence of similar lesions but in milder form. This episode subsided in 4 to 5 days with topical steroid application (double the current dose of prednisone).
## Patient perspective
This patient was promptly informed of this previously unknown adverse event. Treatment options were discussed. She was pleased that the skin lesions were an adverse effect from the medication rather than a new illness. She decided to follow our treatment recommendations and was very satisfied with the result.
# Discussion
Rituximab is extensively used for the treatment of B cell lymphoma and as one of the leading second-line drugs in the treatment of ITP.It has also been used for the treatment of several other autoimmune diseases, including EBA.The side effects of this monoclonal antibody however remain a concern.
Walewski reported that 60% of the first, and 20% of subsequent rituximab infusions were associated with infusion- related reactions in patients with recurrent indolent lymphoma.These adverse effects included mild fever, chills, and occasional skin eruptions. Brown BA reported that infusion-associated events occurred in 25.7% of the multiple sclerosis patients. Reactions were mild-to-moderate; most occurred during the first infusion and most patients went on to receive subsequent doses without any further problems.Errante D reported urticarial reaction occurring 1 h after the start of retuximab infusion which spontaneously resolved after temporary stoppage of infusion. They hypothesized that the urticarial reaction in this particular patient with primary cutaneous B-cell lymphoma was an epiphenomenon of the killing of the B-cells bearing the CD20 antigen in the sites of the disease. This resulted in local cytokine release rather than the occurrence of a hypersensitivity reaction causing the urticarial reaction.In addition to infusion reactions, serum sickness is reported in 1% to 20% of patients, more commonly among those with autoimmune conditions.Diffuse, pruritic, maculopapular rash with generalized edema, associated with aphthous ulcers, diarrhea, arthralgias, and myalgia occurring 48 h after rituximab injection was described by Hellerstedt and Ahmed in a patient with systemic lupus erythematosus.Severe reactions including Stevens-Johnson syndrome and vasculitis have also been observed.EBA is a chronic autoimmune subepidermal blistering disease characterized by circulating and tissue-bound autoantibodies targeting type VII collagen, which is the major component of anchoring fibrils. EBA usually manifests as blisters and erosions on the skin and mucocutaneous areas. EBA is a rare disease. The overall prevalence has yet to be determined. EBA occurs in both genders, at all ages, and in all ethnic groups. Blisters appears to form either through a direct pathway, where the adhesive function of the basement membrane is impaired (mechanobullous type); or through an indirect pathway which involves the complement activation and neutrophil recruitment (inflammatory type).EBA has never been reported as a side effect with rituximab treatment. The mechanism for the occurrence remains unclear. Our patient developed the inflammatory type EBA. We posited that it might represent a delayed type of hypersensitivity reaction. The condition developed the day after rituximab infusion and she had not received any other new drugs. Hence, rituximab was thought to be the likely etiological agent rather than EBA being a coincidental occurrence. In addition, this patient has ITP, a known autoimmune disease, which might make her more susceptible to develop EBA. Fortunately, this patient responded well to steroid therapy.
Laboratory tests to diagnose EBA include H&E staining, immunofluorescence, immunoblotting, immunoelectron microscopy, and ELISA. Biopsy specimens of inflammatory EBA reveal subepidermal blisters infiltrated by various inflammatory cells, including neutrophils, eosinophils, and lymphocytes; whereas those from the mechanobullous EBA often show sparse infiltrate. Histopathological findings alone, cannot confirm a diagnosis of EBA because other subepidermal bullous diseases can have nearly identical appearance. Immunofluorescence studies are necessary to obtain a definitive diagnose. DIF shows linear deposition of IgG, C3, or both along the basement membrane zone. U-serrated immunodeposition pattern in DIF can help to differentiate EBA from other subepidermal bullous disease.Our patient was diagnosed to suffer EBA based on the histopathological findings and the immunofluorescence results.
# Conclusion
EBA has not been previously reported as an adverse effect associated with rituximab therapy. With increased indications for the use of rituximab, it is imperative for clinicians to identify and manage the complications associated with this monoclonal antibody. This case highlights the importance of watching out for blister dermatitis as an early manifestation of rituximab toxicity. Patients should be educated to report any skin lesion that occurs during rituximab therapy; for it has the potential to evolve into more severe lesions if further doses are given. |
The association between vertebrobasilar insufficiency and the risk of dementia: a nationwide register-based retrospective cohort study in Taiwan
BMJ Open publishes all reviews undertaken for accepted manuscripts. Reviewers are asked to complete a checklist review form (http://bmjopen.bmj.com/site/about/resources/checklist.pdf) and are provided with free text boxes to elaborate on their assessment. These free text comments are reproduced below.ARTICLE DETAILSTITLE (PROVISIONAL)The association between vertebrobasilar insufficiency and the risk of dementia: A nationwide register-based retrospective cohort study in Taiwan AUTHORSGENERAL COMMENTSIn this large-scale patient register-based study, the authors examined the association between vertebrobasilar insufficiency (VBI) and risk of dementia and subtypes. It was concluded that patients with VBI appear to have an increased risk of developing dementia. This is an interesting and relevant study. However, I have some concerns that need to be clarified, as specified below. 1. Study population and sample: Data for this study were derived from health insurance databases of NHIRD and LHID (see page 7, please clarify the relationship of NHIRD and LHID). As I understood, although the databases (insurance) covered almost the whole general population, data in the datasets included only those people (outpatients) who sought health care services. Thus, it may not be convincing to call this study as a population-based study. The authors may wish to slightly modify the subtitle as something like "a patient register-based study". 2. Matching factors in the study design: a range of comorbidities (see page 8) were matched in choosing controls for dementia cases. This matching strategy might lead to cases and controls over matched, and thus potentially mask the association between the exposure and outcome, especially given that some of the comorbidities might be on the pathways linking VBI to dementia. 3. Subtype dementia as outcomes: Clinically, Alzheimer's disease (AD) accounts for approximately 50-70% of all dementia cases and vascular dementia accounts for around 20% in the population-based studies. However, in this study AD accounted only for just over 10% of all dementia cases (64 AD cases over 607 all dementia cases, seeTable 2). The authors also acknowledged that the diagnosis of subtype dementias other than non-VD was problematic. Thus, the analysis on subtype dementias may be omitted because the results might be misleading due to unreliable diagnosis and classification of subtype dementias (e.g., AD). 4. Age-stratifying analysis: The authors carried out subgroup analysis by stratifying age into three groups, i.e., <45, 45-64, and ≥65 years of age. It is unclear what the rationale for the agestratifying analysis is. Was any statistical interaction between VBI and age group on dementia risk detected? The authors stated that "…the risk of developing dementia was most prominent in those aged 45-64 or over 65 years " (e.g., page 18, 22). This is actually not true because the point estimation of HR for dementia in people aged <45 years was even greater than those for persons aged 45-64 and those aged 65+ (4.019 vs. 2.953 and 1.752, seeTable 2), and the lack of statistical significance is likely due to limited power in the younger age group (very few cases of dementia). However, the authors may consider to investigate the associations of VBI with early-onset (onset age >65) and late-onset (onset age ≥65) dementia, separately. This is relevant because early-onset and lateonset dementia are thought to have different risk factors.
# Methods
The subject selecting procedure is not clear. The randomisation should be explained whether the selection procedure represents the whole country or not. The diagnosis of the vertebrobasilar insufficiency and dementia is not clearly stated. The criteria for diagnosis should be described.
Also, the interrater variability of the investigators are very important, and the training of the site personal about the study is critical.
# Discussion
Although the authors have used nationwide population-based data with a large sample size to explore the strong temporal relationship between vertebrobasilar insufficiency and dementia, their data has not disclosed the causal relationship.
## General
The paper is acceptable after small revisions including the detailed explanation of the limitations mentioned above. Since the methodological problems seems not to be overcame at the present time, very careful interpretation of the results may be useful for the future studies.
Similarly, "… diagnosis and coding in this study are extremely reliable, …" as a strength vs. "… additional clinical information about VBI or dementia …. was not available, …" as a limitation. 6. Minors: [bib_ref] Posterior Circulation Ischemia: Then, Now, and Tomorrow : The Thomas Willis Lecture--2000, Caplan [/bib_ref] The authors may rephrase the statement "Dementia is one of the most common neurodegenerative …" It should refer to Alzheimer's disease specifically, not dementia in general. [bib_ref] Vertebral Insufficiency: When to Intervene and How?, Jenkins [/bib_ref] Please clarify the relationship between the two databases NHIRD and LHID.
(3) Age may still need to be controlled for as a continuous variable in the analysis given its strong impact on dementia occurrence and the difference in the mean age between cases and controls as shown in (although age is a matching factors). [bib_ref] Vertebrobasilar stenosis predicts high early recurrent stroke risk in posterior circulation stroke..., Gulli [/bib_ref] The authors may justify the analyses according to the length of follow-up time.
## Version 1 -author response
Reviewer: 1 Reviewer Name: Nevzat Uzuner Institution and Country: Eskisehir Osmangazi University, TURKEY Please state any competing interests: None declared Please leave your comments for the authors below Definition Since vertebrobasilar insufficiency is synonymous of the transient ischemic attack of the posterior circulation, it is important to emphasise the temporality of the symptoms and signs. Otherwise, posterior circulation stroke should be considered.
Reply: We appreciate the reviewer's correction and recommendation greatly. VBI typically arises because of the formation of atherosclerotic plaques, resulting in insufficient blood flow through the posterior circulation and is essentially thought to be a transient ischemic attack (TIA) in the posterior circulation [bib_ref] Posterior Circulation Ischemia: Then, Now, and Tomorrow : The Thomas Willis Lecture--2000, Caplan [/bib_ref]. VBI can present with transient and recurrent impairments, including dizziness, instability in standing and walking, double-vision, hearing loss, speech disorders, nauseam, or vomiting [bib_ref] Vertebral Insufficiency: When to Intervene and How?, Jenkins [/bib_ref] [bib_ref] Vertebrobasilar insufficiency, Doss [/bib_ref]. Previous studies have indicated that recurrent VBI may increase the future risk of posterior circulation stroke [bib_ref] Vertebrobasilar stenosis predicts high early recurrent stroke risk in posterior circulation stroke..., Gulli [/bib_ref] [bib_ref] New England Medical Center Posterior Circulation registry, Caplan [/bib_ref]. Furthermore, some VBI patients may suffer from progressive deterioration of memory and higher cortical functions, as well as dementia [bib_ref] Survival and cause of death in Alzheimer's disease and multiinfarct dementia, Mölsä [/bib_ref] [bib_ref] Vertebrobasilar arterial insufficiency with dementia. Controlled trials of treatment with betahistine hydrochloride, Rivera [/bib_ref].
We revised and added above information to the Introduction section (page 5, lines 4, 6 to 13; page 6, line 2).
# Methods
The subject selecting procedure is not clear. The randomisation should be explained whether the selection procedure represents the whole country or not.
Reply: We appreciate the reviewer's correction and recommendation greatly. This present study was conducted using claims data from the National Health Insurance Research Database (NHIRD), which is maintained by the National Health Research Institute (NHRI) in Taiwan. The National Health Insurance (NHI) program of Taiwan is a universal insurance program established in 1995, and it is a mandatory single-payer healthcare system, covering the healthcare costs of more than 99% of Taiwan's residents (approximately 23 million).
In this study, we used the Longitudinal Health Insurance Database (LHID) in Taiwan (20002005), which is a subset of the NHIRD. The LHID contains data of 1 million people, randomly sampled from the NHIRD, and has been demonstrated to be representative the entire population of Taiwan.
We revised and added above information to the Method section (page 7, lines 9, 14 to 22).
The diagnosis of the vertebrobasilar insufficiency and dementia is not clearly stated. The criteria for diagnosis should be described.
Reply: We appreciate the reviewer's correction and recommendation greatly. In this study, the diagnoses of VBI and dementia were based on diagnostic codes entered into the health insurance system database by the physicians.
All diagnoses of VBI were made on the basis of the clinical symptoms and signs with a focus on the cardiovascular and neurological systems and/or supporting imaging studies (Doppler ultrasonography, magnetic resonance angiography, or computed tomography angiography) from outpatient departments of participating hospitals and clinics [bib_ref] Diagnosis and management of vertebrobasilar insufficiency, Stayman [/bib_ref].
The diagnosis of dementia was made based on the diagnostic criteria of the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, after assessing the patients' medical history, daily activities, physical and social function, behavior and cognitive function, blood tests, and brain imaging.
We revised and added above information to the Method section (page 8, lines 12 to 15, 19; page 9, lines 2, 10 to 13) Also, the interrater variability of the investigators are very important, and the training of the site personal about the study is critical.
Reply: We appreciate the reviewer's recommendation greatly. The diagnosis of VBI and dementia was based on diagnostic codes entered into the health insurance system database by the physicians; therefore, additional clinical information about VBI or dementia (imaging data, laboratory and cognitive test data, or prognoses of both diseases) was not available. Thus, we could not determine whether the degree of VBI was associated with dementia severity and other dementia characteristics. Furthermore, the inter-rater variability of the investigators may also exist in this study. In future, protocols for training the site personnel to make a more correct diagnosis of VBI and dementia, and to decrease the inter-rater variability, are needed.
We added above information to the Discussion section (page 18, lines 16, 19 to page 19, line 1).
# Discussion
Although the authors have used nationwide population-based data with a large sample size to explore the strong temporal relationship between vertebrobasilar insufficiency and dementia, their data has not disclosed the causal relationship.
Reply: We appreciate the reviewer's recommendation greatly. In this study we used a longitudinal observational design, with a long follow-up period, which enabled us to assess the temporal relationship between VBI and dementia; however, our findings demonstrated only an association, but not necessarily a causal relationship. Further study is warranted to examine the potential causal relationship between VBI and dementia.
We revised and added above information to the Discussion section (page 19, lines 16 to 20).
## General
The paper is acceptable after small revisions including the detailed explanation of the limitations mentioned above. Since the methodological problems seems not to be overcame at the present time, very careful interpretation of the results may be useful for the future studies.
Reply: We appreciate the reviewer's recommendation greatly. Because of the underlying methodological limitations in this study mentioned above, the present findings in this study should be interpreted and generalized with caution.
We added above information to the Discussion section (page 19, lines 20 to 22).
Reviewer: 2 Reviewer Name: Chengxuan Qiu Institution and Country: Karolinska Institutet, Sweden Please state any competing interests: None.
## Please leave your comments for the authors below
In this large-scale patient register-based study, the authors examined the association between vertebrobasilar insufficiency (VBI) and risk of dementia and subtypes. It was concluded that patients with VBI appear to have an increased risk of developing dementia. This is an interesting and relevant study. However, I have some concerns that need to be clarified, as specified below.
1. Study population and sample: Data for this study were derived from health insurance databases of NHIRD and LHID (see page 7, please clarify the relationship of NHIRD and LHID). As I understood, although the databases (insurance) covered almost the whole general population, data in the datasets included only those people (outpatients) who sought health care services. Thus, it may not be convincing to call this study as a population-based study. The authors may wish to slightly modify the subtitle as something like "a patient register-based study".
Reply: We appreciate the reviewer's correction and recommendation greatly. This present study was conducted using claims data from the National Health Insurance Research Database (NHIRD), which is maintained by the National Health Research Institute (NHRI) in Taiwan. The National Health Insurance (NHI) program of Taiwan is a universal insurance program established in 1995, and it is a mandatory single-payer healthcare system, covering the healthcare costs of more than 99% of Taiwan's residents (approximately 23 million). In this study, we used the Longitudinal Health Insurance Database (LHID) in Taiwan (20002005), which is a subset of the NHIRD. The LHID contains data of 1 million people, randomly sampled from the NHIRD, and has been demonstrated to be representative the entire population of Taiwan. The LHID contains the demographic data, dates of clinical visits, drug prescriptions, and disease diagnosis codes in the format of the International Classification of Disease, Ninth Revision, Clinical Modification (ICD-9-CM). The NHRI reported no statistically significant differences in the sex and age distribution between the LHID subset and the total NHIRD.
In this study, we used data from the outpatient LHID in Taiwan; thus, it may not be accurate to call this study a population-based study. Thus, we have revised the title to "The association between vertebrobasilar insufficiency and the risk of dementia: A nationwide register-based retrospective cohort study in Taiwan".
We revised above information to the Title and the Method section (page 7, line 14 to page 8, line 2).
2. Matching factors in the study design: a range of comorbidities (see page 8) were matched in choosing controls for dementia cases. This matching strategy might lead to cases and controls over matched, and thus potentially mask the association between the exposure and outcome, especially given that some of the comorbidities might be on the pathways linking VBI to dementia. (limitation? ) Reply: We appreciate the reviewer's recommendation greatly. In this study, the comorbidities were matched when choosing controls for dementia cases. Although this matching strategy may have advantages in eliminating the potential influence of confounders, it may be vulnerable to overmatching, especially when the matching variable has some relationship with the outcome. For example, some of the comorbidities might also involve the pathways linking VBI to dementia, which may mask the association between the exposure and outcome.
We added above information to the Discussion section (page 19, lines 10 to 16).
3. Subtype dementia as outcomes: Clinically, Alzheimer's disease (AD) accounts for approximately 50-70% of all dementia cases and vascular dementia accounts for around 20% in the populationbased studies. However, in this study AD accounted only for just over 10% of all dementia cases (64 AD cases over 607 all dementia cases, see . The authors also acknowledged that the diagnosis of subtype dementias other than non-VD was problematic. Thus, the analysis on subtype dementias may be omitted because the results might be misleading due to unreliable diagnosis and classification of subtype dementias (e.g., AD). (supplementary?) Reply: We appreciate the reviewer's recommendation greatly. In this study, the NHIRD dataset was derived from an administrative coding database lacking details of clinical symptoms and signs of dementia, such as the predominant symptoms, mental or mood status, imaging data, and other laboratory results. Thus, the precise diagnosis of other subtypes of the dementia was not available in this study, and the results regarding the subtypes of the dementia should be interpreted with caution. Accordingly, we have revised and (the analysis of the subtypes of dementias has been omitted) and the analysis on the subtypes of dementias was shifted to Supplementary Tables S2 to S5. We have also revised the corresponding information in the Result and Discussion sections. (page 14, lines 6 to 8, 11 to 14, 16; page 19, lines 1 to 6). 4. Age-stratifying analysis: The authors carried out subgroup analysis by stratifying age into three groups, i.e., <45, 45-64, and ≥65 years of age. It is unclear what the rationale for the age-stratifying analysis is. Was any statistical interaction between VBI and age group on dementia risk detected? The authors stated that "…the risk of developing dementia was most prominent in those aged 45-64 or over 65 years" (e.g., page 18, 22). This is actually not true because the point estimation of HR for dementia in people aged <45 years was even greater than those for persons aged 45-64 and those aged 65+ (4.019 vs. 2.953 and 1.752, see , and the lack of statistical significance is likely due to limited power in the younger age group (very few cases of dementia). However, the authors may consider to investigate the associations of VBI with early-onset (onset age >65) and late-onset (onset age ≥65) dementia, separately. This is relevant because early-onset and late-onset dementia are thought to have different risk factors.
Reply: We appreciate the reviewer's correction and recommendation greatly. Although dementia mostly affects older people, those aged ≥ 65 years, it can occur in people aged < 65 years, known as young-onset dementia [bib_ref] T he diagnosis of young-onset dementia, Rossor [/bib_ref]. Therefore, we performed analysis of age as a two-level ordinal variable (with cutoffs at 65 years).
In comparison with age/sex-matched controls, the VBI group had a higher risk of all-cause dementia than the non-VBI group, regardless of age (< 65 years: HR: 2.997, 95% CI: 1.4516.454, p < 0.001; ≥ 65 years: HR: 1.752, 95% CI: 1.5841.937, p < 0.001).
We revised above information to the Abstract, Method, Result and Discussion sections (page 3, lines 18 to 20; page 9, line 23; page 10, lines 6, 7 to 10; page 12, lines 6 to 9; page 14, lines 11 to 14; page 16, lines 5 to 7; page 17, lines 10, 16, 17, 18, 23; page 19, line 25).
## Strengths and limitations (pages 20-21):
The authors may wish to carefully rephrase the strengths and limitations of this study because they appeared to be either contradictory or not true. For instance, on one hand "… while also considering a variety of covariates" was claimed as part of strength", on the other hand as limitation, many "… personal potential confounders, such as lifestyle …… were not available." I notice that education, as one of the most relevant confounders in studying dementia, was not listed in . Similarly, "… diagnosis and coding in this study are extremely reliable, …" as a strength vs. "… additional clinical information about VBI or dementia …. was not available, …" as a limitation.
Reply: We greatly appreciate the reviewer's correction and recommendation. We have revised our strength and limitations as follows:
The major strength of the present study is the use of a large sample size (NHIRD), which provided sufficient statistical power to explore the relationship between VBI and dementia. However, the present study also had certain limitations. First, the diagnosis of VBI and dementia was based on diagnostic codes entered into the health insurance system database by the physicians; therefore, additional clinical information about VBI or dementia (imaging data, laboratory and cognitive test data, or prognoses of both diseases) was not available. Thus, we could not determine whether the degree of VBI was associated with dementia severity and other dementia characteristics. Furthermore, the inter-rater variability of the investigators may also exist in this study. In future, protocols for training the site personnel to make a more correct diagnosis of VBI and dementia, and to decrease the interrater variability, are needed. Second, in this study, the NHIRD dataset was derived from an administrative coding database lacking details of clinical symptoms and signs of dementia, such as the predominant symptoms, mental or mood status, imaging data, and other laboratory results. Thus, the precise diagnosis of other subtypes of the dementia was not available in this study, and the results regarding the subtypes of the dementia should be interpreted with caution. Third, as a retrospective longitudinal study, we had no access to information about some adaptable or personal potential confounders, such as education, lifestyle (smoking or alcohol consumption), physical activity, body mass index, social engagement, family histories, or medications, which were not available in the NHIRD, and which may potentially confound our results. Fourth, in this study, the comorbidities were matched when choosing controls for dementia cases. Although this matching strategy may have advantages in eliminating the potential influence of confounders, it may be vulnerable to overmatching, especially when the matching variable has some relationship with the outcome. For example, some of the comorbidities might also involve the pathways linking VBI to dementia, which may mask the association between the exposure and outcome. Finally, in this study we used a longitudinal observational design, with a long follow-up period, which enabled us to assess the temporal relationship between VBI and dementia; however, our findings demonstrated only an association, but not necessarily a causal relationship. Further study is warranted to examine the potential causal relationship between VBI and dementia. Collectively, because of the underlying methodological limitations in this study mentioned above, the present findings in this study should be interpreted and generalized with caution.
We corrected and added above information to the Discussion section (page 18, line 16 to page 19, line 22). 6. Minors: (1) The authors may rephrase the statement "Dementia is one of the most common neurodegenerative …" It should refer to Alzheimer's disease specifically, not dementia in general.
Reply: We greatly appreciate the reviewer's correction and recommendation. We have rephrased the statement to "Alzheimer's disease (AD) is one of the most common neurodegenerative diseases, …".
We corrected above information to the Introduction section (page 5, lines 14, 16, and 17).
(2) Please clarify the relationship between the two databases NHIRD and LHID.
Reply: We appreciate the reviewer's recommendation greatly. This present study was conducted using claims data from the National Health Insurance Research Database (NHIRD), which is maintained by the National Health Research Institute (NHRI) in Taiwan. The National Health Insurance (NHI) program of Taiwan is a universal insurance program established in 1995, and it is a mandatory single-payer healthcare system, covering the healthcare costs of more than 99% of Taiwan's residents (approximately 23 million). In this study, we used the Longitudinal Health Insurance Database (LHID) in Taiwan (20002005), which is a subset of the NHIRD. The LHID contains data of 1 million people, randomly sampled from the NHIRD, and has been demonstrated to be representative the entire population of Taiwan. The LHID contains the demographic data, dates of clinical visits, drug prescriptions, and disease diagnosis codes in the format of the International Classification of Disease, Ninth Revision, Clinical Modification (ICD-9-CM). The NHRI reported no statistically significant differences in the sex and age distribution between the LHID subset and the total NHIRD. We added above information to the Method section (page 7, line 14 to page 8, line 2). .
(3) Age may still need to be controlled for as a continuous variable in the analysis given its strong impact on dementia occurrence and the difference in the mean age between cases and controls as shown in (although age is a matching factors).
Reply: We appreciate the reviewer's recommendation greatly. Since age has a potential impact on dementia occurrence, we also examined the associations between VBI and the risk of dementia after controlling for age as a continuous variable. The results were similar, as shown in .
We added Supplementary Table S1 and above information to the Result section (page 12, lines 13 to 15).
(4) The authors may justify the analyses according to the length of follow-up time. |
Severe Acute Pancreatitis with Candida Endophthalmitis
Severe acute pancreatitis (SAP) is a risk factor for candidemia. We report a case of candida endophthalmitis in a 67-year-old man who was admitted to a hospital due to SAP with poorly controlled diabetes. After treatment for SAP, he was diagnosed with candidemia and candida endophthalmitis. We chose appropriate antifungal agents based on the results of a bacterial culture test. After treatment, the disappearance of Candida albicans (C. albicans) from the blood stream was confirmed in blood cultures. In addition, exudative plaques consistent with a fungal infection disappeared. After a diagnosis of candidemia is made, it is important to administer appropriate antifungal therapy and perform frequent ophthalmologic examinations.
# Introduction
Although severe acute pancreatitis (SAP) is a benign disease, the mortality rate remains high. Fungal infections are common in patients with infected pancreatic necrosis and pseudocysts. Fungal infections indicate patients with a high risk of mortality in the long term [bib_ref] Fungal infections in patients with infected pancreatic necrosis and pseudocysts: risk factors..., Reuken [/bib_ref]. Many articles have reported that most fungal infections occur late in the clinical course of acute pancreatitis (AP) and are associated with pancreatic fluid collection. There are few reports of fungal infection early in the clinical course. In this case report, a 67-year-old man with poorly controlled diabetes developed candidemia and candida endophthalmitis early in the clinical course of SAP. We suggest that it is important to detect the source of infection and implement appropriate approaches to eliminate fungal infections when SAP patients develop high fever.
## Case report
A 67-year-old man was hospitalized with SAP at a hospital near ours in June 2017. He was transferred to our hospital to receive treatment for SAP 3 days after his initial hospitalization. His past medical history included chronic pan-creatitis and diabetes. He had consumed 5 liters of beer every day for 47 years. A physical examination conducted after his admission to our hospital revealed the following findings: body temperature, 36.6 ; blood pressure, 172/94 mmHg; heart rate, 96 beats/min; and respiratory rate, 21 breaths/min. His abdomen was hard and flat with tenderness and defensiveness. A blood test revealed an increased white blood cell count (WBC, 20,000/μL) and increased serum concentrations of C-reactive protein (CRP, 31.03 mg/dL), amylase (280 IU/L), and lipase (254 U/L). Although he was undergoing insulin treatment at a nearby hospital, his HbA1c (National Glycohemoglobin Standardization Program) level was 9.6% [fig_ref] Table 1: Laboratory Data on Admission [/fig_ref]. Contrast-enhanced computed tomography (CE-CT) revealed low enhancement of the pancreatic parenchyma of the pancreas head with surrounding fluid collection [fig_ref] Figure 1: The contrast-enhanced computed tomography [/fig_ref]. Based on the criteria of the Japanese Ministry of Health, Labour and Welfare, 2008 (2), the patient's prognostic factor score was 2 points (CRP, >15 mg/dL and SIRS), and his CE-CT grade was 2. Thus, we diagnosed the patient with SAP.
He received fluid resuscitation (100 mL/kg, daily), which was started at admission. He was treated with nafamostat mesylate (240 mg, daily) for 5 days and antibiotics (meropenem 1.5 g, daily) for 10 days. He stayed in the intensive care unit (ICU) for 12 days [fig_ref] Figure 2: The early clinical course of SAP [/fig_ref]. His abdominal pain and inflammation improved. However, he developed a high fever 16 days after being hospitalized. The patient did not have a urinary tract infection, pneumonia or infected pancreatic fluid. Candida albicans (C. albicans) was detected in both blood and central venous catheter cultures. The blood level of β-D glucan obtained on the same day was 47.4 pg/mL. We diagnosed the patient with candidemia; thus, ophthalmologic examinations were performed. Exudative plaques consistent with a fungal infection were observed in both retinas after a fundoscopic examination [fig_ref] Figure 3: The fundoscopic and fluorescent fundoscopic findings [/fig_ref]. He was diagnosed with candida endophthalmitis. Fortunately, he did not experience a visual disturbance. We chose appropriate antifungal agents based on the bacterial culture test results; micafungin (MCFG, 150, mg daily), caspofungin (CPFG, 75 mg, daily), and flu-conazole (FLCZ, 800 mg, daily) were administered with intravenous fluids for 59 days [fig_ref] Table 2: The Bacterial Culture Test for C [/fig_ref]. The blood cultures were evaluated weekly, and the disappearance of C. albicans from the blood stream was confirmed. Exudative plaques consistent with a fungal infection disappeared after another fundoscopic examination. The blood levels of β-D glucan became negative (19.8 pg/mL) 50 days after hospitalization [fig_ref] Figure 4: Change in the blood levels of β-D glucan after admission [/fig_ref]. After receiving treatment for candidemia and candida endophthalmitis, he was discharged 95 days after hospitalization without any complications of AP.
# Discussion
Candidemia is a life-threatening infection that is associated with a high rate of mortality and an increased length of hospital stay. The risk factors for candidemia include the use of broad-spectrum antibacterial agents, the use of central venous catheters, receipt of parenteral nutrition, receipt of renal replacement therapy by ICU patients, neutropenia, the use of implantable prosthetic devices, and the receipt of immunosuppressive agents (including glucocorticosteroids, chemotherapeutic agents, and immunomodulators) (3).
In the present case, candidemia occurred at a relatively early phase, and a catheter Candida infection was also proven. It is thought that percutaneous or transcatheter catheter infection occurred and led to candidemia and endophthalmitis. Candidemia may occur from the gastrointestinal tract or due to catheter infection. Because candida is normally distributed throughout the gastrointestinal tract, invasive procedures increase the risk of candidemia [bib_ref] Intravitreal liposomal amphotericin B for treatment of endogenous candida endophthalmitis, Bae [/bib_ref]. AP has been reported to be associated with increased gut mu- Fungal infection is also said to be caused by microbial substitution due to the prolonged prophylactic administration of broad-spectrum antibiotics. SAP carries with it a high risk of fungal infection because the Japanese guidelines for the management of AP (Japanese Guidelines 2015: JPNGL2015) recommend the prophylactic administration of broadspectrum antibiotics to patients with SAP and necrotizing pancreatitis, which may improve their prognosis if antibiotics are administered in the early phase of pancreatitis (within 72 hours of the onset) (5). Additionally, the median time from the onset of AP to the isolation of fungi is generally 7 weeks in pancreatic fluid [bib_ref] Candida lipolytica candidemia as a rare infectious complication of acute pancreatitis: a..., Liu [/bib_ref]. Thus, it is rare to isolate fungi early in the clinical course of AP.
In JPNGL 2015, the effect of the prophylactic administration of antifungal agents in patients with AP is not clear, and the routine administration of these agents is not recommended [bib_ref] Japanese guidelines for the management of acute pancreatitis: Japanese Guidelines, Mayumi [/bib_ref]. De et al. reported that early treatment with fluconazole did not affect the length of hospital stay or mortality of patients with pancreatitis (7). Furthermore, Shorr et al. pointed out that the prophylactic administration of antifungal agents led to an increased frequency of infection with resistant strains of Candida (8).
Feman et al. reported that the incidence of ocular candidiasis in patients with candidemia was only 2% (9). Ocular candidiasis is divided into chorioretinitis and endophthalmitis. Chorioretinitis involves inflammation of the choroid and retina, while endophthalmitis involves inflammation of the vitreous body. The pathogenic pathway of candida endophthalmitis is as follows: candida reaches the choroid and retina through hematogenous spread and proliferates, causing retinal abscesses and subsequent vitreous involvement. Nagao et al. reported that C. albicans as the etiological agent (OR: 3.68, 95%CI: 1.11-12.2, p=0.034) and higher β-D glucan values (OR: 9.99, 95%CI: 2.60-21.3, p=0.001) were identified as statistically significant risk factors for ocular candidiasis in a multivariate analysis (10). Lingappan et al. reported that the incidence of retinal detachment in patients with fungal endophthalmitis was 29% [bib_ref] Endogenous fungal endophthalmitis: causative organisms, management strategies, and visual acuity outcomes, Lingappan [/bib_ref]. Some cases of fungal endophthalmitis may lead to blindness due to retinal detachment.
The primary recommended treatments for candida endophthalmitis are amphotericin (AMPH-B; 0.7-1 mg/kg) with flucytosine (5-FC; 25 mg/kg) every 6 hours, fluconazole (FLCZ; 6-12 mg/kg) daily, and surgical intervention for patients with severe endophthalmitis or vitreitis. Alternative therapy is recommended for patients intolerant of AMPH-B and 5-FC therapy and patients who experience or therapeutic failure with these treatments. The duration of therapy is at least 4-6 weeks, as determined by repeated examinations to verify the resolution. Diagnostic vitreal aspiration is recommended for patients with endophthalmitis of unknown origin (3).
There have been no reports of pancreatitis with candida endophthalmitis. In this case, the patient had multiple risks factors for candidemia. He used broad-spectrum antibacterial agents and central venous catheters and had stayed in the ICU for a long time. It is thought that the combination of his poorly controlled diabetes and central venous catheter infection led to the development of candidemia. We believe that the patient's candidiasis was triggered by a catheter infection, and that this was complicated by candida endophthalmitis through hematogenous spread. Generally, candidemia associated with pancreatitis occurs later in the clinical course because of microbial substitution and infected pancreatic fluid. In emergency treatment for diseases such as SAP, central venous catheters are often inserted. Candida endophthalmitis lacks initial symptoms and should be included in the differential diagnosis of patients with SAP in the early phase of the condition. It is recommended that a fundoscopic examination be quickly performed when candidemia and ocular symptoms are detected in the early phase of SAP.
When we encounter an infection of unknown etiology, we should consider the incidence of candidemia and whether the patients have any risk factors for candidemia. The administration of appropriate antifungal therapy and the frequent performance of ophthalmologic examinations are important for patients diagnosed with candidemia.
The authors state that they have no Conflict of Interest (COI).
[fig] Figure 1: The contrast-enhanced computed tomography (CE-CT) findings on admission. These findings revealed low enhanced pancreatic parenchyma of the pancreas head with surrounding fluid collection. [/fig]
[fig] Figure 2: The early clinical course of SAP. MEPM: meropenem [/fig]
[fig] Figure 3: The fundoscopic and fluorescent fundoscopic findings. Exudative plaques consistent with a fungal infection were observed in both retinas. A, B: Fundoscopic findings. C, D: Fluorescent fundoscopic findings. [/fig]
[fig] Figure 4: Change in the blood levels of β-D glucan after admission. MCFG: micafunginin, CPFG: caspofungin, FLCZ: fluconazole cosal permeability, predisposing patients to septic complications. The translocation of fungal flora from the gastrointestinal tract to extraintestinal sites plays an important role in the pathogenesis of fungal infections in patients with AP. [/fig]
[table] Table 1: Laboratory Data on Admission. [/table]
[table] Table 2: The Bacterial Culture Test for C. albicans. MIC: minimum inhibitory concentration, AMPH-B: amphotericin, 5-FC: flucytosine, MCFG: micafunginin, VRCZ: voriconazole, MCZ: miconazole, FLCZ: fluconazole, ITCZ: Itraconazole [/table]
|
Changes in Physical Activity Following a Genetic-Based Internet-Delivered Personalized Intervention: Randomized Controlled Trial (Food4Me)
Background: There is evidence that physical activity (PA) can attenuate the influence of the fat mass-and obesity-associated (FTO) genotype on the risk to develop obesity. However, whether providing personalized information on FTO genotype leads to changes in PA is unknown.Objective: The purpose of this study was to determine if disclosing FTO risk had an impact on change in PA following a 6-month intervention.Methods:The single nucleotide polymorphism (SNP) rs9939609 in the FTO gene was genotyped in 1279 participants of the Food4Me study, a four-arm, Web-based randomized controlled trial (RCT) in 7 European countries on the effects of personalized advice on nutrition and PA. PA was measured objectively using a TracmorD accelerometer and was self-reported using the Baecke questionnaire at baseline and 6 months. Differences in baseline PA variables between risk (AA and AT genotypes) and nonrisk (TT genotype) carriers were tested using multiple linear regression. Impact of FTO risk disclosure on PA change at 6 months was assessed among participants with inadequate PA, by including an interaction term in the model: disclosure (yes/no) × FTO risk (yes/no).Results: At baseline, data on PA were available for 874 and 405 participants with the risk and nonrisk FTO genotypes, respectively. There were no significant differences in objectively measured or self-reported baseline PA between risk and nonrisk carriers. A total of 807 (72.05%) of the participants out of 1120 in the personalized groups were encouraged to increase PA at baseline. Knowledge of FTO risk had no impact on PA in either risk or nonrisk carriers after the 6-month intervention. Attrition was higher in nonrisk participants for whom genotype was disclosed (P=.01) compared with their at-risk counterparts.Conclusions:No association between baseline PA and FTO risk genotype was observed. There was no added benefit of disclosing FTO risk on changes in PA in this personalized intervention. Further RCT studies are warranted to confirm whether disclosure of nonrisk genetic test results has adverse effects on engagement in behavior change. Trial Registration: ClinicalTrials.gov NCT01530139; http://clinicaltrials.gov/show/NCT01530139 (Archived by WebCite at: http://www.webcitation.org/6XII1QwHz) (J Med Internet Res 2016;18(2):e30)
# Introduction
The prevalence of physical inactivity in Europe and worldwide is high [bib_ref] Global physical activity levels: Surveillance progress, pitfalls, and prospects, Hallal [/bib_ref]. Given that physical inactivity is among the top risk factors for noncommunicable diseases, finding effective ways to achieve long-lasting improvements in physical activity (PA) remains a major challenge [bib_ref] Physical activity and all-cause mortality across levels of overall and abdominal adiposity..., Ekelund [/bib_ref]. While previous intervention strategies have mainly focused on a "one-size-fits-all" approach to change behavior, recent studies have used personalized approaches, such as tailored Web-based interventions [bib_ref] Impact of personalised feedback about physical activity on change in objectively measured..., Godino [/bib_ref] [bib_ref] Effect of a Web-based intervention to promote physical activity and improve health..., Hansen [/bib_ref]. There is inconsistent evidence on whether these personalized approaches are more effective at increasing PA than standard guidelines, and effects, when present, are often small and with short-term efficacy [bib_ref] A systematic review of randomized controlled trials on the effectiveness of computer-tailored..., Broekhuizen [/bib_ref]. Concurrently, there has been a growing interest in using genetic information to personalize lifestyle interventions [bib_ref] Has the revolution arrived?, Collins [/bib_ref]. Although disclosure of such information does not appear to have unintended adverse effects, more randomized controlled trials (RCTs) are needed to establish whether gene-based personalized interventions promote greater behavior change than conventional "one-size-fits-all" interventions [bib_ref] Effects of communicating DNA-based disease risk estimates on risk-reducing behaviours, Marteau [/bib_ref].
In particular, data on whether providing genetic information leads to an increase in PA are lacking.
The fat mass-and obesity-associated (FTO) gene has provided strong evidence of the genetic susceptibility to obesity. Polymorphisms in this gene located in intron 1 and exon 2 have been shown to be consistently and strongly associated with obesity-related markers [bib_ref] The first gene contributing to common forms of human obesity, Loos [/bib_ref] [bib_ref] The bigger picture of FTO: The first GWAS-identified obesity gene, Loos [/bib_ref]. For instance, individuals homozygous for the higher-risk allele, AA, of single nucleotide polymorphism (SNP) rs9939609 in the FTO gene FTO weighed, on average, 3 kg more and had 1.7-fold increased odds of having obesity compared with those homozygous for the lower-risk allele, TT [bib_ref] A common variant in the FTO gene is associated with body mass..., Frayling [/bib_ref]. Moreover, there is increasing evidence that the FTO genetic susceptibility to obesity can be modulated by lifestyle factors, and that PA, for example, may attenuate the effects of the FTO genotype on obesity-related traits [bib_ref] Low physical activity accentuates the effect of the FTO rs9939609 polymorphism on..., Andreasen [/bib_ref] [bib_ref] Interaction of FTO and physical activity level on adiposity in African-American and..., Demerath [/bib_ref] [bib_ref] Assessing the effect of interaction between an FTO variant (rs9939609) and physical..., Jonsson [/bib_ref] [bib_ref] Physical activity attenuates the influence of FTO variants on obesity risk: A..., Kilpeläinen [/bib_ref] [bib_ref] Physical activity and the association of common FTO gene variants with body..., Rampersaud [/bib_ref] [bib_ref] Physical activity attenuates the body mass index-increasing influence of genetic variation in..., Vimaleswaran [/bib_ref]. However, to our knowledge there is no data on whether disclosing information on FTO genotype can motivate individuals to increase their PA. Elucidating whether genetic-based advice can promote improvements in PA behaviors may help in the design of more effective interventions, especially when tailored to individuals who would benefit most from increasing their PA.
As part of the Food4Me study (ClinicalTrials.gov number: NCT01530139)-a Web-based RCT in 7 European countries-we investigated the effects of 3 levels of personalized advice on changes in PA, including a level with genetic information on FTO [bib_ref] Design and baseline characteristics of the Food4Me study: A Web-based randomised controlled..., Celis-Morales [/bib_ref] [bib_ref] Effects of a Web-based personalized intervention on physical activity in European adults:..., Marsaux [/bib_ref]. See Multimedia Appendix 1 for the CONSORT-EHEALTH checklist [bib_ref] CONSORT-EHEALTH: Improving and standardizing evaluation reports of Web-based and mobile health interventions, Eysenbach [/bib_ref]. We found that personalized feedback in general led to greater improvements in self-reported PA, but not in objectively measured PA, compared with standard guidelines [bib_ref] Effects of a Web-based personalized intervention on physical activity in European adults:..., Marsaux [/bib_ref]. However, we did not investigate the effect of disclosing genetic-based information on PA change, and whether the response differs between carriers of a genetic risk and nonrisk carriers. Thus, the aim of these analyses was to assess the impact of knowledge of FTO risk status on change in self-reported and objectively measured PA in Food4Me participants.
# Methods
## Subjects
Subjects were participants of the Food4Me study, a 6-month, Web-based RCT on personalized nutrition and lifestyle conducted in 7 European countries-Germany, Greece, Ireland, the Netherlands, Poland, Spain, and the United Kingdom. As outlined elsewhere [bib_ref] Design and baseline characteristics of the Food4Me study: A Web-based randomised controlled..., Celis-Morales [/bib_ref] , 1607 adults aged ≥18 years were randomized to the study. Exclusion criteria included no or limited access to the Internet, following a prescribed diet, or having altered nutritional requirements because of a medical condition. The local ethics committee of each recruiting center approved the study protocol and all subjects provided informed consent digitally before participating.
## Study design
Participants were randomly allocated to one of the 4 groups-Level 0: standard, nonpersonalized, dietary and PA guidelines; Level 1: dietary and PA advice based on current diet and PA; Level 2: dietary and PA advice based on current diet, PA, and phenotype (eg, waist circumference and blood cholesterol); and Level 3: dietary and PA advice based on current diet, PA, phenotype, and genotype (eg, FTO). The randomization scheme has been described previously [bib_ref] Design and baseline characteristics of the Food4Me study: A Web-based randomised controlled..., Celis-Morales [/bib_ref]. All data were collected remotely following standardized operating procedures. At baseline, participants received study kits by post containing all necessary materials, such as an accelerometer and DNA collection kit (see the Physical Activity Assessment and Genotyping sections below), to perform measurements at home, but used their own scales to measure body weight. Printed instructions were included and demonstration videos were available on the Food4Me website [bib_ref] Design and baseline characteristics of the Food4Me study: A Web-based randomised controlled..., Celis-Morales [/bib_ref].
On the allocated study day and following an 8-hour overnight fast, participants collected a buccal cell sample for DNA; measured their height, weight, and waist circumference; and started wearing an accelerometer. The buccal cell sample was returned to the research center in a prepaid stamped addressed envelope and anthropometric measurement values were self-reported online. Questionnaires to be completed online the same day included the Baecke PA questionnaire (see the Physical Activity Assessment section below). Participants repeated the measurements, except DNA collection, at 3 and 6 months [bib_ref] Design and baseline characteristics of the Food4Me study: A Web-based randomised controlled..., Celis-Morales [/bib_ref].
Following measurements at baseline and 3 months, participants received, at both time points, a personalized (Levels 1-3) or nonpersonalized (Level 0) report, including feedback on PA according to their group. The personalized feedback provided was based on a predefined set of algorithms, including anthropometric, PA (Levels 1-3), phenotypic (Levels 2 and 3), and genotypic (Level 3 only) data. Results in the personalized report were compared with recommendations for each anthropometric, PA (Levels 1-3), and phenotypic (Levels 2 and 3) item, using 3-color graded lines-green: good; amber: improvement recommended; and red: improvement strongly recommended. In addition, Level 3 participants received information in their report about 5 diet-and lifestyle-related genes [bib_ref] Design and baseline characteristics of the Food4Me study: A Web-based randomised controlled..., Celis-Morales [/bib_ref]. For FTO, the message was "A specific variation of this gene is associated with a greater need to maintain a healthy body weight and engage in physical activity. A healthy weight combined with exercise may provide added health benefits for these individuals." Participants were informed whether they were carriers of the risk variant for the FTO SNP rs9939609 (yes or no, if they were genotyped AA or AT, or TT, respectively). Each personalized report (Levels 1-3) also contained a specific message related to body weight and PA. Additionally, for Level 3 participants this specific message referred to FTO. For example, for an AA/AT participant with increased body mass index (BMI), increased waist circumference, and low PA, the message was "We recommend reducing your body weight and waist circumference to a healthy normal range because you have a genetic variation that can benefit by reducing these 2 obesity markers. Also, your physical activity level is too low." Full details of the study design have been published elsewhere [bib_ref] Design and baseline characteristics of the Food4Me study: A Web-based randomised controlled..., Celis-Morales [/bib_ref].
## Physical activity assessment
## Objective physical activity
PA was assessed objectively using the TracmorD triaxial accelerometer (Philips Consumer Lifestyle, the Netherlands) [bib_ref] Estimation of free-living energy expenditure using a novel activity monitor designed to..., Bonomi [/bib_ref]. Participants were instructed to wear the accelerometer every day while awake, except when taking a shower, for the entire duration of the 6-month study. Participants uploaded data every 2 weeks onto the study server via the Internet. Data were recorded with a time-sampling interval of 1 min. A day was considered valid if the participant had worn the TracmorD accelerometer between 10 and 18 h. Wear time was defined as 24 h minus nonwear time. To define nonwear time, we adapted the recommendations of Choi et al [bib_ref] Validation of accelerometer wear and nonwear time classification algorithm, Choi [/bib_ref] to the TracmorD accelerometer. R software version 3.1.2 (The R Foundation)was used for PA data processing.
Daily PA level (PAL)-the ratio of total energy expenditure to basal metabolic rate-was derived from activity counts [bib_ref] Estimation of free-living energy expenditure using a novel activity monitor designed to..., Bonomi [/bib_ref]. Time spent in sedentary behavior-corresponding to <1.5 metabolic equivalents (METs)-and moderate-and vigorous-intensity PA-3 to <6 METs and ≥6 METs, respectively-were calculated based on the application of thresholds for activity energy expenditure (AEE) equivalent to the METs thresholds. Daily AEE was calculated as follows:
Daily AEE = (0.9 × daily PAL -1) × BMR (1) where the daily basal metabolic rate (BMR) is estimated using the Oxford equations developed by Henry, based on sex, age, and weight [bib_ref] Basal metabolic rate studies in humans: Measurement and development of new equations, Henry [/bib_ref]. PA estimates were calculated over a 2-week period at baseline and 6 months. This 2-week assessment period occurred before any feedback was given for the corresponding time point. Sufficient PA data at each time point was defined as having at least 3 valid weekdays and 2 valid weekend days of accelerometer wear during the 2-week period. For individuals with sufficient PA data, mean data per day were calculated based on all valid week and weekend days of the assessment period as follows: Mean = (mean for weekdays × 5 + mean for weekend days × 2) / 7 (2).
For sedentary time and time spent in moderate PA and vigorous PA, weekly estimates were calculated as follows: Mean = (mean for weekdays × 5 + mean for weekend days × 2) (3).
## Self-reported physical activity
At each time point, participants completed the Baecke questionnaire online [bib_ref] A short questionnaire for the measurement of habitual physical activity in epidemiological..., Baecke [/bib_ref] based on their PA during the last month. This short, extensively validated questionnaire [bib_ref] Reliability and validity of three physical activity questionnaires in Flemish males, Philippaerts [/bib_ref] [bib_ref] Doubly labelled water validation of three physical activity questionnaires, Philippaerts [/bib_ref] is composed of 3 sections-work, sport, and nonsport leisure-with indices ranging from 1 to 5 and a sum total (ie, total activity index) ranging from 3 to 15. Scores were calculated at baseline and month 6, according to the questionnaire protocol [bib_ref] A short questionnaire for the measurement of habitual physical activity in epidemiological..., Baecke [/bib_ref].
## Genotyping
Participants collected a buccal cell sample at baseline, using Isohelix SK-1 DNA buccal swabs and Isohelix Dri-capsules (LGC Genomics, Hertfordshire, UK). Samples were returned to the recruiting centers and shipped to LGC Genomics, who extracted the DNA and used competitive allele-specific polymerase chain reaction (KASP) genotyping assays to provide biallelic scoring of SNP rs9939609 in the FTO gene.
## Statistical analyses
Data are presented as means (SD) for continuous variables and as percentages for categorical variables, unless otherwise stated. A chi-square test was used to test if the observed FTO genotype counts were in Hardy-Weinberg equilibrium. To examine if there was an association between PA and FTO genotype, we used baseline data and robust multiple linear regression models, based on computation of SMDM estimatesto account for violation of the normality assumption. FTO genotype was operationalized as risk (AA and AT) and nonrisk (TT).
To study the impact of knowledge of FTO risk status on changes in PA, we used two approaches. In the first approach or primary analysis, we investigated whether personalized advice based on genetic information (ie, FTO risk) was more effective at increasing PA than personalized advice without genetic-based information in FTO risk and nonrisk carriers. In this analysis, we compared Level 3 participants who received personalized advice to increase PA, including disclosure of FTO risk, with participants who received personalized advice to increase PA without any genetic-based information (pooled Levels 1 and 2). As a secondary analysis, we assessed whether personalized advice based on genetic information (ie, FTO risk) was more effective at increasing PA than standard guidelines (ie, nonpersonalized advice) in FTO risk and nonrisk carriers. This analysis compared Level 3 participants with the control group-Level 0, nonpersonalized guidelines. In order to match the characteristics of both groups, participants in the control group were included only if they had insufficient baseline PA (ie, they would have been advised to increase their PA if they had not been in the control group). For both primary and secondary analyses, we used robust multiple regression models, including an interaction term between FTO risk (yes or no) and disclosure of genetic information (yes or no). If there was no significant interaction, we looked at the main effects after removing the interaction term from the model. Models were adjusted for age, sex, country, BMI, season, accelerometer wear time, and baseline PA variable as appropriate. Additional sensitivity analyses were run, stratifying by sex and by tertile of baseline PA variables. Attrition rates between groups were compared using Pearson's chi-square tests. R software version 3.1.2 (The R Foundation)was used to perform all analyses and the significance level was set at P<.05.
# Results
## Attrition rate and compliance
A total of 1607 individuals were randomized into the study (see [fig_ref] Figure 1: Flowchart of study procedures [/fig_ref] and 127 (7.90%) of them dropped out before starting the trial; their characteristics will be reported elsewhere. Genotype and PA data were available for 1279 of the 1480 (86.42%) starters, which were therefore included in the baseline analysis (see [fig_ref] Figure 1: Flowchart of study procedures [/fig_ref]. Although sufficient accelerometer data were defined as having a minimum of 3 valid weekdays and 2 valid weekend days of accelerometer wear, 77.56% (992/1279) of subjects had 10 or more valid days of accelerometer wear at baseline-mean 11.3 days (SD 2.4): 8.2 weekdays (SD 1.9) and 3.2 weekend days (SD 0.8).
Among the 1120 participants who received personalized advice (Levels 1-3), 807 (72.05%) were advised to increase their PA following assessment of baseline PA. Similarly, in the control group (Level 0), 276 of 360 (76.7%) participants would have been advised to increase their PA if the algorithms applied to Levels 1-3 had been applied to the control group (see [fig_ref] Figure 1: Flowchart of study procedures [/fig_ref]. For these participants with inadequate PA, attrition rate was similar between groups (14-15%, P=.45) at month 6 (see [fig_ref] Figure 1: Flowchart of study procedures [/fig_ref]. In the group where FTO risk was disclosed (Level 3), participants with the nonrisk (TT) genotype were more likely to drop out of the intervention than the at-risk (AA/AT) participants-attrition rate 22% (TT) versus 12% (AA/AT); odds ratio (OR) 2.04, 95% CI 0.96-4.29, P=.04. This was also the case when considering all participants in Level 3 (ie, not only those advised to increase their PA)-attrition rate 20% (TT) versus 11% (AA/AT); OR 2.17, 95% CI 1.13-4.17, P=.01 (see [fig_ref] Table 1: Attrition rates after 6 months by intervention level [/fig_ref]. There were no significant differences in attrition rates after 6 months between risk and nonrisk carriers in any other groups.
Although only 157 out of 1083 (14.50%) participants with inadequate PA had dropped out by month 6, compliance with wearing the accelerometer decreased during the study. Thus, 46.45% (503/1083) of subjects had data on FTO genotype, objective PA, and self-reported PA for both baseline and month 6, and were included in the analyses on change in PA (see [fig_ref] Figure 1: Flowchart of study procedures [/fig_ref]. Of these, 85.5% (430/503) and 68.0% (342/503) had 10 days of valid accelerometer wear at baseline and month 6, respectively. Mean number of valid days of accelerometer wear for these participants was 11.9 days (SD 2.1) at baseline-8.6 weekdays (SD 1.7) and 3.3 weekend days (SD 0.7)-and 10.4 days (SD 3.0) at month 6-7.7 weekdays (SD 2.3) and 2.7 weekend days (SD 1.1). This was similar for all intervention groups (data not shown for Levels 0-3). [fig_ref] Table 1: Attrition rates after 6 months by intervention level [/fig_ref]
## Physical activity and fto genotype
The characteristics of the 1279 participants with baseline PA data both from accelerometers and self-reports, as well as data on FTO genotype, are presented in [fig_ref] Table 2: Characteristics of the participants included in baseline analysis [/fig_ref]. Most participants were white, 743 (58.09%) were women, and 588 (45.97%) were overweight or obese. Genotype frequency for FTO rs9939609 did not deviate from Hardy-Weinberg equilibrium (TT=405, TA=641, and AA=233; P=.48). We found no association between objectively measured PAL (P=.35), moderate PA (P=.28), vigorous PA (P=.24), or sedentary time (P=.71) at baseline and FTO risk status (see [fig_ref] Figure 2: Physical activity in FTO rs9939609 risk [/fig_ref] , section a). Similarly, there was no significant difference in baseline self-reported PA between risk and nonrisk carriers (P=.76) (see [fig_ref] Figure 2: Physical activity in FTO rs9939609 risk [/fig_ref] , section b). [fig_ref] Table 3: Changes in physical activity [/fig_ref] displays the PA characteristics of genotyped participants advised to increase their PA at baseline, with objective PA and self-reported PA data at baseline and month 6. There was no significant interaction between disclosure of genetic information and FTO risk status on change in objectively measured or self-reported PA (all P>.25); this is illustrated in . There was also no effect of knowledge of FTO genotype on objectively measured or self-reported PA (all P>.10) (see [fig_ref] Table 3: Changes in physical activity [/fig_ref] and .
## Primary analysis: effect of disclosing fto genotype status on change in physical activity
## Secondary analysis: personalized feedback including disclosure of genetic information compared with standard guidelines
Comparisons between participants in the highest level of personalization (Level 3) who were advised to increase their PA, and control participants (Level 0) who would have been advised to increase PA if they had been in a personalized group, are given in Multimedia Appendix 2. There were no significant interactions between intervention levels and FTO risk status on change in PA. Change in objectively measured PA did not differ significantly between Level 3 and Level 0 participants for both risk and nonrisk carriers. However, Level 3 participants, irrespective of FTO risk status, had greater changes in self-reported PA than Level 0 participants (see Multimedia Appendix 2).
## Sensitivity analyses
Results and conclusions were similar when carrying out the analyses in men and women separately or after stratifying analyses by tertile of baseline PA variables (data not shown). . Effect of knowledge of FTO risk status on change in physical activity (PA) in risk (AA/AT) and nonrisk (TT) carriers. Nondisclosure FTO risk carriers, n=160; nondisclosure FTO nonrisk carriers, n=78; disclosure FTO risk carriers, n=91; disclosure FTO nonrisk carriers, n=39. FTO: fat mass-and obesity-associated gene.
# Discussion
## Principal findings
Our main findings identified that there was no association between objectively measured or self-reported PA and FTO risk status. To our knowledge, our study is the first to investigate the impact of FTO genotype-based feedback on measured change in PA in the context of a personalized lifestyle intervention. We hypothesized that knowledge of carriage of FTO risk would lead to an increase in PA. However, we found no evidence that disclosing such information had any positive or negative effects on PA after a 6-month intervention.
## Comparison with previous work
In the last decade, there has been a growing interest in personalizing lifestyle interventions using genetic tests. This has been done using DNA-based disease risk estimates, primarily in smokers or individuals at risk of certain conditions, such as Alzheimer's disease [bib_ref] Effects of communicating DNA-based disease risk estimates on risk-reducing behaviours, Marteau [/bib_ref]. The hope was that providing such genetic information would motivate recipients to make beneficial behavioral changes beyond what could be achieved without such information. It is unclear whether knowledge of being predisposed to a greater genetic risk of disease would promote positive behavioral change and whether knowledge of only a small genetic risk (ie, a "lower" genetic risk) predisposition would lead to counterproductive behaviors under false reassurances [bib_ref] Deflating the genomic bubble, Evans [/bib_ref]. In their 2010 review, Marteau et al reported no effect of adding DNA-based disease risk estimates compared with a non-DNA-based approach, in terms of smoking cessation, PA, or use of medication/vitamins. A beneficial effect of DNA-based risk estimates on dietary behavior was reported, although no benefit on intention to change dietary behavior was observed [bib_ref] Effects of communicating DNA-based disease risk estimates on risk-reducing behaviours, Marteau [/bib_ref]. Since then, Hollands et al also observed no effect of communicating DNA-based risk assessments for Crohn's disease on smoking cessation, compared with standard risk assessment [bib_ref] Effect of communicating DNA based risk assessments for Crohn's disease on smoking..., Hollands [/bib_ref]. Grant et al reported that diabetes genetic risk counseling did not alter self-reported motivation or adherence to a prevention program in overweight individuals at risk for diabetes [bib_ref] Personalized genetic risk counseling to motivate diabetes prevention: A randomized trial, Grant [/bib_ref]. Although the design of our study was different because we did not aim to recruit individuals specifically at risk of a certain disease, our results are in line with the results of most studies performed so far. Recently, Meisel et al showed that young healthy individuals receiving FTO feedback in their weight control advice felt more prepared to control their weight than subjects receiving weight control advice only. However, this did not translate into behavioral change [bib_ref] Genetic susceptibility testing and readiness to control weight: Results from a randomized..., Meisel [/bib_ref].
Evidence in favor of disclosing genetic information is thus limited. Even the favorable findings for dietary behavior change mentioned above in the review by Marteau et al are weak [bib_ref] Effects of communicating DNA-based disease risk estimates on risk-reducing behaviours, Marteau [/bib_ref].
They are based on only 2 studies [bib_ref] Health behavior changes after genetic risk assessment for Alzheimer disease: The REVEAL..., Chao [/bib_ref] [bib_ref] Genetic Risk Assessment for FH Trial Study Group. Psychological impact of genetic..., Marteau [/bib_ref] , which did not find significant effects when each study was evaluated individually. More recently, Nielsen et al concluded that disclosing genetic information for personalized nutrition resulted in greater improvements in intake of some dietary components compared with general population-based dietary advice. In reality, this was true only for sodium intake, but not for caffeine, vitamin C, or added sugars, which were also studied. In addition, only individuals with the high-risk genotype status for the ACE gene reduced their sodium intake more than controls based on self-reported food intake, not on objective biomarkers of intake [bib_ref] Disclosure of genetic information and change in dietary intake: A randomized controlled..., Nielsen [/bib_ref]. Similarly, Hietaranta-Luoma et al reported that personal genetic information based on ApoE might have positive effects on triglyceride values and waist circumference, but this was observed only in the high-risk ε4+ individuals [bib_ref] Using individual, ApoE genotype-based dietary and physical activity advice to promote healthy..., Hietaranta-Luoma [/bib_ref].
Data suggest that providing genetic test results indicating a higher genetic risk does not lead to fatalism [bib_ref] Effects of communicating DNA-based disease risk estimates on risk-reducing behaviours, Marteau [/bib_ref]. Furthermore, there is no indication that disclosing only a small genetic risk or a lower-risk test result promotes counterproductive behaviors. Similarly, in our study we found no differences in change in PA between individuals aware of their nonrisk FTO status and individuals aware of a risk, or not aware of their genotype. However, we did observe that the attrition rate was significantly greater among individuals informed of their nonrisk FTO status as compared to the other groups. Given the amount and variety of information provided to participants during the Food4Me study, it seems unlikely that this genetic information would be responsible for the higher number of dropouts. Nonetheless, this should be studied further, as it may indicate that such individuals felt the intervention was less relevant for them. Grant et al also reported that subjects receiving lower-risk genetic results showed lower intent to do exercise compared with controls, although there were no differences in terms of attendance to the diabetes prevention program [bib_ref] Personalized genetic risk counseling to motivate diabetes prevention: A randomized trial, Grant [/bib_ref].
Personalized feedback led to greater improvements in self-reported PA, but not objectively measured PA, compared with standard guidelines, as reported previously [bib_ref] Effects of a Web-based personalized intervention on physical activity in European adults:..., Marsaux [/bib_ref].
Discrepancies between self-reported and objectively measured PA have been noted by others. For instance, Wanner et al, in a Web-based tailored PA intervention, reported some improvements in self-reported PA after 6 weeks and 13 months of follow-up, but no differences between individuals in tailored and control groups, and no improvement in objectively measured PA for any group [bib_ref] Effectiveness of active-online, an individually tailored physical activity intervention, in a real-life..., Wanner [/bib_ref]. However, in our study we did find greater improvements in self-reported PA in tailored groups as compared with the controls. It could be that participants desired to comply with recommendations and that receiving more personalized feedback (Levels 2 and 3) increased this desire further. Furthermore, here we show that the bigger improvements in self-reported PA reported earlier are irrespective of FTO genotype, and are not related to knowing one's risk status for FTO. Thus, it is unlikely that subjects with the high-risk variant would feel more pressured to report that they did better, compared with those with the low-risk variant. Finally, we did not observe an association between FTO risk and PA measured objectively or self-reported. This supports studies published thus far that have used mainly self-reported data [bib_ref] Physical activity attenuates the influence of FTO variants on obesity risk: A..., Kilpeläinen [/bib_ref] [bib_ref] FTO genotype, physical activity, and coronary heart disease risk in Swedish men..., Gustavsson [/bib_ref] [bib_ref] Fat mass and obesity-associated (FTO) gene polymorphisms are associated with physical activity,..., Harbron [/bib_ref].
# Strengths and limitations
This study is the first to report the impact of disclosing information on FTO risk status on measured changes in PA. Our PA questionnaire has been validated against doubly labeled water and accelerometry [bib_ref] A short questionnaire for the measurement of habitual physical activity in epidemiological..., Baecke [/bib_ref] [bib_ref] Doubly labelled water validation of three physical activity questionnaires, Philippaerts [/bib_ref] [bib_ref] Comparison of two questionnaires with a tri-axial accelerometer to assess physical activity..., Philippaerts [/bib_ref] , and has been used in large European cohorts before [bib_ref] Comparison of two physical activity questionnaires in obese subjects: The NUGENOB study, Tehard [/bib_ref] [bib_ref] The Diet, Obesity and Genes (Diogenes) Dietary Study in eight European countries..., Larsen [/bib_ref]. However, self-reports introduce large measurement error [bib_ref] A comparison of direct versus self-report measures for assessing physical activity in..., Prince [/bib_ref] and the Baecke questionnaire is no exception [bib_ref] Comparison of self-reported with objectively assessed energy expenditure in black and white..., Walsh [/bib_ref]. Thus, a strength of this study was the objective assessment of PA using triaxial accelerometers. Although accelerometers underestimate certain activities, such as cycling, swimming, or resistance training, the TracmorD model used in this study has been validated against doubly labeled water [bib_ref] Estimation of free-living energy expenditure using a novel activity monitor designed to..., Bonomi [/bib_ref] and it has been shown to be reliable and accurate [bib_ref] A comparison of energy expenditure estimation of several physical activity monitors, Dannecker [/bib_ref] [bib_ref] Daily physical activity assessment with accelerometers: New insights and validation studies, Plasqui [/bib_ref] [bib_ref] Validating measures of free-living physical activity in overweight and obese subjects using..., Valenti [/bib_ref].
By design, we recruited individuals interested in taking part in a personalized intervention on nutrition and lifestyle, which is less representative than a European-wide survey. Nonetheless, our participants were broadly representative of the European adult population, most of whom had adequate nutrient intakes but could benefit from improved dietary choices and greater PA [bib_ref] Profile of European adults interested in Internet-based personalised nutrition: The Food4Me study, Livingstone [/bib_ref]. Given that Food4Me was an intervention that targeted multiple dietary and lifestyle behaviors, the genetic results might have also been diluted by the amount of information provided. Moreover, the genetic feedback was a positive reinforcement. Participants with the higher-risk genotype would only benefit more by reducing their weight or increasing their PA. It is possible that the impact would have been greater if participants had been made more aware of the links between obesity and lifelong ill health. Furthermore, genetic feedback provided by health professionals skilled in genetic counseling might have been more effective that written feedback. However, this would have been more expensive and outside the scope of this study, which was designed to test the effects of an Internet-delivered intervention. Such interventions are thought to offer considerable advantages in terms of reach, scalability, and sustainability [bib_ref] Personalising nutritional guidance for more effective behaviour change, Celis-Morales [/bib_ref]. Attrition rates (~15%) were as expected and compliance with the measurements was good, except for wearing the monitor.
Only half of the participants had accelerometer data for both baseline and month 6-whereas >75% had self-reported PA data at both time points-which limited the size of the sample analyzed in the PA analyses. It is possible that wearing the monitor for 6 months was too demanding for the amount of feedback given. It may be important for future studies that participants be able to visualize their activity levels, in real time, whenever desired (eg, on an accompanying website). Improvements in activity measurement may reduce participants' confusion and/or frustration. Having personalized coaches available, who can also operate online, may have motivated participants to wear their accelerometer and to improve their PA, although this also means extra costs. For the sole purpose of assessment, better compliance may be obtained by sending out monitors and collecting them back directly after assessment [bib_ref] Evolution of accelerometer methods for physical activity research, Troiano [/bib_ref]. In spite of this, our sample size was acceptable, and the results did not change when looking at all self-reported PA data available.
# Conclusions
There was no added benefit of knowledge of FTO risk on change in PA in this intervention study. Although there were no differences in outcome measures between participants informed of a nonrisk and those informed of a risk, or those not informed of their FTO risk status, the nonrisk subjects were more likely to drop out of the study by 6 months. More studies are needed to confirm whether disclosure of lower-risk genetic test results has adverse effects on engagement in behavioral changes. Before that, more effort should be devoted to identify the features necessary to engage individuals, how to frame the feedback, and how to coach effectively, especially those at risk, to reduce health inequalities.
[fig] Figure 1: Flowchart of study procedures. Participants in Level 0 (controls) received standard, nonpersonalized guidelines during the intervention, whereas participants in Levels 1-3 received personalized advice. PA: physical activity; FTO: fat mass-and obesity-associated gene. [/fig]
[fig] Figure 2: Physical activity in FTO rs9939609 risk (AA/AT, n=874) and nonrisk (TT, n=405) carriers. FTO: fat mass-and obesity-associated gene. [/fig]
[table] Table 1: Attrition rates after 6 months by intervention level. [/table]
[table] Table 2: Characteristics of the participants included in baseline analysis. [/table]
[table] Table 3: Changes in physical activity (PA) from baseline to month 6 for participants receiving personalized advice to increase their PA.Self-reported PA: total activity index FTO: fat mass-and obesity-associated gene. b PA: physical activity. c PAL: physical activity level. [/table]
|
Interleukin-17 promotes proliferation, migration, and invasion of trophoblasts via regulating PPAR-γ/RXR-α/Wnt signaling
To investigate the effect of Interleukin 17 (IL-17) on the invasive capacity of trophoblast cells and the underlying mechanism, we collected placental tissues samples from pregnant women with preeclampsia (PE) and healthy pregnant women. The expression levels of IL-17 mRNA and protein in tissue samples were determined using qRT-PCR and Western blot, respectively. Cell viability and cell proliferation was determined using CCK-8 assay, and colony formation assay, respectively. Cell migration and invasion capacity were determined using transwell cell migration assay. Our results showed that the mRNA expression of IL-17 was significantly increased in PE patients and may be used as a sensitive biomarker for PE (P < 0.01). IL-17 overexpression promoted cell viability, migration, and invasion of human extravillous trophoblast cell line, HTR8/SVneo; however, IL-17 knockdown inhibited these effects. Additionally, IL-17 activated PPAR-γ/RXR-α signaling pathway, which promoted proliferation, migration, and invasion of trophoblast cells. Moreover, PPAR-γ/RXRα heterodimers activated Wnt signaling. In conclusion, our study provides evidence that IL-17 is overexpressed in PE and promotes proliferation, migration and invasion of trophoblast cells via activating PPAR-γ/RXR-α/Wnt signaling.ARTICLE HISTORY
# Introduction
Preeclampsia (PE) is an idiopathic pregnancyspecific syndrome characterized by proteinuria. PE is a major cause of miscarriage and fetal death in pregnant women [bib_ref] Making sense of pre-eclampsia: two placental causes of pre-eclampsia, Staff [/bib_ref] [bib_ref] Screening for preeclampsia by maternal factors and biomarkers at 11-13 weeks' gestation, Tm [/bib_ref]. Epidemiological studies show that the incidence of PE is about 3-5% globally and 2-8% in Country [bib_ref] Genetic and non-genetic risk factors for pre-eclampsia: an umbrella review of systematic..., Giannakou [/bib_ref]. Early diagnosis and intervention is important to reduce adverse events caused by PE. Recently, spiral artery recasting has been shown to be closely related to the occurrence and development of PE. Spiral artery recasting is characterized by trophoblast invasion and secondary activation of apoptosis of vascular endothelial smooth muscle cells. Changes in invasiveness of trophoblasts are considered to be an important pathological basis for the occurrence of PE [bib_ref] Impact of road traffic pollution on pre-eclampsia and pregnancyinduced hypertensive disorders, Pedersen [/bib_ref]. Therefore, understanding the molecular mechanisms underlying the invasiveness of trophoblasts is of vital importance.
Interleukin 17 (IL-17) is a pro-inflammatory factor secreted by Th17 cells. IL-17 is highly inflammatory and can activate inflammation by regulating the expression of other inflammatory factors, such as IL-6 and IL-8 [bib_ref] The orphan nuclear receptor RORgammat directs the differentiation program of proinflammatory IL-17+..., Ivanov [/bib_ref]. Studies have shown that IL-17 is closely related to trophoblast cell infiltration [bib_ref] Increased circulating interleukin-17 levels in preeclampsia, Molvarec [/bib_ref]. However, the specific molecular mechanism through which IL-17 regulates and enhances trophoblast invasion has not been elucidated.
Retinoic acids (RAs), active products of vitamin A, play a key role in embryogenesis as well as cell differentiation, migration, and invasion [bib_ref] Retinoic acid and its derivatives in skin, Szymański [/bib_ref]. RA signaling through its receptors (such as RARA, RARB, RARG, and the retinoic X receptor RXRA) is closely associated with the development of fetal placenta [bib_ref] Expression and regulation of retinoic acid receptor responders in the human placenta, Huebner [/bib_ref]. For example, increase in decidual RAR-α causes retardation of placental growth. RXR-α is overexpressed in cytotrophoblasts differentiation and its knockdown increases the invasion of extravillous cytotrophoblasts into uterine tissues [bib_ref] The role of PPAR-gamma/RXR-alpha heterodimers in the regulation of human trophoblast invasion, Fournier [/bib_ref]. Decrease in RAs contributes to the pathogenesis of PE [bib_ref] Retinoic acid is a negative regulator of sFLT1 expression in decidual stromal..., Deepak [/bib_ref]. However, the exact role of RA signaling in PE has not been elucidated.
Thie aim of this study was to investigate the role of IL-17 in PE and the underlying mechanisms. We hypothesized that IL-17 might promote the proliferation, migration and invasion of trophoblast cells via activating PPAR-γ/RXR-α/Wnt signaling.
# Materials and methods
# Materials
## Study participants
Clinical samples were collected from pregnant women with PE and healthy control at Shijiazhuang People's Hospital. The inclusion criteria for the PE group included: (1) the relevant diagnostic criteria for PE; (2) Good mental state;
(3) the patient voluntarily signed the informed consent form. The exclusion criteria for the PE group included: (1) artificial insemination, or multiple pregnancy; (2) absence of clinical data; (3) patients with chronic hypertension or other cardiovascular diseases. (4) patients with inflammatory condition. The inclusion criteria for the healthy control NC group included: (1) normal indexes of all physical examinations; (2) singleton pregnancy; (3) no past history of eclampsia. The exclusion criteria for NC group included: (1) refusal to sign an informed consent form; (2) pregnancyinduced hypertension or gestational diabetes [bib_ref] Association between maternal infections and preeclampsia: a systematic review of epidemiologic studies, Rustveld [/bib_ref]. And the Characteristic information of the PE patients and healthy controls were collected from the Shijiazhuang People's Hospital. This study was approved by Ethics committee of the Shijiazhuang People's Hospital.
## Reagent
The following reagents were purchased from the indicated manufacturers: human extravillous trophoblast cell line HTR8/SVneo (in early pregnancy) from ATCC (Manassas, VA, USA), DMEM high sugar medium from Shandong Boko Biological Industry Co., Ltd (City, Country), fetal bovine serum from Shandong Boko Regenerative Medicine Co., Ltd (City, Country), trypsin from Shenzhen Love Biotechnology Co., Ltd (Shenzhen, China), TRIzol reagent from Shanghai Bohu Biotechnology Co., Ltd (Shanghai, China), UltraSYBR mixture from Shanghai Bangjing Industrial Co., Ltd (Shanghai, China), tissue lysate from Shanghai Yuchun Biotechnology Co., Ltd (Shanghai, China), ReverTra Ace qPCR-RT kit from Dongyang Textile (Shanghai) Biotechnology Co., Ltd (Shanghai, China), Lipo3000 TM reagent from Beijing Noble Technology Co., Ltd (Beijing, China), Rabbit antihuman GAPDH monoclonal antibody from Shanghai Lei Hao Information Technology Co., Ltd (Shanghai, China), and HRP conjugated goat antirabbit IgG from Mingxiu (Shanghai, China) Biotechnology Co., Ltd (Shanghai, China).
# Method
## Specimen collection
Clinical samples (1 cm 3 ) were collected from the participants. The tissues were immediately cleaned and dried before being frozen and stored at −80ºC for detection of IL-17 expression.
## Cell culture and transfection
HTR8/SVneo cells were cultured in DMEM media supplemented with 10% fetal bovine serum in a standard humidified incubator (37ºC, 5% CO 2 ). Cells were passaged after reaching 90% confluency. Cells were seeded at a density of 1 × 10 5 cell per well in 6-well plates and treated with 1 µM BMS493 (an inhibitor of pan-RAR receptors).
Cells were later transfected with siRNA-IL-17, IL-17 overexpression plasmids or the negative control using Lipofectamine® 3000 (Invitrogen, Waltham, MA, USA) and were used for further experiments after 48 h.
## Quantitative rt-pcr
Total RNA was extracted from the tissue samples and cells (following transfection) using TRIzol. The concentration and purity of the total RNA were assessed from OD 260/280 readings (ratio >1.8) using a spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA). The RNA (1 µg) was used to synthesize cDNA using SYBR Premix Ex Taq (Takara), as per manufacturer's instructions. PCR was performed using Power SYBR® Green PCR Master Mix (TaKaRa). The reaction system: 2 µL cDNA, 0.4 µL each of forward and reverse primers, 0.4 µL ROX reference dye, 10 µL SYBR Premix Ex Taq, and 6.8 µL double-distilled H 2 O. The reaction conditions were as follows: the reaction has a total of 40 cycles, of which 65ºC 1 min 10 min 95ºC 15 s min 95ºC. GAPDH was used as the internal reference. Relative mRNA expression was calculated using the 2 −ΔΔCq method [bib_ref] Monitoring gene expression: quantitative real-time rt-PCR, Wagner [/bib_ref]. The primer sequences are shown in .
# Western blot analysis
As described by previous study [bib_ref] Monitoring gene expression: quantitative real-time rt-PCR, Wagner [/bib_ref] , Total protein was collected from the cells by using radioimmunoprecipitation assay lysis buffer (Beyotime, Jiangsu, China). Protein concentration was determined using BCA kit and separated by SDS-PAGE (40 μg per lane). The proteins were then transferred to PVDF membranes. After blocking with 5% skimmed milk for 2 h, the membranes were incubated with primary antibodies overnight at 4ºC. The membranes were incubated with goatanti-rabbit secondary antibody (Mingxiu (Shanghai) Biotechnology Co., Ltd). All the antibodies (unless specified otherwise) were purchased from Abcam (Cambridge, UK) and diluted at 1:1000. Subsequently, the bands were visualized using an ECL chemiluminescence kit and analyzed using ImageJ 3.0 software (NIH, Bethesda, MD, USA).
## Luciferase assay
Cells were seeded into 24-well plate and later cotransfected with reporter plasmids containing Wnt promoter and pcDNAPPAR-ɤ/si-PPAR-ɤ for 48 h. The luciferase activity was determined using a luciferase kit (Promega, Madison, WI, USA). Luciferase activity was normalized to Renilla luciferase activity according to a previous study [bib_ref] Down-regulation of lncRNA DNAJC3-AS1 inhibits colon cancer via regulating miR-214-3p/LIVIN axis, Han [/bib_ref].
## Transwell cell migration assay analysis
According to a previous study [bib_ref] MicroRNA miR-330-3p suppresses the progression of ovarian cancer by targeting RIPK4, Cai [/bib_ref] , following transfection, cells were collected, resuspended at a density of 5 × 10 4 cells per ml and seeded (1 mL) per well in upper chambers of 24-well plates precoated with Matrigel. The lower chamber was filled with 1 mL FBS. After 24 h, cells that migrated or invaded into the lower chamber were fixed with 4% paraformaldehyde and subsequently stained with 0.1% crystal violet. The cells were then visualized with an inverted microscope.
# Statistical analysis
The data were processed using SPSS 22.0 software (IBM Corp, Armonk, NY, USA) and expressed as mean ± SD. All data in this study conformed to normal distribution. The data were analyzed by Student t-test and ANOVA followed by Tukey's test. P < 0.05 was considered statistically significant.
# Results
IL-17 is overexpressed in PE. Moreover, overexpression of IL-17 promoted the proliferation, migration, and invasion of trophoblast cells by regulating PPAR-ɤ/RXR-α/Wnt signaling. Therefore, IL-17 may be a potential therapeutic target for PE.
## Il-17 was overexpressed in preeclampsia
The expression of IL-17 mRNA was significantly higher in the PE group than in the HC group (P < 0.01). The results of the ROC curve showed that the AUC area of IL-17 in the PE group was 0.9463 (95% confidence interval: 0.8995 ~ 0.9930, P < 0.001) [fig_ref] Figure 1: IL-17 was overexpressed in PE patients [/fig_ref]. Moreover, high expression of IL-17 was associated with systolic and diastolic blood pressure, gestational age, and proteinuria [fig_ref] Table 2: Characteristic information of the PE and healthy patients [/fig_ref].
## Overexpression of il-17 promoted proliferation, migration, and invasion of trophoblast cells
To investigate the role of IL-17 in PE, we analyzed the cellular functions of trophoblast following transfection with IL-17 overexpression plasmids. As shown in . The sequence of primers used in this study. [fig_ref] Figure 2: IL-17 promoted proliferation, migration, and invasion of trophoblast cells [/fig_ref] and b, the mRNA and protein expression of IL-17 increased significantly (P < 0.001), suggesting that the cells were successfully transfected. Overexpression of IL-17 promoted cell viability and colony formation (P < 0.01, [fig_ref] Figure 2: IL-17 promoted proliferation, migration, and invasion of trophoblast cells [/fig_ref] and d). This was consistent with the results from transwell cell migration assay; upregulation of IL-17 significantly increased the migration and invasion ability of trophoblast cells (P < 0.01, [fig_ref] Figure 2: IL-17 promoted proliferation, migration, and invasion of trophoblast cells [/fig_ref]. Additionally, upregulation of IL-17 increased the mRNA and protein expression of MMP-3, MMP-9, and Vimentin (P < 0.01, figure 2f and g).
[formula] 5′-3′ IL-17 F AACGCCGAGGCCAATAACTTTC R AGGGTCCTCATTGCGGCTCAGA MMP-2 F CACTTTCCTGGGCAACAAAT R CTCCTCAATGCCCTTGATGT MMP-3 F TCGGTGGCTTCAGTACCT R CCTCCTCCCAGACCTTC Vimentin F ATGACCGCTTCGCCAACTAC R CGGGCTTTGTCGTTGGTTAG PPAR-γ F GGGATCAGCTCCGTGGATCT R TGCACTTTGGTACTCTTGAAGTT RXR-α F CGACCCTGTCACCAACATTTGC R GAGCAGCTCATTCCAGCCTGCC GAPDH F GGAGCGAGATCCCTCCAAAAT R GGCTGTTGTCATACTTCTCATGC [/formula]
## Knockdown of il-17 suppressed proliferation, migration, and invasion of trophoblast cells
To investigate the role of IL-17 in PE, cells were transfected with siRNA-IL-17. As shown in [fig_ref] Figure 3: IL-17 knockdown suppressed the proliferation, migration, and invasion of trophoblast [/fig_ref] , IL-17 expression was significantly decreased in siRNA-IL-17 group (P < 0.01), which was more remarkable in siRNA-IL-17 1#, (P < 0.001). Therefore, siRNA-IL-17 1# was used in the following experiments. IL-17 knockdown significantly suppressed cell viability, colony formation, migration, and invasion of trophoblast cells, HTR8/ SVneo (P < 0.01, [fig_ref] Figure 3: IL-17 knockdown suppressed the proliferation, migration, and invasion of trophoblast [/fig_ref]. Additionally, IL-17 knockdown decreased the mRNA and protein expression of MMP3, MMP9, and Vimentin (P < 0.01, [fig_ref] Figure 3: IL-17 knockdown suppressed the proliferation, migration, and invasion of trophoblast [/fig_ref] and g).
## Il-17 activated ppar-ɤ/rxr-α signaling
RA signaling plays a crucial in the pathologically altered placentas in PE patients [bib_ref] Expression and regulation of retinoic acid receptor responders in the human placenta, Huebner [/bib_ref]. Therefore, we determined the expression of PPAR-ɤ/RXR-α in trophoblast cells. As shown in [fig_ref] Figure 4: IL-17 activated PPAR-γ/RXR-α signaling pathway [/fig_ref] , the mRNA and protein expression of PPAR-ɤ/RXR-α was upregulated in cells overexpressing IL-17 and down regulated in cells with silenced IL-17 (P < 0.01).
## Activated rxr-α promoted proliferation, migration, and invasion of trophoblast cells
Rescue assays were performed to verify the role of RXR-α in PE. As shown in [fig_ref] Figure 5: Activation of RXR-α promoted proliferation, migration, and invasion of trophoblast cells [/fig_ref] , the expression of RXR-α was significantly increased by its agonist, Magnaldehyde B (MB) (P < 0.01). Moreover, MB significantly increased cell viability, colony formation, migration, and invasion of trophoblast as well as increased the mRNA and protein expression of MMP3, MMP9, and Vimentin (P < 0.01, [fig_ref] Figure 5: Activation of RXR-α promoted proliferation, migration, and invasion of trophoblast cells [/fig_ref]. [fig_ref] Figure 6: PPAR-ɤ/RXR-α heterodimers transcriptionally activated Wnt2 signaling pathway [/fig_ref] and b show the binding sites of the binding motif of PPAR-ɤ/RXR-α heterodimers. The binding sites were further verified by luciferase assay (P < 0.01, [fig_ref] Figure 6: PPAR-ɤ/RXR-α heterodimers transcriptionally activated Wnt2 signaling pathway [/fig_ref]. Moreover, the mRNA and protein expression of WNT2 increased following overexpression of PPAR-ɤ and decreased following PPAR-ɤ silencing [fig_ref] Figure 6: PPAR-ɤ/RXR-α heterodimers transcriptionally activated Wnt2 signaling pathway [/fig_ref] and f).
## Ppar-ɤ/rxr-α heterodimers transcriptionally activated wnt2 signaling
# Discussion
In the present study, we found that IL-17 was overexpressed in PE patients. Upregulation of IL-17 promoted proliferation, migration, and invasion of trophoblast cells. Additionally, IL-17 interacted with PPAR-ɤ/RXR-α to activate Wnt2/β-catenin signaling pathway. However, knockdown of IL-17 suppressed the observed proliferation, migration, and invasion of trophoblast cells. Therefore, IL-17 may be a potential biomarker for PE. The development of fetal placenta is closely associated with abnormal accumulation of cytokines, such as interleukins (ILs), tumor necrosis factors (TNFs), and interferons (INFs) [bib_ref] The role of interleukins in preeclampsia: a comprehensive review, Bellos [/bib_ref] [bib_ref] Immune cell infiltration landscape and immune marker molecular typing in preeclampsia, Meng [/bib_ref] [bib_ref] Fetal HLA-G mediated immune tolerance and interferon response in preeclampsia, Wedenoja [/bib_ref]. IL-17, a member of the interleukin family, functions as a pro-inflammatory factor in spondyloarthritis [bib_ref] The role of the IL-23/IL-17 pathway in the pathogenesis of spondyloarthritis, Tsukazaki [/bib_ref] , psoriasis [bib_ref] Anti IL-17 in psoriasis, Ly [/bib_ref] , and atherosclerosis [bib_ref] The Impact of IL-17 in atherosclerosis, Lu [/bib_ref]. Moreover, the potential role of IL-17 in inflammation and autoimmune diseases make limited contributions to the establishment of the diversity, stability, and plasticity in placenta. The imbalance of IL-17/IL-35 or IL-17/IL-10 in the placenta induces PE and low fetal weight in placental malaria [bib_ref] Expression imbalance of IL-17/IL-35 in peripheral blood and placental tissue of pregnant..., Lu [/bib_ref] [bib_ref] Low fetal weight is directly caused by sequestration of parasites and indirectly..., Fitri [/bib_ref]. Moreover, IL-17-induced oxidative stress is a key factor that induces hypertension during pregnancy, which may further contribute to the development of the fetal placenta [bib_ref] IL-17-mediated oxidative stress is an important stimulator of AT1-AA and hypertension during..., Dhillion [/bib_ref]. In the present study, we found that IL-17 was overexpressed in PE and can therefore be used a sensitive biomarker. Overexpression of IL-17 promoted proliferation, migration, and invasion of trophoblast cells, while, IL-17 knockdown induced the degradation of trophoblast.
Generally, about 90% of the research on IL-17 focuses on its potential role in inflammation and autoimmune pathways. Few studies have investigated the role of IL-17 in placental development and pregnancy. The release of IL-17 produces natural killer cells and enhances proliferation and invasion of human trophoblast cells [bib_ref] Interleukin-17 signaling mediates cytolytic natural killer cell activation in response to placental..., Travis [/bib_ref] [bib_ref] Decidual stromal cells recruit Th17 cells into decidua to promote proliferation and..., Wu [/bib_ref]. The accumulation of IL-17 released by Th17 promotes the apoptosis of granular cells in vitro and induces premature ovarian failure in vivo [bib_ref] hPMSC transplantation restoring ovarian function in premature ovarian failure mice is associated..., Yin [/bib_ref]. Therefore, these results suggested that IL-17 plays a negative role in placental development and pregnancy, and that the role of IL-17 varies in different cells. Interestingly, IL-17 soluble receptor C infusion impedes Th17 function and inhibits the development of PE [bib_ref] Administration of interleukin-17 soluble receptor C suppresses TH17 cells, oxidative stress, and..., Cornelius [/bib_ref]. Therefore, suppressing the release of IL-17 in trophoblast cells may be a promising intervention strategy for PE.
The alteration of RAs is crucial factor for the development of PE. For instance, low levels of RA contribute to the pathogenesis of PE [bib_ref] Retinoic acid is a negative regulator of sFLT1 expression in decidual stromal..., Deepak [/bib_ref]. These effects are regulated by RA signaling. RA receptor responder 1 is hypermethylated in choriocarcinoma of the placenta. VEGF and RARβ interact with VEGF to promote intramembranous absorption and fetal vasculature on the placental surface [bib_ref] Retinoic acid pathway regulation of vascular endothelial growth factor in ovine amnion, Cheung [/bib_ref]. RXR-α, an RA receptor, promotes the differentiation of human trophoblast. Moreover, high levels of RXR-α increase the risk of PE [bib_ref] Gene variants and haplotypes of Vitamin D biosynthesis, transport, and function in..., Ghorbani [/bib_ref]. In the present study, we found that IL-17 increased the expression of RXR-α, which promoted proliferation, migration, and invasion of trophoblast cells. Additionally, RXRs can form hetero-or homodimers that bind to specific DNA elements [bib_ref] Retinoic acid receptors and retinoid X receptors: interactions with endogenous retinoic acids, Allenby [/bib_ref]. Previous studies have reported the role of PPAR-ɤ/RXR-α heterodimers in placental development [bib_ref] The role of PPAR-gamma/RXR-alpha heterodimers in the regulation of human trophoblast invasion, Fournier [/bib_ref] [bib_ref] PPARγ/RXRα heterodimers are involved in human CGβ synthesis and human trophoblast differentiation, Tarrade [/bib_ref]. In the present study, IL-17 activated PPAR-ɤ/RXR-α signaling to enhance the invasion capacity of trophoblast cells.
Wnt signaling pathway is a key regulator of embryogenesis [bib_ref] LRP6 regulates Rab7-mediated autophagy through the Wnt/β-catenin pathway to modulate trophoblast cell..., Li [/bib_ref]. Wnt pathway is involved in blastocyst formation and trophoblast development [bib_ref] Inactivation of nuclear Wnt-beta-catenin signaling limits blastocyst competency for implantation, Xie [/bib_ref] [bib_ref] Cdx2 is essential for axial elongation in mouse development, Chawengsaksophak [/bib_ref]. The potential role of Wnt signaling in has attracted attention [bib_ref] Wnt/β-catenin signaling in development and disease, Clevers [/bib_ref]. Depletion of Wnt2 attenuates branching and placental labyrinth formation [bib_ref] Targeted disruption of the Wnt2 gene results in placentation defects, Monkley [/bib_ref]. Inactivation of Wnt signaling by LIM and SH3 Protein 2 suppresses the migration and invasion of trophoblast [bib_ref] LASP2 inhibits trophoblast cell migration and invasion in preeclampsia through inactivation of..., Chen [/bib_ref]. Wnt2 is upregulated during syncytiotrophoblast differentiation, and this aberrant expression induces the development of PE in vivo and in vitro [bib_ref] Primate-specific miR-515 family members inhibit key genes in human trophoblast differentiation and..., Zhang [/bib_ref] [bib_ref] Bisphenol A exposure alters placentation and causes preeclampsia-like features in pregnant mice..., Ye [/bib_ref]. In the present study, PPAR-ɤ/RXR-α heterodimers transcriptionally activated Wnt2/β-catenin signaling pathway. Therefore, IL-17 may promote proliferation, migration, and invasion of trophoblast cells via activation of PPAR-ɤ/RXR-α/Wnt signaling pathway.
# Conclusion
In conclusion, IL-17 is overexpressed in PE. Moreover, overexpression of IL-17 promoted the proliferation, migration, and invasion of trophoblast cells by regulating PPAR-ɤ/RXR-α/Wnt signaling. Therefore, IL-17 may be a potential therapeutic target for PE.
[fig] Figure 1: IL-17 was overexpressed in PE patients. (a) The mRNA expression of IL-17 in PE patients. (b) AUC analysis of IL-17 expression. **P < 0.01. [/fig]
[fig] Figure 2: IL-17 promoted proliferation, migration, and invasion of trophoblast cells. Transfection efficiency of IL-17 determined using (a) qRT-PCR, and (b) Western blot. (c) Cell viability of trophoblast cells. (d) Colony formation capacity of trophoblast cells. (e) Migration and invasion capacity of trophoblast cells. Expression of MMP3, MMP9, and Vimentin at (f) the mRNA level, and (g) the protein level. **P < 0.01, ***P < 0.001. [/fig]
[fig] Figure 3: IL-17 knockdown suppressed the proliferation, migration, and invasion of trophoblast. Transfection efficiency of IL-17 determined using (a) qRT-PCR, and (b) Western blot. (c) Cell viability of trophoblast cells. (d) Colony formation capacity of trophoblast cells. (e) Migration and invasion capacity of trophoblast cells. Expression of MMP3, MMP9, and Vimentin at (f) the mRNA level, and (g) the protein level. **P < 0.01, ***P < 0.001. [/fig]
[fig] Figure 4: IL-17 activated PPAR-γ/RXR-α signaling pathway. (a) mRNA expression of PPAR-γ. (b) mRNA expression of RXR-α. (c) Protein expression of PPAR-γ/RXR-α. **P < 0.01, ##P < 0.01. [/fig]
[fig] Figure 5: Activation of RXR-α promoted proliferation, migration, and invasion of trophoblast cells. (a) mRNA expression of RXR-α. (b) Transfection efficiency of IL-17 determined by Western blot. (c) Cell viability of trophoblast cells. (d) Colony formation of trophoblast cells. (e) Migration and invasion ability of trophoblast cells. Expression of MMP3, MMP9, and Vimentin at (f) the mRNA level, and (g) the protein level. **P < 0.01, #P < 0.05, ##P < 0.01. [/fig]
[fig] Figure 6: PPAR-ɤ/RXR-α heterodimers transcriptionally activated Wnt2 signaling pathway. (a) The binding motif of PPAR-ɤ. (b) The binding sites of PPAR-ɤ predicted by JASPAR. (c, d) The binding sites verified by luciferase assay. (e, f) mRNA and protein expression of Wnt signaling pathway. **P < 0.01, ##P < 0.01. [/fig]
[table] Table 2: Characteristic information of the PE and healthy patients. [/table]
|
HPV32‐related Heck’s disease in a chronic graft‐versus‐host disease patient with long‐term successful KTP laser treatment: A rare case report
[bib_ref] In situ detection of human papillomavirus types 13 and 32 in focal..., Henke [/bib_ref] [bib_ref] A case report of focal epithelial hyperplasia (Heck's disease) with PCR detection..., Ozden [/bib_ref] [bib_ref] Association of HLA-DR4 (DRB1*0404) with human papillomavirus infection in patients with focal..., Garcia-Corona [/bib_ref] [bib_ref] Focal epithelial hyperplasia associated with human papillomavirus 13 and common human leukocyte..., Akoglu [/bib_ref] [bib_ref] A case report of focal epithelial hyperplasia (Heck's disease) with PCR detection..., Ozden [/bib_ref]
## | case report
A 68-year-old post-transplant patient presented to our dental clinic as part of a comprehensive evaluation of his cGVHD with the chief complaint of proliferating white patches in his mouth. During the initial visit, we identified suspicious oral lesions and needed to rule out malignant potential. The patient was a 68-year-old Caucasian man with cGVHD, previously treated for acute myelogenous leukemia (AML, translocation 9:11 and trisomy 8) with an allogeneic matched related donor hematopoietic stem cell transplant. Transplant was preceded by 4 days of busulfan/ fludarabine conditioning and followed by two cyclophosphomide doses. Following treatment of acute skin GVHD, systemic immunosuppression was stopped, and at 9 months post-transplant, the patient developed cGVHD of the lung, oral cavity, skin, and eye. Systemic immunosuppression resumed, and at 18 months post-transplant, a single flat white plaque was noted on the maxillary labial mucosa. It was excised and pathologist evaluated. H&E staining revealed acanthosis with expansion and fusion of the rete ridges along with scattered nuclear fragmentation consistent with FEH. Patient and donor HLA type were HLA DRB1*1501, ruling out genetic links between HLA and HPV-induced FEH in this case. At 5 years post-transplant, the patient presented to our hospital dental clinic during a comprehensive multi-specialty assessment of cGVHD that included consent and research sampling on study NCT00092235 (clinicaltrials.gov) which conforms to the principles of the Declaration of Helsinki. Evaluation of the oral cavity revealed multiple, sessile whitish papules involving the maxillary and mandibular labial and bilateral buccal mucosa . Biopsy of the buccal mucosal from an area of mild lichenoid patterning was read as normal squamous mucosa, supporting the absence of oral mucosal cGVHD. Given the elevated risk of oral squamous cell carcinoma in cGVHD patients, the sentinel maxillary labial mucosal lesion was also biopsied at 5 years post-transplant and was read as hyperplastic squamous mucosa with parakeratosis and few dyskeratotic cells and perinuclear clearing suspicious for HPV viral cytopathic changes . Rete ridges were fused with scattered nuclear fragmentation consistent with FEH. Immunohistochemistry staining was positive for endothelial marker CD31 . Due to the relatively rapid proliferation of these lesions rather than the occurrence and regression typical of FEH, the differential diagnosis for these lesions included infection or coinfection with more aggressive strains of HPV such as HPV16 that could represent a higher risk of malignant transformation for an already at-risk patient.
In situ hybridization (ISH) testing for high/low-risk HPV was done at the Mayo Clinic and ruled out HPV-16 and HPV-18 while demonstrating inconclusive staining for HPV-6/11 with high background . To determine the true HPV strain, we proceeded to sequence the virus using genomic DNA isolated from tissue using the Qiagen Blood and Tissue kit. DNA was amplified using Rolling Circle Amplification (RCA). [bib_ref] A sequence-independent strategy for detection and cloning of circular DNA virus genomes..., Rector [/bib_ref] The RCA product was digested with BamHI and ligated into the matching pML2d site. The cloned viral DNA was sequenced by primer walking (Eurofins Genomics) and determined to be HPV32, represented as a linear diagram in [fig_ref] FF I G U R E 4: I G U R E 2 Tissue analysis of a biopsy of... [/fig_ref]. The HPV32 sequence was submitted to GenBank (KT236450.1) and was 99% identical to a previously banked sequence (GenBank X74475.1), along with greater than 99% nucleotide identity to all other banked partial sequences [fig_ref] FF I G U R E 4: I G U R E 2 Tissue analysis of a biopsy of... [/fig_ref].
With no standardized treatment protocol specific for HPV32-mediated FEH, initial treatment followed a standard protocol for oral HPV lesions: imiquimod 5% cream applied three times weekly. After 4 weeks of treatment, clinical manifestations appeared unchanged . Due to the presence of blood vessels within the structural fronds of the papillomatous lesions, we hypothesized that a 532 nm potassium titanyl phosphate (KTP) laser could be an effective therapy for the lesions. [bib_ref] Leukocyte migration and adhesion, Imhof [/bib_ref] Oxyhemoglobin, which absorbs visible light at 542 and 577 nm, is the primary chromophore in blood vessels. Once absorbed by oxyhemoglobin, light energy converts into thermal energy causing photocoagulation, mechanical injury, and finally thrombosis and occlusion of the blood vessel. Thus, delivery of laser energy at or near absorption peaks for oxyhemoglobin may be an effective means to selectively ablate highly vascular tissues. A similar hemostatic laser was reported to be effective in a pediatric case of FEH. [bib_ref] Laser excision of focal epithelial hyperplasia (Heck's Disease): a rare case report, Nallanchakrava [/bib_ref] For our patient, one course of 532 nm potassium titanyl phosphate (KTP) laser therapy in the clinic under topical anesthesia was completed . No adverse effects were noted. FEH lesions had predominately cleared by 1-month post-treatment and remained clear through 5 years post-treatment (most recent follow-up). Thus, treatment with a 532 nm KTP laser in this case was safe and effective at clearing HPV32 oral papillomatous lesions in a patient with cGVHD.
# | discussion
Focal epithelial hyperplasia is an uncommon disease that follows mucosal HPV13 or HPV32 infection with sessile papule development ranging from minor to severe that affects the oral cavity and anal/genital mucosae. It is more commonly detected in young Native or Indigenous populations of the Americas.Children with FEH are more likely to be carrying HPV13, but adult FEH cases are more frequently linked to HPV32. Often, FEH is left untreated as, particularly in children, it is often a self-limiting disease. [bib_ref] Surgical management of nonmalignant lesions of the mouth, Woods [/bib_ref] In cases with cosmetic or other concerns, surgical excision may be done. To date, FEH has not been described in the literature in adult patients with cGVHD, suggesting a previously unreported population susceptible to this uncommon disease. This case includes a patient who was post-transplant, on long-term steroid therapy for severe lung cGVHD, with an altered immune system that likely predisposed the patient to persistent HPV infection.
A clear first line therapeutic has not been developed for patients with FEH. Traditional treatment options include surgery/cryosurgery, Toll-like receptor agonists, and interferon-β. 14-16 Imiquimod, a topical agent, was shown to successfully treat FEH in three cases but was ineffective here. [bib_ref] Treatment of focal epithelial hyperplasia with topical imiquimod: report of three cases, Yasar [/bib_ref] Topical interferon-β cleared infection in a 4-year-old Turkish boy. 14 While surgical excision has been shown to be effective, it requires removing more healthy tissue than necessary particularly for widespread lesions. Quantum molecular resonance scalpel surgery and diode laser (808 nm) removal have had success with the potential to selectively ablate papillomatous tissue, resulting in reduced scarring. [bib_ref] Diode laser therapy for Heck's disease associated with HPV13 infection, Bombeccari [/bib_ref] [bib_ref] A case of Heck's disease treated with quantum molecular resonance scalpel, Sarraj [/bib_ref] In conclusion, we report the first KTP laser treatment of a chronic graft-vs-host disease patient with HPV32 positive FEH.
[fig] F: I G U R E 1 Oral photographs of widespread focal epithelial hyperkeratosis or Heck's disease in a patient with cGVHD. (A), upper lip; (B), lower right labial mandibular ridge; (C), left buccal mucosa; (D) lower lip. Original photographs taken at NIH [/fig]
|
Network meta-analysis of infliximab biosimilars for the treatment of rheumatoid arthritis
A c c e p t e d M a n u s c r i p t Purpose. This article assesses the relative efficacy and safety of infliximab biosimilars in treatment of patients with rheumatoid arthritis (RA).Methods.A frequentist, random-effects network meta-analysis was performed to evaluate evidence from randomized controlled trials that examined the use of infliximab biosimilars for treatment of patients with RA. PubMed and MEDLINE and other sources were searched for reports evaluating rates of response to treatment with the reference product (infliximab) vs an infliximab biosimilar.The primary efficacy outcome of interest was the rate of attainment of ACR20 (ie, 20% improvement in American College of Rheumatology core measures). The primary safety outcome was the rate of treatment-related serious adverse events (SAEs). Data were extracted by the primary author, and an assessment for risks of methodological bias was performed for each evaluated study.Results. Five studies that enrolled a total of 2,499 patients were included. Overall comparisons using odds ratios and 95% confidence intervals (CIs) did not indicate statistically significant differences in response to treatment with biosimilar agents relative to each other or the infliximab reference product. ORs for ACR20 response for biosimilars vs infliximab were as follows: 1.475 (95% CI, 0.940-2.315) for infliximab-axxq, 1.259 (95% CI, 0.854-1.855) for infliximab-dyyb, 0.865 (95% CI, 0.5511.358) for infliximab-qbtx, and 0.832 (95% CI, 0.506-1.367) for infliximab-abda. Similar findings were observed in reported SAE rates among patients treated with the various biosimilars.Conclusion. ACR20 response appears to be comparable and nonsignificantly different between infliximab biosimilars. In the absence of any meaningful differences in safety or efficacy, biosimilar cost may be the deciding factor in choosing a treatment or agent for formulary inclusion.
A c c e p t e d M a n u s c r i p t Am J Health-Syst Pharm. 2021;78:xxx-xxx Tumor necrosis factor (TNF) is a proinflammatory endogenous cytokine that plays a role in many inflammatory conditions, including rheumatoid arthritis (RA), ankylosing spondylitis, Crohn's disease, and ulcerative colitis. [bib_ref] IBM Corporation; 2020, Infliximab [/bib_ref] [bib_ref] AGA clinical practice guidelines on the management of moderate to severe ulcerative..., Feuerstein [/bib_ref] [bib_ref] American College of Rheumatology Guidelines for the Treatment of Rheumatoid arthritis, Singh [/bib_ref] TNF inhibitors may decrease symptoms and slow the progression of disease in these patients. [bib_ref] AGA clinical practice guidelines on the management of moderate to severe ulcerative..., Feuerstein [/bib_ref] [bib_ref] American College of Rheumatology Guidelines for the Treatment of Rheumatoid arthritis, Singh [/bib_ref] Current guidelines from the American College of Rheumatology (ACR) recommend using TNF inhibitors in patients who do not respond to monotherapy with first-line disease-modifying antirheumatic drugs, such as methotrexate. [bib_ref] American College of Rheumatology Guidelines for the Treatment of Rheumatoid arthritis, Singh [/bib_ref] Numerous TNF inhibitors have been approved by the US Food and Drug Administration (FDA), with the oldest being infliximab (Remicade, Janssen Biotech) which came to market in 1998.In recent years, several biosimilars to infliximab have been developed, including infliximab-dyyb (Inflectra, Pfizer), infliximab-abda (Renflexis, Merck), infliximab-qbtx (Ixifi, Pfizer), and infliximabaxxq (Avsola, Amgen).Biosimilars are highly similar to their respective originator reference products, with only minimal clinical differences in safety, purity, or potency. [bib_ref] Biosimilars in rheumatology: understanding the rigor of their development, Goel [/bib_ref] Although noninferiority or equivalency studies of biosimilars and their originator products are conducted, biosimilars are not often directly compared with each other. Thus, there may be unrecognized differences in efficacy or safety between biosimilar agents themselves. This lack of evidence makes it challenging for clinicians, payers, and healthcare organizations to make decisions about drug formularies and clinical care when choosing between biosimilars.
To date, no indirect or head-to-head studies comparing all infliximab biosimilar agents against one another have been published. Therefore, a network meta-analysis may be useful because it allows for multiple comparisons across a range of interventions, even in the absence of direct evidence. [bib_ref] The PRISMA extension statement for reporting of systematic reviews incorporating network meta-analyses..., Hutton [/bib_ref] The objective of the study described here was to evaluate the comparative efficacy and safety of FDA-approved infliximab biosimilars for the treatment of RA using a network meta-analysis framework. M a n u s c r i p t
# Methods
The study was designed and is reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) extension guidelines for network metaanalyses. [bib_ref] The PRISMA extension statement for reporting of systematic reviews incorporating network meta-analyses..., Hutton [/bib_ref] Study selection and eligibility criteria. Studies were included if they met the following parameters: a randomized controlled trial (RCT) study design; inclusion of patients diagnosed with RA; evaluation of an FDA-approved infliximab biosimilar in patients who had an incomplete response to methotrexate; and reporting of efficacy in terms of ACR20 response rate (ie, 20% improvement in American College of Rheumatology core measures). There was no minimum sample size requirement for inclusion of a study in the meta-analysis. Studies were excluded if they had a quasi-RCT design, included duplicate data (ie, were repeat publications), or involved patients who were not naïve to infliximab therapy (eg, studies that evaluated switching from Remicade to an infliximab biosimilar). These switching studies were excluded because most were extensions of the original published trials; therefore, it was believed that including them could bias the analysis due to counting of the same patient cohorts multiple times within the network.
Outcome measures. The primary efficacy endpoint was the ACR20 response rate. The ACR20 response rate was chosen because it is often the primary measure of RA treatment efficacy across published studiesand is recognized as a core measure of disease activity by ACR. 12 ACR20 is a dichotomous endpoint whose achievement requires 20% or greater improvement in tender and swollen joint counts (per an assessment of prespecified joints), as well as an improvement of ≥20% in 3 of 5 other areas: inflammation (as evidenced by an elevated erythrocyte sedimentation rate or C-reactive protein level), patient assessment of disease activity (based on Likert response scale or specific questions within broader self-assessment instruments), physician assessment of disease activity (based on horizontal visual analog scale or Likert scale assessment), patient assessment of pain (based on horizontal visual analog scale or Likert scale assessment), and patient assessment of A c c e p t e d M a n u s c r i p t physical function (using any physical function scale, such as the Arthritis Impact Measurement Scale or Health Assessment Questionnaire-Disability Index [HAQ-DI]). [bib_ref] The American College of Rheumatology preliminary core set of disease activity measures..., Felson [/bib_ref] The primary safety endpoint was the rate of occurrence of any serious adverse event (SAE).
Although SAEs are not always well defined in clinical trials, reporting of SAEs generally follows the definition outlined by FDA, which includes a life-threatening event, death, hospitalization, disability, permanent damage, and other outcomes that may jeopardize the patient or require medical intervention. 14,15 A comparison of rates of specific adverse reactions was not deemed to be feasible due to differences in how they were generally codified and reported in the articles. independent evaluation was needed to validate questionable article inclusions. After initial screening, full reports for all titles that appeared to meet the inclusion criteria, as well as those whose appropriateness for inclusion was uncertain, were obtained. The full-text reports were then screened against the inclusion criteria. The following data were extracted using a standardized collection form: study design parameters, sample size, key patient demographics (age, markers of disease severity, etc), information regarding interventions (drug dose, frequency, etc), details related to study efficacy endpoints (primary efficacy outcome, follow-up period, prespecified equivalence A c c e p t e d M a n u s c r i p t margin, intention-to-treat vs per-protocol methodology, results for primary outcome, etc), and key information related to safety endpoints (primary safety outcome, frequencies and types of of adverse effects [AEs], type of adverse effect, etc). Of note, the abstracted data from the intentionto-treat population in each study were used in the primary efficacy and safety evaluation.
Validity assessment. The Cochrane Risk of Bias Tool (version 2) was used to assess the quality of studies based on the following criteria: randomization process, deviations from intended interventions, missing outcome data, measurement of the outcome, and selection of reported results.Quality of evidence in each domain was then rated as low, high, or unclear. A qualitative synthesis of clinical and methodological heterogeneity was also conducted, as well as an assessment of transitivity. [bib_ref] Network meta-analysis: an introduction for clinicians, Rouse [/bib_ref] [bib_ref] Network meta-analysis: application and practice using Stata, Shim [/bib_ref] Statistical analysis. Stata version 16.1 (StataCorp LLC, College Station, TX) was used to perform the network meta-analysis using the "mvmeta" and "network" commands, a method that follows a frequentist approach. [bib_ref] Network meta-analysis: an introduction for clinicians, Rouse [/bib_ref] [bib_ref] Network meta-analysis: application and practice using Stata, Shim [/bib_ref] [bib_ref] Network meta-analysis, White [/bib_ref] The "network meta inconsistency" command within Stata was used to assess model consistency. A pairwise odds ratio (OR) and 95% confidence interval (CI) was determined to measure the association between a treatment and its efficacy. Results were considered statistically significant if the 95% CI did not contain the value 1. The Confidence in Network Meta-Analysis (CINeMA) Tool (version 1.9.1), which also uses a frequentist approach, was used to generate a corresponding league table comparing the relative effects between agents, including indirect pairwise comparisons between biosimilars. [bib_ref] CINeMA: An approach for assessing confidence in the results of a network..., Nikolakopoulou [/bib_ref] A random-effects model was used for the CINeMA analysis. This same approach was used to perform prespecified sensitivity analyses to evaluate the impact of follow-up period differences on ACR20 response.
The efficacy and safety of each biosimilar agent in the treatment of RA was then arranged in the order of the probability of being ranked as the best-performing agent and displayed via a rankogram. Information on relative effects was converted to a probability that a treatment would be the best, second best, third best, fourth best, or worst by ranking each therapy according to the surface under the cumulative ranking curve (SUCRA). The SUCRA represents a numeric summary of A c c e p t e d M a n u s c r i p t the overall rank distribution associated with each treatment and supplements graphical representations for a given outcome. The SUCRA value is 1 when a treatment is certain to be the best and 0 when a treatment is certain to be the worst. [bib_ref] Indirect treatment comparison, Chaimani [/bib_ref] [bib_ref] Approaches to interpreting and choosing the best treatments in network meta-analyses, Mbuagbaw [/bib_ref] Results Study characteristics. The initial literature search yielded 294 results. After removing duplicates, 192 article titles and abstracts were reviewed, which resulted in 181 exclusions. The most common reason for exclusion was a non-RCT article type. The remaining 11 articles underwent fulltext review and, of these, 6 were excluded . Thus, a total of 5 studies that cumulatively enrolled 2,499 patients met inclusion criteria and were included in the analysis. A summary of these studies is provided in . A c c e p t e d M a n u s c r i p t Risk of bias. All 5 of the included studies were determined to have a low overall risk of bias (eFigure 1). Although all studies were randomized, authors of 3 of the 5 studies failed to provide specific details on how the random allocation sequence was generated; however, based on reviewer judgment, this was not deemed to have resulted in a high risk of bias given that these were large, multicenter trials run by experienced clinical trial teams. This decision was acceptable under the Cochrane Risk of Bias Tool (version 2) methodology.Comparative efficacy. In general, individual study results for efficacy demonstrated equivalence of each biosimilar to the infliximab reference product in terms of ACR20 response . Infliximab-axxq 24 and infliximab-dyyb 27,28 were the only products associated with a higher ACR20 response rate than the reference product, whereas infliximab-qbtx 25 and infliximab-abda 26 both were associated with lower response rates (albeit still within the prespecified equivalence margins).
Notably, the initial risk difference for infliximab-axxq exceeded the upper bound of the prespecified equivalence margin, such that superiority might be considered; however, this study was not specifically designed to test for superiority, and a post hoc analysis reduced the CI to within the equivalence range. [bib_ref] Comparative clinical efficacy and safety of the proposed biosimilar ABP 710 with..., Genovese [/bib_ref] As described in , results of the network meta-analysis did not indicate any significant differences in ACR20 response between biosimilar agents relative to each other or the infliximab reference product. Although point estimates exceeded 1 in several comparisons, all 95% CIs crossed 1 and there was wide CI overlap between treatments. Relative to the infliximab reference product, the OR for ACR20 achievement was 1.47 (95% CI, 0.94-2.32) with use of infliximab-axxq;, 1.259 (95% CI, 0.854-1.855) for infliximab-dyyb, 0.865 (95% CI, 0.551-1.358) for infliximab-qbtx, and 0.832 (95% CI, 0.506-1.367) for infliximab-abda. These findings were consistent with results of a predefined sensitivity analysis to evaluate the impact of follow-up period differences on ACR20 response. A c c e p t e d M a n u s c r i p t Results of the cumulative ranking probability analysis suggested that infliximab-axxq has a 63.4% chance of being the best-performing agent, while the cumulative probability of it being at least second best is 92% and eFigure 2). The numerical summary of the rank distribution for each treatment (ie, SUCRA ranking probabilities) was 0.9 for infliximab-axxq, 0.7 for infliximab-dyyb, 0.4 for the infliximab reference product, and 0.2 for both infliximab-qbtx and infliximab-abda.
Comparative safety. In general, the rates of AEs differed across individual studies. Overall AE rates were slightly lower in the evaluation of infliximab-axxq by Genovese and colleagues 24 and substantially higher in the evaluation of infliximab-dyyb by Takeuchi and colleagues, 28 in which over 80% of patients in both the biosimilar and reference product arms experienced an AE. The rate of SAEs followed a similar trend, with lower rates reported in the article by Genovese and colleagues [bib_ref] Comparative clinical efficacy and safety of the proposed biosimilar ABP 710 with..., Genovese [/bib_ref] and higher rates in the evaluation by Takeuchi and colleagues. 28 Additional details are provided in . Unfortunately, differences in data reporting made it difficult to compare rates of specific AEs; however, the most commonly reported AEs across the studies were infusion reactions, hypersensitivity, infection, and elevated liver enzymes. Results of our meta-analytic analysis do not suggest any statistically significant differences in SAE rates between biosimilars .
# Discussion
The study was conducted to compare the relative clinical effectiveness of infliximab and its FDA-approved biosimilars. A network meta-analysis design was chosen to perform an indirect treatment comparison given the lack of head-to-head comparative data on the 5 agents. Overall, the point estimates and 95% CIs of our ACR20 efficacy results suggest that there are no clinically meaningful differences between agents. Although infliximab-axxq was found to have the highest probability of being the best at achieving ACR20 response based on SUCRA ranking probabilities, this result must be interpreted with caution. Specifically, SUCRA rankings may exaggerate small differences in relative effects, especially when based on data from a limited network [bib_ref] Approaches to interpreting and choosing the best treatments in network meta-analyses, Mbuagbaw [/bib_ref] ; this is because SUCRA analysis does not consider the magnitude of the difference in effects. Therefore, it is A c c e p t e d M a n u s c r i p t possible that the first-ranked treatment may be either slightly better or vastly better than the second-ranked treatment. Additionally, SUCRA values do not account for the possibility of random chance in the model. [bib_ref] Approaches to interpreting and choosing the best treatments in network meta-analyses, Mbuagbaw [/bib_ref] Similar to the primary efficacy results, the results of analysis of SAEs showed comparable and nonsignificant differences between products. However, these safety results should also be interpreted with caution given that the follow-up period varied between studies from 22 to 54 weeks. [bib_ref] Comparative clinical efficacy and safety of the proposed biosimilar ABP 710 with..., Genovese [/bib_ref] Likewise, definitions of SAEs were not uniformly defined across studies.
Overall, our findings further support the use of biosimilars in practice and add additional credence to the clinical similarity of the evaluated products. Biosimilars must undergo a rigorous comparison against the reference product in order to gain regulatory approval. Specifically, manufacturers must demonstrate biosimilarity through a totality-of-the-evidence approach, which includes structural and functional analytical studies, pharmacokinetic and pharmacodynamic evaluations, and, if necessary, immunogenicity studies, animal studies, and other comparative clinical trials.However, individual biosimilars are not directly compared with one another. Thus, although a biosimilar may be determined to be equivalent or noninferior to a reference product (depending on the study design), this does not preclude comparative differences between biosimilar agents. Thus, the results of this evaluation may help clinicians, payers, and healthcare organizations make decisions about the choice of biosimilar agent. With all other things being equal, product cost and contractual opportunities may become the primary deciding factor when selecting an infliximab biosimilar for formulary inclusion.
To our knowledge, this is the only published study using a network meta-analytic technique to compare all 4 infliximab biosimilars with one another for the treatment of RA. Previous literature has mostly reported on direct comparisons of biosimilars with the reference product in RCTs. Aside from these head-to-head trials, 2 recent meta-analyses comparing biosimilars to the infliximab reference product have been published. In a study by Bae and colleagues, 29 pooled outcomes data for infliximab-abda and infliximab-dyyb plus methotrexate were compared against data on both use of infliximab plus methotrexate and use of a placebo plus methotrexate. Although individual A c c e p t e d M a n u s c r i p t biosimilars were not compared in that study, results indicated no major difference in ACR20 response between the pooled biosimilars and infliximab. Similarly, a study by Graudal and colleagues 30 found that the individual treatment effects of infliximab-abda and infliximab-dyyb were comparable to the reference product's in terms of RA progression; however, these biosimilars were not directly compared.
The results of our meta-analysis should be considered in the context of several limitations. Importantly, the network of trials was sparse, with only one study being identified for each of the biosimilars except for infliximab-dyyb, for which 2 studies were identified; this limited the robustness of our evaluation and the indirect comparisons performed. Additionally, while the included studies were all highly similar with regard to methodology and design, follow-up periods differed significantly. Long-term data were available only for infliximab-dyyb and infliximab-abda, which could have impacted the applicability of our findings. To address this limitation, a sensitivity analysis was conducted to evaluate ACR20 response at 30 weeks instead of 54 weeks, which was the follow-up period reported in the 2 longer-term studies. The results of the sensitivity analysis were consistent with findings of the primary efficacy analysis, indicating that follow-up period duration did not greatly impact the results (see eTable 2). Lastly, the primary efficacy analysis focused on only one outcome, ACR20, and assessed only efficacy data from RCTs. Investigators who conduct future network meta-analyses may consider evaluating additional outcomes or incorporating real-world effectiveness data from observational studies.
# Conclusion
Results of a network meta-analysis suggest that infliximab biosimilars are generally comparable with regard to ACR20 response and SAEs. In the absence of any meaningful differences in safety or efficacy, cost may be the deciding factor when choosing a biosimilar agent for formulary inclusion.
# Disclosures
The authors have declared no potential conflicts of interest. |
The reciprocal influences of prognosis between two types of surgical interventions and early breast cancer patients with diverse luminal subtypes
Background: To investigate and compare the effects of breast-conserving therapy (BCT) and mastectomy on the disease recurrence and long-term survival in early-stage luminal breast cancer and the difference in prognosis across diverse luminal subtypes receiving single surgical modality.Methods: The databases of PubMed and Embase were retrieved to select eligible trials that were published from inception to 13 November 2018. The clinical trials that offered the details about recurrent disease and/or survival in luminal tumors underwent BCT or mastectomy met the inclusion criteria (n=24). With the random-or fixed-effect model basing on heterogeneity Chi 2 test with its significant level of P < .1, pooled odds ratio (OR) with its 95% CI, and P value were identified for endpoints.Results: The analyzed data were constituted of 25 qualified trials with 13,032 unique women suffered from luminal cancers. The fixed-effect models were utilized. On the LRR regarding BCT versus mastectomy, the pooled data indicated no significant difference in luminal carcinomas (OR, 0.84; 95%CI, 0.43-1.64; P = .61; n = 867). In BCT cohort, the pooled data showed that there were some significant benefits favoring luminal A over luminal B in LR (OR, 0.61; 95%CI, 0.46-0.81; P = .0007; n = 5406), DM (OR, 0.53; 95%CI, 0.41-0.69; P < .00001; n = 4662), DFS (OR, 0.59; 95%CI, 0.36-0.96; P = .03; n = 776) and OS (OR, 0.65; 95%CI, 0.42-0.99; P = .05; n = 1149), but not in LRR (OR, 0.74; 95%CI, 0.48-1.13; P = .16; n = 3732), coupled with luminal A/B over luminal-HER2 in LRR (OR, 0.43; 95%CI, 0.25-0.76; P = .004; n = 890), DM (OR, 0.56; 95%CI, 0.35-0.90; P = .02; n = 1396), DFS (OR, 0.47; 95%CI, 0.27-0.83; P = .009; n = 532); in mastectomy cohort, there were apparent advantages of LRR (OR, 0.58; 95%CI, 0.36-0.92; P = .02; n = 1768), LR (OR,0.56; 95%CI, 0.38-0.83; P = .004; n = 1209), DM (OR, 0.58; 95%CI, 0.40-0.84; P = .004; n = 652) and OS (OR, 0.62; 95%CI, 0.43-0.89; P = .009; n = 652) in luminal A vs luminal B.Conclusion: For early luminal breast cancer, the equality of LRR was achieved in BCT and mastectomy. In comparison, luminal A cancers benefit the most improved tumor re-appearence and survival in luminal diseases regardless of the option of surgical modality, whereas luminal-HER2 is affected by the worst clinical outcomes in them who follows BCT.Abbreviations: BCT = breast-conserving therapy, CI = confidence interval, DFS = disease-free survival, DM = distant metastasis, ER = estrogen receptor, HER2 = human epidemic growth factor receptor 2, LR = local recurrence, LRR = localregional relapse, OR = odds ratio, OS = overall survival, pCR = pathological complete response, PFS = progress free survival, PR = progesterone receptor, TNBC = triple negative breast cancer.
# Introduction
The discernible landscape that radiotherapy following breastconserving therapy (BCT) and mastectomy achieve equivalent disease-free survival (DFS) and overall survival (OS) for early breast cancer has been well confirmed by a myriad of large randomized controlled trials. [bib_ref] Twenty-year follow-up of a randomized trial comparing total mastectomy, lumpectomy, and lumpectomy..., Fisher [/bib_ref] [bib_ref] Twenty-year follow-up of a randomized study comparing breast-conserving surgery with radical mastectomy..., Veronesi [/bib_ref] Approximately 5% to 10% of operable patients undergo mastectomy and 10% to 20% of early invasive breast cancer receiving BCT will gradually develop a tumor recurrence within 10 years, [bib_ref] Prognosis of patients with local recurrence after mastectomy or conservative surgery for..., Fodor [/bib_ref] [bib_ref] The time-course of metastases from breast cancer after mastectomy and breast-conserving surgery..., Fodor [/bib_ref] [bib_ref] Long-term results of a randomized trial comparing breast-conserving therapy with mastectomy: European..., Van Dongen [/bib_ref] thus increasing the risk of distant metastasis (DM) and mortality. In this context, BCT has become an adequate surrogate of local regional treatment for mastectomy in patients with early breast cancer in that it maximum downsizes the physiological and psychological burden of sacrificing the breast.
Even if the gross mass tumor is successfully excised and the surgical margin is negative, as the breast cancer is multifocal, [bib_ref] Histologic multifocality of Tis, T1-2 breast carcinomas. Implications for clinical trials of..., Holland [/bib_ref] thus will remaining microscopic residual lesions, if left untreated, 30% to 40% of these women are still in the detriment of disease recrudesce. [bib_ref] Twenty-year follow-up of a randomized trial comparing total mastectomy, lumpectomy, and lumpectomy..., Fisher [/bib_ref] [bib_ref] Effects of radiotherapy and of differences in the extent of surgery for..., Clarke [/bib_ref] Many signaling pathways are relative to the reappearance of breast tumor, including the destruction of estrogen-receptor-related signaling pathways, and the amplification or overexpression of proto-oncogenes such as human epidemic growth factor receptor 2 (HER2). [bib_ref] Molecular biology and genetics of breast cancer development: a clinical perspective, Buchholz [/bib_ref] Recently, according to the expression level of these receptors and tumor grade, breast cancer can be divided into the following five subtypes: luminal A, estrogen receptor (ER) positive or progesterone receptor (PR) positive and HER2 negative with grade 1 or 2; luminal B, ER positive or PR positive and HER2 negative with grade 3; luminal-HER2, ER positive or PR positive and HER2 positive; HER2-enriched, ER negative, PR negative and overexpression of HER2; triple negative breast cancer (TNBC), negativity of ER, PR and HER2. [bib_ref] Gene expression profiling predicts clinical outcome of breast cancer, Van't Veer [/bib_ref] [bib_ref] Repeated observation of breast tumor subtypes in independent gene expression data sets, Sorlie [/bib_ref] [bib_ref] Molecular classification of breast carcinomas by immunohistochemical analysis: are we ready?, Tang [/bib_ref] Historically, the luminal breast cancers are always prone to benefit a favorable prognosis, with a recurrent tumor rate 2 to 3 times less than HER2-amplified and TNBC that are wildly recognized as the high-risk tumors. [bib_ref] Breast cancer subtype approximated by estrogen receptor, progesterone receptor, and HER-2 is..., Nguyen [/bib_ref] [bib_ref] Breast cancer subtypes and the risk of local and regional relapse, Voduc [/bib_ref] In 2012, a systematic review was implemented by Lowery and colleagues who investigated and compared the local regional relapse (LRR) of breast cancer patients with different molecular subtypes after BCT or mastectomy, [bib_ref] Locoregional recurrence after breast cancer surgery: a systematic review by receptor phenotype, Lowery [/bib_ref] reaffirming this conception. Inadequately, this ingenious study fails to investigate whether the surgical decisions could impact on the disease recrudesce and long-term survival of luminal cancers and what the influence of different luminal subtypes undergo single modality of surgical intervention on these clinical outcomes. To settle this issue, herein, we performed a meta-analysis to establish and compare the prognosis in two types of treatment scenarios followed by luminal disease and the difference across various luminal subtypes, including luminal-HER2 if applicable, who were treated with either BCT or mastectomy.
# Methods
## Search strategy
Using the accurate retrieval strategy: luminal AND (mastectomy OR (breast conserving surgery) OR (breast preserving surgery) OR (breast conservation surgery) OR (breast conserving treatment) OR (breast preservation treatment) OR (breast conservation treatment) OR (breast conserving therapy) OR (breast preserving therapy) OR (breast conservation therapy)) AND ((breast cancer) OR (breast tumor) OR (breast neoplasms)) AND ((local regional recurrence) OR (local regional relapse) OR (pathological complete response) OR (overall survival) OR (disease free survival) OR (progress free survival) OR (disease metastasis) OR (metastasis free survival) OR LRR OR pCR OR OS OR DFS OR PFS), electronic searches were conducted in databases of PubMed and Embase. In the course of the retrieval procedure, no restrictions were required. Citation searching was finished as of 9 November 2018.
## Inclusion criteria
Clinical trials; Early stage breast cancer female patient with luminal phenotype; Delineating the outcomes of disease recurrence and/or longtime survival in BCT arm and/or mastectomy arm; The precise number of events or event ratio coupled with total sample size was provided.
## Exclusion criteria
Not published in English; Review, case report, conference abstract, or conference paper; Be incapable of meeting the inclusion criteria.
The retrieved citations were independently screened by two coauthors (Qian Wu and Zhumin Su) on the basis of titles, abstracts and full-texts, and only the satisfactory studies that met the inclusion criteria were reserved. Provided comprising of any discordance, it was addressed by discussion.
# Data abstraction
With the application of Excel vision 2016, the following information was respectively abstracted by two reviewers (Qian Wu and Biyuan Zhang) from eligible trails: the first author, original nation, publication year, study duration, median followup, the regimen of adjuvant therapy and radiotherapy, total sample size as well as the number of events. If some inconformities surfaced, they were resolved by the third reviewer (Lijiu Zhang).
# Statistical analysis
The crude odds ratio (OR) with its 95% confidence interval (CI), and P value regarding to all valid relapsing disease and survival Medicine benefit for each included study was calculated. Provided that the number of events was not described, its computation was obtained in light of the endpoint percentage or other information seen in the publication. The Heterogeneity Chi 2 test with its significant level of P < .1 was employed to evaluate the heterogeneity among different studies. [bib_ref] Methods for Meta-analysis in Medical Research, Sutton [/bib_ref] A random-effect model was used to integrate the data when heterogeneity test appeared to no statistically significant (P < .1); if with a drastically different situation, a fixed-effect model was utilized. [bib_ref] Methods for Meta-analysis in Medical Research, Sutton [/bib_ref] The publication bias was assessed by creating a funnel plot with its 95% CI. If the data were uniformly arranged at the left and right of the plot, it denoted no major asymmetry that signified a likelihood of publication bias. The difference of eligibility criteria, sample size, potential bias, as well as treatment regimen in selected trials was discussed, and their qualities were estimated in terms of the instrument provided by Jadad and colleagues [bib_ref] Assessing the quality of reports of randomized clinical trials: is blinding necessary?, Jadad [/bib_ref] (Supplemental , http://links.lww.com/MD/C883 in Appendix, page 6).
All statistical tests were analyzed by using Revman Manager software version 5.3 and tracked once more.
# Results
# Search results
Following the path of the systematic retrieval, we collected 753 potential citations, of which classified as duplications (n = 123), reviews (n = 35), case reports (n = 13), as well as conference abstracts (n = 229) coupled with conference papers (n = 7) were deleted. The remainder articles (n = 346) were selected by applying title and abstract screening. After the previous procedure, a total of 53 studies were entered into full-text scrutinization, and 28 of them were excluded by virtue of no association between luminal cancer with surgical paradigm (n = 19), containing males (n = 1), no prognosis (n = 2), without comparison of luminal subtypes in mastectomy arm (n = 4) or in BCT arm (n = 2). Ultimately, 25 eligible clinical trials [bib_ref] Breast cancer subtypes and the risk of local and regional relapse, Voduc [/bib_ref] with 13,032 unique patients were involved for data extraction after removal of ineligible studies that were unable to meet the inclusion criteria. The PRISMA flow diagram was outlined in [fig_ref] Figure 1: The flow diagram of selecting trials [/fig_ref]. The details of radiotherapy and adjuvant therapy details in analyzed studies.
## Localregional recurrence
As demonstrated in [fig_ref] Figure 2: The comparison of LRR between BCT and mastectomy received by luminal tumors [/fig_ref] , the pooled data with respect to treatment modality of BCT compared to mastectomy for early breast cancer indicated that no statistically The fix-effect model was leveraged for all analyses owing to no heterogeneity among selected studies.
## Publication bias
The funnel plots were drawn, as provided in eFigure 3a-o, http:// links.lww.com/MD/C883 (Appendix, page 9-16), in which the analyzed data were evenly distributed at the left and right sides of the plots, indicating no advent of major asymmetry that might give rise to heterogeneity, indeed, which was not detected in each meta-analysis.
# Discussion
As mentioned earlier that the equivalency of OS and DFS was certified in early breast tumor patient who underwent the treatment scenarios of mastectomy and BCT, our results further corroborate that the two paradigms also achieve equivalent LRR in luminal cancer. Additionally, breast cancer women with luminal A subtype are entitled to a degressive disease reappearence and reinforced survival benefit, compared to luminal B. Whilst both subtypes considered as an entirety in comparison to luminal-HER2, the latter is underway to be in disadvantageous.
Breast cancer exhibits diversity in tumor invasiveness [bib_ref] Repeated observation of breast tumor subtypes in independent gene expression data sets, Sorlie [/bib_ref] [bib_ref] Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications, Sorlie [/bib_ref] and responsiveness to systematic therapy [bib_ref] Breast cancer molecular subtypes respond differently to preoperative chemotherapy, Rouzier [/bib_ref] on the basis of different molecular phenotypes. At present, the limited data are inefficient to validate that molecular subtype is a robust factor in predicting LRR. [bib_ref] Supervised risk predictor of breast cancer based on intrinsic subtypes, Parker [/bib_ref] Although some studies have sought to evaluate the result of this problem, their conclusions are still suspect in that the heterogeneity of patient populations and surgical strategies deserves to be in consideration (for instance, some only centers on patients treated with BCT, while others involve women under mastectomy). [bib_ref] Breast cancer subtype approximated by estrogen receptor, progesterone receptor, and HER-2 is..., Nguyen [/bib_ref] [bib_ref] Treatment and survival outcome for molecular breast cancer subtypes in black women, Ihemelandu [/bib_ref] [bib_ref] Locoregional relapse and distant metastasis in conservatively managed triple negative early-stage breast..., Haffty [/bib_ref] [bib_ref] Triple-negative breast cancer: clinical features and patterns of recurrence, Dent [/bib_ref] Our results suggested that the similar LRR was captured in breast tumors with luminal phenotype regardless of the surgical strategy, and in luminal A compared to luminal B when treated with BCT. However, there was an exception of comparison between both subtypes who underwent mastectomy, representing that a significantly improved LRR benefit favored luminal A over luminal B. As the reduplicative observations that the uppermost survival dividend is experienced in luminal cancer, [bib_ref] Gene expression profiling predicts clinical outcome of breast cancer, Van't Veer [/bib_ref] [bib_ref] Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications, Sorlie [/bib_ref] these results further consolidates them and are consistent with the study of Voduc and colleagues, [bib_ref] Breast cancer subtypes and the risk of local and regional relapse, Voduc [/bib_ref] Notably, based on the consistent appreciation that postoperative radiotherapy can reduce the risk of tumor metastasis, disease recurrence and mortality, [bib_ref] Postoperative radiotherapy in high-risk postmenopausal breast-cancer patients given adjuvant tamoxifen: Danish Breast..., Overgaard [/bib_ref] [bib_ref] Locoregional radiation therapy in patients with high-risk breast cancer receiving adjuvant chemotherapy:..., Ragaz [/bib_ref] Wu extracted patients who received radiotherapy after surgery for subgroup analysis and uncovered luminal A with a significantly reduced rate of DM by comparison with luminal B (26.9 vs 45.5, P < .05). Consequently, to sum up the aforementioned results, hormone therapy may be a crucial factor affecting the rate of DM in luminal patients accepting BCT, and postoperative radiotherapy may pave a superior way to lower DM for women affected by luminal A breast cancer in paradigm of mastectomy.
The preferable prognosis with reference to OS and DPS is always attached importance by people who suffer from carcinomas. Our study found that there was a more promising OS in luminal A breast cancer compared with luminal B under both treatment modality, alongside a favorable DFS in whom diagnosed with luminal A disease choosing BCT as the treatment intervention, which is in agreement with some similar studies. A longish-term clinical trial initiated by Kaiser and colleagues who took the 10-year OS as an endpoint and realized a statistically significant result that luminal B cancer women who received BCT had disadvantageous survival compared with luminal A (83.2% vs 89.1%; P = .04). [bib_ref] Intraoperative tumor bed boost with electrons in breast cancer of clinical stages..., Kaiser [/bib_ref] Moreover, the optimal DPS in breast cancer women with luminal A subtype (87.4% vs luminal B of 82.6%, P = .04; vs luminal-HER2 of 74.8%, P = .006) was confirmed by the study of Jia et al. who retrospective analyzed 405 breast cancer patients with luminal phenotype underwent BCT. [bib_ref] HER2-enriched tumors have the highest risk of local recurrence in Chinese patients..., Jia [/bib_ref] Therefore, it is reasonable to believe that an improved survival is beneficial to early-stage breast cancer patient with luminal A in any surgical strategies.
It is acknowledged that there are some limitations in this article. First, albeit no publication bias, the inclusion criteria exerted restriction on publication in English might lead to selection bias. Second, for the purpose to omit mixing diverse therapies and avert heterogeneity among included trails, only articles with similar arms were analyzed, thus causing only 2 to 4 eligible studies with small sample sizes in some meta-analyses, which might lead to result bias. Third, as reviewed above, hormone therapy and postoperative radiotherapy were essential agents that impacted on prognosis in luminal disease. However, due to finite information in the included trials, we did not conduct a subgroup analysis on them that might imply some impressive findings.
Despite of these limitations, this novel meta-analysis with a large volume of sample size amply evidences that LRR of luminal breast cancers not vary with the alternative of surgical decisionmaking, but with the usage of certain surgical intervention, either BCT or mastectomy, the best prognosis and the minimum rate of tumor relapse preferred luminal A tumor. In the future, the results of subgroup analysis in terms of luminal tumor whether receives hormone therapy and postoperative irradiation will be also in expectation.
# Conclusion
For early luminal breast cancer, the similar LRR appears after implementation of BCT and mastectomy. In addition, no matter how the treatment intervention is selected, patients with luminal A subtype experience the most reformative clinical prognosis of luminal diseases; whereas in women following BCT, the luminal-HER2 cancers are injured by the poorest outcomes.
# Author contributions
Data curation: Zhumin Su.
[fig] Figure 1: The flow diagram of selecting trials. BCT = breast conserving therapy, MAS = mastectomy. 2. Characteristics of analyzed trials The original nations of included studies were China (n = 3), the United States (n = 5), Denmark (n = 1), the Netherlands (n = 1), Canada (n = 4), Japan (n = 2), France (n = 2), Austria (n = 1), Australia (n = 1), Singapore (n = 1), Italy (n = 1), Turkey (n = 2), and South Korea (n = 1). The publication dates of them ranged from 2008 to 2018. The sample sizes ranged from 45 to 2202 (median: 442). In light of the assortments of disease relapse and long-dated survival, the eligible trials were stratified into 5 categories: the LRR trials (n = 16), in which concurrent focused on mastectomy and BCT (n = 4); the local recurrence (LR) trials (n = 8); the DM trials (n = 10); the DFS trials (n = 3); and the OS [/fig]
[fig] Figure 2: The comparison of LRR between BCT and mastectomy received by luminal tumors. (A) LRR in luminal cancers; (B) LRR in luminal A cancers; (C) LRR in luminal B cancers. BCT = breast conserving therapy, MAS = mastectomy. [/fig]
[fig] Figure 3: The comparison of LRR across luminal tumors with different molecular subtype undergoing mono-surgical strategy. (A) luminal A vs luminal B following breast conservation therapy; (B) luminal A/B vs luminal-HER2 following breast conservation therapy; (C) luminal A vs luminal B following mastectomy. [/fig]
|
Toluidine blue versus frozen section for assessment of mucosal tumor margins in oral squamous cell carcinoma
Background: When the resected specimen is sent for intraoperative margin assessment, all margins are grossly checked, and selected margins undergo a frozen section (FS) examination. Therefore, there is a possibility of sampling error. This study evaluated the effectiveness of using toluidine blue (TB) as an intraoperative triage screening tool to detect positive mucosal margins of the resected specimens of oral squamous cell carcinoma (OSCC) and serve as a guide for FS sampling. Methods: Surgical samples of 30 consecutive patients with biopsy-proven OSCC were included in the study. A total of 140 mucosal margins were analyzed intraoperatively by TB and FS, the results were compared with the final histopathology.Results: Of the 140 examined mucosal tumor margins, 14 stained positives with TB, six were true-positives, eight were falsepositives, and there were no false-negatives, as confirmed by final histopathology of the same margins. The diagnostic performance measures were sensitivity 100.0%; specificity 94.0%; positive predictive value (PPV) 42.9%; negative predictive value (NPV) 100.0%; and accuracy 94.3% (95% CI: 89.0-97.5%). For FS, there were three true-positives, three false-negatives, and no false-positives. The diagnostic performance measures were sensitivity 50.0%; specificity 100.0%; PPV 100.0%; NPV 97.8%; and accuracy 97.9% (95% CI: 93.9-99.6%).Conclusion: TB is less specific but more sensitive than FS for detecting positive mucosal margins of resected OSCC. Screening the tumor mucosal margins with TB before FS sampling may help identify more tumor-bearing margins.
# Background
Oral cancer is the eleventh most common malignancy in the world [bib_ref] Global incidence and risk factors of Oral Cancer, Ghantous [/bib_ref]. Oral squamous cell carcinoma (OSCC) is the most common oral cavity malignancy and represents more than 90.0% of oral cancers [bib_ref] Global epidemiology of oral and oropharyngeal cancer, Warnakulasuriya [/bib_ref]. The main treatment modality for managing OSCC is surgical resection [bib_ref] Treatment guidelines and patterns of care in oral cavity squamous cell carcinoma:..., Fujiwara [/bib_ref]. A critical issue of surgical oncology is achieving complete removal of a tumor at the primary site with a negative margin. Failure to identify residual neoplastic tissue results in positive surgical margins correlated with local recurrence and poor patient outcomes [bib_ref] The prognostic implications of the surgical margin in oral squamous cell carcinoma, Sutton [/bib_ref] [bib_ref] Influence of condition of surgical margins on local recurrence and diseasespecific survival..., Mcmahon [/bib_ref].
Several diagnostic methods have been used for intraoperative identification of tumor-involved margins, including visualization and palpation of the resected margins, touch imprint cytology, microendoscope, fluorescent techniques, Raman spectroscopy, narrow-band imaging, and optical coherence tomography [bib_ref] Surgical margins and its evaluation in oral cancer: a review, Ravi [/bib_ref] [bib_ref] A novel integrated platform for the identification of surgical margins in oral..., Baj [/bib_ref] [bib_ref] Surgical margins in oral squamous cell cancer: intraoperative evaluation and prognostic impact, Mannelli [/bib_ref]. These methods have their challenges and limitations in terms of appropriate performance, practicality, and cost-effectiveness.
A frozen section (FS) is still used in most oncological centers as the standard-of-care means for intraoperative margin assessment. Although it is a reliable method for identifying residual tumor in the sampled tissue, only selected margins-guided by gross examination-are analyzed. Therefore, there is a possibility of sampling error.
Toluidine blue (TB) is a metachromatic stain that is easily available, economical, and has a high affinity for DNA and RNA. It rapidly stains malignant and premalignant cells, but not normal mucosa. Several studies have demonstrated the ability of TB to detect oral cavity premalignant lesions and OSCC [bib_ref] Final evaluation of tolonium chloride rinse for screening of high-risk patients with..., Mashberg [/bib_ref] [bib_ref] Analysis of oral lesion biopsies identified and evaluated by visual examination, chemiluminescence..., Epstein [/bib_ref] [bib_ref] A personalized computational model predicts cancer risk level of oral potentially malignant..., Wang [/bib_ref]. However, its use as a screening tool for tumor-involved margins after surgical excision of OSCC has not been extensively studied.
In the literature, only three studies addressed TB intraoperatively. In one study, TB was used intraorally before resection to determine the extent of the lesions [bib_ref] The role of vital tissue staining in the marginal control of oral..., Kerawala [/bib_ref]. In the other two studies, TB was used intraorally to evaluate the tumor bed margins after excision of primary squamous cell carcinomas (SCC) [bib_ref] A comparative analysis of toluidine blue with frozen section in oral squamous..., Junaid [/bib_ref] [bib_ref] The role of toluidine blue in assessing margin status after resection of..., Portugal [/bib_ref] ; Portugal et al. [bib_ref] The role of toluidine blue in assessing margin status after resection of..., Portugal [/bib_ref] studied the role of TB in assessing margin status after resection of SCC of the upper aerodigestive tract (UADT). The authors applied TB directly to the remaining unresected mucosa. They reported a 100.0% sensitivity in the detection of tumor-involved margins with few false-positives. Junaid et al. [bib_ref] A comparative analysis of toluidine blue with frozen section in oral squamous..., Junaid [/bib_ref] evaluated TB staining of tumor bed margins after excision of primary OSCC. Their results indicated a 100.0% sensitivity for detecting positive margins and a specificity of 97.0%.
There are drawbacks to tumor-bed driven margin assessment. In short, tumor bed biopsies are so small, fragmented, unoriented, and not representative of the actual margin status (derived from the main resection specimen) [bib_ref] Early squamous cell carcinoma of the oral tongue: comparing margins obtained from..., Chang [/bib_ref] [bib_ref] Intraoperative margin assessment in early Oral squamous cell carcinoma, Chiosea [/bib_ref]. Recently, Kain et al. performed a systematic review on the surgical margins in oral cavity SCC. This review provides support for the practice of specimen-driven margin assessment when using FS analysis to improve the utility of the results [bib_ref] Surgical margins in oral cavity squamous cell carcinoma: current practices and future..., Kain [/bib_ref].
Due to the drawbacks of tumor-bed driven margin assessment, we decided to test the use of TB in the assessment of the mucosal margin status of a resected OSCC specimen (specimen-driven margin).
The aim of this study was to evaluate the effectiveness of using TB as an intraoperative triage screening tool to detect positive mucosal margins of the resected specimens of OSCC and serve as a guide for FS sampling.
# Methods
This prospective diagnostic accuracy study was designed according to the STARD checklist and was conducted at the Faculty of Dentistry and National Cancer Institute (NCI), Cairo University, Egypt, from July 2018 to June 2019. The accuracies of TB and FS were compared with the final histopathology of the same margins. Thirty consecutive patients with biopsy-proven OSCC who were undergoing primary excision were included in the study. The patients were recruited regardless of age, sex, ethnicity, and tumor stage or grade. The exclusion criteria were patients with a history of head and neck radiotherapy, those with previous treatment (surgery and radio−/ chemotherapy) for current OSCC because increased fibrosis and scar tissue can lead to mechanical retention of TB, making it difficult to interpret staining results. This study was approved by the Research Ethical Committee at the Faculty of Dentistry, Cairo University (number: . This study operated in compliance with the Helsinki Declaration, and written informed consent was obtained from all patients.
All surgical procedures were performed by an experienced head and neck surgeons at Head and Neck Surgical Oncology Unit, NCI, Egypt. After the surgical excision of the tumor, the surgeon marked the surgical margins by surgical sutures and then immediately sent the resected specimen to the Surgical Pathology Unit.
The mucosal margins were evaluated at two stages: During the first stage, the tissues were assessed intraoperatively (in the laboratory as the operation progressed in the operating room) using: 1) TB. The mucosal margins of the resected tumor were stained with TB according to the following protocol:
One percent TB solution staining was conducted as described by Mashberg [bib_ref] Final evaluation of tolonium chloride rinse for screening of high-risk patients with..., Mashberg [/bib_ref]. Briefly, irrigating the tumor margins was performed first using 1% acetic acid followed by normal saline. The margins were then gently dried with gauze. One percent toluidine blue solution was applied using a cotton swab and left in place for 30 s . Thereafter, the tissue was once again irrigated with 1% acetic acid (for eliminating the excess of stain) followed by normal saline. The margins stained in royal blue were labeled as positive, while those stained light blue or unstained were labeled as negative [bib_ref] Toluidine blue uptake in potentially malignant oral lesions in vivo: clinical and..., Gandolfo [/bib_ref].
2) Hematoxylin and eosin (H&E) stained FS. A minimum of four samples are shaved off the mucosal margins (specimen margin)-anterior, posterior, medial (or superior), and lateral (or inferior). Additional samples were taken if deemed valuable. Any margin that was positive by TB was included in FS sampling.
The FS interpretation was performed by another histopathologist who was blinded to the result of the TB staining (TB staining does not affect FS interpretation due to the known instability of TB in fixation and dehydration solutions [bib_ref] Toluidine blue uptake in potentially malignant oral lesions in vivo: clinical and..., Gandolfo [/bib_ref]. Even if staining persisted after processing, it would be masked by the H&E stain).
The second stage (final histopathological) involved evaluation of FS samples (FS remnants) taken intraoperatively, as described above, following formalin fixing and paraffin embedding and H&E staining. The final histopathological interpretation was performed by another histopathologist who was blinded to the TB and FS results. Resection margins which showed severe dysplasia, carcinoma in situ, or invasive carcinoma, were labeled as positive [bib_ref] Significance of positive margins in oral cavity squamous carcinoma, Loree [/bib_ref] [bib_ref] Frozen sections and complete resection in oral cancer surgery, Tirelli [/bib_ref]. The severity of the dysplasia was based on the 2017 WHO grading system [bib_ref] Oral potentially malignant disorders and oral epithelial dysplasia, Reibel [/bib_ref].
## Sample size calculation
A previous paper by Junaid et al. [bib_ref] A comparative analysis of toluidine blue with frozen section in oral squamous..., Junaid [/bib_ref] reported that TB staining had a sensitivity and specificity of 100.0 and 97.0%, respectively. Using a 95% confidence interval and a 5% significance level, 30 patients were required. The sample size was calculated by Arifin, W. N. (2017).
# Statistical method
The data were coded and entered using the Statistical Package for the Social Sciences (SPSS) version 25. Data are summarized using means, standard deviations, minimums, and maximums for quantitative data, and frequency (counts), and relative frequency (percentages) for categorical data. Standard diagnostic indices, including sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and diagnostic efficacy were calculated as described by Galen RS [bib_ref] Predictive value and efficiency of laboratory testing, Galen [/bib_ref]. For comparing categorical data, the Chi-square (χ 2 ) test was performed. The exact test was used instead when the expected frequency was less than 5 [bib_ref] Biostatistics 103: qualitative data -tests of independence, Chan [/bib_ref]. The receiver operator curve (ROC) was constructed and an area under curve analysis was performed to compare the margin status between TB and FS. A p-value of less than 0.05 was considered statistically significant.
# Results
Of the 30 patients enrolled in this study, 19 (63.3%) patients were males, and 11 (36.7%) were females. The mean age was 54.4 ± 12.3 years (23-75).
The oral cavity subsites were tongue (n = 13; 43.3%), lip (n = 4; 13.3%), buccal mucosa (n = 4; 13.3%), retromolar area (n = 4; 13.3%), lower alveolar ridge (n = 3; 10.0%), and floor of the mouth (n = 2; 6.7%). Regarding the T category of the lesions, according to the seventh edition of the staging of head and neck cancer by the American Joint Committee of Cancer, 3 (10.0%) were T1, 17 (56.7%) were T2, 3 (10.0%) were T3, and 7 (23.3%) were T4.
A total of 140 mucosal margins were analyzed intraoperatively by TB and FS. The results were compared with the final histopathological results of the same margins. Of the 140 examined mucosal surgical margins, there were six (4.3%) positive margins with invasive carcinoma in the final histopathological examination, all these margins were stained positively by TB, and only three (50.0%) of these margins were detected by FS.
For TB, 14/140 (10.0%) margins (14 margins from 11 patients; in three patients, there were two positive margins a Topical application of 1% TB to the mucosal margins of resected retromolar SCC, b another Segmental mandibulectomy specimen (after eliminating the excess of stain); red arrows indicated positive TB staining, green arrow indicated negative TB staining per patient) were positively stained. However, among these 14 margins, only six (42.9%) were true-positives as confirmed by the final histopathology of the same margins, and eight (57.1%) margins were false-positive [fig_ref] Table 1: The results of TB and FS compared with final histopathology results [/fig_ref]. There were no false-negatives as confirmed by the final histopathology of the same margins, indicating that TB was able to identify all true-positive margins. Thus, the sensitivity was 100.0% (95% CI: 54.1-100.0%). Eight margins (8/140, 5.7%) were false-positive, as confirmed on the final histopathology, indicating specificity of 94.0% (95% CI: 88.6-97.4%). The diagnostic accuracy of TB was 94.3% (95% CI: 89.0-97.5%) with a PPV of 42.9% (95% CI: 27.7-59.5%) and NPV of 100.0% [fig_ref] Table 2: Accuracy of TB and FSAbbreviations [/fig_ref]. The sensitivity and NPV remained unchanged in all T categories. In contrast, specificity and PPV were decreased in T3 and T4 tumors [fig_ref] Table 3: Sensitivity, specificity, reliability [/fig_ref].
For the FS, there were three true-positives [fig_ref] Figure 2: Microscopic photo of FS showing [/fig_ref] , 134 true-negatives [fig_ref] Figure 2: Microscopic photo of FS showing [/fig_ref] , three false-negative margins, and no false-positive margins, as confirmed in the final histopathology of the same margins [fig_ref] Figure 3: Microscopic photo of final histopathology showing [/fig_ref] , [fig_ref] Table 1: The results of TB and FS compared with final histopathology results [/fig_ref]. The sensitivity of the test was 50.0% (95% CI: 11.8-88.2%) and specificity 100.0% (95% CI: 97.3-100.0%). PPV was 100.0% and NPV was 97.8% (95% CI: 95.2-99.0%). The diagnostic accuracy of FS was 97.9% (95% CI: 93.9-99.6%) [fig_ref] Table 2: Accuracy of TB and FSAbbreviations [/fig_ref]. The specificity remains unchanged in all T categories, while NPV decreases with T4 [fig_ref] Table 3: Sensitivity, specificity, reliability [/fig_ref]. Regarding the ROC [fig_ref] Figure 4: Receiver operator curve showing area under curve for FS =75 [/fig_ref] analysis for positive margins, we found that the area under the curve for FS was 75.0% (95% CI 49.0-100.0%) with a sensitivity of 50.0% and a specificity of 100.0% (p-value = 0.039). The area under the curve for TB was 97.0% (95%CI 94.2%-99.8%) with a sensitivity of 100.0% and specificity of 94.0% (p < 0.001).
# Discussion
TB is a metachromatic and affordable stain with a high affinity for DNA and RNA. High DNA and RNA content in actively growing tissue-such as malignant proliferation and wider intercellular canals compared to normal epithelial cells-is responsible for staining malignant lesions [bib_ref] VELscope versus toluidine blue for detection of dysplastic changes in oral keratotic..., Belal [/bib_ref]. Since the 1960s, TB has been used in vivo as a screening tool. It has demonstrated its ability to generate malignant and premalignant cell stains, but not normal mucosa [bib_ref] In vivo staining properties of oral cancer, Shedd [/bib_ref].
The ability of TB to detect oral and oropharyngeal carcinoma is well documented. The sensitivity and specificity rates for use of TB as a screening tool for oral premalignant and malignant lesions ranged from 77.0-100.0% and 65.5-100.0%, respectively [bib_ref] The toluidine blue test in lesions of the oral cavity, Myers [/bib_ref] [bib_ref] Sensitivity and specificity of OraScan (R) toluidine blue mouthrinse in the detection..., Warnakulasuriya [/bib_ref] [bib_ref] Utility of toluidine blue test in accessing and detecting intraoral malignancies, Singh [/bib_ref] [bib_ref] Toluidine blue and Lugol's iodine application in the assessment of oral malignant..., Epstein [/bib_ref]. Onofre et al.and Warnakulasuriya et al. [bib_ref] Sensitivity and specificity of OraScan (R) toluidine blue mouthrinse in the detection..., Warnakulasuriya [/bib_ref] reported that TB was 100.0% sensitive for detecting oral carcinoma without false-negative results. Epstein et al. [bib_ref] The utility of tolonium chloride rinse in the diagnosis of recurrent or..., Epstein [/bib_ref] reported that TB was 96.7% sensitive for detecting recurrence or second primary cancers in patients previously treated for UADT malignancies.
TB can more accurately assess the superficial extent of a lesion before its excision, thus adding confidence in tumor excision within safe surgical margins. It can also be used to assess margin status after tumor resection [bib_ref] Biologic endoscopy': optimization of upper aerodigestive tract cancer evaluation, Piazza [/bib_ref]. However, only two studies addressed the use of TB for assessing tumor bed margins after excision of primary SCC [bib_ref] A comparative analysis of toluidine blue with frozen section in oral squamous..., Junaid [/bib_ref] [bib_ref] The role of toluidine blue in assessing margin status after resection of..., Portugal [/bib_ref] , and its use in assessing specimendriven margins has not been studied.
We examined use of TB as a triage-screening tool for detecting positive mucosal margins of the main OSCC specimen (specimen-driven margin), thereby guiding FS sampling. We compared the results of TB to FS (standard of care) and final histopathology (gold standard) of the same margins.
TB had a sensitivity of 100.0% and a specificity of 94.0% with a NPV of 100.0% and a PPV of 42.9%. Our results are in general agreement with previous studies [bib_ref] A comparative analysis of toluidine blue with frozen section in oral squamous..., Junaid [/bib_ref] [bib_ref] The role of toluidine blue in assessing margin status after resection of..., Portugal [/bib_ref] regarding the ability of TB to identify all positive mucosal tumor margins with no false-negatives and a slightly lower specificity.
In this study, 14 margins were stained positively with TB. Six were true-positive, as confirmed by the final histopathological examination of the same margins. The eight false-positive margins slightly reduced the TB specificity and our results were lower than the findings of Junaid et al. [bib_ref] A comparative analysis of toluidine blue with frozen section in oral squamous..., Junaid [/bib_ref] who reported a TB specificity of 97.0%. Six of these eight false-positive margins were attributed to surgical traumatic handling of the mucosa. This led to the exposure of submucosal connective tissue which facilitated TB uptake by the nuclear material of the injured exposed cells. In the remaining two false-positive margins, TB staining was caused by extensive inflammatory cell foci. These were observed during the FS examination and in the final histopathological examination of the same margins. At the site of inflammation, there is an increase in both cellular activity and mechanical retention, which facilitating TB retention [bib_ref] The use of toluidine blue in the detection of premalignant and malignant..., Cancela-Rodriguez [/bib_ref].
Although our PPV was high (42.9%) compared to the 27.2% in the study by Junaid et al. [bib_ref] A comparative analysis of toluidine blue with frozen section in oral squamous..., Junaid [/bib_ref] , it is still suboptimal. PPV is inversely proportional to the number of false-positive margins, which were most often related to the traumatic handling of the mucosa during the resection. This indicates that careful tissue handling during excision may reduce false-positive results and increases the PPV. However, TB staining had NPV of 100.0%, indicating that it identified all positive mucosal tumor margins. When considering the T category of the tumor, the false-positive margins were two in T2, one in T3, and five in T4. Similar to Junaid's results [bib_ref] A comparative analysis of toluidine blue with frozen section in oral squamous..., Junaid [/bib_ref] , most of the false-positives were in the T4. All five false-positive T4 stains were attributed to mucosal trauma sustained during the intraoperative handling of large-size tumors. This exposed the submucosal connective tissue and facilitated retention of the TB.
In his study, Portugal et al. [bib_ref] The role of toluidine blue in assessing margin status after resection of..., Portugal [/bib_ref] were reported that TB staining of the surrounding unresected mucosa identified three cases of a second primary tumor that were not previously identified. We cannot comment on identifying second primary tumors as this was not detected in our study.
FS identified three of six positive margins, indicating a sensitivity of 50.0%. These results are close to those of Mair et al.who reported that FS had a sensitivity of 45.4%. Our findings were inconsistent with some previous studies that reported FS sensitivity of 72.0-100.0% [bib_ref] A comparative analysis of toluidine blue with frozen section in oral squamous..., Junaid [/bib_ref] [bib_ref] Frozen section evaluation of margin status in primary squamous cell carcinomas of..., Layfield [/bib_ref] [bib_ref] Accuracy of frozen sections in oral cancer resections, an experience of a..., Abbas [/bib_ref] [bib_ref] Refining the utility and role of frozen section in head and neck..., Du [/bib_ref]. The three false negatives that reduced the sensitivity of FS in our study were mainly attributed to interpretation errors rather than sampling errors (because all margins that were positive by TB were included in the FS sampling). The interpretation errors occurred as a result of inherent artifacts caused by ice crystal formation, or by distortion of the tissue architecture during the freezing process. These artifacts included darkened nuclei and nuclei angulation changes. These factors make it more difficult to recognize malignant cells and thus complicates the interpretation of the FS.
All positive FS margins were also positive in the final histopathological examination of the same margins with no false-positives, indicating a 100.0% specificity. Our results were similar to those of Junaid et al. [bib_ref] A comparative analysis of toluidine blue with frozen section in oral squamous..., Junaid [/bib_ref] who found that FS had a specificity of 100.0%. When comparing margin discrepancies based on the T category, we found two false-negative margins in T4 and one falsenegative in T2. Notably, our study had a small sample size and it was not possible to determine if the intraoperative false-negative results of the FS specimens and T category depend on each other. This would require comparing the FS results within only one T category.
A triage test is used to rule out disease (not to rule in disease), and therefore needs very high sensitivity [bib_ref] STARD 2015 guidelines for reporting diagnostic accuracy studies: explanation and elaboration, Cohen [/bib_ref]. The result of this study showed that TB is highly sensitive and can be used as a triage-screening tool for positive OSCC mucosal margins. Moreover, TB can reduce the time and cost of FS. The number of FS biopsies can be limited to the positive stained mucosal margins plus a deep (non-mucosal) margin.
Although TB has a high sensitivity for positive OSCC mucosal margins, this test is not without limitations. TB is less-specific than FS. Also, resected tumor margin scars that result from previous chemotherapy, radiotherapy, or surgery may cause TB retention, potentially producing false-positive results. Also, TB can only be used to assess mucosal margins and has no diagnostic value for deep (non-mucosal) margins of resected specimens since it is readily absorbed by connective tissue.
# Conclusion
TB is less-specific but more sensitive than FS for detecting positive mucosal margins of resected OSCC. Screening the tumor mucosal margins with TB before FS sampling may help identify more tumor-bearing margins.
[fig] Figure 2: Microscopic photo of FS showing; a invasive squamous epithelial cells within connective tissue stroma (positive margin H&E ×100). b epithelial surface and connective tissue stroma (negative margin with inflammation, H&E ×200) [/fig]
[fig] Figure 3: Microscopic photo of final histopathology showing; a invasive squamous epithelial cells within connective tissue stroma (positive margin, H&E × 200). b normal epithelial surface and connective tissue stroma (negative margin, H&E × 200) [/fig]
[fig] Figure 4: Receiver operator curve showing area under curve for FS =75.0% (95%CI 49.0-100.0%) with sensitivity = 50.0% and specificity = 100.0%, p value = 0.039. Area under curve for TB =97% (95%CI 94.2-99.8%) with sensitivity = 100.0% and specificity = 94.0%, p-value< 0.001. Green and blue lines represent FS and TB respectively [/fig]
[table] Table 1: The results of TB and FS compared with final histopathology results [/table]
[table] Table 2: Accuracy of TB and FSAbbreviations: CI Confidence Interval, FS frozen section, NPV negative predictive value, PPV positive predictive value, TB toluidine blue [/table]
[table] Table 3: Sensitivity, specificity, reliability (PPV and NPV) and accuracy of TB and FS for all T categories [/table]
|
Metal-Dependent DNA Recognition and Cell Internalization of Designed, Basic Peptides
A fragment of the DNA basic region (br) of the GCN4 bZIP transcription factor has been modified to include two His residues at designed i and i+4 positions of its N-terminus. The resulting monomeric peptide (brHis 2 ) does not bind to its consensus target DNA site (5′-GTCAT-3′). However, addition of Pd(en)Cl 2 (en, ethylenediamine) promotes a high-affinity interaction with exquisite selectivity for this sequence. The peptide−DNA complex is disassembled by addition of a slight excess of a palladium chelator, and the interaction can be reversibly switched multiple times by playing with controlled amounts of either the metal complex or the chelator. Importantly, while the peptide brHis 2 fails to translocate across cell membranes on its own, addition of the palladium reagent induces an efficient cell internalization of this peptide. In short, we report (1) a designed, short peptide that displays highly selective, major groove DNA binding, (2) a reversible, metal-dependent DNA interaction, and (3) a metal-promoted cell internalization of this basic peptide.
# ■ introduction
Protein expression depends on the concerted action of transcription factors (TFs), which are specialized proteins that specifically bind to regulatory DNA sequences, thereby modulating the transcription of downstream genes. Over the past few years, there has been a great interest in the development of miniaturized models of TFs capable of reproducing their DNA-binding properties, as they might open interesting biomedical opportunities. 2 These peptides interact with the DNA mainly through the major groove and, thus, might provide a compelling alternative to classic DNA binding agents that interact by intercalation or by insertion into the minor groove. 3 However, the design and preparation of this type of minimalistic peptides has proven extremely challenging. 2, [bib_ref] DOI: 10.1002/ ejoc, Boga [/bib_ref] The most successful examples have been inspired by the DNA binding domain of bZIP proteins such as GCN4, an archetypal member of this family of TFs that specifically binds to palindromic ATF/CREB (5′-ATGAC-GTCAT-3′) or AP1 (5′-ATGA(c)TCAT-3′) sites as a leucine zipper-mediated dimer of uninterrupted α-helices. The DNA interaction occurs through the N-terminal basic region (br), which folds into an αhelix upon insertion in the major groove of the target DNA site.Following the seminal work by Kim and co-workers in 1990,a number of groups have demonstrated that the complete leucine zipper of the GCN4 DNA binding domain could be replaced by artificial dimerizing elements without significantly compromising the DNA recognition properties of the system.In contrast, monomeric GCN4 peptides display very low DNA binding affinity,probably because the high entropic cost associated with the folding of the peptide chain into the αhelical conformation, required for DNA binding, is not compensated by the enthalpic gain of a monomeric interaction. 10,11 A couple of reports have shown that introducing lactam bridges or covalent staples can promote DNA binding, although the affinity and selectivity of these modified systems are rather modest.Better results have been obtained by grafting key DNA binding residues from the bZIP basic regions into stable peptide scaffolds that support the αhelical conformation. In an alternative approach, our group has demonstrated that tethering the peptides to prosthetic minor groove binders that establish accessory stabilizing interactions can also lead to stable DNA complexes. 14 Furthermore, we have shown that such tethering can also be achieved through a nickel-mediated coordination with bisbenzamidine−bipyridine conjugates. 15 Tethering metalbased intercalators to peptides has also allowed cooperative binding to DNA. 16 Despite these important advances, an efficient and selective major groove DNA interaction by simple, minimalist peptides, featuring natural amino acids, has not been demonstrated.
Herein we show that a designed bis-histidine grafted peptide featuring 23 amino acids of the basic region of GCN4 binds its consensus DNA site (5′-GTCAT-3′) with high affinity and specificity when treated with a slight excess of [Pd(en)Cl 2 . The palladium complex works as a stapling agent that favors the required α-helical folding and, ultimately, the insertion of the peptide into the major groove of the target DNA [fig_ref] Figure 1: Top [/fig_ref].
Noticeably, the interaction can be reversibly switched by controlling the relative amounts of the palladium complex and a metal sequestering agent. Overall, we report the shortest, highaffinity, sequence-selective DNA binding natural peptide described so far, the first on−off, cyclically switchable peptide DNA binder, and the use of a palladium complex to control a selective DNA interaction.
Importantly, we have also found that the palladium clip triggers the cellular uptake of otherwise nonpenetrating peptides. To the best of our knowledge, this represents the first demonstration of a metal-promoted cell internalization process.
■ RESULTS AND DISCUSSION Design and Synthesis. As reference for our design we used a fragment of the GCN4 basic region comprising residues Asp 226 to Arg 249 . 5,17 Inspection of the X-ray structure of the GCN4/DNA complex revealed that residues Leu 230 and Arg 234 are oriented toward the solvent on the outer face of the α-helix, and their substitution by His residues should not affect the DNA contacting surface of the peptide. 3a,18 Thus, we synthesized the peptide brHis 2 , featuring two His residues, as well as the natural basic region peptide (br) and the control peptide brHis, in which only one residue is replaced by His (Arg 234 → His).
DNA Binding of the Peptide brHis 2 . The DNA binding properties were studied by electrophoretic mobility assays (EMSA) in polyacrylamide gel under nondenaturing conditions, using SybrGold as the DNA stain. As expected, incubation of the peptide brHis 2 with the double-stranded (ds) oligonucleotide GTCAT containing the consensus binding site . To our knowledge, reversible cyclic DNA binding of such minimalist synthetic peptides lacks any precedent. [bib_ref] There is an earlier report of a Co(II)-responsive GCN4 dimer featuring iminodiacetic..., Mosquera [/bib_ref] The metal-promoted interaction is highly selective; therefore, mutation in a single position of the target site was enough to completely abolish the DNA binding [fig_ref] Figure 2: DNA binding properties of brHis 2 [/fig_ref] and [fig_ref] Figure 1: Top [/fig_ref]. It is also important to note that neither the natural GCN4 basic region (br) nor the peptide containing a single His mutation (brHis) gives rise to new bands in the presence of palladium [fig_ref] Figure 1: Top [/fig_ref] , which confirms the requirement of the two coordinating histidines.
The above data are consistent with the proposed DNA binding by a peptide monomer. Binding by a dimeric species is unlikely, as it would require a specific arrangement of two basic regions that are difficult to envision. Nevertheless, to fully discard the presence of such dimeric species binding to the DNA target, we carried out control experiments with a disulfide dimer of the basic region peptide [br 2 (SS)], similar to that used by Kim et al. As shown in [fig_ref] Figure 3: Comparative EMSA experiment with monomeric brHis 2 and br 2 [/fig_ref] , the interaction of this dipeptide with a DNA featuring either its consensus dimeric target site (AP-1) or the monomeric half site (GTCAT) leads to slower migrating bands than those observed for the DNA complex with [(brHis 2 )Pd(en)] 2+ . The retarded band observed in lane 8 with AP1, in the presence of brHis 2 and the Pd reagent, is consistent with the insertion of two monomeric peptides in their respective half-sites. ESI/MS analysis of the peptide-Pd(en) 2 complexes [fig_ref] Figure 1: Top [/fig_ref] and UV/vis experiments with Pd(bipy)Cl 2 in the presence of the target DNA [fig_ref] Figure 1: Top [/fig_ref] also support the formation of DNA complexes with monomeric peptides (see the Supporting Information).
DNA Binding Is Only Observed in the Presence of Specific Metal Complexes. At this point we checked whether other metal ions could also promote DNA binding, or if the induction of the DNA interaction was exclusive for the palladium reagent. As shown in [fig_ref] Figure 4: DNA properties of brHis 2 with different metal salts [/fig_ref] , mixing the target dsDNA GTCAT with brHis 2 in the presence of several Ni 2+ , Zn 2+ , Cu 2+ , Co 2+ , or Fe 2+ salts failed to give rise to new bands in the EMSA [fig_ref] Figure 4: DNA properties of brHis 2 with different metal salts [/fig_ref]. It is important to highlight the failure of NiCl 2 , as this metal complex had been successfully used for promoting bivalent major−minor groove interactions in cooperation with bisbenzamidine−bipyridine conjugates. However, it is not useful in the case of the more challenging monovalent peptide−DNA interactions.
Not surprisingly, trans-[Pd(PPh 3 ) 2 Cl 2 ] also failed to induce DNA binding, while Pd(bipy)Cl 2 was roughly as effective as Pd(en)Cl 2 [fig_ref] Figure 1: Top [/fig_ref]. We also proved that an analogous complex featuring nitrate instead of chloride ligands Pd(bipy)-(NO 3 ) 2 is also a very efficient binding trigger [fig_ref] Figure 1: Top [/fig_ref]. Therefore, it appears that square planar cis-Pd(II) complexes present the ideal coordination properties to promote an effective folding, and interaction to the target DNA.Notably, similar PtCl 2 complexes were ineffective, most probably because of the low reactivity and kinetic stability of the precursor salts [fig_ref] Figure 4: DNA properties of brHis 2 with different metal salts [/fig_ref]. However, electrophilic Au(III) complexes, particularly [Au(bipy)Cl 2 ] + , which exhibit a square planar geometry similar to that of Pd(II), can also promote the DNA binding [fig_ref] Figure 4: DNA properties of brHis 2 with different metal salts [/fig_ref]. This result is quite appealing as, to the best of our knowledge, gold complexes had never been used in related bio-supramolecular strategies.
Spectroscopic Characterization of the Palladium-Promoted Interaction between brHis 2 and the DNA. In order to obtain more information on the DNA binding process, we performed fluorescence anisotropy titrations by adding brHis 2 to a solution containing 20 μM cis-[Pd(en)Cl 2 ] and a tetramethylrhodamine (TMR)-labeled ds-oligonucleotide containing the target sequence. The addition of brHis 2 led to a progressive increase in the fluorescence anisotropy of the TMR label. The data could be fitted to a simplified 1:1 binding model by formally considering the complex [(brHis 2 )(Pd(en))] 2+ as a single DNA binding unit, with an apparent K D ≈ 24 ± 17 nM [fig_ref] Figure 5: Left [/fig_ref].
As expected for a poorly structured peptide, the circular dichroism (CD) spectrum of brHis 2 presents a relatively weak negative signal at 222 nm, even in the presence of the consensus DNA GTCAT [fig_ref] Figure 5: Left [/fig_ref]. Addition of Pd(en)Cl 2 to the mixture promoted a significant increase in the negative ellipticity at 222 nm, which is consistent with the folding of the peptide into an α-helix [fig_ref] Figure 5: Left [/fig_ref] right, thick solid line).In accordance with the results obtained by EMSA, addition of DEDTC to the mixture resulted in a drastic decrease in the helicity of the peptide [fig_ref] Figure 5: Left [/fig_ref] right, narrow solid line), which correlates with the disruption of the DNA complex. Not surprisingly, NiCl 2 or Pt(en)Cl 2 complexes did not promote the peptide α-helical folding in the presence of the target DNA [fig_ref] Figure 1: Top [/fig_ref]. Interestingly, adding Pd(en)Cl 2 to brHis 2 , in the absence of the target DNA, gives rise only to a relatively small increase in helicity [fig_ref] Figure 1: Top [/fig_ref]. This result raised the question of whether the palladium is coordinating the peptide in the absence of the DNA template. UV−vis spectra and HPLC-MS analysis of the mixture of the peptide and the palladium complex were fully consistent with the formation of , in 10 mM phosphate buffer pH 7.5 and 100 mM of NaCl. The contribution of DNA to the CD spectrum has been subtracted for clarity. Mean residue molar ellipticity (MRE) was calculated with respect to the 24-mer brHis 2 . the expected metal-clipped peptide [fig_ref] Figure 1: Top [/fig_ref]. Furthermore, fluorescence anisotropy experiments carried out by adding Pd(en)Cl 2 to a solution of a TMR-labeled brHis 2 resulted in a progressive decrease in the anisotropy of the TMR fluorophore at 559 nm, which agrees with the peptide acquiring a more compact conformation [fig_ref] Figure 1: Top [/fig_ref].
The above data indicate that the Pd(II) complex does coordinate the histidine residues, but the peptide only acquires a high proportion of helicity after addition of the target DNA. The palladium clip is not inducing a permanent helical conformation as traditional helix staples, 10,24 but rather it enhances the local helical propensity, which thus facilitates the subsequent folding in the presence of the DNA. 25,10,24 Therefore, the system maintains the folding-upon-binding behavior of the parent GCN4 transcription factor, which might be the reason that the [(brHis 2 )Pd(en)] 2+ complex shows such an excellent DNA selectivity. Internalization Studies: brHis 2 Only Translocates in the Presence of Pd(II). A key factor when considering potential biological applications of this type of peptides has to do with their ability to cross cell membranes. It is known that basic peptides derived from transcription factors, such as the Antennapedia, can be efficiently internalized. Moreover, previous studies with covalently stapled peptides suggest that the conformational restriction associated with the stapling favors the cell uptake, 12,28 and therefore we were curious to know if the palladium clipping might also influence the cell penetration ability of brHis 2 .
Incubation of mammalian Vero or HeLa cells with TMR-brHis 2 and 2-fold washing with phosphate-buffered saline (PBS) revealed that despite the basic character of the peptide, it shows a rather poor internalization. 29 Gratifyingly, and also quite surprisingly, addition of 1 equiv of Pd(en)Cl 2 led to the appearance of bright intracellular emission, which was mainly localized in endosomes [fig_ref] Figure 1: Top [/fig_ref].Importantly, neither the TMR-labeled basic region peptide (TMR-br) nor the single mutated basic region (TMR-brHis) showed appreciable cell translocations upon incubation using similar conditions [fig_ref] Figure 2: DNA binding properties of brHis 2 [/fig_ref].
While obtaining exact information on the molecular mechanism of the cell internalization process will require a detailed study, we have carried out preliminary experiments with modified peptides that shed light on the structural requirements of the peptide for an effective metal-switched uptake. Thus, we found that the positively charged TMR fluorophore is not required for the metal-induced internalization, so that when this fluorophore was replaced by the anionic fluorescein [fig_ref] Figure 3: Comparative EMSA experiment with monomeric brHis 2 and br 2 [/fig_ref] , FITC-brHis 2 ), the resulting peptide replicates the palladium-induced internalization displayed by the parent TMR-brHis 2 . Similarly, the metal switch is perfectly operative in specifically modified peptides, such as the double mutant peptides Leu 247 → Ala, Gln 248 → Ala or Thr 236 → Ala, Ala 238 → Thr, as well as the single mutant Ala 238 → Thr. Importantly, the cell entrance of longer peptides that include amino acid insertions, such as one containing a Leu-Ser fragment between Ala 229 and His 230 , can be also regulated by the Pd(II) reagent. Not surprisingly, the cell uptake is sensitive to the presence of arginine residues, which is in consonance with the well-known role of these amino acids in cellpenetrating peptides.Thus, the double mutant Arg 245 → Ser; Arg 249 → Ala peptide failed to translocate across the plasmatic membrane, even in the presence of the palladium clip [fig_ref] Figure 3: Comparative EMSA experiment with monomeric brHis 2 and br 2 [/fig_ref].
Although unraveling the exact structural requirements for the cell uptake switch requires further work, these early results indicate that the internalization requires the metal clip and the presence of several arginine residues in the sequence, but is highly tolerant to other changes. The underlying internalization mechanism is also unknown, but it is likely that the clipping effect of the palladium promotes a change in the amphiphilicity of the system, which favors the membrane crossing. Our peptides will require further engineering to promote endosomal escape; however the above results represent the first examples of a metal-triggered cell internalization of designed basic peptides and should foster further research in this area. 32
# ■ conclusion
In summary, we have introduced a new approach for achieving a highly selective recognition of dsDNA with a minimal (23 amino acids long), synthetic peptide that is entirely made of natural residues. Key for the success of the approach is the introduction of histidines in suitable i, i+4 positions in the sequence and the addition of square-planar, Pd(II) or Au(III) reagents. In contrast with other covalently stapled α-helical peptides, our palladium−peptide complex is mostly unstructured in solution and acquires a predominantly helical conformation only in the presence of the target DNA sequence. The kinetic lability of the metal coordination facilitates the disassembly of the supramolecular structure upon addition of external agents that sequester the palladium.
Importantly, the DNA binding and disassembly processes can be repeated several times, therefore providing for a fully reversible switchable system. Also importantly, we show that while the apo-peptide presents a very poor cellular uptake, the . Fluorescence micrographies of HeLa cells. Bright-field images are superimposed to the red emission channel after incubation with 5 μM TMR-brHis 2 for 30 min at 37°C (left) and the same experiment in the presence of 5 μM Pd(en)Cl 2 (right). The complex was preincubated (1:1) in water for 10 min before the addition. The cells were washed twice with PBS before being observed in a fluorescence microscope. All incubations were made in Dulbecco's modified Eagle medium completed with 10% fetal bovine serum. TMR: tetramethylrhodamine. . Sequences of modified peptides tested in the metal-switched cell uptake process, highlighting the mutated residues (blue) and whether the peptide is internalized (green dot) or does not show an observable translocation under the same conditions (red dot). addition of the palladium reagent triggers an efficient internalization. This is the first demonstration of triggering the cell penetration of a short, basic peptide using a metal complex; this tactic promises to find further applications in dynamic cellular delivery.
## ■ associated content
## * s supporting information
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/jacs.7b07422.
Peptide synthesis and characterization; detailed protocols for EMSA experiments, fluorescence anisotropy, circular dichroism parameters, and control experiments; cell internalization and cytotoxicity studies (PDF)
[fig] Figure 1: Top: Schematic structure of the supramolecular assembly into specific DNA sites of a modified GCN4 basic region featuring two His residues (His 230 and His 234 , brHis 2 ) in the presence of a metal clip. Bottom: Sequences of the natural GCN4 basic region and the mutated peptides brHis 2 and brHis. Mutated residues are in bold. [/fig]
[fig] Figure 2: DNA binding properties of brHis 2 . (a) Incubation with the target ds-oligonucleotide GTCAT and switching the DNA binding by addition of DEDTC. (b) Incubation with mutated dsDNA MUT. Concentration of the components when present: 75 nM dsDNA, 2 μM brHis 2 , 20 μM Pd(en)Cl 2 . (a) Lane 1: dsDNA; lane 2: dsDNA and brHis 2 ; lane 3: previous mixture and Pd(en)Cl 2; lane 4: mixture in lane 3 after addition of 50 equiv of DEDTC; lane 5: addition of 60 equiv of Pd(en)Cl 2 to the mixture in lane 4; lane 6: mixture in lane 5 after addition of 150 equiv of DEDTC; lane 7: addition of 200 equiv of Pd(en)Cl 2 to the mixture in lane 6. All the equiv are expressed relative to brHis 2 . Samples were resolved on a 10% nondenaturing polyacrylamide gel and 0.5× TBE buffer over 40 min at 25°C and stained with SyBrGold (5 μL in 50 mL of 1× TBE) for 10 min, followed by fluorescence visualization. Oligonucleotide sequences (only one strand shown, brHis 2 binding site in bold): GTCAT: 5′-CGC GTCAT AATTGAGAG CGC-3′; MUT: 5′-CGC GTGAT AATTGAGAG CGC-3′. [/fig]
[fig] Figure 3: Comparative EMSA experiment with monomeric brHis 2 and br 2 (SS). Lanes 1−4: 75 nM GTCAT; lanes 2, 3: 2 μM brHis 2 ; lane 3: 20 μM Pd(en)Cl 2 ; lane 4: 1.2 μM br 2 (SS); lanes 5−8: 75 nM AP-1; lane 6: 1.2 μM br 2 (SS); lanes 7, 8: 2 μM brHis 2 ; lane 8: 20 μM Pd(en)Cl 2 . The dsDNA AP-1 features two binding sites in neighbor sequences. We have used this ds-oligonucleotide: 5′-TGGAG ATGA cg TCAT CTCGT-3′ (only one strand shown). Each peptide of the disulfide br 2 (SS) features the sequence of peptide br(Figure 1), with the C-terminal Arg (249) substituted by Gly-Gly-Cys. [/fig]
[fig] Figure 4: DNA properties of brHis 2 with different metal salts. Lane 1: 75 nM GTCAT; all other lanes: 75 nM GTCAT, 2 μM brHis 2 , 20 μM metal complexes (same conditions as in Figure 2a and b). Note: The gold dichloride complexes present a positive charge. [/fig]
[fig] Figure 5: Left: Representative fluorescence anisotropy titration of a 25 nM solution of the TMR-labeled target ds-oligonucleotide TMR-GTCAT (TMR-5′-CGCGTCATAATTGAGAGCGC-3′, only one strand shown) in the presence of 20 μM Pd(en)Cl 2 and increasing concentrations of brHis 2 . The best fit to a 1:1 binding model is also shown. Measurements were made in 20 mM Tris-HCl buffer pH 7.5 and 100 mM NaCl. Right: CD of a 5 μM solution of brHis 2 (dotted line), after the subsequent addition of 1 equiv of the target dsDNA (dashed line), of the same solution after the addition of cis-Pd(en)Cl 2 (thick solid line), and after addition of DEDTC demonstrating the reversibility (narrow solid line). All experiments were carried out at 25°C [/fig]
|
Good Clinical Teachers Likely to be Specialist Role Models: Results from a Multicenter Cross-Sectional Survey
Context: Medical educational reform includes enhancing role modelling of clinical teachers. This requires faculty being aware of their role model status and performance. We developed the System for Evaluation of Teaching Qualities (SETQ) to generate individualized feedback on previously defined teaching qualities and role model status for faculty in (non) academic hospitals.Objectives: (i) To examine whether teaching qualities of faculty were associated with their being seen as a specialist role model by residents, and (ii) to investigate whether those associations differed across residency years and specialties.Methods & Materials: Cross-sectional questionnaire survey amongst 549 Residents of 36 teaching programs in 15 hospitals in the Netherlands. The main outcome measure was faculty being seen as specialist role models by residents. Statistical analyses included (i) Pearson's correlation coefficients and (ii) multivariable logistic generalized estimating equations to assess the (adjusted) associations between each of five teaching qualities and 'being seen as a role model'.Results: 407 residents completed a total of 4123 evaluations of 662 faculty. All teaching qualities were positively correlated with 'being seen as a role model' with correlation coefficients ranging from 0.49 for 'evaluation of residents' to 0.64 for 'learning climate' (P,0.001). Faculty most likely to be seen as good role models were those rated highly on 'feedback' (odds ratio 2.91, 95% CI: 2.41-3.51), 'a professional attitude towards residents' (OR 2.70, 95% CI: 2.34-3.10) and 'creating a positive learning climate' (OR 2.45, 95% CI: 1.97-3.04). Results did not seem to vary much across residency years. The relative strength of associations between teaching qualities and being seen as a role model were more distinct when comparing specialties.Conclusions: Good clinical educators are more likely to be seen as specialist role models for most residents.
# Introduction
Medical education, and in particular, graduate medical education continues to face many challenges. One of the major challenges is the issue of role modelling by clinician educators in delivering high-quality clinical training to residents on their journey to becoming high-performing physicians. Role modelling is at the heart of building physicians who exhibit the knowledge, attitudes, behaviors and identity of a 'good professional' [bib_ref] Role modelling -making the most of a powerful teaching strategy, Cruess [/bib_ref] [bib_ref] Role modelling in physicians' professional formation: reconsidering an essential but untapped educational..., Kenny [/bib_ref] [bib_ref] Observation, reflection, and reinforcement: surgery faculty members' and residents' perceptions of how..., Park [/bib_ref]. Role models not only help shape our future physicians, they also influence their career choices and predict residents' satisfaction with their training [bib_ref] Career choice influence in Australian anesthesists, Roberts [/bib_ref] [bib_ref] Learning behind the scenes: perceptions and observations of role modelling in pediatric..., Balmer [/bib_ref]. Given the widespread interest for enhancing the effectiveness of medical education the scanty attention for role modelling is remarkable. This evidently has to do with the rather elusive character of role modelling. Learning from role models occurs through observation and reflection [bib_ref] Role modelling -making the most of a powerful teaching strategy, Cruess [/bib_ref] [bib_ref] Learning behind the scenes: perceptions and observations of role modelling in pediatric..., Balmer [/bib_ref] and is not a straight forward process. Rather it is a complex mix of conscious and unconscious learning in which explicit, theoretical knowing and implicit or tacit knowing are being transferred from the clinical teacher -the 'master' -to the trainee [bib_ref] Recognizing tacit knowledge in medical epistemology, Henry [/bib_ref]. Although many clinical teachers are involved in today's residency training residents do not choose to pattern the activities or behaviors of all of them. Not all faculty are perceived by residents as good specialist role models, defined as a person whose skills and behavior residents desire to emulate. In fact various studies report that many faculty -up to more than 50% -are not perceived as role models or are seen as negative role models by residents [bib_ref] Role modelling -making the most of a powerful teaching strategy, Cruess [/bib_ref] [bib_ref] Role modelling in physicians' professional formation: reconsidering an essential but untapped educational..., Kenny [/bib_ref] [bib_ref] Attributes of excellent attending-physician role models, Wright [/bib_ref]. This situation should arouse at least some concern about the quality of medical education. As laid down in the Compact Between Resident Physicians and Their Teachers, a declaration of the fundamental principles of graduate medical education, residents deserve faculty to be role models and faculty commit themselves to act like them. Clearly, in today's attention to educational reform enhancing role modelling as a teaching strategy is appropriate if not necessary [bib_ref] Role modelling in physicians' professional formation: reconsidering an essential but untapped educational..., Kenny [/bib_ref]. Even experienced faculty may have a limited insight in their strengths and weaknesses as teachers, in residents' perception of their behaviour and in the potential for improving their teaching skills and possibly their effectiveness as role models. Improving role modelling requires -at an individual level -that faculty are aware of their role model status and performance, reflect upon their experiences and participate in staff development when deemed necessary [bib_ref] Role modelling -making the most of a powerful teaching strategy, Cruess [/bib_ref] [bib_ref] Role modelling in physicians' professional formation: reconsidering an essential but untapped educational..., Kenny [/bib_ref]. To support this process we developed a system (named System for Evaluation of Teaching Qualities or SETQ) for generating individualized feedback for faculty in academic and other teaching medical institutions [bib_ref] An instrument for the assessment of the training qualities of clinician-educators, Lombarts [/bib_ref] [bib_ref] Development of a system for the evaluation of the teaching qualities of..., Lombarts [/bib_ref]. From the literature and our own conversations with residents we know they do recognize the multiple roles that faculty embody and display -often simultaneously -in performing their daily activities: as teachers and as care givers [bib_ref] Identifying characteristics that students, interns and residents look for in their role..., Elzubeir [/bib_ref] [bib_ref] Residents' perceptions of the ideal teacher -a qualitative study, Boor [/bib_ref] [bib_ref] An alternative approach to defining the role of clinical teacher, Ullian [/bib_ref] [bib_ref] How medical students and residents describe the roles and charateristics of their..., Paukert [/bib_ref] [bib_ref] What makes a good clinical teacher in medicine? A review of the..., Sutkin [/bib_ref]. These roles should be distinguished from the person [bib_ref] Role modelling in physicians' professional formation: reconsidering an essential but untapped educational..., Kenny [/bib_ref] [bib_ref] Residents' perceptions of the ideal teacher -a qualitative study, Boor [/bib_ref] [bib_ref] An alternative approach to defining the role of clinical teacher, Ullian [/bib_ref] [bib_ref] How medical students and residents describe the roles and charateristics of their..., Paukert [/bib_ref]. There is a wealth of literature describing the characteristics of professionals effectively fulfilling these roles [bib_ref] How medical students and residents describe the roles and charateristics of their..., Paukert [/bib_ref] [bib_ref] What makes a good clinical teacher in medicine? A review of the..., Sutkin [/bib_ref] [bib_ref] Excellence in role modelling: insight and perspectives from the pros, Wright [/bib_ref]. What is lacking is empirical evidence demonstrating if and how these roles are related. This study explores the association between faculty being seen as specialist role models and their teaching performance as clinical educators. The first objective of this study was to examine whether the teaching qualities of faculty as evaluated by residents were related to the faculty being considered role models by the residents who themselves are future specialists. The second objective was to investigate if and how any relations between teaching qualities and being seen as role models differed across specialties and residency years.
# Methods
The System for Evaluation of Teaching Qualities (SETQ)
We developed the System for Evaluation of Teaching Qualities (or SETQ) to assess, monitor, feedback and possibly improve the teaching performance of clinician educators or faculty in residency programs in academic medical settings [bib_ref] An instrument for the assessment of the training qualities of clinician-educators, Lombarts [/bib_ref] [bib_ref] Development of a system for the evaluation of the teaching qualities of..., Lombarts [/bib_ref]. The SETQ is an integral and cyclical system of (i) two internet-based specialtyspecific instruments for evaluating faculty's teaching qualitiesone instrument completed by residents and another by faculty themselves, (ii) individualized quantitative and qualitative feedback reporting to each faculty, and (iii) follow-up discussion with the respective program director for individualized maintenance or improvement support. A core purpose of the SETQ study is to assist faculty in increasing their self insight into their teaching. This is aided by residents' evaluations in order to improve faculty's teaching performance within the Dutch graduate medical education system. SETQ was successfully introduced at one academic medical center in the Netherlands and has been adopted by many other residency programs across the Netherlands. The SETQ study aims to provide detailed longitudinal investigation of faculty development in terms of their teaching performance. Over time, the study hopes to report on the personal and contextual determinants of faculty's teaching performance. It will also research the educational and quality of care consequences of the teaching qualities of clinician educators in academic medicine.
The SETQ instruments were initially modeled on the Stanford Faculty Development Program (SFDP26) instrument developed in the United States [bib_ref] Improving clinical teaching. Evaluation of a national dissemination program, Skeff [/bib_ref] [bib_ref] Factorial validation of a widely disseminated educational framework for evaluating clinical teachers, Litzelman [/bib_ref] [bib_ref] Validation of a global measure of faculty's clinical teaching performance, Williams [/bib_ref]. We have described elsewhere the initial development, translation, back-translation, discussion, and specialty-specific adaption of the instruments [bib_ref] An instrument for the assessment of the training qualities of clinician-educators, Lombarts [/bib_ref] [bib_ref] Development of a system for the evaluation of the teaching qualities of..., Lombarts [/bib_ref]. We developed two instruments per specialty: one resident-completed instrument and one faculty self-evaluation instrument. For each specialty, both the residents' and faculty instruments were almost identical except that the resident-completed instrument additionally contained open-ended questions asking the residents to list the core strengths of each faculty they evaluate as well as some improvement points for the faculty. Although the instruments were specialty-specific, they still shared 22 core items aimed at measuring the faculty's performances on: creating conducive learning climate (7 items), their professional attitude toward residents (3 items), communicating learning goals (4 items), evaluating residents' knowledge and skills (4 items), and giving feedback to residents (4 items). Each item had a 5-point response: strongly disagree, disagree, neutral, agree and strongly agree. The 22 items could be grouped into five stable composite-scales for the resident-completed instruments across all specialties. The results of psychometric analysis have been reported elsewhere [bib_ref] An instrument for the assessment of the training qualities of clinician-educators, Lombarts [/bib_ref] [bib_ref] Development of a system for the evaluation of the teaching qualities of..., Lombarts [/bib_ref]. In this study, we used data from the resident-completed instrument.
## Setting and study population
This multisite study was conducted at 36 teaching programs, of which 18 are being offered by a large academic medical center and 18 by 14 other teaching hospitals in the Netherlands. All 549 residents from 18 (sub-)specialties were invited by email to participate in the SETQ and evaluate all 743 faculty. Each resident received personal login details to access the relevant instruments via a dedicated, secure and password protected webbased portal. The formative purpose of the exercise and anonymity of use and eventual faculty feedback and reporting were emphasized. Residents were free to choose which faculty members to evaluate in their respective specialty and could evaluate many faculty members. Under Dutch law this study did not require ethical approval by the institutional review committee.
## Outcome: considered role models for residents as future specialists
The study outcome was each resident's response to the singular global evaluation question ''…this faculty is a role model for me as a future [specialist]''. Just like the core items on the instruments, the response here was measured on 5-point scale ranging from strongly disagree to strongly agree. This outcome was dichotomized as follows: 0 being the reference for responses ''strongly disagree or disagree or neutral'' versus 1 for ''agree or strongly agree''.
## Main predictors: teaching qualities of faculty
The main predictors consisted of the 5 composite-scales of teaching qualities of faculty, namely learning climate, professional attitude toward residents, communication of goals, evaluation of residents' knowledge and skills, and feedback. The teaching quality 'learning climate' refers to an environment in which self directed learning of residents is promoted, faculty adequately manage teaching encounters and are well prepared for teaching activities. Displaying a 'professional attitude towards residents' encompasses establishing a positive respectful ambiance of the faculty-resident relationship. 'Communication of learning goals' refers to the educator's ability to express expectations and establish and prioritize goals regarding the residents' skills, knowledge and/or attitudes resulting from a teaching interaction. 'Evaluation' describes the assessment of residents' achievements, and lastly, 'feedback' focuses on providing a resident with information for the purpose of improving his or her performance. Each compositescale was the averaged score of the items that loaded primarily on it as shown in [fig_ref] Table 1: Number of Study Participants and Number of Residents' Evaluations [/fig_ref]. This averaging ensured that the value of each composite-scale, like the loading items, ranged from 1 to 5 with 5 capturing the best possible teaching quality on that composite-scale.
## Covariates
Several covariates were used in the analysis. Specialty -with internal medicine as reference -was adjusted for only in models involving the pooled analysis of all specialties. Sex and residency year (in all models except the residency-specific models for second aim of this paper) of the participating residents were the other covariates included in the statistical analysis descried below.
# Statistical analysis
First, we described the study sample using standard descriptive statistics. Second, for all specialties combined per residency year, and all residency years combined per specialty separately, we used Pearson's correlation coefficient to quantify the correlations between each composite-scale of teaching qualities and the outcome variable. These correlations gave a first insight into the unadjusted relationships between teaching qualities and faculty being seen as role models. Additionally, for easy visualization, the associations between each teaching composite-scale and being seen as a role model are displayed as scatterplots. Third, for a more definitive analysis of the first and second study objectives, for all specialties combined and each specialty separately, we used logistic generalized estimating equations to relate all five composite-scales of teaching qualities to the outcome, adjusted for the abovementioned covariates. These regression equations allowed for appropriate adjustment for cross-clustering because, like many healthcare surveys [bib_ref] Psychometric properties of the Dutch version of the Hospital-level Consumer Assessment of..., Arah [/bib_ref] [bib_ref] Made in the USA: the import of American Consumer Assessment of Health..., Delnoij [/bib_ref] [bib_ref] Using multilevel modeling to assess case-mix adjusters in consumer experience surveys in..., Damman [/bib_ref] , evaluation scores from this study were nested within residents and faculty jointly, with some crossing among residents and faculty. That is, some faculty members were evaluated by some or all of the same residents in their residency program. This nested crossing implied that scores of any particular faculty member would correlate better within that faculty than across faculty members but at the same time there would be some correlations among scores given by the same resident to different faculty. Due to smaller sample sizes for some specialties such as neurosurgery, plastic surgery, orthopedic surgery and ophthalmology, we present combined specialty results grouping medical ones separately from the surgical specialties. To see how the associations varied across residency years, the analysis was repeated for all specialties stratified by residency years ranging from the first year to the sixth year. Interns (registered physicians who have not entered residency training yet) were classified separately. We note that not all specialties have six-year residencies in the Netherlands. Estimated odds ratios and their 95% confidence limits were reported.
In sensitivity analysis, we analyzed the outcome and main predictors as continuous scales and also log-transformed them to check the impact of choice of functional form and variable definition on our results. Further sensitivity analysis looked at the impact of different cut-points used in dichotomizing the outcome, for instance. Since there were no material differences in the findings of the different approaches, we report the results of the analysis where the outcome was a dichotomy (0 = strongly disagree or disagree or neutral'' versus 1 = ''agree or strongly agree'') and main predictors were assumed to be continuous scales. This
# Results
Of the 549 invited residents 407 ultimately participated in the SETQ study, yielding a response rate of 74.1%. 56,3% of the residents were female. Residents evaluated 662 (89.1%) faculty members by completing a total of 4123 evaluations. There were about 10 evaluations per resident and approximately 6 evaluations per faculty member. These numbers varied between the 19 participating specialties. The mean number of evaluations per resident did not differ significantly between male and female residents nor between residents from different residency years, with the exception of interns who evaluated significantly less (on average 8) faculty members. [fig_ref] Table 1: Number of Study Participants and Number of Residents' Evaluations [/fig_ref] gives an overview of the numbers of participants and the residents' response rates per specialty. [fig_ref] Figure 1: Scatterplots of each of five teaching qualities and being seen as role... [/fig_ref] shows the scatterplots for the association of each composite-scale of teaching skills with faculty being seen as role models. [fig_ref] Table 2: Correlations between Faculty's Teaching Qualities and Their Being Seen as Specialist Role... [/fig_ref] displays the correlations between composite-scales of teaching qualities of faculty and their being seen as role models. For all specialties combined, all five composite scales were correlated with faculty being seen as role models by residents. The correlation coefficients ranged from 0.493 for 'evaluation of residents' toward 0.634 for 'professional attitude toward residents' and 0.637 for 'learning climate' (P,0.001). Among first thru third year residents from all specialties, the highest correlations were observed between 'professional attitude towards residents' and the outcome. For the fourth thru sixth year residents the teaching aspects 'learning climate' and 'feedback' showed the strongest correlations. 'Evaluation of residents' and 'communication of goals' displayed the smallest correlation with faculty being seen as role models. The specialty-specific correlations between each composite-scale of teaching qualities and faculty being seen as role models pointed to 'learning climate', 'professional attitude toward residents' and 'feedback' having the largest correlations with faculty being seen as role models in many specialties such as internal medicine, pediatrics, cardiology, radiology, general surgery and anesthesiology. [fig_ref] Table 3: Adjusted Odds Ratios [/fig_ref] shows the mutually adjusted odds ratios for associations between teaching qualities and faculty being seen as role models for all specialties combined and each specialty separate. [fig_ref] Table 4: Order of the Importance of Teaching Scales in Predicting Faculty Seen as... [/fig_ref] summarizes these results in terms of the ranking of the relative importance of the teaching qualities. 'Learning climate', 'professional attitude toward residents' and 'feedback' had the highest odds ratios in the analysis combining all specialties. The same three teaching qualities were the most important predictors of faculty being seen as role models in internal medicine, anesthesiology, radiology, pediatrics, ENT, general surgery, and obstetrics and gynecology. In cardiology, neurology and general surgery 'evaluation of residents' knowledge and skills' was found to be a more important predictor than ' feedback' (for neurology and surgery) or 'professional attitude' (cardiology). When comparing the combined medical and surgical specialties, 'learning climate' and 'feedback' were the best predicting teaching qualities for the medical specialties (odds ratios being 2.91 and 2.59 respectively) and for the surgical specialties the most important teaching qualities were 'feedback' and 'professional attitude' (odds ratios 3.55 and 2.92 respectively).
The results of the per residency year analysis shown in table 5 and summarized in table 6 indicate that the teaching qualities ranked somewhat differently in their influence on whether residents considered faculty as role models for different residency years. Yet, there were three noticeable patterns across all residency years. The first is that 'feedback' was consistently found to be the most important predictor for all residency years, except the third year. The second is that this was for all years followed by either 'learning climate' (second, fourth, fifth and sixth year residents) or 'professional attitude towards residents' (interns and first year residents). The third is that, consequently, 'evaluation of residents' and 'communication of goals' tended to have the least predictive power with the outcome among most residency years except the third year.
# Discussion
## Summary of main findings
This study provides strong empirical evidence of good clinical teachers being perceived as specialist role models by residents. Faculty most likely to be seen as good specialist role models are those rated highly on the teaching qualities 'giving residents feedback', 'creating a positive learning climate' and 'a professional attitude towards residents'. Residents' views do not seem to vary much across residency years with regards to the relative importance assigned to one or more of these three teaching qualities and they even fully agree on the finding that clinical educators who perform well on the teaching quality 'feedback' are most likely seen as specialist role models. The different relative rankings in teaching qualities are more distinct when comparing specialties.
# Strengths and weaknesses
Findings should be considered in the light of potential study strengths and limitations. This study adds to the literature on role models and clinical teaching skills by moving beyond descriptive to empirical analysis of associations. We note, however, that our findings may be limited to the Netherlands, thus necessitating specific evaluations in other countries and health systems. This study involved faculty form 15 teaching hospitals whereby 55% were attending physicians at one academic medical center, thus potentially limiting the findings to this study population. Furthermore, as in many western health care systems, postgraduate medical education is also being modernized in the Netherlands. Although the introduction of competency-based residency training programs accelerates the development of clinician-educators, residents may not yet be exposed to new teaching requirements like the communication of learning goals. This may account for the higher level of ''don't know responses'' for the items loading on the communication of learning goals scale. Excluding this scale from the teaching qualities did not alter the substantive findings but could have equally introduced a bias by excluding the necessary adjustment for interscale correlations. Hence, we chose to maintain the scale for the reported data analysis in this study. Finally, it could be argued that faculty seen as role model specialists were perceived as better teachers rather than the other way around. Nonetheless, we point out that theoretically the opposite is more plausible since residents were students who were first exposed to the teaching skills of their faculty while their perception of faculty as their role model developed over time. So, even when those residents recall their role models as exhibiting superior teaching skills, our speculation that excellent clinical teachers were seen as role models will still be largely valid. Furthermore, in administering the instruments, residents were first asked to report on the faculty's actual teaching skills before stating the extent to which the evaluated faculty were seen as role models.
## Explanation and interpretation
Role modelling is the teaching method most preferred by residents [bib_ref] Observation, reflection, and reinforcement: surgery faculty members' and residents' perceptions of how..., Park [/bib_ref] [bib_ref] Residents' Perceptions of Professionalism in Training and Practice: Barriers, Promoters, and Duty..., Ratanawongsa [/bib_ref]. In exploring the phenomenon of successful role models research consistently finds that teaching qualities of physicians are of critical value in being seen as a role model by residents [bib_ref] Residents' perceptions of the ideal teacher -a qualitative study, Boor [/bib_ref] [bib_ref] An alternative approach to defining the role of clinical teacher, Ullian [/bib_ref] [bib_ref] How medical students and residents describe the roles and charateristics of their..., Paukert [/bib_ref] [bib_ref] What makes a good clinical teacher in medicine? A review of the..., Sutkin [/bib_ref]. Most studies reach this conclusion based on qualitative research approaches, typically consulting residents to describe their role models and/or faculty to reflect on being a role model [bib_ref] Residents' perceptions of the ideal teacher -a qualitative study, Boor [/bib_ref] [bib_ref] An alternative approach to defining the role of clinical teacher, Ullian [/bib_ref] [bib_ref] How medical students and residents describe the roles and charateristics of their..., Paukert [/bib_ref]. Our study goes further than most studies in providing strong empirical evidence for the previously reported qualitative finding that good clinical educators are more likely to be regarded as specialist role models. We hereto statistically explored the phenomenon of specialist role models in two ways. In the first step we looked at the bivariate relationship between being seen as a specialist role model and each of five teaching qualities defined previously in the context of the SETQ system for evaluating faculty as clinical teachers. Most of the five teaching qualities were positively and significantly related to being seen as a role model, with correlation coefficients falling within the range of 0.40 to 0.75 for all specialties combined and for the various residency years. Some of the specialty specific analyses did not yield strong correlations, i.e. for neurosurgery, ophthalmology, pathology and rehabilitation medicine, most likely due to the small sample sizes. The correlation coefficients reported indicate that the association between the teaching qualities and being a role model is moderate to strong lending some support to our informed assumption that teaching performance of faculty is part of the phenomenon of being a specialist role model. In the second step, we used the technique of linear regression to also describe the relation between teaching performance and being a role model. The listed adjusted odds ratios represent the odds of scoring highly on one or more teaching qualities and the outcome measure of being perceived as a role model. They indicate that faculty most likely to be seen as good specialist role models are those rated highly on the teaching qualities 'giving residents feedback', 'creating a positive learning climate' and 'a professional attitude towards residents'. On the other hand, communication of learning goals' and 'evaluation of residents' are the least distinct predictors for being seen as a specialist role model for residents as future specialists. We may argue that these findings suggest that the latter two skills are more or less restricted to the clinical teaching situation: once in practice there is neither a need nor the required culture to set learning goals or to peer assess each other's knowledge or skills. In contrast, giving feedback, displaying a professional attitude and creating a positive learning climate are all qualities crucial for the high performance of every practicing physician and so need to be transferred from the teaching environment into future practice. Residents in our study seem to be able to distinguish between faculty's teaching qualities relevant for training them now and skills they desire to emulate for future practice. Residents' views do not seem to vary much across residency years with regards to the relative importance assigned to one or more of these three teaching qualities; they agree on the finding that clinical educators who perform well on the teaching quality 'feedback' are most likely seen as specialist role models. Many studies stress that feedback is crucial for delivering effective teaching [bib_ref] AMEE guide no. 27: effective educational and clinical supervision, Kilminster [/bib_ref]. This study now finds that it is also important for being seen as a role model by residents across residency years. The different relative rankings in teaching qualities are more distinct when comparing specialties. Scoring well on 'giving feedback' clearly stands out as a strong predictor for being seen as a role model for surgical residents, which was also reported by Maker et al [bib_ref] Are you a surgical role model?, Maker [/bib_ref]. This may reflect their need for concrete guidance or and their tenacity to analyze feedback -be it confirmative or corrective -to improve their performance. For medical residents 'creating a positive learning climate' is considered the most important predictor for being perceived as a role model. As the core items of this teaching quality are i.e. participation in discussions, bringing up problems and keeping up to date with the literature, this finding could reflect residents' need to be trained in acquiring knowledge and critical clinical reasoning, typically seen as qualities needed as an internal medicine specialist.
## Implications for medical education, research and policy
Explicit strategies are needed to support all faculty to become role models for residents. This study suggests that an effective and more active way to improve role modelling may be to enhance teaching skills. This requires that faculty are aware of their teaching performance. We found that systems such as SETQ can be instrumental in increasing faculty's self-insight by systematically providing them feedback on their teaching qualities and their role model status. Receiving feedback should be followed up bypreferably guided -reflection [bib_ref] Effects of multisource feedback and a feedback facilitator on the influence behavior..., Seifert [/bib_ref] [bib_ref] Improving clinical teaching skills using the parallel process model, Marvel [/bib_ref] [bib_ref] To the point: medical education reviews evaluation in context: assessing learners, teachers,..., Metheny [/bib_ref] and participation in staff development [bib_ref] Role modelling -making the most of a powerful teaching strategy, Cruess [/bib_ref]. Future assessments need to demonstrate actual improvements in teaching and role modelling effectiveness.
# Conclusions
Role modelling is an integral aspect of medical education and residents rightfully deserve good role models. Our study shows that good clinical educators are more likely to be seen as specialist role models for most residents. This opens up opportunities for improving role modelling as a teaching strategy since many of the predictors of being seen as a role model are teaching behaviors and skills that can be acquired.
## Supporting information
[fig] Figure 1: Scatterplots of each of five teaching qualities and being seen as role model specialist. doi:10.1371/journal.pone.0015202.g001 [/fig]
[table] Table 1: Number of Study Participants and Number of Residents' Evaluations. [/table]
[table] Table 2: Correlations between Faculty's Teaching Qualities and Their Being Seen as Specialist Role Models. [/table]
[table] Table 3: Adjusted Odds Ratios (95% Confidence Interval) for Associations between Faculty's Training Qualities and their Being Seen as Specialist Role Models. [/table]
[table] Table 4: Order of the Importance of Teaching Scales in Predicting Faculty Seen as Role Model Specialists; Results per Specialty (Summarizing Results ofTable 3). [/table]
[table] Table 5: Odds Ratios (OR) for the Adjusted Associations Between Teaching Qualities and Faculty Seen as Specialist Role Models across Different Residency Years. For each residency year, there was simultaneous adjustment for all composite-scales, specialty, and resident's sex. doi:10.1371/journal.pone.0015202.t005 [/table]
[table] Table 6: Descending Order of the Importance of Teaching Scales in Predicting Faculty Seen as Role Model Specialists; Results per Year of Training (Summarizing Results ofTable 5). [/table]
[table] Table S1: Psychometric Properties of the Five Composite-Scales of the SETQ. Instruments from 19 Medical and Surgical Specialties. (DOC) [/table]
|
Cervicofacial nocardiosis in children
# Discussion
Primary cutaneous nocardiosis and the nocardial lymphocutaneous syndrome mimicking sporotrichosis have been described in adult patients. 6-~ Among children, however, primary nocardial skin and lymph node infections have been reported infrequently, 3-5 and Nocardia is unappreciated as an etiology of lymphadenitis. Common etiologies for cervical lymphadenitis in children include Staphylococcus aureus, Group A streptococci, and atypical mycobacteria. The presence of a pustule in association with cervical lymphadenitis suggests other potential etiologies as well, including tularemia, plague (Yersinia pestis), cutaneous diphtheria, cat scratch disease, cervicofacial actinomycosis, and sporotrichosis.
Among 347 nocardia isolates from man, 83.3% were N. asteroides, 6.9% were N. brasiliensis, 2.9% were N. curiae and 6.3% were identified as Nocardia sp? N. asteroides isolates were associated with pulmonary, systemic, or primary central nervous system infections, whereas more than 50% of the N. brasiliensis isolates were from skin or soft tissue infection. N. brasiliensis can be associated with mycetomas, abscesses, pustules, and cutaneous abscesses, and over 90% of cutaneous disease with lymphatic involvement of an extremity in adults ("sporotrichoid" lymphocutaneous syndrome ~) has involved this species. Nocardial isolates were speciated by the Houston City Health Department? ~ N. brasiliensis was isolated from two of the three children reported here with the cervicofacial syndrome and from one of the two previously reported cases in the literature, thus supporting the observation that N. brasiliensis is the most common Nocardia sp. involving the skin, especially when lymph node involvement also is present.
A careful surveillance of Nocardia isolates from hospital laboratories and the Houston City Health Department (which serves as a reference laboratory for the city), and of cases referred to the Infectious Disease Services of the various hospitals has been maintained in Houston since 1975. During that time 20 cases of primary cutaneous nocardiosis were identified. Three of these were in children and these were the only cases that involved the face. During the same time period only two children were seen with pulmonary nocardia infection (R. J. Wallace, Jr., M.D., personal communication). A review of cutaneous isolates of Nocardia reported to the State Health Laboratory in Austin, Texas, over a three-year period revealed 24 additional cases of primary cutaneous disease, of which three were in children. Two of these involved the extremities; the third involved the face. As had been noted with the cases from Houston, none of the adult patients had cervicofacial involvement. (Joe Steadham, Ph.D., personal communication).
Nocardia sp are found in the soil worldwide, and no direct man-to-man or animal-to-man transmission has been documented. Although a history of trauma was not found in our patients (as it has been with most cases reported in adults), the isolation ofNocardia sp. from both pustule and regional lymph nodes suggests that the organism was introduced into the skin from the environment and involved the regional lymph nodes via lymphatic spread. The frequent involvement of the nose and cheek as the primary site of involvement in children but not in adults is unexplained. Perhaps the playing habits of children, contamination with soil, and the blood and lymphatic supply of this area contribute to this characteristic clinical picture.
Nocardia sp. grow well aerobically on blood or chocolate agar at 37~ however, it may take from two to five days for colonies to be visible. Gram stain is important since the presence of gram-positive, branching rods should alert the laboratory to hold the plates for one week rather than the standard 48 to 72 hours. Since the Gram stain is often negative, all plates should be held for one week?
Each of our patients received trimethoprim/sulfamethoxazole orally for one to three months and had prompt resolution of fever and lymphadenitis, and gradual improvement in the pustule. Sulfonamides have been the drug of choice for nocardial infections. Recent clinical and laboratory assessment of trimethoprim/sulfamethoxazole indicates that this drug is highly effective for the treatment of nocardiosis; in vitro synergy between the two drugs is usually apparent when ratios of 1:5 trimethoprim/sulfamethoxazole or better are achieved. H Side effects are unusual but one patient developed a reversible leukopenia while receiving therapy.
All three patients were in good health prior to their infection and subsequently have remained well, with no unusual or recurrent infections. All patients had normal white blood cell function as determined by chemiluminescence. Thus, it appears that the development of cervicofacial nocardiosis can occur in normal children. It is not necessary to undertake an exhaustive evaluation of the patient's immunologic status if there is a prompt chnical response to therapy.
Pediatricians should be aware that Nocardia sp. in children may present as a cervicofacial syndrome and cause cervical adenitis. In particular, patients who are not
## Volume 99
Brief clinical and laboratory observations responding to "standard" therapy should be suspected of having a nocardial infection. Gram stain of pustules and aspirates, and incubation of material for culture on blood agar for one week will identify Nocardia sp. Trimethoprim/sulfamethoxazole given orally results in rapid clinical improvement. Prompt diagnosis and treatment will eliminate the need for inappropriate parenteral antibiotics, and should eliminate the need for surgical drainage. THE CELLULAR COMPONENT of chyle contains predominantly lymphocytes of the T-cell type?' 2 The removal of thoracic duct lymph has been shown to reduce cellular immunity in human beings 3,4 and to deplete thymic-dependent lymphoid tissues in calves? To establish if a deficiency in T-lymphocyte numbers occurs in patients with chylothoraces following thoracic surgery, we examined T-lymphocyte numbers in three patients who developed postoperative chylothoraces. transposition of the great vessels, respectively, were studied . Patient 3 required a second surgical procedure for revision of the initial correction eight months after the first surgery. He again developed a postoperative chylothorax. Controls matched for age, surgical procedure, and postoperative time under similar stress as the paitents were also studied.
## Transient t-cell depression in postoperative chylothorax
# Materials and methods
Lymphocyte studies. Peripheral blood. Absolute lymphocyte numbers were determined from the percentage of the total leukocyte count as seen on the complete blood count, using a Coulter electronic cell counter, and differential coatings of Wright-stained smears.Peripheral blood lymphocytes were isolated from venous blood on Ficoll-Hypaque gradients. T-lymphocyte numbers were determined by producing E-rosettes with sheep erythrocytes. 7 B-lymphocyte numbers were determined by incubating the lymphocytes with fluoresceinlabeled anti-Ig and determining the percentage of fluorescent cells, r
Chile. Ten milliliters of chyle drainage were centrifuged at 900 g (200 rpm) for 25 minutes and washed four |
Orchestrated translation specializes dinoflagellate metabolism three times per day
[bib_ref] Phased protein synthesis at several circadian times does not change protein levels..., Markovic [/bib_ref]
## Figure s7: comparison of rna levels and rpf read counts.
A heat map comparing RNA levels (determined from a previous RNA Seq experiment [bib_ref] The Lingulodinium circadian system lacks rhythmic changes in transcript abundance, Roy [/bib_ref] , right) and translation rates (determined from RPF counts, left) shows the large amplitude in translation is not due to changes in RNA levels (A). Both data sets were normalized to their respective means before plotting. The max/min ratio of values (zero values excluded) for translation rates (blue) and RNA levels (red) for 3326 transcripts showing significant differences (B). showing peak translation from each of the three temporal windows (ZT0, ZT12 and ZT16) were compared to translation rates (blue lines, right axis) as determined by RPF counts. RNA level changes are never greater than two-fold. Transcript accession numbers are provided in Methods. One time sample for qPCR did not yield high quality RNA and was not included in subsequent analyses. RPF read counts (orange bars, 3 times; orange lines, 12 times) were superimposed on protein levels previously determined by LS-MS/MS (blue lines) for thirteen proteins previously shown to vary over the daily cycle [bib_ref] Label-free MS/MS analyses of the dinoflagellate Lingulodinium identifies rhythmic proteins facilitating adaptation..., Bowazolo [/bib_ref]. Sequences found to be significantly enriched using Fisher's exact test for the ZT6 samples were placed into GO categories. An arbitrary cutoff of 5 sequences was applied to simplify the presentation.
## Figure s13 fatty acid metabolism.
Heat maps of RPF in the three times (bottom left) and the twelve times (bottom right) experiments show two groups of enzymes. One group, whose synthesis occurs around ZT0, contains only enzymes used for fatty acid metabolism. The other, with peak expression later during the day, contains enzymes with an additional involvement in another pathway, such as amino acid metabolism or fatty acid degradation. Heat maps of RPF in the three times (bottom left) and the twelve times (bottom right) experiments show two groups of enzymes. One group, whose synthesis occurs around midday, contains two enzymes used for starch biosynthesis. The other group, with peak expression later during the day, contains enzymes used for starch degradation and conversion of UDP-glucose to glucose-1-P. Heat maps of RPF in the three times (left) and the twelve times (right) experiments show most proteins have peak expression late during the night. This pattern is consistent with a requirement for forming the cellulosic thecal plates which are constructed during mitosis around ZT1.
## Figure s19 purine metabolism
A schematic view of the enzymes involved in purine biosynthesis shows both unregulated enzymes that are found in the transcriptome (yellow circles) and those whose synthesis is regulated (red circles) (A). The three times experiment (B) contains many more sequences synthesized during the day, while the twelve times experiment (C) contains more transcripts translated at the end of the day.
## Figure s20 pyrimidine metabolism.
A schematic view of the enzymes involved in pyrimidine biosynthesis shows unregulated enzymes in the transcriptome (yellow circles) and transcripts whose synthesis is regulated (red circles) (A). The two experiments (B, C) both contains more transcripts translated at the onset of night.
## Figure s21: pyruvate metabolism
Heat maps for the 3 times (bottom left) and 12 times experiment (bottom right) show two patterns of protein synthesis, one occurring around ZT10 and the other around ZT22. While all enzymes found are involved in pyruvate metabolism, only those uniquely involved in pyruvate metabolism show a peak at ZT10. Enzyme with a peak expression around ZT22 can also be involved in glycolysis, the TCA cycle, glyoxylate metabolism or fatty acid biosynthesis. Note that peak read counts for some sequences in the 2 hr experiments shown in B and D have lower values than in the 3 times experiment and thus appear white when the z-scores are calculated with all data (as shown here) instead of red when z-scores are calculated for the two data sets separately (as in . However, the timing of peak read counts is unaffected by how the zscores are calculated. A schematic view of amino acid biosynthesis (redrawn from KEGG map 01230). Non-regulated enzymes that are found in the transcriptome are shown as yellow circles, while regulated enzymes in the transcriptome are shown as red circles. Amino acids are shown as gray squares and other intermediates as white squares. A similar pattern of regulated enzymes is seen in the 12 times experiment. Schematic views of selected regions of amino acid biosynthesis (as in are shown to illustrate the regulated enzymes (shown by EC number; A,C,E,G) and their read counts (B,D,F,H) for four different pathways. Enzymes numbers in red catalyze potential rate limiting steps.
[fig] Figure S2: Size distribution of Ribosome Protected Fragments (RPF).Sequences were trimmed to remove sequencing adaptors, and the proportion of sequences of different lengths recorded in each of the nine samples. [/fig]
[fig] Figure S3: Comparison of different read lengths. [/fig]
[fig] Figure S4: Similarity of triplicate samples to one another. RPF read counts obtained after mapping to the transcriptome were tested for similarity to all other samples using correlation coefficients (A) and principal components analysis (B). [/fig]
[fig] Figure S5: Read distribution for LBP.Reads corresponding to triplicate LBP samples (top, ZT6; middle, ZT14; bottom, ZT20) assembled with the transcript sequence. At ZT14, when LBP synthesis rates are high, reads are spread out over the length of the coding sequence. Total reads are indicated at right. [/fig]
[fig] Figure S6: Comparison of RPF read counts and 35 S-methionine in vivo labeling. [/fig]
[fig] Figure S8: Comparison of translation rates and mRNA accumulation levels for selected examples. RNA levels determined by RT-qPCR (orange bars, left axis) for two examples of transcripts [/fig]
[fig] Figure S9: Comparison of RPF read counts with diurnal variations in protein levels. [/fig]
[fig] Figure S10: Principal GO categories for enriched RPF at ZT6. [/fig]
[fig] Figure S11: Heat maps of RPF for proteins related to the light reaction of photosynthesis.Heat maps showing combined photosystem proteins, light harvesting proteins, and ferredoxin in the three times (left) and twelve times (right) experiments. [/fig]
[fig] Figure S12: Pentose phosphate shunt. Heat maps of RPF for pentose phosphate enzyme expression show two different patterns in both the three times (bottom left) and the 12 times (bottom right) experiment. The enzyme with peak expression around ZT22 are only involved in the pentose phosphate pathway (red circles in upper and bottom right). The enzymes interconverting fructose-6-P to fructose -1,6-diP (1-phosphotransferase, EC 2.7.1.90) shows peak expression around ZT10. [/fig]
[fig] Figure S14: Starch metabolism. [/fig]
[fig] Figure S15: Cellulose metabolism. [/fig]
[fig] Figure S16: Fructose/Mannose Metabolism. Heat maps of RPF pentose phosphate enzyme expression show two different patterns in both the three times (bottom left) and the 12 times (bottom right) experiment. The enzymes with peak expression around ZT22 are only involved in the pentose phosphate pathway (red circles in upper and bottom right). The enzyme interconverting fructose-6-P to fructose -1,6-diP (PFP, EC 2.7.1.90, red square) has a different pattern of protein synthesis in both experiments (as in figures 3, S9). [/fig]
[fig] Figure S17: Sequences common between experiments 1 and 2 in processes regulated around ZT0. The pathway (A, C) and heat map of RPF (B, D) for regulated sequences found in carbon fixation (A, B) and glycolysis (C, D). Sequence order is the same for experiment 1 (left side) and experiment 2 (right side) in the heat maps (B, D). All data points in each row were used to calculate the row z-score. [/fig]
[fig] Figure S18: Principal GO categories for enriched RPF at ZT14. Legend as forFigure S7. [/fig]
[table] Table S1: Number of reads before and after digestion of an ZT14 sample test sample with two different RNases. [/table]
[table] Table S2: Number of reads before and after trimming (Experiment 1). [/table]
[table] Table S3: Number of reads before and after trimming (Experiment 2). [/table]
[table] Table S4: Average RPF counts for diurnally regulated proteins at three times. [/table]
[table] Table S6: Consensus motifs detected in transcripts whose synthesis is highest near ZT0, at ZT12, or at ZT16. [/table]
|
Structure–function characterization of the mono- and diheme forms of MhuD, a noncanonical heme oxygenase from Mycobacterium tuberculosis
## Edited by ruma banerjee
MhuD is a noncanonical heme oxygenase (HO) from Mycobacterium tuberculosis (Mtb) that catalyzes unique heme degradation chemistry distinct from canonical HOs, generating mycobilin products without releasing carbon monoxide. Its crucial role in the Mtb heme uptake pathway has identified MhuD as an auspicious drug target. MhuD is capable of binding either one or two hemes within a single active site, but only the monoheme form was previously reported to be enzymatically active. Here we employed resonance Raman (rR) spectroscopy to examine several factors proposed to impact the reactivity of mono-and diheme MhuD, including heme ruffling, heme pocket hydrophobicity, and amino acid-heme interactions. We determined that the distal heme in the diheme MhuD active site has negligible effects on both the planarity of the His-coordinated heme macrocycle and the strength of the Fe-N His linkage relative to the monoheme form. Our rR studies using isotopically labeled hemes unveiled unexpected biomolecular dynamics for the process of heme binding that converts MhuD from mono-to diheme form, where the second incoming heme replaces the first as the His75-coordinated heme. Ferrous CO-ligated diheme MhuD was found to exhibit multiple Fe-C-O conformers, one of which contains catalytically predisposed H-bonding interactions with the distal Asn7 residue identical to those in the monoheme form, implying that it is also enzymatically active. This was substantiated by activity assays and MS product analysis that confirmed the diheme form also degrades heme to mycobilins, redefining MhuD's functional paradigm and further expanding our understanding of its role in Mtb physiology.
Degradation of heme is a critically important physiological process in many forms of life catalyzed by heme oxygenases (HOs). Canonical HOs degrade heme to biliverdin, releasing carbon monoxide and free ferrous iron [fig_ref] Figure 1: Structures of heme and its degradation products achieved through different heme oxygenase... [/fig_ref] [bib_ref] Features of the reaction of heme degradation catalyzed by the reconstituted microsomal..., Yoshida [/bib_ref] , and this activity was considered a paradigm for all heme degradation until the recent discovery of some bacterial HOs that catabolize heme to different products. MhuD from Mycobacterium tuberculosis degrades heme to mycobilins, which retain the α-meso carbon as an aldehyde at the ring cleavage site [bib_ref] A new way to degrade heme: The Mycobacterium tuberculosis enzyme MhuD catalyzes..., Nambu [/bib_ref].
IsdG and IsdI from Staphylococcus aureus are other examples of noncanonical HOs that degrade heme to staphylobilins, releasing Fe(II) and formaldehyde instead of CO [bib_ref] The IsdG-family of haem oxygenases degrades haem to a novel chromophore, Reniere [/bib_ref]. The distinct products of canonical and noncanonical HOs suggest different enzymatic mechanisms, presumably involving different reactive intermediates. In pathogens, the main function of HOs is to harvest the iron needed to survive and sustain infections from host heme molecules [bib_ref] Heme utilization by pathogenic bacteria: Not all pathways lead to biliverdin, Wilks [/bib_ref]. The essential role of MhuD in the Mtb heme uptake pathway has identified it as an auspicious target for development of new antitubercular drugs and treatment strategies [bib_ref] Insights on how the Mycobacterium tuberculosis heme uptake pathway can be used..., Owens [/bib_ref].
MhuD and IsdG/I proteins form homodimers of subunits with ferredoxin-like α/β-sandwich folds and mostly hydrophobic distal active sites containing only one polar amino acid residue (asparagine) and no ordered water molecules [bib_ref] Unusual diheme conformation of the heme degrading protein from Mycobacterium tuberculosis, Chim [/bib_ref] [bib_ref] Crystallographic and spectroscopic insights into heme degradation by Mycobacterium tuberculosis MhuD, Graves [/bib_ref] [bib_ref] Ruffling of metalloporphyrins bound to IsdG and IsdI, two hemedegrading enzymes in..., Lee [/bib_ref]. Conversely, canonical HOs have an overall α-helical fold and a highly polar distal heme pocket containing two ordered water molecules that are essential for its enzymatic activity [bib_ref] Quantum mechanical/molecular mechanical study of mechanisms of heme degradation by the enzyme..., Chen [/bib_ref] [bib_ref] H NMR detection of immobilized water molecules within a strong distal hydrogen-bonding..., Syvitski [/bib_ref]. Interestingly, MhuD is capable of binding either one or two hemes within the same active site, a feature that is unique among all known HOs [fig_ref] Figure 2: X-ray crystal structures showing the active sites of monoheme and diheme forms... [/fig_ref]. The heme in the monoheme form of MhuD is coordinated to the proximal His-75 residue. In the diheme form, the additional heme was proposed to be in the distal active site stacked planar upon the His-ligated heme and coordinated indirectly with Asn-7 via a chloride ion [bib_ref] Unusual diheme conformation of the heme degrading protein from Mycobacterium tuberculosis, Chim [/bib_ref]. Asn-7 was suggested to be key to the enzymatic activity of MhuD by providing H-bonding interactions to stabilize and regiospecifically orient reactive intermediates in the monoheme form [bib_ref] A new way to degrade heme: The Mycobacterium tuberculosis enzyme MhuD catalyzes..., Nambu [/bib_ref] , an interaction that structural data suggested to be blocked by the distal heme in diheme MhuD [bib_ref] Unusual diheme conformation of the heme degrading protein from Mycobacterium tuberculosis, Chim [/bib_ref]. Additionally, the heme macrocycle in monoheme MhuD was reported to exhibit unusually extensive ruffling similar to IsdG/I proteins, as opposed to more planar heme structures in diheme MhuD and canonical HOs. This distortion was implied to contribute to different heme degradation mechanisms of noncanonical HOs [bib_ref] Crystallographic and spectroscopic insights into heme degradation by Mycobacterium tuberculosis MhuD, Graves [/bib_ref] [bib_ref] Measurement of heme ruffling changes in MhuD using UV-vis spectroscopy, Graves [/bib_ref] [bib_ref] A ferric-peroxo intermediate in the oxidation of heme by IsdI, Takayama [/bib_ref].
It was shown that MhuD degrades heme by unprecedented sequential mono-and dioxygenation reactions [bib_ref] Unique coupling of mono-and dioxygenase chemistries in a single active site promotes..., Matsui [/bib_ref]. The initial studies showed that the monoheme form was enzymatically active, while diheme was inactive [bib_ref] Unusual diheme conformation of the heme degrading protein from Mycobacterium tuberculosis, Chim [/bib_ref]. This brought into question the functional role of the second heme in the catalysis of MhuD. It was proposed that MhuD has much higher affinity for the first heme than the second, implying that monoheme functions as a heme degrader in low intracellular heme conditions and when heme concentrations rise, MhuD binds the second heme, and its diheme form acts as a heme storage or regulatory protein [bib_ref] The affinity of MhuD for heme is consistent with a heme degrading..., Thakuri [/bib_ref]. However, more recent studies showed that MhuD did not evolve to preferentially bind one heme molecule and that the diheme form is comparably favored, implicating a more complex functional role of diheme MhuD in Mtb physiology [bib_ref] MhuD from Mycobacterium tuberculosis: Probing a dual role in heme storage and..., Matthews [/bib_ref].
Resonance Raman (rR) spectroscopy is an ideal technique for studying the active site environment of heme proteins, providing a wide range of structural information regarding oxidation, spin, and coordination states of heme iron, geometries of heme substituent groups and their H-bonding interactions, deformations of the heme plane, and the disposition of endogenous and exogenous heme axial ligands [bib_ref] Non-planar heme deformations and excited state displacements in horseradish peroxidase detected by..., Huang [/bib_ref] [bib_ref] Resonance Raman spectra of heme proteins and model compounds, Kincaid [/bib_ref] [bib_ref] Resonance Raman spectroscopy as a structural probe of the cytochrome P450 enzymatic..., Mak [/bib_ref] [bib_ref] Hydrogen bonding of ironcoordinated histidine in heme proteins, Rousseau [/bib_ref] [bib_ref] Ligand-protein interactions in nitric oxide synthase, Rousseau [/bib_ref] [bib_ref] Resonance Raman Spectra of Heme and Metalloproteins, Spiro [/bib_ref]. As such, rR studies are particularly well suited for revealing structural differences between mono-and diheme MhuD as they were proposed to exhibit differences in heme planarity [bib_ref] Unusual diheme conformation of the heme degrading protein from Mycobacterium tuberculosis, Chim [/bib_ref] [bib_ref] Crystallographic and spectroscopic insights into heme degradation by Mycobacterium tuberculosis MhuD, Graves [/bib_ref]. rR measurements of ferrous His-coordinated heme proteins reveal the strength of the Fe-proximal ligand bond [bib_ref] Hydrogen bonding of ironcoordinated histidine in heme proteins, Rousseau [/bib_ref]. Additionally, rR studies of ferrous carbonmonoxy adducts of heme proteins provide an effective probe of the polarity and crowding of the distal active site environment [bib_ref] CO as a vibrational probe of heme protein active sites, Spiro [/bib_ref] [bib_ref] Role of the axial ligand in heme-CO backbonding; DFT analysis of vibrational..., Vogel [/bib_ref] , making the CO adduct ideal for monitoring the MhuD heme pocket in the presence or absence of an additional heme.
# Results
## Preparation of mhud protein
The presence of a terminal His-tag on MhuD was previously shown to complicate the determination of hemebinding affinity [bib_ref] The affinity of MhuD for heme is consistent with a heme degrading..., Thakuri [/bib_ref]. To obtain the most accurate model of MhuD protein without any additional amino acids to its native primary structure from affinity tags or protease cleavage sites, we employed a fusion of MhuD at its C-terminus to a cleavable intein, DNA gyrase subunit A from Mycobacterium xenopi (Mxe GyrA), which contains a chitinbinding domain that allows affinity purification prior to its removal in its entirety from MhuD. This expression and purification process produced relatively high yields of apo-MhuD protein, e.g., 20 mg per liter of cell culture and >99% purity determined by SDS-PAGE [fig_ref] Figure 1: Structures of heme and its degradation products achieved through different heme oxygenase... [/fig_ref] and Image Lab software (Bio-Rad).
Reconstituting apo-MhuD protein with heme is also not a trivial endeavor. Since MhuD can bind two hemes in the same active site, special measures must be taken to ensure reliable and consistent heme:protein stoichiometry during reconstitution of monoheme samples to prevent formation of diheme, which was shown to occur even when incubated with heme in a 1:1 molar ratio [bib_ref] MhuD from Mycobacterium tuberculosis: Probing a dual role in heme storage and..., Matthews [/bib_ref]. Therefore, MhuD samples in this study were reconstituted using a CN-CO replacement method that has provided an apparent solution to this challenging problem. The effectiveness of this method in producing pure ferric monoheme MhuD without contamination by residual diheme populations is evidenced by an almost 1:1 molar ratio of heme to protein despite being incubated with approximately fourfold excess of hemindicyanide. This new technique used to generate mono-and diheme samples with 1:1 and 2:1 heme:protein stoichiometric ratios, respectively, was crucial to accurately characterize the structural differences between the two forms of MhuD. Additional results and detailed discussion for the preparation of MhuD protein samples are provided in the Supporting information.
## Ferric state
The UV-vis electronic absorption spectra of ferric monoand diheme MhuD at pH 7.5 are shown in [fig_ref] Figure 3: UV-vis spectra of mono-and diheme MhuD and free heme in different oxidation... [/fig_ref]. Monoheme has a Soret band maximum at 405 nm and is red-shifted from that of diheme at 396 nm. Both forms exhibit Q-bands at 562 nm, 586 nm, and 605 nm, the last of which is substantially more prominent in the diheme sample. For both forms, changes in pH had very minimal effect on the Soret and no changes in the positions of the Q-bands [fig_ref] Figure 2: X-ray crystal structures showing the active sites of monoheme and diheme forms... [/fig_ref] and [fig_ref] Table 1: Data summary for ferrous-CO ligated mono-and diheme MhuD and free heme with... [/fig_ref]. A more thorough discussion of UV-vis data for MhuD in the ferric state is provided in the Supporting information.
The rR spectra of ferric mono-and diheme MhuD at pH 7.5 are shown in [fig_ref] Figure 4: rR spectra of mono-and diheme MhuD in the ferric state at pH... [/fig_ref]. The vibrational modes were assigned based on data reported for HO-1, myoglobin, and heme model compounds [bib_ref] Resonance Raman and EPR spectroscopic studies on heme-heme oxygenase complexes, Sun [/bib_ref] [bib_ref] Heme-heme oxygenase complex: Structure and properties of the catalytic site from resonance..., Takahashi [/bib_ref] [bib_ref] Assignment of protoheme resonance Raman spectrum by heme labeling in myoglobin, Hu [/bib_ref] [bib_ref] Consistent porphyrin force field. 3. Out-of-plane modes in the resonance Raman spectra..., Li [/bib_ref]. The high-frequency (HF) spectra of mono-and diheme MhuD show spin state marker bands ν 3 and ν 2 at 1492 cm −1 and 1575 cm −1 , respectively, indicating that both forms are predominantly 5-coordinated high spin (5cHS). A second set of ν 3 and ν 2 modes are observed at 1505 cm −1 and 1587 cm −1 , respectively, associated with a minor contribution of 6-coordinated low spin (6cLS). The 6cLS state is most reasonably associated with the presence of a small fraction of heme iron coordinated to a water molecule that presumably has a hydroxide character caused by H-bonding interactions with the distal Asn-7 residue. The diheme spectrum shows an increase in relative intensity (with respect to ν 4 ) of the 5cHS spin state markers, ν 3 and ν 2 at 1492 cm −1 and 1574 cm −1 , respectively, while maintaining a similar relative intensity of the minor 6cLS state. This implies that the spin state of the His-coordinated heme remains the same in both the mono-and diheme active site, but the latter contains an additional 5-coordinated heme molecule. The rR spectra of ferric MhuD samples at different pH [fig_ref] Figure 3: UV-vis spectra of mono-and diheme MhuD and free heme in different oxidation... [/fig_ref] and show that the 5cHS state remains dominant in acidic and alkaline conditions for both MhuD forms, contrary to the 6cHS H 2 O-bound to 6cLS OH − -bound transition observed in canonical HOs and IsdG/I proteins [bib_ref] Resonance Raman and EPR spectroscopic studies on heme-heme oxygenase complexes, Sun [/bib_ref] [bib_ref] Unique electronic structures of the highly ruffled hemes in heme-degrading enzymes of..., Takahashi [/bib_ref]. These results indicate that MhuD contains a predominantly hydrophobic distal active site environment.
The low-frequency (LF) region of rR spectra of heme proteins is rich with out-of-plane (oop) heme deformation modes, as well as modes associated with the heme peripheral groups and endogenous and exogenous axial ligands [bib_ref] Resonance Raman spectra of heme proteins and model compounds, Kincaid [/bib_ref] [bib_ref] Assignment of protoheme resonance Raman spectrum by heme labeling in myoglobin, Hu [/bib_ref] [bib_ref] Consistent porphyrin force field. 3. Out-of-plane modes in the resonance Raman spectra..., Li [/bib_ref]. Surprisingly, the LF spectra of ferric mono-and diheme samples differ only slightly, the main difference being the change in the propionate bending modes. Monoheme contains δ(C β -C c -C d ) at 382 cm −1 , and it shifts to 375 cm −1 in the diheme spectrum, implying that the propionate groups in monoheme form stronger H-bonds with one or more active site amino acids than those in diheme MhuD [bib_ref] Phe393 mutants of cytochrome P450 BM3 with modified heme redox potentials have..., Chen [/bib_ref]. This 7 cm −1 shift of δ(C β -C c -C d ) illustrates the alteration of protein-heme interactions associated with the presence of the additional distal heme molecule, which reportedly induces a protein conformational change that almost triples the active site volume of diheme MhuD relative to its monoheme form [bib_ref] Structure of a Mycobacterium tuberculosis heme-degrading protein, MhuD, variant in complex with..., Chao [/bib_ref].
Several out-of-plane (oop) modes are seen in the LF region, namely γ 7 at 313 cm −1 , γ 12 at 496 cm −1 , γ 21 at 551 cm −1 , and γ 11 at 716 cm −1 . Oop modes, especially those of B 1u symmetry such as γ 11 and γ 12 , are expected to be more strongly activated for proteins with ruffled heme than those with planar heme [bib_ref] Non-planar heme deformations and excited state displacements in horseradish peroxidase detected by..., Huang [/bib_ref] [bib_ref] Structural characterization of synthetic and protein-bound porphyrins in terms of the lowest-frequency..., Jentzen [/bib_ref]. Interestingly, no significant differences in activation of oop modes were detected between mono-and diheme MhuD, the latter of which actually has slightly greater enhancement of γ 7 and γ 12 . This is surprising since previous crystallographic studies reported substantially different degrees of oop distortions of the hemes in mono-(1.4 Å) and diheme (0.7 Å) MhuD [bib_ref] Unusual diheme conformation of the heme degrading protein from Mycobacterium tuberculosis, Chim [/bib_ref] [bib_ref] Crystallographic and spectroscopic insights into heme degradation by Mycobacterium tuberculosis MhuD, Graves [/bib_ref].
## Ferrous state
The rR spectra of the ferrous state of mono-and diheme MhuD were measured using the 441.6 nm excitation line to exclusively enhance modes associated with the Hiscoordinated heme, while the other ferrous heme in the distal active site of diheme was effectively rR silent (Soret at 383 nm, [fig_ref] Figure 3: UV-vis spectra of mono-and diheme MhuD and free heme in different oxidation... [/fig_ref]. The HF ferrous rR spectra are shown and discussed in [fig_ref] Figure 4: rR spectra of mono-and diheme MhuD in the ferric state at pH... [/fig_ref].
Shown in [fig_ref] Figure 5: rR spectra of mono-and diheme MhuD in the ferrous state [/fig_ref] , the ν(Fe-N His ) stretching mode of monoheme MhuD is seen at 218 cm −1 , consistent with previously published data for monoheme MhuD and rat HO-1 [bib_ref] Heme-heme oxygenase complex: Structure and properties of the catalytic site from resonance..., Takahashi [/bib_ref] , and within 1 cm −1 of those of IsdG and IsdI [bib_ref] Unique electronic structures of the highly ruffled hemes in heme-degrading enzymes of..., Takahashi [/bib_ref]. This frequency is indicative of neutral imidazole character of the proximal heme ligand, His-75. Interestingly, ν(Fe-N His ) in the diheme spectrum is also seen at 218 cm −1 , indicating that the presence of the distal heme in the active site does not affect [bib_ref] Crystallographic and spectroscopic insights into heme degradation by Mycobacterium tuberculosis MhuD, Graves [/bib_ref]. Panel B shows ferric diheme MhuD with both hemes in the 5-coordinated high spin state (PDB ID 3HX9) with the protein secondary structure displayed in light green [bib_ref] Unusual diheme conformation of the heme degrading protein from Mycobacterium tuberculosis, Chim [/bib_ref]. In both structures, the side chain functional groups of the following amino acids are displayed: Asn-7, His-25, His-28, His-75, and His-78. The proximal heme in both forms is shown coordinated to His-75, despite the presence of several other histidine residues surrounding the active site. The α2 helix is kinked in monoheme but is extended in the diheme form to expand the volume of the active site to accommodate an additional heme. The Asn-7 residue is shown within range of electrostatic interactions with the exogenous cyanide ligand in monoheme, and the Cl − ion (green sphere) coordinated axially to the distal heme in the diheme form.
the strength of the proximal heme Fe-N His linkage. Additionally, as in ferric samples, no differences are observed in the degree of nonplanar heme deformations between mono-and diheme MhuD. The assignment of the ν(Fe-N His ) stretching mode in the ferrous monoheme spectra was confirmed using isotopically labeled hemes [bib_ref] Aggregation of ferrihaems. Dimerization and protolytic equilibria of protoferrihaem and deuteroferrihaem in..., Brown [/bib_ref] Fe-and 58 Fe-PPIX [fig_ref] Figure 5: rR spectra of mono-and diheme MhuD in the ferrous state [/fig_ref] , i-iii), which shifted the mode to 220 cm −1 and 216 cm −1 , respectively. Likewise, ν(Fe-N His ) is observed at 220 cm −1 and 216 cm −1 in the spectra of the 54 Fe 54 Fe-diheme and 58 Fe 58 Fediheme samples [fig_ref] Figure 5: rR spectra of mono-and diheme MhuD in the ferrous state [/fig_ref] , respectively, confirming the 4 cm −1 difference between these two iron isotopes is maintained in the spectra of diheme samples (full LF spectra are shown in [fig_ref] Figure 5: rR spectra of mono-and diheme MhuD in the ferrous state [/fig_ref].
The samples with mixed iron isotopes, [bib_ref] Spin state and axial ligand bonding in the hydroxide complexes of metmyoglobin,..., Feis [/bib_ref] Fe 54 Fe-and 54 Fe 58 Fe-diheme, are named in the order of which the hemes were added to apoMhuD, e.g., the 58 Fe 54 Fe-diheme sample was prepared first as ferric [bib_ref] Spin state and axial ligand bonding in the hydroxide complexes of metmyoglobin,..., Feis [/bib_ref] Fe-monoheme and then excess [bib_ref] Aggregation of ferrihaems. Dimerization and protolytic equilibria of protoferrihaem and deuteroferrihaem in..., Brown [/bib_ref] Fe-PPIX was added to form diheme (see Experimental procedures). This strategy was employed to ensure that all samples were completely bound with one heme of a particular isotope prior to addition of the second and was necessary to facilitate reliable insight into the dynamics associated with binding of the second heme molecule in the same active site. The 58 Fe 54 Fe-diheme sample [fig_ref] Figure 5: rR spectra of mono-and diheme MhuD in the ferrous state [/fig_ref] , right) exhibits the ν(Fe-N His ) stretching mode at 220 cm −1 , the same frequency as that of [bib_ref] Aggregation of ferrihaems. Dimerization and protolytic equilibria of protoferrihaem and deuteroferrihaem in..., Brown [/bib_ref] Fe-monoheme instead of the expected frequency of 216 cm −1 , which would be observed if [bib_ref] Spin state and axial ligand bonding in the hydroxide complexes of metmyoglobin,..., Feis [/bib_ref] Fe-PPIX was still coordinated to His-75. This implies that the second incoming heme replaces the first as the His-ligated heme. The experiment was repeated for the 54 Fe 58 Fe-diheme sample and ν(Fe-N His ) was observed at 217 cm −1 , roughly the same frequency as that of the 58 Fe-monoheme sample, further confirming the His-coordinated heme switching mechanism during diheme MhuD formation.
## Ferrous-co ligated state
The UV-vis spectrum of ferrous-CO monoheme MhuD exhibits a Soret maximum at 417 nm and the spectrum of diheme MhuD appears to have contributions of features observed in the CO-ligated free heme spectrum; e.g., the Soret peak of free heme at 407 nm is at the same wavelength as the lower wavelength shoulder seen on the Soret band of the diheme MhuD spectrum [fig_ref] Figure 3: UV-vis spectra of mono-and diheme MhuD and free heme in different oxidation... [/fig_ref].
The full range rR spectra of mono-and diheme MhuD samples, as well as free heme, are shown in [fig_ref] Figure 6: Deconvoluted rR spectra of ferrous-CO adducts of mono-and diheme MhuD and free... [/fig_ref] and discussed in Supporting information. The most notable differences in the spectra of mono-and diheme MhuD lie in the 500 to 600 cm −1 and 1800 to 2000 cm −1 regions, where the modes associated with the Fe-C-O fragment are typically observed for His-ligated proteins [bib_ref] CO as a vibrational probe of heme protein active sites, Spiro [/bib_ref]. Careful spectral deconvolutions in the LF and HF regions were performed as described in the Supporting information to determine the real frequencies of modes associated with the Fe-C-O fragments.
Shown in [fig_ref] Figure 6: Deconvoluted rR spectra of ferrous-CO adducts of mono-and diheme MhuD and free... [/fig_ref] are the deconvoluted spectra for [bib_ref] Simultaneous determination of hemes a, b, and c from pyridine hemochrome spectra, Berry [/bib_ref] Fe-PPIX mono-and diheme MhuD and free heme samples in the 450 to 610 cm −1 range for 12 C 16 O, 13 C 16 O and 13 C 18 O isotopes. The ν(Fe-C) stretching mode in the spectrum of monoheme MhuD is seen at 525 cm −1 for the naturally abundant isotopes, and it shifts to 521 cm −1 and 515 cm −1 upon substitution with [bib_ref] Quantum mechanical/molecular mechanical study of mechanisms of heme degradation by the enzyme..., Chen [/bib_ref] Experimental evidence for heme degradation by diheme MhuD revealed the presence of additional heme modes at 497 cm −1 , 551 cm −1 , and 587 cm −1 . The spectral parameters of these modes, such as their frequencies, bandwidths, and relative intensities, remained largely unchanged between the spectra of samples with isotopically labeled CO, as described in the Supporting information.
Interestingly, the deconvoluted spectra of diheme MhuD show multiple Fe-C-O conformers, denoted as A, B, and C, in both the LF and HF regions [fig_ref] Figure 6: Deconvoluted rR spectra of ferrous-CO adducts of mono-and diheme MhuD and free... [/fig_ref] , D-F). Conformers A and C have ν(Fe-C) and ν(C-O) modes located at the same frequencies and exhibit identical CO isotopic shifts and bandwidths as monoheme MhuD and free heme [fig_ref] Figure 6: Deconvoluted rR spectra of ferrous-CO adducts of mono-and diheme MhuD and free... [/fig_ref] , G-I), respectively (extracted data reported in Tables S3-S17). A third Fe-C-O conformer, conformer B, is unique to diheme MhuD and its ν(Fe-C) mode is seen at 505 cm −1 shifting to 499 cm −1 and 487 cm −1 upon 13 C 16 O and 13 C 18 O substitution, respectively. The corresponding ν(C-O) mode for conformer B is located at 1949 cm −1 and downshifts by 92 cm −1 with 13 C 18 O isotopic replacement. Shown in [fig_ref] Figure 7: Inverse correlation plot of ν [/fig_ref] is an inverse correlation plot of ν(Fe-C) and ν(C-O) frequencies of MhuD and various other heme proteins and model compounds [bib_ref] CO as a vibrational probe of heme protein active sites, Spiro [/bib_ref] [bib_ref] Role of the axial ligand in heme-CO backbonding; DFT analysis of vibrational..., Vogel [/bib_ref] [bib_ref] Heme-heme oxygenase complex: Structure and properties of the catalytic site from resonance..., Takahashi [/bib_ref] [bib_ref] Unique electronic structures of the highly ruffled hemes in heme-degrading enzymes of..., Takahashi [/bib_ref] [bib_ref] Identification of histidine 25 as the heme ligand in human liver heme..., Sun [/bib_ref]. The ν(Fe-C)/ν(C-O) pairs of MhuD conformers A and B fall on the line of 6-coordinated hemes with histidine as a proximal ligand, confirming the neutral character of the imidazole group of both conformers [bib_ref] CO as a vibrational probe of heme protein active sites, Spiro [/bib_ref]. Conformer A, which is present in both mono-and diheme MhuD, is located in the upper left region of the correlation line indicating strong positive polarization [bib_ref] CO as a vibrational probe of heme protein active sites, Spiro [/bib_ref] [bib_ref] Role of the axial ligand in heme-CO backbonding; DFT analysis of vibrational..., Vogel [/bib_ref] , likely from H-bonding interactions with the Asn-7 residue. Similar interactions were seen between the CN − ligand and Asn-7 in the ferric-CN MhuD crystal structure [bib_ref] Crystallographic and spectroscopic insights into heme degradation by Mycobacterium tuberculosis MhuD, Graves [/bib_ref]. Notably, conformer A lies on the correlation line remarkably close to the CO adducts of IsdG/I proteins that have Asn at very similar positions, implying a similar interaction for all three proteins [bib_ref] Unique electronic structures of the highly ruffled hemes in heme-degrading enzymes of..., Takahashi [/bib_ref]. Since the distal asparagine residue is proposed to be responsible for directing the heme reactive oxygen species to the appropriate meso-carbons to form hydroxyheme in the catalytic mechanisms of MhuD and IsdG/I proteins, conformer A likely represents the configuration adopted by catalytically active Fe-O-O fragments [bib_ref] A ferric-peroxo intermediate in the oxidation of heme by IsdI, Takayama [/bib_ref] [bib_ref] Unique coupling of mono-and dioxygenase chemistries in a single active site promotes..., Matsui [/bib_ref] , implying diheme MhuD retains enzymatic activity. Conformer B of diheme MhuD lies toward the middle of the correlation line, closer to wild-type myoglobin and rat HO-1, which experience much weaker H-bonding interactions [bib_ref] CO as a vibrational probe of heme protein active sites, Spiro [/bib_ref] [bib_ref] Heme-heme oxygenase complex: Structure and properties of the catalytic site from resonance..., Takahashi [/bib_ref]. Conformer C falls on the correlation line characteristic of 5-coordinated heme model compounds, representing the CO-ligated heme in the distal active site, which is not ligated to the protein. Its position on the correlation plot is very similar to that of the H25A mutant of hHO-1 where the proximal heme ligand, histidine, was replaced with alanine, providing another example of a 5-coordinated species within a protein active site [bib_ref] Identification of histidine 25 as the heme ligand in human liver heme..., Sun [/bib_ref].
The consistency of the deconvoluted spectral patterns of the Fe-C-O conformers was further confirmed using MhuD samples containing isotopically labeled hemes. The deconvoluted CO spectra of samples with [bib_ref] Aggregation of ferrihaems. Dimerization and protolytic equilibria of protoferrihaem and deuteroferrihaem in..., Brown [/bib_ref] Fe-PPIX are shown in [fig_ref] Figure 7: Inverse correlation plot of ν [/fig_ref] and [bib_ref] Spin state and axial ligand bonding in the hydroxide complexes of metmyoglobin,..., Feis [/bib_ref] Fe-PPIX in [fig_ref] Figure 8: Deconvoluted rR spectra of ferrous-CO adducts of diheme MhuD samples containing mixed... [/fig_ref]. For MhuD samples with mixed heme iron isotopes, 54 Fe 58 Fe-and 58 Fe 54 Fe-diheme, the deconvoluted spectra are shown in [fig_ref] Figure 8: Deconvoluted rR spectra of ferrous-CO adducts of diheme MhuD samples containing mixed... [/fig_ref]. The 54 Fe 58 Fediheme sample exhibits ν(Fe-C) modes associated with conformers A and B at 522 cm −1 and 502 cm −1 , respectively, the same frequencies as those in the 58 Fe 58 Fe-diheme sample. Conformer C, however, has ν(Fe-C) at 532 cm −1 , the same frequency as that of the 54 Fe 54 Fe-diheme sample, meaning the 54 Fe-heme now occupies the distal active site. The analogous, but reverse case is observed for the 58 Fe 54 Fe-diheme sample. These results reinforce the proposal derived from studies of the ferrous state (vide supra), that during formation of diheme MhuD, the incoming second heme replaces the originally Hiscoordinated heme and displaces it to the distal active site. It is noted that the reliability of the deconvolution methodology presented here is reinforced by the fact that the ratios of ν(Fe-C) peak areas between conformers A, B, and C in the spectra of diheme samples are consistent between each of the variations of CO and Fe isotopes (Tables S3-S17). The frequencies and isotopic shifts of ν(Fe-C), ν(C-O) and δ(Fe-C-O) for all samples measured herein are listed in [fig_ref] Table 1: Data summary for ferrous-CO ligated mono-and diheme MhuD and free heme with... [/fig_ref].
## Activity assays
In order to test the hypothesis that diheme MhuD is enzymatically active, several activity assays were performed and monitored by UV-vis spectroscopy. The natural redox partner for MhuD has yet to be identified, so the assays were carried out using electron donor substitutes, ascorbate [fig_ref] Figure 9: UV-vis absorption spectra of ascorbate activity assays for mono-and diheme MhuD [/fig_ref] and NADPH-cytochrome P450 oxidoreductase (POR, [fig_ref] Figure 9: UV-vis absorption spectra of ascorbate activity assays for mono-and diheme MhuD [/fig_ref]. The ascorbate assays were performed in the presence of superoxide dismutase and catalase to minimize nonenzymatic heme degradation, as confirmed by a control assay with free heme [fig_ref] Figure 1: Structures of heme and its degradation products achieved through different heme oxygenase... [/fig_ref]. Shown in [fig_ref] Figure 9: UV-vis absorption spectra of ascorbate activity assays for mono-and diheme MhuD [/fig_ref] , the absorbance of the Soret band diminishes over time for both mono-and diheme MhuD, indicating that heme is being degraded by both forms. As the reaction begins, both MhuD forms show an initial increase in the Q-band at 560 nm and shift of the Soret to 408 nm as oxy-MhuD accumulates, consistent with previously published ascorbate activity assays for monoheme MhuD [bib_ref] Unusual diheme conformation of the heme degrading protein from Mycobacterium tuberculosis, Chim [/bib_ref] [bib_ref] Crystallographic and spectroscopic insights into heme degradation by Mycobacterium tuberculosis MhuD, Graves [/bib_ref] [bib_ref] Measurement of heme ruffling changes in MhuD using UV-vis spectroscopy, Graves [/bib_ref] [bib_ref] Unique coupling of mono-and dioxygenase chemistries in a single active site promotes..., Matsui [/bib_ref]. This species decreases in absorbance as the reaction progresses and forms the products, mycobilins-a and b with bands at 340 nm and 550 nm. The absorbance spectrum of free mycobilin-a has bands at 345 nm and 565 nm, and those for mycobilin-b are at 336 nm and 555 nm [bib_ref] A new way to degrade heme: The Mycobacterium tuberculosis enzyme MhuD catalyzes..., Nambu [/bib_ref]. A plot of Soret Δabsorbance versus reaction time for mono-and diheme MhuD is shown in [fig_ref] Figure 1: Structures of heme and its degradation products achieved through different heme oxygenase... [/fig_ref] and discussed in Supporting information. The activity of MhuD was consistent with different electron donors as POR also enabled heme degradation by both forms [fig_ref] Figure 9: UV-vis absorption spectra of ascorbate activity assays for mono-and diheme MhuD [/fig_ref] , consistent with published POR assays for monoheme [bib_ref] Unusual diheme conformation of the heme degrading protein from Mycobacterium tuberculosis, Chim [/bib_ref] , but contradictory to that reported for diheme MhuD [bib_ref] Unusual diheme conformation of the heme degrading protein from Mycobacterium tuberculosis, Chim [/bib_ref]. To ensure that the observed activity of diheme MhuD is not attributable to the presence of a small fraction of monoheme form generated during gel filtration, which could potentially strip one of the heme molecules from the diheme active site, an additional ascorbate assay of MhuD (4 μM) in the presence of a large excess of heme (20 μM) was performed [fig_ref] Figure 1: Structures of heme and its degradation products achieved through different heme oxygenase... [/fig_ref]. It is clear that under Experimental evidence for heme degradation by diheme MhuD these conditions MhuD still degrades all heme molecules to mycobilins, as evidenced by the decrease in the Soret band and increase in absorbance of the bands around 340 nm and 550 nm.
While it is well established that monoheme MhuD degrades heme to mycobilin products [bib_ref] Unique coupling of mono-and dioxygenase chemistries in a single active site promotes..., Matsui [/bib_ref] [bib_ref] A dynamic substrate is required for MhuD-catalyzed degradation of heme to mycobilin, Thakuri [/bib_ref] , the enzymatic activity of diheme MhuD reported here for the first time calls for characterization of its heme degradation products. The crude products of the ascorbate assay were analyzed by electrospray ionization-mass spectrometry (ESI-MS) and the data [fig_ref] Figure 1: Structures of heme and its degradation products achieved through different heme oxygenase... [/fig_ref] revealed that the major product of diheme MhuD catalysis is also mycobilin with m/z 611.25, like that of monoheme. A minor product with m/z 583.25 was detected at nearly identical relative abundance as previously reported data for the monoheme MhuD reaction and was identified as biliverdin [bib_ref] A dynamic substrate is required for MhuD-catalyzed degradation of heme to mycobilin, Thakuri [/bib_ref]. Therefore, the products of heme degradation by diheme MhuD are highly consistent with that of monoheme in that its major products are mycobilins.
# Discussion
## Structure-function correlation of mono-and diheme mhud
rR spectroscopy was employed in this study to investigate several features of the active site environment that dictate the reactivity of mono-and diheme MhuD such as heme (non) planarity, hydrophobicity of the distal active site, strength of the Fe-proximal ligand bond, and active site amino acid interactions with heme distal ligands and peripheral groups. The rR data shown here provide for the first time a uniform picture of heme (non)planarity in mono-and diheme forms of MhuD, i.e., the His-coordinated heme molecules in both forms have a similar extent of nonplanar deformations. Additionally, the ferric state spectra confirmed that MhuD has a largely hydrophobic distal active site that is void of an ordered water molecule network like that possessed by canonical HOs [bib_ref] Quantum mechanical/molecular mechanical study of mechanisms of heme degradation by the enzyme..., Chen [/bib_ref] [bib_ref] H NMR detection of immobilized water molecules within a strong distal hydrogen-bonding..., Syvitski [/bib_ref] , suggesting distinct modes of oxygen activation. The ferrous state rR spectra indicated that the strength of the Fe-N His linkage, which has great implications for heme reactivity [bib_ref] Hydrogen bonding of ironcoordinated histidine in heme proteins, Rousseau [/bib_ref] [bib_ref] Probing structure-function relations in hemecontaining oxygenases and peroxidases, Dawson [/bib_ref] , is unaffected by the presence of an additional heme in the active site, i.e., it is the same for mono-and diheme forms. In fact, the ferrous rR spectra of both mono-and diheme are essentially identical, e.g., there are no differences in activation of out-of-plane modes that would indicate varying degrees of heme ruffling or alterations to the geometry of heme peripheral groups. These data are important because they not only reflect a large and malleable active site of MhuD but imply that the His-coordinated heme might retain its inherent reactivity in both mono-and diheme forms.
While one can envision that binding of the second heme could be preceded by a conformational change of the protein structure to allow binding of the newly added heme on the distal side of the His-coordinated heme, a dramatically different pathway was revealed in the rR spectra of ferrous diheme samples containing mixed heme iron isotopes where the second heme was shown to replace the first as the Hiscoordinated species. This implies that binding of the second heme is preceded by a, perhaps allosteric, protein conformational change that prompts the cleavage of the original Fe-N His bond and movement of the α2 helix to extend the proximal heme pocket and allow the now free His-75 residue to bind the incoming second heme while the original heme remains in the distal active site. Alternatively, the heme-binding mechanism could be a result of conformational selection due to a flexible active site and a relatively labile Fe-N His bond. While the X-ray crystal structures displayed in [fig_ref] Figure 2: X-ray crystal structures showing the active sites of monoheme and diheme forms... [/fig_ref] show only one histidine-ligated state where His-75 acts as the sole endogenous heme ligand, there could be other conformational states where the heme is coordinated to alternative His residues. There are several histidines surrounding the MhuD active site; His-78, for example, is located only three residues away in a flexible loop region. Therefore, conformational selection would be plausible wherein the second incoming heme binds to one of the other histidine residues in a particular state either before, after, or simultaneously with the cleavage of the Fe-N His bond of the original heme to ultimately displace it to the distal active site (with no endogenous ligand). Regardless of whether the heme displacement occurs via induced fit or conformational selection mechanisms, our data unambiguously reveal that this unusual heme switching process takes place. These results demonstrate the unique capability of rR spectroscopy to unveil subtle, but physiologically important biomolecular dynamics processes that are not easily assessed using other biophysical methods.
The rR spectra of ferrous-CO adducts presented here provide important new insight into the structural differences between mono-and diheme MhuD. Until now, it was not known if either of the heme molecules in the active site of the diheme protein is capable of binding exogenous ligands. More importantly however, the rR spectra revealed two Hiscoordinated Fe-C-O conformers in diheme MhuD, one of which retains H-bonding interactions with Asn-7, identical to that seen in monoheme MhuD (conformer A), and the second (conformer B) in which these interactions are significantly disrupted. The third Fe-C-O conformer in diheme MhuD, conformer C, is associated with the 5-coordinated heme that remains in the distal heme pocket and is intermittently positioned to alter the H-bonding interactions of conformer A to generate conformer B. It is also noted based on the inspection of relative intensity ratios of modes associated with conformers A and B, that the former is still dominant in diheme MhuD, indicating that the second heme does not completely block the access of the heme exogenous ligands to the key Asn-7 residue. This is, in fact, quite surprising because the crystal structure of diheme MhuD suggests that such interactions would be completely impeded by the presence of the distal heme in the active site [bib_ref] Unusual diheme conformation of the heme degrading protein from Mycobacterium tuberculosis, Chim [/bib_ref]. Such a stark contrast in the structural implications derived from the spectroscopic data collected in solution here as compared with that collected on crystalline forms is a clear indication of the inherent highly flexible nature of the MhuD heme pocket in which the heme molecules have more conformational mobility to maintain the interactions between the His-coordinated heme and the distal active site. In fact, careful inspection of the X-ray crystal structure for diheme MhuD [fig_ref] Figure 2: X-ray crystal structures showing the active sites of monoheme and diheme forms... [/fig_ref] shows that the Asn-7 residue is actually located above the edges of the heme macrocycle, rather than directly above the heme iron. Such positioning requires less dramatic movement of the distal heme to enable the interaction between Asn-7 and the His-ligated heme exogenous ligands. It is reasonable to assume that the diheme crystal structure captures only one of multiple conformational states present in solution, likely showing that which is associated
## Experimental evidence for heme degradation by diheme mhud
with Fe-C-O conformer B where these interactions are blocked or weakened. Furthermore, the fact that diheme MhuD has a substantially large relative population of the catalytically predisposed conformer A apparently contradicts the proposal of the diheme protein being inactive, i.e., the activity of diheme might be slightly diminished relative to monoheme, but not entirely lost.
## Catalytic activity
The data presented here revealed that both mono-and diheme forms of MhuD are enzymatically active. Not only does diheme MhuD degrade the His-coordinated heme, but also the distal heme in its active site as well. Interestingly, after degrading the bound heme molecules in its active site, MhuD proceeds to bind the excess heme present in solution and convert it to mycobilin [fig_ref] Figure 1: Structures of heme and its degradation products achieved through different heme oxygenase... [/fig_ref]. Previous studies reported relatively high affinity of the protein for the products, suggesting that removal of mycobilins from the MhuD active site is likely mediated by another protein [bib_ref] Structure of a Mycobacterium tuberculosis heme-degrading protein, MhuD, variant in complex with..., Chao [/bib_ref]. Our data, however, indicate that such proteins are not required for successive heme degradation reactions, as discussed in detail in the Supporting information.
The catalytic activity of diheme MhuD shown here is in stark contrast with previously reported data [bib_ref] Unusual diheme conformation of the heme degrading protein from Mycobacterium tuberculosis, Chim [/bib_ref]. The main difference between the previous work and the studies presented here is that the former involved MhuD samples containing a His-tag on the C-terminus, whereas the protein in our study has no additional amino acids to the native MhuD sequence. In that work, the reasonable assumption was made that the His-tag would have a negligible effect on the properties and activity of MhuD, as is the case for most proteins. It seems, however, that for the particular case of MhuD, the presence of the His-tag does have a substantial effect on hemebinding properties and quaternary structure of the protein.
A recent study showed that the His-tag interfered with measurement of the dissociation constants for diheme MhuD, but not monoheme, resulting in a K D disparity of three orders of magnitude between the tagged and tagless protein [bib_ref] The affinity of MhuD for heme is consistent with a heme degrading..., Thakuri [/bib_ref]. Furthermore, it was also reported using size-exclusion chromatography that His-tagged diheme MhuD exists in various higher-order oligomeric states of the protein, while the monoheme form is strictly in dimeric state [bib_ref] The TB structural genomics consortium: A decade of progress, Chim [/bib_ref]. When the diheme samples in our study were applied to a Sephadex G-75 column (GE Healthcare), no such separation of oligomeric states was observed. Similarly, analytical ultracentrifugation studies recently showed that diheme MhuD with the His-tag removed by TEV protease exists only as a dimeric protein [bib_ref] MhuD from Mycobacterium tuberculosis: Probing a dual role in heme storage and..., Matthews [/bib_ref]. The aggregation of tagged diheme MhuD would likely impede access of electrons to heme that are required in the catalytic cycle and explain its apparent inactivity as reported previously. This is not an isolated case of a His-tag altering the activity and heme-binding affinity of heme proteins involved in the heme uptake and utilization pathways of pathogenic bacteria. Rv0203 is one such protein that, like MhuD, is involved in the heme uptake pathway of Mtb. The His-tag was shown to alter heme-binding affinity for Rv0203 [bib_ref] Characterization of heme ligation properties of Rv0203, a secreted heme binding protein..., Owens [/bib_ref]. Another example is HupZ from Streptococcus pyogenes for which the His-tag induced heme stacking and higher-order oligomeric states of the protein that altered its enzymatic function [bib_ref] Heme binding to HupZ with a C-terminal tag from group A streptococcus, Traore [/bib_ref].
## Functional implications
Since its discovery, the evolutionary significance of the capability for MhuD to bind either one or two hemes in the same active site has been theorized [bib_ref] Unusual diheme conformation of the heme degrading protein from Mycobacterium tuberculosis, Chim [/bib_ref] [bib_ref] MhuD from Mycobacterium tuberculosis: Probing a dual role in heme storage and..., Matthews [/bib_ref] [bib_ref] The TB structural genomics consortium: A decade of progress, Chim [/bib_ref]. It was proposed that diheme MhuD functions primarily as a heme storage protein during periods of high intracellular heme concentration to prevent iron-mediated toxicity, a function comparable to that of the iron storage role of Mtb ferritin proteins, BfrA and BfrB [bib_ref] Unusual diheme conformation of the heme degrading protein from Mycobacterium tuberculosis, Chim [/bib_ref]. Conversely, at times of low heme concentration, one of the heme molecules in the diheme active site could be released or transferred to other heme proteins to enable heme degradation by monoheme MhuD in order to harvest the iron for use in several cellular processes. While these proposals seem reasonable, it appears implausible in conditions with an influx of heme that MhuD would lose the functionality to degrade and actively decrease its intracellular levels and prevent cytotoxicity. The studies of IsdG/I may provide clues to the evolutionary significance of mono-and diheme forms of MhuD. IsdG/I both degrade heme to staphylobilins, presumably by the same enzymatic mechanism. However, IsdG has a secondary function not possessed by IsdI; it inhibits the ferrochelatase CpfC protein involved in the heme biosynthesis pathway of S. aureus [bib_ref] Staphylococcus aureus haem biosynthesis and acquisition pathways are linked through haem monooxygenase..., Videira [/bib_ref]. Thus, IsdG mediates the intracellular heme concentration by regulating the pathways of both heme biosynthesis and degradation. Perhaps the physiological roles displayed by the two separate S. aureus proteins, IsdG/I, are evolutionary features combined into one protein for Mtb; i.e., monoheme MhuD, like IsdI, functions as strictly a heme degrader, whereas diheme functions analogously to IsdG as both a heme degrader and a regulatory protein to exert coupled control over heme degradation and biosynthesis pathways in Mtb. [bib_ref] Simultaneous determination of hemes a, b, and c from pyridine hemochrome spectra, Berry [/bib_ref] Fe-protoporphyrin IX (hemin-chloride) was purchased from Sigma-Aldrich and the isotopically labeled hemes, [bib_ref] Aggregation of ferrihaems. Dimerization and protolytic equilibria of protoferrihaem and deuteroferrihaem in..., Brown [/bib_ref] Feand 58 Fe-protoporphyrin IX, from Frontier Scientific. BL21 (DE3) E. coli cells and chitin resin were obtained from New England Biolabs. Human NADPH-cytochrome P450 oxidoreductase (POR) was obtained from ProSpec, and human superoxide dismutase from Bio Basic. The 13 C 16 O and 13 C 18 O gas isotopes and all other materials, unless mentioned otherwise, were purchased from Sigma Aldrich.
## Experimental procedures
# Materials
## Mhud protein expression and purification
The use of cleavable inteins with affinity domains for protein purification has been successfully applied to several systems [bib_ref] Single-column purification of free recombinant proteins using a self-cleavable affinity tag derived..., Chong [/bib_ref] [bib_ref] Utilizing the C-terminal cleavage activity of a protein splicing element to purify..., Chong [/bib_ref] [bib_ref] Semisynthesis of cytotoxic proteins using a modified protein splicing element, Evans [/bib_ref] [bib_ref] Semisynthesis of proteins by expressed protein ligation, Muir [/bib_ref] [bib_ref] The roles of C-terminal domain and type III domains of chitinase A1..., Watanabe [/bib_ref]. The full-length MhuD protein was expressed and purified while fused at its C-terminus to a cleavable intein-DNA gyrase subunit A from M. xenopi (Mxe GyrA). The gene encoding the MhuD-Mxe GyrA fusion protein was ordered from ATUM in the pD454-SR vector containing a T7 promotor and a gene for ampicillin resistance. A transformation of the pD454-SR vector into BL21(DE3) E. coli cells was performed via heat shock method. Yeast extract tryptone (2xYT) media with 100 mg/l ampicillin was used for inoculation of expression cultures. Cells were grown at 37 C while shaking at 220 rpm to an OD 600 value of 0.6 before protein expression was induced by addition of 1 mM IPTG and temperature was lowered to 25 C. The cells were harvested 4 h after inducing protein expression and the cell pellets were frozen at −80 C.
The affinity purification using Mxe GyrA intein was performed according to procedures adapted from the IMPACT Kit (New England Biolabs). The cells were lysed by sonication on ice in 20 mM Tris-Cl, 50 mM NaCl, pH 8.5 with added DNAase, 1 mM PMSF, and 1 mM EDTA. Cell particulate was removed by centrifugation and lysate was applied to a column with chitin resin at 4 C. The column was then washed with 20 mM Tris-Cl, 0.5 M NaCl, pH 8.5 to remove contaminating proteins, followed by equilibration with 20 mM Tris-Cl, 50 mM NaCl, 50 mM β-mercaptoethanol, pH 8.5, and stopping flow to induce intein thiolysis. Cleaved MhuD was eluted after incubating the proteins on the column for 24 h βmercaptoethanol was removed from the eluted protein sample by buffer exchange into 100 mM borate, pH 9.1 before freezing at −80 C.
## Reconstitution of apo-mhud with heme: cn-co replacement method
Hemin chloride was dissolved in 0.3 M NaOH and diluted to 500 μM with 100 mM borate, 50 mM KCN, pH 9.1, and pH was readjusted to 9.1 with HCl [bib_ref] Reconstitution of myoglobin from apoprotein and heme, monitored by stopped-flow absorption, fluorescence..., Kawamura-Konishi [/bib_ref] [bib_ref] Metalloporphyrins. VI. Cycles of changes in systems containing heme, Shack [/bib_ref]. The hemin-dicyanide was added dropwise in fourfold molar excess to 100 μM apo-MhuD in 100 mM borate, 40 mM KCN, pH 9.1 and incubated overnight at 4 C. The sample was then concentrated and applied to two successive Bio-Gel P6 columns (Bio-Rad) equilibrated with 100 mM borate, 20 mM KCN, pH 9.1, and 100 mM potassium phosphate, pH 7.5 (Raman buffer) to remove excess heme and cyanide, respectively.
The monoheme-CN MhuD sample was placed in a septumsealed glass vial and degassed under carbon monoxide gas, while stirring, for 20 min before reduction by approximately threefold molar excess of freshly prepared sodium dithionite (50 mM) with a gas-tight syringe. CO gas continued to flow over the sample while stirring for an additional 5 min before transferring the sample to a glove box under argon atmosphere where it was applied anaerobically to a Bio-Gel P6 column equilibrated with Raman buffer saturated with CO gas to remove the excess dithionite and displaced cyanide. The ferrous-CO adduct of MhuD was then oxidized back to ferric state by addition of 100-fold molar excess of potassium ferricyanide, which was then removed on a Bio-Gel P6 column equilibrated with Raman buffer. The same method was applied to prepare monoheme MhuD samples containing the isotopically labeled hemes, [bib_ref] Aggregation of ferrihaems. Dimerization and protolytic equilibria of protoferrihaem and deuteroferrihaem in..., Brown [/bib_ref] Fe-or [bib_ref] Spin state and axial ligand bonding in the hydroxide complexes of metmyoglobin,..., Feis [/bib_ref] Fe-PPIX.
To generate diheme MhuD samples, a fivefold molar excess of hemin in 100 mM potassium phosphate, 30 mM caffeine, pH 7.5 was added to the ferric monoheme MhuD samples prepared using the CN-CO replacement method. Caffeine was used to prevent formation of heme dimers in the titration solution [bib_ref] The hemeregulatory motif of nuclear receptor Rev-erbβ is a key mediator of..., Carter [/bib_ref] [bib_ref] Caffeine derivatives of haematin compounds, Gallagher [/bib_ref] [bib_ref] Ligand-binding in porphyrin systems, Gallagher [/bib_ref]. Excess heme was removed on a Bio-Gel P6 column equilibrated with Raman buffer. The same method was employed to prepare diheme MhuD samples containing the isotopically labeled hemes. To generate the 54 Fe 54 Fe-and [bib_ref] Spin state and axial ligand bonding in the hydroxide complexes of metmyoglobin,..., Feis [/bib_ref] Fe 54 Fe-diheme samples, [bib_ref] Aggregation of ferrihaems. Dimerization and protolytic equilibria of protoferrihaem and deuteroferrihaem in..., Brown [/bib_ref] Fe-PPIX was added in fivefold molar excess to the ferric [bib_ref] Aggregation of ferrihaems. Dimerization and protolytic equilibria of protoferrihaem and deuteroferrihaem in..., Brown [/bib_ref] Fe-and 58 Fe-monoheme samples, respectively. Similarly, the 58 Fe 58 Fe-and 54 Fe 58 Fe-diheme samples were prepared by adding [bib_ref] Spin state and axial ligand bonding in the hydroxide complexes of metmyoglobin,..., Feis [/bib_ref] Fe-PPIX to the ferric [bib_ref] Spin state and axial ligand bonding in the hydroxide complexes of metmyoglobin,..., Feis [/bib_ref] Feand [bib_ref] Aggregation of ferrihaems. Dimerization and protolytic equilibria of protoferrihaem and deuteroferrihaem in..., Brown [/bib_ref] Fe-monoheme samples, respectively. The eluted protein samples were analyzed by Bradford and pyridine hemochrome assays to determine protein and heme concentrations, respectively, as described in detail in Supporting information.
## Preparation of samples for rr measurements
All MhuD samples for rR measurements were concentrated to 150 to 200 μM protein concentration in Raman buffer (100 mM potassium phosphate, pH 7.5). In total, 100 μl of each sample was transferred to a Wilmad 5 mm Economy NMR tube for rR measurements. The samples for ferric state studies at different pH were exchanged into the appropriate buffers: 100 mM potassium phosphate pH 5.5 or 100 mM borate pH 9.1. To prepare the ferrous and ferrous-CO samples, NMR tubes containing ferric samples in Raman buffer were septum-sealed to ensure an anaerobic environment. Samples were degassed under flow of argon for 25 to 30 min. For ferrous-CO samples, the NMR tubes were then filled with the appropriate isotopes of CO gas. The samples were then reduced by addition of approximately threefold molar excess of anaerobically prepared 50 mM sodium dithionite using a gas-tight syringe and rR spectra were measured immediately.
## Resonance raman measurements
The ferric and Fe 2+ -CO adducts of MhuD were measured using the 406.7 nm and 413.1 nm excitation lines, respectively, obtained from an Innova 302C Kr + laser (Coherent Inc). The ferrous MhuD samples were excited using the 441.6 nm laser line from an IK5351R-D He-Cd laser (Kimmon Koha). Scattered light was collected using a 1250M-Series II Spectrometer (Horiba Scientific) fitted with a Pylon:400B CCD detector (Princeton Instruments). The width of the entrance slit on the spectrometer was 150 μm and a grating with 1200 grooves/ mm was used. The ferric and ferrous MhuD samples were measured using 10 mW laser power. The power on the Fe 2+ -CO MhuD samples was kept at 1 to 2 mW to prevent photodissociation of the CO ligand. The NMR tubes containing samples were spun throughout measurements to prevent localized heating, protein degradation, and photodissociation of CO adducts. Measurements were performed in the 180 backscattering geometry at room temperature using a cylindrical lens to focus the laser beam as a line image on the sample. Fenchone and acetone-D 6 were used as standards for calibration, and spectra were processed using GRAMS/32 AI software (Galactic Industries).
## Rr spectral deconvolution
The ν(Fe-C) stretching and δ(Fe-C-O) bending modes in the LF region of rR spectra of MhuD Fe 2+ -CO adducts overlap with several heme modes, which complicates the accurate determination of their frequencies. Similarly, the positions of the ν(C-O) stretching modes in the HF region are difficult to discern because they have relatively weak intensities. Therefore, in order to accurately determine the frequencies and extent of the isotopic shifts of these modes, the averaged spectra in the LF region, and difference patterns in the HF region were deconvoluted using a curve fitting procedure. The curve fitting application in GRAMS/32 AI software was employed, and the spectra were fitted with mixed 75%/25% Lorentzian/Gaussian functions. The overall peak fitting protocol was adapted from a previously published procedure [bib_ref] Defining CYP3A4 structural responses to substrate binding. Raman spectroscopic studies of a..., Mak [/bib_ref] and is described in detail in Supporting information.
## Uv-vis spectroscopy and activity assays
UV-vis electronic absorption spectra were measured between 200 and 800 nm with a 1.0 nm data interval and scan rate of 600 nm/min using an Agilent Cary 60 UV-Vis Spectrophotometer. Ferrous and ferrous-CO samples of MhuD, as well as free heme samples, were prepared by placing diluted ferric samples into septum-sealed quartz cuvettes and degassing them under flow of argon while stirring for 25 to 30 min. Samples were then reduced by titration with anaerobically prepared sodium dithionite until the Soret and Q-bands no longer changed in the ferrous spectra. Samples were then placed under flow of CO gas with stirring for 5 to 10 min before measuring the ferrous-CO spectra.
The ascorbate and POR activity assays were both performed in open air at ambient temperature (20 C) in 10 mm pathlength quartz cuvettes. The ascorbate assays were performed with 5 μM mono-or diheme MhuD in 100 mM potassium phosphate, pH 7.5 with 1250 units/ml catalase, 75 units/ml superoxide dismutase, 10 mM EDTA, and the reactions were initiated by addition of 10 mM sodium ascorbate. Spectra were measured every 10 or 15 min for 3.5 h. The ascorbate assay of 10 μM free hemin-chloride was performed under the same conditions, and spectra were measured every 15 min for 2 h. The NADPH-cytochrome P450 oxidoreductase (POR) activity assays were performed with 5 μM mono-or diheme MhuD in 100 mM potassium phosphate, pH 7.5 with 1250 units/ml catalase, 75 units/ml superoxide dismutase, 10 mM EDTA, 100 μM NADPH, and reactions were initiated by addition of 50 nM POR. Spectra were measured intermittently over the course of 7 h.
## Esi-ms of heme degradation products
To analyze the crude products of heme degradation by MhuD samples, turnover reactions were performed with slight modifications to facilitate the reaction rates and minimize product decomposition prior to analysis. The reaction mixtures contained 10 μM mono-or diheme MhuD in 50 mM potassium phosphate pH 6.0 with 1500 units/ml catalase, 565 units/ml superoxide dismutase, 10 mM EDTA, and reactions were initiated by addition of 10 mM sodium ascorbate. Reactions were incubated at 37 C and protected from light. After halting reactions by addition of HCl, the products were extracted into dichloromethane and washed three times with deionized water. The crude products were evaluated via ESI-MS. Samples were analyzed by direct injection and infused to the MS with 90% MeCN with 0.1% formic acid and 10% water with 0.1% formic acid (v/v) using a Thermo Vanquish high performance liquid chromatography system (HPLC). ESI-MS data were acquired using a Thermo Q-Exactive Orbitrap operated in full scan, positive ion mode, with a scan range of 560 to 640 m/z and 140,000 mass resolution. High-resolution MS data were analyzed using Xcalibur Qual Browser, where extracted ion currents (EICs) were obtained (using a 5 ppm mass tolerance) for m/z values corresponding to biliverdin (583.2541 m/z), mycobilin (611.2489 m/z), and heme (616.1766 m/z).
## Data availability
All data are contained within this article.
Supporting information-This article contains supporting information .
[fig] Figure 1: Structures of heme and its degradation products achieved through different heme oxygenase pathways. The canonical HO product, α-biliverdin is labeled in red. The noncanonical HO products mycobilin (mycobilin-a) from MhuD and staphylobilin (5-oxo-δ-bilirubin) from IsdG/I pathways are labeled in blue and green, respectively. [/fig]
[fig] Figure 2: X-ray crystal structures showing the active sites of monoheme and diheme forms of MhuD. Panel A shows the 6-coordinated low spin ferric-CN ligated monoheme MhuD structure (PDB ID 4NL5) with the protein secondary structure displayed in light blue [/fig]
[fig] Figure 3: UV-vis spectra of mono-and diheme MhuD and free heme in different oxidation and ligation states at pH 7.5. Spectra are shown for apo-(gray), monoheme (red), and diheme (blue) MhuD in the ferric state (A), as well as monoheme MhuD (red), diheme MhuD (blue), and free heme (green) in the ferrous (B) and ferrous-CO ligated (C) states. Spectra are normalized to the Soret peak. [/fig]
[fig] Figure 4: rR spectra of mono-and diheme MhuD in the ferric state at pH 7.5. Shown are the HF (top) and LF (bottom) regions of ferric monoheme (A and C) and diheme (B and D) MhuD measured with 406.7 nm excitation line at laser power of 10 mW. [/fig]
[fig] Figure 5: rR spectra of mono-and diheme MhuD in the ferrous state. Shown is the LF region of 56 Fe-PPIX-bound monoheme (A) and diheme (B) MhuD at pH 7.5 measured with 441.6 nm excitation line at laser power of 10 mW. The ν(Fe-N His ) mode is expanded for mono-(i-iii) and diheme (iv-vii) MhuD samples bound with different combinations of 54 Fe and 58 Fe isotopically labeled hemes. [/fig]
[fig] Figure 6: Deconvoluted rR spectra of ferrous-CO adducts of mono-and diheme MhuD and free heme. Shown is 56 Fe-monoheme MhuD with 12 C 16 O (A), 13 C 16 O (B), 13 C 18 O (C) isotopes, 56 Fe 56 Fe-diheme MhuD with 12 C 16 O (D), 13 C 16 O (E), 13 C 18 O (F) isotopes, and free 56 Fe-PPIX with 12 C 16 O (G), 13 C 16 O (H), 13 C 18 O (I) isotopes. The insets in (A), (D), and (G) show the 12 C 16 O -13 C 18 O difference spectrum in the HF region for each respective sample. In each spectrum the black line represents the experimental data, and the red line shows the trace fitted with a mixed 75%/25% Lorentzian/Gaussian function. The black and white fitted peaks with dotted lines represent heme modes and δ(Fe-C-O) modes. The ν(Fe-C) and ν(C-O) modes for Fe-C-O conformers A-C are represented by dark, medium, and light blue peaks, respectively. [/fig]
[fig] Figure 7: Inverse correlation plot of ν(Fe-C) and ν(C-O) frequencies for ferrous-CO adducts of heme proteins. Shown are MhuD conformers (dark, medium, and light blue circles), ratHO-1 (29), IsdG/I (dark green triangles) (32), H25A hHO-1 (orange square) (36), myoglobin variants (light green triangles)(26), and various cytochrome P450s (red diamonds) and heme model compounds (yellow squares)(27). [/fig]
[fig] Figure 8: Deconvoluted rR spectra of ferrous-CO adducts of diheme MhuD samples containing mixed heme Fe isotopes. Shown are 54 Fe 58 Fe-diheme MhuD with 12 C 16 O (A) and 13 C 18 O (B) isotopes, and 58 Fe 54 Fe-diheme MhuD with 12 C 16 O (C) and 13 C 18 O (D) isotopes. The insets in (A) and (C) show the 12 C 16 O -13 C 18 O difference spectrum in the HF region for each respective sample. For each spectrum, the black line represents the experimental data, and the red line represents the trace fitted with a mixed 75%/25% Lorentzian/Gaussian function. The black and white fitted peaks with dotted lines represent heme modes and δ(Fe-C-O) modes. The ν(Fe-C) and ν(C-O) modes are color-coordinated according to the isotope of heme iron, with54 Fe-CO conformers represented by dark, medium, and light green peaks, and 58 Fe-CO conformers by violet, magenta, and red peaks. [/fig]
[fig] Figure 9: UV-vis absorption spectra of ascorbate activity assays for mono-and diheme MhuD. The monoheme (A) and diheme (B) MhuD assays were carried out at room temperature. Reaction mixtures contained 5 μM protein concentration of mono-or diheme MhuD in 100 mM potassium phosphate, pH 7.5 with 1250 units/ml catalase, 75 units/ml superoxide dismutase, 10 mM EDTA, and the reactions were initiated by addition of 10 mM sodium ascorbate. [/fig]
[table] Table 1: Data summary for ferrous-CO ligated mono-and diheme MhuD and free heme with isotopically labeled hemes [/table]
|
Genetic Interaction Analysis of TCF7L2 for Biochemical Recurrence after Radical Prostatectomy in Localized Prostate Cancer
Backgroud: Accumulated evidence has demonstrated a significant role of the Wnt pathway in human prostate cancer. We hypothesize that genetic variants in the Wnt pathway effector, Transcription factor 7-like 2 (TCF7L2), may influence clinical outcomes in prostate cancer. Methods: We comprehensively selected 12 tagged single-nucleotide polymorphisms (SNPs) to capture majority of common variants across TCF7L2, and genotyped in 458 localized prostate cancer patients treated with radical prostatectomy (RP). Kaplan-Meier analysis, Cox proportional hazard model, and survival tree analyses were performed to identify significant SNPs that correlated with biochemical recurrence (BCR) after surgery. Results: A higher-order SNP-SNP interaction profile consisting of TCF7L2 rs7094463, rs10749127, and rs11196224 was significantly associated with BCR (P trend = 0.001). After adjusting for possible confounders, the genetic profile remained significant (P trend = 0.007). None of the studied SNPs were individually associated with BCR. Conclusions: Our results support a genetic interaction in the TCF7L2 SNPs as a predictor of disease recurrence after curative RP in localized prostate cancer patients.
# Introduction
Transcription factor 7-like 2 (TCF7L2), also known as TCF4, is an important effector of the canonical Wnt signaling pathway. Activation of the Wnt pathway leads to an increase of β-catenin stabilization in the cytoplasm and subsequent accumulation in the nucleus, where β-catenin acts as coactivator for the T Ivyspring International Publisher cell factor/lymphoid enhancer-binding factor (TCF/LEF) transcription factor family to stimulate transcription of numerous target genes involved in cellular proliferation, apoptosis, and invasion [bib_ref] Wnt signalling in stem cells and cancer, Reya [/bib_ref]. Dysregulation of the Wnt pathway is highly associated with cancer initiation and progression, including prostate cancer [bib_ref] Aberrant expression of the growth factor Wnt-5A in human malignancy, Iozzo [/bib_ref]. The growth of both normal cells in the prostate gland and prostate cancer cells relies on androgen/androgen receptor (AR) signals. It was recently shown that AR is also a target gene for the TCF7L2/β-catenin complex [bib_ref] Complex regulation of human androgen receptor expression by Wnt signaling in prostate..., Yang [/bib_ref]. Considering the crosstalk between Wnt and AR signaling pathways, together with the growth regulatory role of TCF7L2, we evaluated the influence of the genetic variants in TCF7L2 on disease recurrence in localized prostate cancer patients receiving curative radical prostatectomy (RP).
# Materials and methods
## Patient recruitment and data collection
This study included 458 Taiwanese patients who underwent RP as initial therapy for localized prostate cancer, as described previously [bib_ref] Individual and cumulative association of prostate cancer susceptibility variants with clinicopathologic characteristics..., Bao [/bib_ref] [bib_ref] Prognostic significance of prostate cancer susceptibility variants on prostate-specific antigen recurrence after..., Huang [/bib_ref] [bib_ref] Clinical significance of runt-related transcription factor 1 polymorphism in prostate cancer, Huang [/bib_ref] [bib_ref] Prognostic significance of cyclin D1 polymorphisms on prostate-specific antigen recurrence after radical..., Yu [/bib_ref]. Briefly, patients diagnosed with histologically confirmed prostate cancer were recruited from four medical centers in Taiwan: National Taiwan University Hospital, Kaohsiung Medical University Hospital, E-Da Hospital, and Kaohsiung Veterans General Hospital. Demographic, clinical, and follow-up data were obtained from the medical records. Biochemical recurrence (BCR) was defined as two consecutive prostate-specific antigen (PSA) values of at least 0.2 ng/mL [bib_ref] Defining the ideal cutpoint for determining PSA recurrence after radical prostatectomy. Prostate-specific..., Freedland [/bib_ref] [bib_ref] Genetic variants in microRNAs and microRNA target sites predict biochemical recurrence after..., Huang [/bib_ref]. All participants provided written consent, and the local ethics committees approved the research protocol.
## Single-nucleotide polymorphisms (snp) selection and genotyping
Genomic DNA was extracted from peripheral blood with the QIAamp DNA Blood Maxi Kit (Qiagen, Valencia, CA, USA) according to the manufacturer's protocol, and stored until the time of study. We utilized a tagging SNP approach to investigate all the genetic variability in the TCF7L2. Tagging SNPs were selected using the Tagger algorithm available through Haploview, using pairwise SNP selection with r 2 ≥0.8 and minor-allele frequencies (MAF) ≥0.2 from the HapMap population data for Han Chinese in Beijing, China (CHB) [bib_ref] Efficiency and power in genetic association studies, De Bakker [/bib_ref] [bib_ref] A second generation human haplotype map of over 3.1 million SNPs, International Hapmap [/bib_ref]. We identified 15 tagging SNPs for TCF7L2, but three SNPs that failed at Sequenom assay design were excluded. Genotyping was carried out at the National Center for Genome Medicine, Taiwan, using the Sequenom iPLEX matrix-assisted laser desorption/ionization time-of-flight mass-spectrometry technology. All 12 SNPs were in Hardy-Weinberg equilibrium (P > 0.05) and had average genotyping call rate ≥0.93. For quality control, we randomly selected 10 samples for duplicates, and the concordance rate was >0.99 for all SNPs assayed.
# Statistical analysis
Patient clinicopathologic characteristics were summarized as either the numbers and percentages of patients, or the median and interquartile range of values. The association between patient characteristics with BCR was assessed by the log-rank test. Individual SNPs were initially assessed using the log-rank test for the three genetic models of inheritance: dominant (common homozygotes versus variant allele carrying genotypes), recessive (common allele carrying genotypes versus variant homozygotes), and additive (P for trend). Higher order SNP-SNP interactions were evaluated using survival tree analysis by STREE software (http://c2s2.yale.edu/software/ stree/), which uses recursive partitioning to identify subgroups of patients with similar risk of disease recurrence. Kaplan-Meier analysis with log-rank test was then used to estimate the survivals between each of the terminal subgroups and categorized into low-, medium-, and high-risk groups. Multivariate Cox proportional hazards regression analyses were used to assess the effect of genetic interaction profile in TCF7L2 on BCR, with or without adjusting for known prognostic factors, including age, PSA at diagnosis, pathologic Gleason score, stage, surgical margin, and lymph node metastasis, as previously described [bib_ref] Prognostic significance of prostate cancer susceptibility variants on prostate-specific antigen recurrence after..., Huang [/bib_ref]. The Statistical Package for the Social Sciences software, version 22.0.0 (IBM, Armonk, NY, USA), was used for other statistical analyses. A two-sided P value of <0.05 was considered statistically significant.
# Bioinformatics analysis
HaploReg v2 [bib_ref] HaploReg: a resource for exploring chromatin states, conservation, and regulatory motif alterations..., Ward [/bib_ref] and the Encyclopedia of DNA Elements (ENCODE) [bib_ref] ENCODE data in the UCSC Genome Browser: year 5 update, Rosenbloom [/bib_ref] data were used to identify the regulatory potential of the region adjoining the SNPs.
# Results
We identified 184 (40.2%) patients experienced BCR in localized prostate cancer receiving RP during a median follow-up time of 54 months [fig_ref] Table 1: Clinical characteristics of the study cohortAbbreviations [/fig_ref]. PSA at diagnosis, pathologic Gleason score, stage, surgical margin, and lymph node metastasis, were all significantly associated with BCR (P < 0.001).
Of the 12 tagged SNPs we analyzed to capture majority of the variants in TCF7L2, none of the SNPs showed noteworthy associations with BCR under additive, dominant, and recessive models using the log-rank tests (all P > 0.05, Supplementary Material: [fig_ref] Table 1: Clinical characteristics of the study cohortAbbreviations [/fig_ref]. Therefore, higher order SNP-SNP interactions in modulating the risk of disease recurrence were further explored by survival tree analysis. We identified three tagged SNPs, rs11196224, rs7094463, and rs10749127, potentially having interactions, and the resulting tree structure was comprised of four terminal groups with low-, medium-, and high-risk of BCR according to the log-rank tests [fig_ref] Figure 1: Higher order SNP-SNP interactions among TCF7L2 polymorphisms for BCR in localized prostate... [/fig_ref]. The median BCR-free survival of patients at low risk has not been reached during the follow-up. In comparison, median survival time was 82 months in the medium-risk group and the hazard ratio (HR) was 10.5 [95% confidence interval (CI) 1.47-75.0, P = 0.019, Table 2 and [fig_ref] Figure 1: Higher order SNP-SNP interactions among TCF7L2 polymorphisms for BCR in localized prostate... [/fig_ref] ]. The median survival time was only 53 months for high-risk patients and the HR was 16.1 (95% CI 2.06-125, P = 0.008; P for trend = 0.001).
In multivariate analysis, adjusting for age at diagnosis, PSA, pathologic Gleason score, stage, surgical margin, and lymph node metastasis, the genetic interaction profile remained significant. In comparison to the low-risk group, the medium-risk group presented a 6.29-fold increased risk of disease progression (95% CI 0.87-45.4, P = 0.068, [fig_ref] Table 2: Cox proportional hazards analysis of TCF7L2 genetic interaction profiles with BCR in... [/fig_ref] , and the high-risk group had a 10.5-fold increased risk (95% CI 1.34-83.0, P = 0.025; P for trend = 0.007). These data indicated that the genetic interaction profile in TCF7L2 provided additional predictive information beyond the conventional risk factors to influence prostate cancer outcomes. Functional annotations from the ENCODE data for all correlated variants within the linkage disequilibrium (LD) blocks (r 2 >0.8) containing the three interacting SNPs, rs7094463, rs10749127, and rs11196224, are shown in [fig_ref] Figure 2: Expanded view of the ENCODE data for the LD blocks containing the... [/fig_ref]. The rs7094463 and seven additional linked SNPs are situated at a locus with histone modification patterns characteristic of promoter in several cell types. In addition, the regulatory motif of forkhead box (Fox) transcription factors was predicted to be altered by rs7094463 [fig_ref] Figure 2: Expanded view of the ENCODE data for the LD blocks containing the... [/fig_ref] and Supplementary Material: [fig_ref] Table 2: Cox proportional hazards analysis of TCF7L2 genetic interaction profiles with BCR in... [/fig_ref]. The tagged SNP rs10749127 and several rs11196224-linked SNPs are situated at a locus with histone modification patterns characteristic of enhancers, and possibly alter multiple regulatory motifs [fig_ref] Figure 2: Expanded view of the ENCODE data for the LD blocks containing the... [/fig_ref] and Supplementary Material: and S4). This suggests that these variants are theoretically functional and might explain the association of TCF7L2 with disease progression.
# Discussion
TCF7L2 is the effector of the Wnt signaling pathway, and it combines with β-catenin to form an active nuclear complex that induces the expression of target genes involved in cellular proliferation, apop-tosis, and invasion. Many studies have identified the association between TCF7L2 SNPs, rs7903146 or rs12255372 (in LD with each other), and the risk of several types of cancer [bib_ref] Transcription factor 7-like 2 (TCF7L2) variant is associated with familial breast cancer..., Burwinkel [/bib_ref] [bib_ref] Variation in TCF7L2 and increased risk of colon cancer: the Atherosclerosis Risk..., Folsom [/bib_ref] [bib_ref] No association between a candidate TCF7L2 variant and risk of breast or..., Goode [/bib_ref] [bib_ref] Association of the TCF7L2 polymorphism with colorectal cancer and adenoma risk, Hazra [/bib_ref] , including prostate cancer [bib_ref] Evaluation of a variant in the transcription factor 7-like 2 (TCF7L2) gene..., Agalliu [/bib_ref]. However, these SNPs have a MAF of <0.05 in Asian populations, compared to >0.25 in other ethnic groups, thus limiting power to detect an association in Asian patients. In this study, we comprehensively selected 12 tagged SNPs to capture common genetic variability (MAF ≥0.2) in the TCF7L2, and determined their prognostic values. We showed that a genetic interaction profile consisting of the TCF7L2 rs7094463, rs10749127, and rs11196224 correlates with disease recurrence in prostate cancer patients treated with RP. The genetic interaction analysis takes the complexity of interacting SNPs into account, and this approach provides a novel way to use genetic variants in TCF7L2 to predict prostate cancer outcomes.
ENCODE data indicated that rs7094463 is located at a locus with histone modification patterns characteristic of promoter and a RNA polymerase II binding region [fig_ref] Figure 2: Expanded view of the ENCODE data for the LD blocks containing the... [/fig_ref] and Supplementary Material: [fig_ref] Table 2: Cox proportional hazards analysis of TCF7L2 genetic interaction profiles with BCR in... [/fig_ref]. rs10749127 and rs11196224 coincide with regions of open chromatin, which probably corre-spond to the enhancers of TCF7L2 [fig_ref] Figure 2: Expanded view of the ENCODE data for the LD blocks containing the... [/fig_ref] and Supplementary Material: . In addition, multiple regulatory motifs were predicted to be altered by these SNPs. Therefore, it is plausible that these SNPs might influence TCF7L2 expression by altering the transcription factor binding sites. Further mechanistic studies are necessary to determine whether these significant SNPs have functional activity in the Wnt pathway or in the clinical outcome of prostate cancer patients.
In conclusion, this is the first study to explore the interaction between TCF7L2 SNPs in relation to the clinical outcomes for prostate cancer. The genetic interaction analysis relies on the data mining to identify the best model for the data, potentially leading to that the optimal results only showed in the initial test cohort. However, testing for the interaction among SNPs is probably more rational than testing for each individual SNP since the response to treatment is a complex phenomenon. If validated in independent studies, our results might be applicable to future modeling of clinical outcomes for prediction of disease recurrence in localized prostate cancer patients. [fig_ref] Table 1: Clinical characteristics of the study cohortAbbreviations [/fig_ref]. http://www.medsci.org/v12p0243s1.pdf Abbreviations TCF7L2, transcription factor 7-like 2; BCR, biochemical recurrence; RP, radical prostatectomy; SNP, single-nucleotide polymorphism; HR, hazard ratio; CI, confidence interval; PSA, prostate-specific antigen.
# Supplementary material
[fig] Figure 1: Higher order SNP-SNP interactions among TCF7L2 polymorphisms for BCR in localized prostate cancer patients. (A) Survival tree analysis identifies the interactions among the three polymorphisms. (B) Kaplan-Meier curves of BCR-free survival based on the survival tree analysis. Numbers in parentheses indicate the number of patients. [/fig]
[fig] Figure 2: Expanded view of the ENCODE data for the LD blocks containing the three interacting SNPs. ENCODE data showed evidence of promoter/enhancer elements coinciding with the variants linked with the three interacting SNPs, rs7094463, rs10749127, and rs11196224. The H3K4Me1, H3K4Me3, and H3K27Ac tracks show the genome-wide levels of enrichment of the mono-methylation of lysine 4, tri-methylation of lysine 4, and acetylation of lysine 27 of the H3 histone protein, as determined by the ChIP-seq assays. These levels are thought to be associated with enhancer and promoter regions. Chromatin State Segmentation track displays chromatin state segmentations by integrating ChIP-seq data using a Hidden Markov Model for eight different cell types. The chromatin state regions predicted for promoters and enhancers are highlighted. DNase clusters track shows DNase hypersensitivity areas. Tnx Factor track shows regions of transcription factor binding of DNA, as assayed by ChIP-seq experiments. [/fig]
[table] Table 1: Clinical characteristics of the study cohortAbbreviations: IQR, interquartile range; PSA, prostate-specific antigen; CI, confidence interval. *P value was calculated by the log-rank test for disease recurrence. †Median follow-up time and 95% CIs were estimated with the reverse Kaplan-Meier method. [/table]
[table] Table 2: Cox proportional hazards analysis of TCF7L2 genetic interaction profiles with BCR in localized prostate cancer patients treated with RP Abbreviations: BCR, biochemical recurrence; RP, radical prostatectomy; NR, not reached; HR, hazard ratio; CI, confidence interval. *HRs were adjusted for age, PSA, Gleason score, stage, surgical margin, and lymph node metastasis. P < 0.05 are in boldface. [/table]
|
DE-STRESS: a user-friendly web application for the evaluation of protein designs
De novo protein design is a rapidly growing field, and there are now many interesting and useful examples of designed proteins in the literature. However, most designs could be classed as failures when characterised in the lab, usually as a result of low expression, misfolding, aggregation or lack of function. This high attrition rate makes protein design unreliable and costly. It is possible that some of these failures could be caught earlier in the design process if it were quick and easy to generate information and a set of high-quality metrics regarding designs, which could be used to make reproducible and data-driven decisions about which designs to characterise experimentally. We present DE-STRESS (DEsigned STRucture Evaluation ServiceS), a web application for evaluating structural models of designed and engineered proteins. DE-STRESS has been designed to be simple, intuitive to use and responsive. It provides a wealth of information regarding designs, as well as tools to help contextualise the results and formally describe the properties that a design requires to be fit for purpose.
# Introduction
There has been rapid development in the field of de novo protein design over recent years, with more groups producing increasingly ambitious designs with complex behaviour, often applied in cellular environments.
Despite the great promise of de novo protein design, it remains the domain of highly specialist research groups, as there are significant barriers blocking broader adoption as a methodology. One major challenge is that only a fraction of designs adopt stable, folded structures when expressed, and it can be difficult to identify these models using the metrics calculated during the design process alone. This is especially challenging for designs with complex requirements that are needed for targeted applications.
Here, we present DE-STRESS (DEsigned STRucture Evaluation ServiceS), a user-friendly web application for evaluating structural models of designed and engineered proteins. We aim to provide the user with as much information as possible about their designs before they select sequences to characterise experimentally.
# Methods and results
The DE-STRESS application consists of a simple and intuitive user-interface, written in Elm/JavaScript, and a backend web stack, consisting of Gunicorn/Flask/GraphQL/Post-greSQL. The interface has three main sections that the user can explore: designs, reference sets and specifications.
On the Designs page, users can upload models of proteins (in PDB format) to the DE-STRESS server, where all the included metrics will be calculated for each design. Once the metrics have been calculated, an overview of the whole batch of designs is provided. Detailed information can be viewed for each design, as well as a comparison to the active Reference Set and Requirement Specification (vide infra).
On the Reference Sets page, users can define a set of known protein structures from the PDB, which can be used as a basis of comparison for their designs. We have precalculated the metrics included in DE-STRESS for the biological units, as defined by the PDBe (http://ftp.e bi.ac.uk/pub/databases/pdb/data/biounit/), of 82 010 protein structures. The remaining structures in the PDB either did not contain protein, contained formatting errors in the PDB file or, in the case of large structures, failed to return results within a reasonable timeframe. Similar restrictions have been placed on models that can be uploaded to the webserver, in order to ensure server stability. Using these data, the user can define their own reference sets by submitting a list of PDB accession codes, enabling them to compare their designs to relevant structures. Additionally, two default reference sets are provided as an example, based on high-quality structures from Top500and. Finally, users can also define reference sets from models that they have uploaded to the server, allowing the creation of reference sets from unpublished structures.
Once a reference set has been defined, aggregated metrics are presented alongside the metrics for the user's designs. All the data used to generate the reference sets are available to search and download, programmatically and interactively, through a GraphQL API available at the following url: https:// pragmaticproteindesign.bio.ed.ac.uk/big-structure/graphql.
Reference sets are intended to provide context for the metrics returned by DE-STRESS, and while we have provided some example reference sets of high-quality crystal structures, manually defining a reference set is much more useful. For example, if you are designing coiled-coil proteins, comparing your designs to similar structures is likely to be more useful than a general set of proteins, as the metrics and other information, such as sequence composition, will be directly comparable. Furthermore, it is worth noting that designs with metrics that fall outside of the range observed for known protein structures may still express, fold and be functional. In certain instances, the designer may be actively selecting designs with metrics that fall outside of those observed for natural proteins; in fact, this is very often the purpose of designing proteins de novo.
The specifications page allows the user to define 'Requirement Specifications', which encapsulate the properties their designs should have in order to be fit for purpose. The user can define complex rules, such as nested conditional properties, that can be used to filter designs, alongside associated metadata. We plan to expand the role of the specifications in the future, allowing the user to capture more information about their design intent and export the specification to be used by other programmes.
A variety of external software packages are used by the DE-STRESS web server to calculate metrics for uploaded models, with detailed and up-to-date information on the version and command used to run the software provided in the supplementary material and glossary page of the application. Basic information is extracted using ISAMBARD, such as the isoelectric point and composition of the sequence, as well as implementations of metrics from the literature, such as packing densityand hydrophobic fitness. In addition to these metrics, DE-STRESS applies a range of scoring functions that have either been developed specifically for protein design or have been applied to design proteins, such as BUDE FF, EvoEF2, Rosettaand DFIRE2. Finally, we use Aggrescan3Dto calculate an aggregation propensity score for protein structures, as aggregation is a common failure mode for designed and engineered proteins. For detailed information regarding the metrics included see supplementary tables 2-11. These metrics are presented on the Design Details page, alongside a visualisation of the model, using the NGL JavaScript library, and other information such as secondary structure assignment using DSSP.
A privacy first approach has been taken when implementing DE-STRESS. No login is required to use the application and no data regarding the user, or their designs, are stored on our server. Designs are submitted directly to an in-memory The scoring convention indicates whether the score is considered more favourable if it is lower (−ve) or higher (+ve).
job queue, with no associated metadata, and the results are returned directly to the user. All data regarding the user's designs are stored locally on the device used to access the website and can be exported to a CSV file for further analysis.
With this architecture, we aim to give the user confidence in submitting their designs to the server. However, if they would like to take further steps to ensure that no one could access their data, they can run a local instance of the web application, which we have made as simple as possible by containerising the application. We envisage that, for many users, DE-STRESS will be useful for generating descriptive information and statistics that could be manually examined to choose designs that meet the needs of their application. Beyond this, the datasets that DE-STRESS creates could be useful for automatic identification of highquality models using data-driven methods. As a simple example of this, we attempted to discern experimentally determined structures from decoys that are used to benchmark proteinfolding algorithms. The dataset and the associated scripts for performing this analysis are available on GitHub: https://githu b.com/wells-wood-research/stam-m-wood-c-de-stress-2021.
Using the DE-STRESS web application, we generated and exported metrics for a random sample of 9 experimentallydetermined structures, along with 360 decoys (40 per structure) from a set that was previously generated by 3DRobot. 3DRobot produces structurally diverse and well-packed decoy structures that are difficult to discern from experimentally determined structures with simple metrics, such as percentage of secondary structure or radius of gyration, as well as more complex metrics such as statistical potentials. We also included alternative structures for each of these proteins, identified through the structural similarity search tool on the RSCB PDB, to ensure that the findings were robust to minor variations in the structure.
Firstly, before performing PCA, various metrics were excluded from the data set. Categorical and discrete variables were excluded as PCA requires continuous variables, and metrics that were constant for a structure were excluded. This is because these metrics provided no information that could be used to distinguish between the decoys and experimentally determined structures. In addition to this, as the energy values are dependent on the size of the protein structure, we normalised for length by dividing all the energy values by the number of residues in the structure. A full list of the metrics included in the analysis is shown in supplementary table 12. Before performing PCA, the metrics were normalised using min-max normalisation. The first two principal components, which explained 60% of the variance in the data, were plotted against each other to explore how well the DE-STRESS metrics differentiated the experimentally determined structures from the decoys. The additional crystallographic structures were also included in these plots.
From the plots in, we see that the experimentally determined structures, shown as stars, are generally distinct from the decoys and usually have the largest values for both principal components 1 and 2. In addition to this, the additional crystallographic structures, shown as squares, are close to the experimentally determined structures used to generate the 3DRobot set. This analysis suggests that the DE-STRESS metrics are capturing properties of the structural models that can be used to distinguish 'native-like' structures from decoys.
To understand which metrics differentiated the experimentally determined structures from the decoys, we examined the relative contribution of individual DE-STRESS metrics to the first two principal components. It was found that the major contributors to PC1 were the short-range and long-range hydrogen bonding terms from the Rosetta forcefield and the total BUDE forcefield energy value. The major contributors to PC2 were the average score of Aggrescan3D, the steric component of the BUDE forcefield and a solvation term from the Rosetta forcefield. Longrange hydrogen bonding was highlighted by the developers of 3DRobot as a key differentiating feature between native structures and decoys, and maintaining this property was key to creating native-like decoys. However, while analysis performed in the 3DRobot publication shows that their decoys performed significantly better in this regard than previous decoy sets, our results indicate that there is still some detectable difference between the experimentally determined structures and the decoys. The fact that Aggres-can3D is a key contributor to PC2 might indicate that the surface properties of the decoys are significantly different from the native structures, and this could indicate a potential route to improve the process used to create decoy models.
# Conclusions
DE-STRESS enables both non-experts and seasoned protein designers to rapidly evaluate their designs, providing a framework for making reproducible, data-driven decisions about which designs to take forward for experimental characterisation. While some protocols and applications have been developed to meet similar needs as DE-STRESS , none of them have the same breadth of metrics and tools, all packaged in a user-friendly web application. However, the metrics included in the initial version of DE-STRESS are certainly not exhaustive: general methods for assessing the quality of a protein model, as well as more specialised methods for assessing design quality, such as analysis of covariation similarity with homologs, could be useful additions in future releases. We believe that our analysis of decoy structures demonstrates that the metrics included in this initial version of DE-STRESS are useful and can be used to identify high-quality models, but further experimental work is required to determine whether this translates into a reduced failure rate of designs taken into the lab, which would meet our ultimate aim of increasing the efficiency of the protein-design process and making it more accessible and reliable as a technique.
## Availability
DE-STRESS is available for non-commercial use, without registration, through the following website: https://pragma ticproteindesign.bio.ed.ac.uk/de-stress/. Source code for the application is available on GitHub: https://github.com/wellswood-research/de-stress. The data used to generate reference sets is available through a GraphQL API, with the following URL: https://pragmaticproteindesign.bio.ed.ac.uk/big-structu re/graphql.
# Funding
This work is supported by an Engineering and Physical Sciences Research Council Fellowship (EP/S003002/1 to C.W.W.); the United Kingdom Research and Innovation (grant EP/S02431X/1 to M.S.), UKRI Centre for Doctoral Training in Biomedical AI at the University of Edinburgh, School of Informatics; the Wellcome Trust-University of Edinburgh Institutional Strategic Support Fund (ISSF3).
## Supplementary data
Supplementary data are available at PEDS online. |
High abundance of virulence gene homologues in marine bacteria
Marine bacteria can cause harm to single-celled and multicellular eukaryotes. However, relatively little is known about the underlying genetic basis for marine bacterial interactions with higher organisms. We examined whole-genome sequences from a large number of marine bacteria for the prevalence of homologues to virulence genes and pathogenicity islands known from bacteria that are pathogenic to terrestrial animals and plants. As many as 60 out of 119 genomes of marine bacteria, with no known association to infectious disease, harboured genes of virulence-associated types III, IV, V and VI protein secretion systems. Type III secretion was relatively uncommon, while type IV was widespread among alphaproteobacteria (particularly among roseobacters) and type VI was primarily found among gammaproteobacteria. Other examples included homologues of the Yersinia murine toxin and a phage-related 'antifeeding' island. Analysis of the Global Ocean Sampling metagenomic data indicated that virulence genes were present in up to 8% of the planktonic bacteria, with highest values in productive waters. From a marine ecology perspective, expression of these widely distributed genes would indicate that some bacteria infect or even consume live cells, that is, generate a previously unrecognized flow of organic matter and nutrients directly from eukaryotes to bacteria.
# Introduction
Pathogenic bacteria have a profound impact on plants, animals and humans. Thus, understanding the origin and distribution of genes that convey pathogenicity is a central issue. These virulence genes represent virulence factors that are used by the bacteria for attachment to and degradation of eukaryotic cells. In particular, bacterial protein secretion systems are crucial for pathogenicity as they allow export of virulence factors. The molecular maps that describe the interaction between bacterial pathogens and their hosts have largely been generated through research on infectious disease and its causative agents. The extent to which homologues to such virulence factors are present in environmental bacteria is less known [bib_ref] Virulence genes in environmental strains of Vibrio cholerae, Chakraborty [/bib_ref] [bib_ref] Genomic characterization of non-O1, non-O139 Vibrio cholerae reveals genes for a type..., Dziejman [/bib_ref] [bib_ref] Type VI secretion: a beginner's guide, Bingle [/bib_ref]. Interestingly, such genes are widespread in soil bacteria, where they supposedly allow bacteria to interact with eukaryotes. Analogously [bib_ref] Ecological genomics of marine roseobacters, Moran [/bib_ref] suggested that aquatic heterotrophic prokaryotes could prosper at the expense of marine eukaryotes using the same principles as known pathogens.
The typical genome composition of a pathogenic bacterium consists of a core genome shared with related commensals, yet augmented with varying genetic elements that appear to fall into a few strategic categories [bib_ref] Pathogenicity islands and their role in bacterial virulence and survival, Hochhut [/bib_ref] [bib_ref] Pathogenicity islands: a molecular toolbox for bacterial virulence, Gal-Mor [/bib_ref] , for example, coding for the capacity of adhesion to host surfaces through the expression of adhesins to enable colonization, for molecules that interfere with intrinsic host cell functions, and for the ability to deliver such molecules (effector proteins) upon contact with their cognate host target. Often, the virulence genes are contained on extended genetic continuums termed pathogenicity islands (PAIs, see [bib_ref] Pathogenicity islands in bacterial pathogenesis, Schmidt [/bib_ref] for review). The PAIs vary in size from a few kilobases to large chromosomal regions, and their base composition and codon usage often differ from the host genome. They are frequently integrated in proximity to tRNA or insertion sequences, or are flanked by direct repeats. This implies that PAIs can move between different bacterial species through horizontal gene transfer [bib_ref] Pathogenicity islands in bacterial pathogenesis, Schmidt [/bib_ref].
The original sources of virulence genes and PAIs, however, remain to be elucidated. It has been suggested that virulence genes in soil bacteria represent a reservoir for virulence genes found in human, animal, plant and insect pathogens. Similarly, one could expect that the long evolutionary history of marine bacteria, preceding terrestrial life by almost three billion years, would make them an interesting target in the search for virulence genes. Marine bacterioplankton experience a dynamic environment and show diverse interactions with other plankton organisms. For example, there are numerous examples of marine bacteria causing harm to algae (see [bib_ref] Algicidal bacteria in the sea and their impact on algal blooms, Mayali [/bib_ref] for a review). Recent comparative genome analyses of three members of the marine Roseobacter clade by have provided important insights into the lifestyles of bacteria in this clade. Similar to known pathogens, the Roseobacter clade genomes consist of a core set of genes and a much larger number of additional genes present in only a single or few genomes. Interestingly, a number of virulence genes were found in the Roseobacter genomes and these were suggested to permit roseobacters to 'directly capture organic matter from eukaryotic cells' [bib_ref] Ecological genomics of marine roseobacters, Moran [/bib_ref]. The authors in this case suggest a 'mix-and-match' genome arrangement that would allow adaptation to a wide diversity of ecological niches.
In this study, we carried out an extensive in silico analysis to examine the prevalence of homologues of virulence genes and PAIs, known from bacteria that are pathogenic to terrestrial animals and plants, in whole-genome sequenced marine bacteria. Further, the frequency of these genes in the ocean was estimated by analysis of the metagenome data set from the Global Ocean Sampling expedition [bib_ref] The Sorcerer II Global Ocean Sampling Expedition: Northwest Atlantic through Eastern Tropical..., Rusch [/bib_ref].
# Results and discussion
The search and identification of virulence gene homologues in marine bacteria were based on three approaches: (i) BLAST analysis to generate amino acid sequence similarity values, (ii) classification of proteins within PFAM, TIGRFAM or COG domain structure libraries and (iii) gene arrangement, that is, conservation of operon structure of defined PAIs (see Methods section for details). We found virulence genes in 60 out of 119 analysed bacteria [fig_ref] Table 1: Secretion systems and PAIs in genome sequenced marine bacteria [/fig_ref] and, in addition, we found numerous other genes and operons encoding putative adhesins, surface structures and protein secretion, and chemotactic sensory systems, which may also play important roles in marine bacterial behaviour.
## Protein secretion systems
Most pathogenic bacteria need to produce and export virulence factors, enzymes or toxins, in order to invade a host cell. To secrete nucleic acids or proteins through the outer membrane, Gram-negative bacteria apply specialized translocation systems. Predicted homologues to genes of known virulence-associated types III, IV, V and VI secretion systems were found in a large portion of the analysed marine bacteria [fig_ref] Table 1: Secretion systems and PAIs in genome sequenced marine bacteria [/fig_ref].
The type III secretion system (T3SS), or 'injectisome', exports proteins from the bacterial cytoplasm into eukaryotic cells, and helps mediating the interaction of bacterial symbionts or pathogens with their host [bib_ref] The type III secretion injectisome, Cornelis [/bib_ref]. The exported effector molecules can, for example, enable tissue invasion and/or inhibition of host defences [bib_ref] The bacterial injection kit: type III secretion systems, Mota [/bib_ref]. We found T3SSs in six marine bacteria [fig_ref] Table 1: Secretion systems and PAIs in genome sequenced marine bacteria [/fig_ref] , where putative T3SS genes were located in genome regions encompassing at least 17-43 genes (Sphingomonas sp. SKA58 and Vibrio alginolyticus 12G01 respectively). In type III secretion, specific ATPases are required for the actual secretion event; and the marine bacterial T3SS genome regions contained type III secretion ATPases with amino acid sequence similarities between 44% and 82% to the corresponding ATPase in Yersinia pestis CO92. A T3SS similar to that found in Y. pestis CO92 was found in V. alginolyticus 12G01, in accordance with previous reports [bib_ref] Functional characterization of two type III secretion systems of Vibrio parahaemolyticus, Park [/bib_ref]. The T3SSs are found in a wide variety of gammaproteobacteria, but appear less common in alphaproteobacteria [bib_ref] Bioinformatics, genomics and evolution of non-flagellar type-III secretion systems: a Darwinian perspective, Pallen [/bib_ref] [bib_ref] Type III secretion: more systems than you think, Troisfontaines [/bib_ref]. We found putative T3SSs in three alphaproteobacteria, where previously not described. One of these systems, found associated with a large plasmid in Sphingomonas sp. SKA58, apparently shared features typical for both T3SSs and the structurally related flagellar motor.
The type IV secretion system (T4SS) is involved in DNA and/or protein transfer into prokaryotic and eukaryotic cells, and accounts for translocation of virulence factors in a number of bacterial pathogens [bib_ref] The versatile bacterial type IV secretion systems, Cascales [/bib_ref]. Studies of the model plant pathogen Agrobacterium tumefaciens (alphaproteobacteria) have provided molecular detail on the VirB/D4 T4SS [bib_ref] Biogenesis, architecture, and function of bacterial type IV secretion systems, Christie [/bib_ref]. found putative T4SSs in seven genome-sequenced roseobacters (some of which were also included in the present study). Among the diverse marine bacteria studied here, putative T4SSs of the VirB/D4 type were exclusively found in alphaproteobacteria (18 genomes), with one to three vir gene operons per genome [fig_ref] Table 1: Secretion systems and PAIs in genome sequenced marine bacteria [/fig_ref]. This secretion system was amply distributed among members of the Roseobacter clade, but was also present in members of the families Erythrobacteraceae and Caulobacteraceae.
Virulence gene homologues in marine bacteria 1349
[formula] Leeuwenhoekiella blandensis MED217 Bact - - - - 1 - Microscilla marina ATCC 23134 Bact - - - - 1 - Algoriphagus sp. PR1 Bact - - - - 1 - Kordia algicida OT-1 Bact - - - - 1 - Blastopirellula marina DSM3645 Pla - - - 1 a - - Alphaproteobacterium BAL199 a - - - 2 - - Erythrobacter litoralis HTCC2594 a - 1 b 1 c - 1 - Erythrobacter sp. NAP1 a - 1 b 2 d - - - Erythrobacter sp. SD-21 a - 1 - - - - Sphingomonas sp. SKA58 a 1 1 - - - - Brevundimonas sp. BAL3 a - 1 1 - - - Nitrobacter sp. Nb311A a - - - - 1 - Fulvimarina pelagi HTCC2506 a - 1 - - - - Stappia aggregata IAM 12614 a 1 1 1 1 - - Pseudovibrio sp. JE062 a 1 - - 2 - - Roseobacter sp. SK209-2-6 a - - - 1 - - Oceanicola batsensis HTCC2597 a - 2 1 1 - - Rhodobacterales bacterium HTCC2654 a - 3 - - 1 - Roseovarius nubinhibens ISM a - 1 - - - - Roseovarius sp. TM1035 a - 1 - - - - Roseovarius sp. 217 a - 3 1 d 1 - - Roseobacter sp. MED193 a - 2 - 1 - - Roseobacter litoralis Och 149 a - - - 1 - - Phaeobacter gallaeciensis BS107 a - 1 - - - - Roseovarius sp. HTCC2601 a - - 1 2 - - Roseobacter sp. CCS2 a - - 1 - - - Rhodobacterales bacterium HTCC2083 a - - - 1 - - Sagittula stellata E-37 a - 1 - 1 - - Oceanibulbus indolifex HEL-45 a - 3 - - - - Sulfitobacter sp. NAS-14.1 a - 2 - - - - Roseobacter sp. GAI101 a - - - 1 - - Rhodobacterales bacterium Y4I a - - - 1 - - Rhodobacterales KLH11 a - 1 3 - 1 - Pelagibacter ubique HTCC1062 a - - - - - - Methylophilales bacterium HTCC2181 b - - 1 d - - - Betaproteobacteria sp. KB13 b - - 1 - - - Marinomonas sp. MED121 g - - - 2 - - Marinobacter sp. DG893 g - - - 1 - - Marinobacter sp. ELB17 g - - - 1 - - Methylophaga sp. DMS010 g - - 1 - - - Alcanivorax sp. DG881 g - - 1 1 - - Limnobacter sp. MED105 g 1 - 3 1 - - Moritella sp. PE36 g - - - 2 - - Nitrococcus mobilis Nb-231 g - - - 1 1 - Oceanobacter sp. RED65 g - - 2 1 - - Gammaproteobacterium HTCC2080 g - - - 1 - - Alteromonadales TW-7 g - - - 1 - - Pseudoalteromonas tunicata D2 g - - - 1 e 1 - Psychromonas sp. CNPT3 g - - - 1 - - Reinekea blandensis MED297 g - - - 1 - - Stenotrophomonas sp. SKA14 g - - 1 d 1 - - Photobacterium sp. SKA34 g - - - 1 - 1 Vibrio angustum S14 g - - - 1 - 1 Vibrio fischeri MJ11 g - - - 2 - - Vibrio shilonii AK1 g - - - 1 - - Vibrio alginolyticus 12G01 g 1 - - 2 - - Vibrio sp. MED222 g - - - 1 - - Vibrio splendidus 12B01 g - - - 1 - - Vibrionales bacterium SWAT-3 g - - - 3 - - Vibrio campbellii AND4 g 1 - - 1 1 - Plesiocystis pacifica SIR-1 d - - - 1 1 - a. Lacks ImcF. b. Lacks VirB8. c. Lacks POTRA. d. Lacks ShlB. [/formula]
e. Lacks ImcF and ImpA. The presence of the corresponding genomic island is indicated by digits denoting the number of occurrences; '-' denotes not detected. Bact, Bacteroidetes; a, Alphaproteobacteria; b, Betaproteobacteria; g, Gammaproteobacteria; Pla, Planctomycetes.
The two-partner secretion (TPS) system, a variant of the type V secretion system (T5SS), typically consists of a secreted TpsA protein (generally an adhesin) and a TpsB protein that facilitates the secretion of TpsA [bib_ref] Twopartner secretion in Gram-negative bacteria: a thrifty, specific pathway for large virulence..., Jacob-Dubuisson [/bib_ref]. In Bordetella pertussis, the causative agent of whooping cough, the TPS system secretes the filamentous haemagglutinin, which mediates adhesion to ciliated cells of the upper respiratory tract [bib_ref] The crystal structure of filamentous hemagglutinin secretion domain and its implications for..., Clantin [/bib_ref]. Other examples of TPS-mediated adhesion include cell-root interaction of Pseudomonas putida KT2440 [bib_ref] A two-partner secretion system is involved in seed and root colonization and..., Molina [/bib_ref] and the binding to red-blood cells in Proteus mirabilis . Operons similar to TPS were found in 16 marine bacteria [fig_ref] Table 1: Secretion systems and PAIs in genome sequenced marine bacteria [/fig_ref]. In Oceanobacter sp. RED65 we found two copies of TPS operons. The first TPS operon encoded two different putative haemagglutinins/adhesines, one having a similarity of 41% conserved amino acids to the filamentous haemagglutinin of B. pertussis. The second TPS operon encoded a protein with 38% similarity to HmwA filamentous haemagglutinin/adhesin of the human pathogen Haemophilus influenzae. The predicted size of the three variants of the putative excreted virulence factors in strain RED65 ranged from 80.3 to 306.1 kDa.
The type VI secretion system (T6SS) represents another pathway for protein translocation [bib_ref] A virulence locus of Pseudomonas aeruginosa encodes a protein secretion apparatus, Mougous [/bib_ref] [bib_ref] Identification of a conserved bacterial protein secretion system in Vibrio cholerae using..., Pukatzki [/bib_ref]. The genes of the T6SS constitute a genomic island important for the virulence/ antivirulence of a large number of animal and plant pathogens [bib_ref] Identification of a unique IAHP (IcmF associated homologous proteins) cluster in Vibrio..., Das [/bib_ref]. Searches specific for core conserved T6SS proteins revealed the presence of putative virulence factor homologues in 23 gammaproteobacteria and 12 alphaproteobacteria [fig_ref] Table 1: Secretion systems and PAIs in genome sequenced marine bacteria [/fig_ref]. Proteins of the T6SS operons in the marine bacteria showed amino acid sequence similarities between 29 and 85% to the corresponding T6SS proteins in Salmonella enterica ssp. I (the Sci-operon; [bib_ref] The Salmonella enterica subspecies I specific centisome 7 genomic island encodes novel..., Folkesson [/bib_ref]. Additionally, the planctomycete Blastopirellula marina DSM3645 harbours 12 typical T6SS genes, divided into two genome locations. The T6SS genomic islands carry a core set of genes found in almost all of the examined genomes and several additional genes [fig_ref] Figure 1: Gene organization of T6SSs in five representatives of the marine bacteria compared... [/fig_ref]. Focusing on this core set of genes, the T6SS island of Roseobacter sp. MED193 showed the highest similarity to the S. enterica ssp. I T6SS genes, being 29-71% and 46-84% at the nucleotide and amino acid levels respectively (see [fig_ref] Table 2: Frequencies of putative virulence genes in the Global Ocean Sampling metagenomic data... [/fig_ref]. The closest match to the V. cholerae O1 T6SS was found in Marinomonas sp. MED121 (54-85% identical, 79-92% conserved amino acids). That the closest matches were not found in the phylogenetically closest relatives might be an indication of the T6SS genomic island as a functional unit, which may have been acquired from other species through horizontal gene transfer. Genes encoding proteins known to be exported by the V. cholerae T6SS, that is, the Hcp (SciK/M) and VgrG proteins, are found within the operon [bib_ref] Identification of a conserved bacterial protein secretion system in Vibrio cholerae using..., Pukatzki [/bib_ref] , and had predicted homologues in all of the marine T6SS islands identified [fig_ref] Figure 1: Gene organization of T6SSs in five representatives of the marine bacteria compared... [/fig_ref]. The genes vasA (sciC) and vasK (sciS), necessary for T6SS-mediated secretion in V. cholerae V52 [bib_ref] Identification of a conserved bacterial protein secretion system in Vibrio cholerae using..., Pukatzki [/bib_ref] , belonged to the core set of genes found in all the marine T6SS islands, whereas the vasH gene, also important for T6SS function in Aeromonas hydrophila [bib_ref] Molecular characterization of a functional type VI secretion system from a clinical..., Suarez [/bib_ref] , had a homologue only in six marine gammaproteobacteria (Photobacterium sp. SKA34, V. angustum S14, Marinobacter spp. DG893 and ELB17, Psychromonas sp. CNPT3 and Marinomonas sp. MED121). Taken together, our detection of T6SS in 37 out of 119 marine bacterial genomes, where previously unrecognized, corroborate and extend the findings of that this secretion system is 'widespread in nature and not confined to known pathogens'.
## 'antifeeding islands'
Serratia entomophila causes amber disease in larva of the grass grub Costelytra zealandica, a plant pest in New Zealand. The pathogenicity of this bacterium, for example, strain A1MO2, derives in part from an 'antifeeding effect' that prevents larvae from feeding [bib_ref] Cloning Serratia entomophila antifeeding genes -a putative defective prophage active against the..., Hurst [/bib_ref]. The bacterial genes responsible for the antifeeding effect reside in a locus containing genes that bear similarity to phage structural genes. In the marine bacterium Vibrio campbellii AND4 [fig_ref] Figure 2: Gene organization of the 'antifeeding' prophage-like island in marine bacteria [/fig_ref] , a genomic island was detected that contained 16 out of 18 genes in the S. entomophila operon constituting the 'antifeeding island' [bib_ref] Cloning Serratia entomophila antifeeding genes -a putative defective prophage active against the..., Hurst [/bib_ref]. Amino acid sequence similarities of these putative encoded virulence factors ranged from 43% to 83% . Similar genomic islands were found in another 10 marine bacteria [fig_ref] Table 1: Secretion systems and PAIs in genome sequenced marine bacteria [/fig_ref] [fig_ref] Figure 2: Gene organization of the 'antifeeding' prophage-like island in marine bacteria [/fig_ref] , which contained from 8 to 13 of the 18 genes in the S. entomophila operon. Notably, the genes afp2-4 and afp7, identified as essential for the antifeeding effect provided by S. entomophila, were present in all the 'antifeeding island' copies identified, although some other conserved genes were missing. It was recently shown that the phage-like gene products of the 'antifeeding operon' in Photorhabdus asymbiotica ATCC43949 form contractile 'syringe' structures, and it was suggested that the phage structural proteins could be used to inject toxins into eukaryotic cells .
# Yersinia phospholipase-related toxin
The rat flea is the natural vector that spreads Y. pestis, the causative agent for plague. In the flea, Y. pestis cells growing in the midgut prevent the flea from feeding and increase its attempts to bite, thereby increasing the spread of Y. pestis. The Yersinia 'murine toxin' (ymt), a phospholipase D homologue [bib_ref] Murine toxin of Yersinia pestis shows phospholipase D activity but is not..., Hinnebusch [/bib_ref] , is essential for infection of the flea gut. When transferred to Escherichia coli the ymt gene enables E. coli, which normally lacks this gene, to establish a chronic infection of the flea [bib_ref] Role of Yersinia murine toxin in survival of Yersinia pestis in the..., Hinnebusch [/bib_ref]. Predicted homologues to the Y. pestis ymt gene were found in the marine bacteria Marinomonas sp. MED121 and Vibrio sp. MED222 with a similarity of 64-69% conserved amino acids. Indeed, some bacteria of the Vibrionaceae family are well known to associate to marine arthropods, for example, copepods [bib_ref] Ecological relationships between Vibrio cholerae and planktonic crustacean copepods, Huq [/bib_ref] , possibly indicating a role for the marine ymt genes in this context.
## Novel putative pais
A frequently used strategy of invasive pathogenic bacteria is to interfere with actin polymerization in the eukaryotic target cell, for example, through interference with Rho GTPases or by directly targeting actin. One such inhibitor of actin polymerization is the mono(ADP-ribosyl)transferase SpvB from the enteropathogenic bacterium S. enterica serovar Dublin, which is known to act both on mammalian and Acanthamoeba polyphaga actin polymerization [bib_ref] Salmonella enterica SpvB-mediated ADP-ribosylation as an activator for host cell actin degradation, Tezcan-Merdol [/bib_ref]. The ability to immo- bilize eukaryotic cells could be useful also for marine bacteria when infecting eukaryotes. Predicted homologues of the S. enterica SpvB protein were found in the two Vibrionaceae bacteria Photobacterium sp. SKA34 and Vibrio angustum S14. In Photobacterium sp. SKA34 there were three copies of the spvB-like genes (with 41-42% conserved amino acids) in a cluster of six genes, where the other three genes in the operon showed weak similarities to anthrax lethal factor (41% conserved over 360 amino acids), tetanus (53% conserved over 101 amino acids) and botulinum toxins (36% conserved over 328 amino acids) respectively. This cluster is located in a region (~80 kb, 66 genes) also containing genes with high sequence similarities (i.e. > 40% conserved amino acids) to genes encoding insect toxins, autotransporter adhesins with Ig-domains, a secreted endonuclease, EmpA metallo-protease of the type found in fish-pathogen V. anguillarum and a haemolysin/lecithinase produced by V. cholerae. Also, other virulence factors, such as putative cytastatin, chitinase, collagenase and penicillin V acyclase genes, as well as a PapC/FimD type usher porin involved in pili biogenesis, were found in this genomic region. V. angustum S14 carries a similar genomic region but with some gene rearrangements [fig_ref] Figure 1: Gene organization of T6SSs in five representatives of the marine bacteria compared... [/fig_ref] , which might be due to that the region is framed and divided by several transposase and phage-type integrase genes. This suggests that the region is part of a mobile genetic element. Similar to many PAIs [bib_ref] Pathogenicity islands in bacterial pathogenesis, Schmidt [/bib_ref] , also this genomic island is located next to a cluster of tRNA genes.
## Environmental virulence genes
We found a high number of predicted virulence gene homologues in the genomes of cultured marine bacteria, indicating that bacteria carrying such genes can be readily isolated from sea water. However, this observation may not necessarily reflect the occurrence of potential virulence genes among the uncultivated majority of marine bacterioplankton. Therefore, to achieve a first estimate of the prevalence of virulence genes in the ocean we searched the environmental DNA data set generated by the Sorcerer II Global Ocean Sampling Expedition (GOS) including data from 39 marine planktonic samples, a hypersaline lagoon and a lake [bib_ref] The Sorcerer II Global Ocean Sampling Expedition: Northwest Atlantic through Eastern Tropical..., Rusch [/bib_ref]. The GOS data set was searched for selected protein sequences from the T4SSs and T6SSs, as well as the (ADP-ribosyl)-transferase of the novel island in Photobacterium SKA34 and the 'antifeeding' island Afp2. In [fig_ref] Table 2: Frequencies of putative virulence genes in the Global Ocean Sampling metagenomic data... [/fig_ref] we present the frequencies of these genes normalized to the metagenomic library sizes with marine samples grouped according to high (> 1 mg chl a l -1 ) and low (< 1 mg chl a l -1 ) chlorophyll a concentrations. The investigated virulence genes showed a three to four times higher abundance in productive marine waters (58 genes per 10 9 bp) and in the hypersaline lagoon (57 genes per 10 9 bp) compared with open ocean sites (12 genes per 10 9 bp). The freshwater samples showed an intermediate value (22 genes per 10 9 bp) with a notable exception in that the T6SS genes were lacking.
The frequency of the typical single-copy recA gene was used to estimate the abundance of the virulence genes as a fraction of the total number of genomes [fig_ref] Table 2: Frequencies of putative virulence genes in the Global Ocean Sampling metagenomic data... [/fig_ref] [bib_ref] Bacterial classifications derived from RecA protein sequence comparisons, Karlin [/bib_ref] [bib_ref] The Sorcerer II Global Ocean Sampling Expedition: Northwest Atlantic through Eastern Tropical..., Rusch [/bib_ref]. From this analysis, we estimate that the investigated putative virulence genes were present in 8-12% of the genomes from the productive marine localities and the hypersaline lagoon, but were less common in more oligotrophic marine localities and in freshwater (2-3%; [fig_ref] Table 2: Frequencies of putative virulence genes in the Global Ocean Sampling metagenomic data... [/fig_ref]. These values may be underestimates due to queries being limited to only a few selected virulence genes or overestimates due to the presence of several virulence genes/copies per bacterial a. > 1 mg chl a l -1 . b. < 1 mg chl a l -1 . Marine samples were divided into productive (> 1 mg chl a l -1 , 12 samples) and more oligotrophic (< 1 mg chl a l -1 , 27 samples) regions based on chl a data published in . Samples from a hypersaline lagoon (Floreana Island, Ecuador) and from freshwater (Lake Gatun, Panama) were also included. Numbers give gene frequencies per billion bases sequenced. Numbers within parentheses give total number of hits in the CAMERA Global Ocean Sampling Database (http://camera.calit2.net/). genome. Nevertheless, the GOS subset of randomly sequenced environmental DNA support our finding from whole-genome sequences that virulence gene homologues have a widespread occurrence in marine bacteria, particularly in productive waters with high densities of algae. In this scenario, expression of virulence genes as a means to obtain resources could provide a selective advantage for opportunistic bacteria. In contrast, one would not expect to find homologues of virulence genes in typically oligotrophic marine bacteria like Pelagibacter ubique that is characterized by small size and a slow growing single-cell life strategy [bib_ref] SAR11 clade dominates ocean surface bacterioplankton communities, Morris [/bib_ref] [bib_ref] Oligotrophic bacterioplankton with a novel single-cell life strategy, Simu [/bib_ref] [bib_ref] Genome streamlining in a cosmopolitan oceanic bacterium, Giovannoni [/bib_ref]. Accordingly, we found no predicted homologues of virulence genes in any of the three sequenced strains of P. ubique [fig_ref] Table 1: Secretion systems and PAIs in genome sequenced marine bacteria [/fig_ref].
## Concluding remarks
In light of our findings it is reasonable to question if presence of virulence gene homologues fits with what is known about marine bacteria. Unfortunately, a comprehensive understanding of their life strategies is largely lacking. However, some interesting patterns have emerged. For example, several of the Roseobacter clade bacteria appear to be associated with phytoplankton [bib_ref] Response of Alteromonadaceae and Rhodobacteriaceae to glucose and phosphorus manipulation in marine..., Allers [/bib_ref] [bib_ref] Ecological genomics of marine roseobacters, Moran [/bib_ref] , and their containing T4SSs may provide a means to retrieve resources from these eukaryotes. For fast-growing gammaproteobacteria, such as Vibrio and Alteromonas species, labile organic carbon substrates, for example, amino acids or glucose, stimulates growth [bib_ref] Succession of pelagic marine bacteria during enrichment: a close look at cultivation-induced..., Eilers [/bib_ref] [bib_ref] Differential growth response of colony-forming a-and g-Proteobacteria in dilution culture and nutrient..., Pinhassi [/bib_ref]. Also organic matter released during intense algal blooms stimulates the growth of V. cholerae O1 in sea water [bib_ref] Growth of Vibrio cholerae O1 in red tide waters off California, Mourino-Pérez [/bib_ref]. At this stage it can only be speculated that the use for specific protein secretion systems and other virulence gene homologues enable, or even is a prerequisite, for a fruitful transition from a free-living lifestyle to life attached to phytoplankton or crustaceans. We screened for only a subset of known virulence genes and PAIs, and thus our data should represent a conservative estimate of the number of potential virulence genes in marine bacteria. Still, the presence of virulence genes does not necessarily imply that these bacteria are pathogens (e.g. on humans or fish). Even though our sequence analysis shows that marine bacteria have genetic homologues of virulence genes in known pathogens, the function of the expressed proteins may differ between organisms. For example, marine bacteria could use secretion systems and adhesins to attach to each other, other bacteria or to floating particles. However, it is plausible that such traits could also be useful to attach to and exploit compromised, dying or dead single-celled or multicellular eukaryotes. In so doing, virulence genes and PAIs could constitute the genetic tools for marine bacteria to cause harm to algae, and thereby contribute to explaining some of the ample records of such behaviour [bib_ref] Algicidal bacteria in the sea and their impact on algal blooms, Mayali [/bib_ref]. Alternatively, or in addition, 'virulence' genes may play an important role in protecting bacteria against protozoan predation [bib_ref] Off the hook -how bacteria survive protozoan grazing, Matz [/bib_ref]. These authors also proposed that microbial virulence factors associated with human disease could originate from such antipredator defence.
Interestingly, it was recently suggested that hostpathogen interactions should be studied in a wider ecological and evolutionary perspective to better understand the life strategy of pathogenic bacteria [bib_ref] Bacterial pathogenomics, Pallen [/bib_ref]. Functions that have evolved under long time in nature may have been recruited through horizontal gene transfer to perform similar or different functions by more recently emerging pathogenic bacterial species. The precise functions of the putative virulence gene homologues reported here remain unknown until experimental genetics and physiological response experiments can confirm their native function; for example, are they used to support host association, mutualism or commensalism or infection? Whatever the case may be, our analysis show that several marine bacteria contain virulence gene homologues that could potentially be used to attach to, and take advantage of, eukaryotic cells in a manner similar to that used by known pathogens (i.e. as an intrusive way to obtain organic matter and nutrients). Such functions could be important for understanding the ecology and evolution of marine bacteria.
In conclusion, we found a widespread occurrence of virulence gene homologues in a phylogenetically diverse set of genome sequenced marine bacteria. Further, our analysis of environmental DNA suggests that several of these genes are widely distributed in the oceans. Consequently, marine bacteria may constitute a reservoir for virulence genes found in human, animal, plant and insect pathogens as previously suggested for virulence genes in soil bacteria. From a marine ecology perspective, expression of these genes would indicate that some bacteria infect or even consume live cells. This would result in a previously unrecognized flow of organic matter and nutrients directly from eukaryotes to bacteria [fig_ref] Figure 2: Gene organization of the 'antifeeding' prophage-like island in marine bacteria [/fig_ref].
## Experimental procedures
## Analyses of genomes of marine bacteria
Whole-genome sequencing was done by the J. Craig Venter Institute (JCVI; sequences and strain information available at https://moore.jcvi.org/moore/) within the Gordon and Betty Moore Foundation Marine Microbiology Initiative (http:// www.moore.org/marine-micro.aspx). Open reading frames were predicted and annotated using the annotation pipeline at JCVI, which integrates results from hidden Markov models and BLAST. Functional prediction was carried out using COG analysis and motif finding. Non-coding RNA prediction was done using tRNA-scanSE and BLAST. All automatically annotated genes of interest were inspected and verified manually by BLAST [bib_ref] Basic local alignment search tool, Altschul [/bib_ref] , COG, PFAM and TIGRFAM analyses. If the annotation was judged to be correct, the analyses were further extended by repeating the analyses on neighbouring genes and by searching for elements (tRNA genes, IS sequences, transposons, phage/plasmid replicons and deviating GC content) indicative of mobile genetic regions. When a putative genomic island had been identified, homologous islands were screened for in other marine bacteria, where in addition to the aforementioned analysis, attention was also paid to gene order and operon structure in homologous genomic islands to verify that the putative island was indeed a conserved genomic unit.
An initial analysis was made on the genome sequences from marine bacterial isolates that we contributed to the Marine Microbiology Initiative and made available to us directly after sequencing. These genomes were: Neptuniibacter caesariensis MED92, Marinomonas sp. MED121, Dokdonia sp. MED134, Polaribacter sp. MED152, Roseobacter sp. MED193, Leeuwenhoekiella blandensis MED217, Vibrio sp. MED222, Reinekea sp. MED297, Oceanobacter sp. RED65, Photobacterium sp. SKA34, Loktanella vestfoldensis SKA53 and Sphingomonas sp. SKA58. In a first step, to find potential virulence gene homologues, we screened the automatic JCVI annotation by searching for virulence and secretion system genes or by looking for signs of defect prophage and IS elements indicating the presence of mobile genetic elements.
We also conducted BLASTP searches using genes of known virulence factors from animal and human pathogens as query sequences. In particular, we looked for tyrosine kinases, ADP-ribosyltransferases, various known and suspected haemolysins using characterized virulence genes from S. enterica (e.g. spvB, hlyA) and Y. pestis . For each of the virulence gene systems analysed, the original pathogen genomes where the virulence genes are present were included in the screening as positive controls to confirm that the methodology used could detect the relevant genes. Standard BLASTP was made with BLASTP 2.2.14 with no filtering, expectancy: 10, word size: 3, scoring matrix: BLOSUM 62, gap costs: opening 11, extension 1. Virulence factor homologues were identified on the criteria of an E-value Յ 10 -6 and sequence similarity > 30%. Type III secretion system. The T3SSs were located using a combination of PFAMs and TIGRFAMs that target peptides associated with T3SSs and are essential for virulence (references), for example, TIGR02500 (YscD/HrpQ), PF05932 (CesT), TIGR02497 (YscI/HrpB), TIGR02499 (HrpE/YscL) and PFAMs associated with protein exportation, that is, PF01514 (YscJ/FliF), PF00813 (inner membrane P protein FliP), PF00771 (Flagellar/Hr/Invasion Proteins Export Pore), PF01313 (export protein FliQ, family 3) and PF01312 (type III secretion exporter). Bacterial genomes with hits against several of these domains were manually inspected to exclude operon structures containing genes involved in flagellar biosynthesis. Type IV secretion system. The T4SSs were screened for among the marine bacterial genomes using PF04610 (TrbL/ VirB6 plasmid conjugal transfer protein) and PF04335 (VirB8 protein) and an E-value Յ 10 -6 . Genomes with positive hits for these PFAMs were manually inspected for presence of the T4SS core conserved protein classes VirB4, VirB6, VirB8, VirB9, VirB10, VirB11, VirD4 [bib_ref] Protein homology network families reveal step-wise diversification of type III and type..., Medini [/bib_ref]. Operon structures with genes for these seven proteins were thus detected using BLASTP with known VirB proteins from A. tumefaciens and PFAM and COG matches. The vir gene homologues encoding these proteins were all present in the genomes scored as T4SS positive and presented in [fig_ref] Table 1: Secretion systems and PAIs in genome sequenced marine bacteria [/fig_ref]. Type V secretion system. We screened for TPS (or T5bSS) systems using a combination of the PFAMs PF05860 (filamentous haemagglutinin, N-terminal domain), PF08479 (polypeptide transport-associated, ShlB-type) and PF03865 (haemolysin activator HlyB) and an E-value Յ 10 -6 . Positive hits were manually inspected to confirm that the two components of the TPS system were located in adjacent open reading frames.
Type VI secretion system. The T6SSs were identified essentially as described by , using the seven T6SS-specific PFAM domains PF05591 (DUF770), PF05936 (DUF876), PF05943 (DUF877), PF05947 (DUF879), PF06996 (DUF1305), PF06761 (ImcF-related) and PF06812 (ImpA, N-terminal domain) and an E-value Յ 10 -6 . Genomes were scored as positive for T6SS based on the presence of these markers combined. Some of the genomes had two or more paralogues of the genes containing these PFAMs.
Yersinia murine toxine was searched for among the marine bacterial genomes using BLASTP with the Ymt amino acid sequence from Y. pestis strain CO92 (GenBank Accession Number NP_395420) and an E-value Յ 10 -6 and sequence identity values > 30%.
In summary, functional annotation was based on: (i) sequence similarity values in BLAST searches over the entire length of the queried genes (unless specifically stated otherwise in the text), all sequence comparisons were made at the protein level, sometimes complemented by DNA sequence analysis, (ii) classification of proteins within the COG, PFAM or TIGRFAM domain structure libraries [bib_ref] CD-Search: protein domain annotations on the fly, Marchler-Bauer [/bib_ref] and (iii) gene arrangement/complement, that is, presence in the same genetic context in the marine bacteria as in known pathogens, that is, within the operon structure of defined PAIs.
## Gene frequencies in the global ocean sampling data
We used BLAST analysis, using virulence factor sequences from the marine bacterial genomes as queries, to find and identify putative virulence factors in the CAMERA Global Ocean Sampling Database (http://camera.calit2.net/). Proteins from the following organisms were used to search the Global Ocean Sampling data [bib_ref] The Sorcerer II Global Ocean Sampling Expedition: Northwest Atlantic through Eastern Tropical..., Rusch [/bib_ref] : T5SS TspB homologue: Oceanobacter sp. RED65, T6SS SciB and SciS homologues: Roseobacter sp. MED193, 'Antifeeding island' Afp2 homologue: L. blandensis MED217, Yersinia murine toxin homologue: Vibrio sp. MED222 (APD-ribosyl)transferase homologue: Photobacterium sp. SKA34. Matches to Burkholderia were excluded from the analysis as these sequences may stem from contaminants [bib_ref] The Sorcerer II Global Ocean Sampling Expedition: Northwest Atlantic through Eastern Tropical..., Rusch [/bib_ref]. Photorhabdus virulence cassettes confer injectable insecticidal activity against the wax moth. J Bacteriol 188: 2254-2261.
## Supporting information
Additional Supporting Information may be found in the online version of this article: [fig_ref] Figure 1: Gene organization of T6SSs in five representatives of the marine bacteria compared... [/fig_ref]. Novel PAI in Photobacterium sp. SKA34 and Vibrio angustum S14. Genes with more than single copies, showing homologies to different classes of virulence genes, are marked in different colours (see legend below figure). The genome regions span from SKA34_08238 to SKA34_08563 and VAS14_15892 to VAS14_16212. [fig_ref] Figure 2: Gene organization of the 'antifeeding' prophage-like island in marine bacteria [/fig_ref]. Conceptual model of the carbon flow in the marine pelagic. Grey lines represent release of dissolved organic carbon as the result of several different mechanisms including virus lysis. The red lines depict possible added effects of putative virulence genes on the carbon fluxes in the marine planktonic food web. [fig_ref] Table 1: Secretion systems and PAIs in genome sequenced marine bacteria [/fig_ref]. List of genomes included in the study. Genomes marked in boldface contained some/several of the virulence genes or PAIs that were screened for. Genomes were sequenced through the Moore Foundation Initiative in Marine Microbiology. Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.
[fig] Figure 1: Gene organization of T6SSs in five representatives of the marine bacteria compared with those of Salmonella enterica subsp. I, Rhizobium leguminosarum biovar trifolii, and Vibrio cholerae O1 biovar eltor str. N16961. Predicted gene homologues are depicted with the same colours and patterns. The following genetic regions are shown: Photobacterium sp. SKA34 (SKA34_05615 to SKA34_05715), Oceanobacter sp. RED65 (RED65_00815 to RED65_00895), Reinekea sp. MED297 (MED297_18458 to MED297_18883), Vibrio sp. MED222 (MED222_13880 to MED222_13980), Marinomonas sp. MED121 (#1: genes MED121_07350 to MED121_07465; #2: MED121_11870 to MED121_11955), Roseobacter sp. MED193 (MED193_00960 to MED193_01055). [/fig]
[fig] Figure 2: Gene organization of the 'antifeeding' prophage-like island in marine bacteria. Genes that are predicted homologues to S. entomophila 'antifeeding' genes are shown with the same patterns as these. Unfilled boxes indicate genes that do not have homologues in S. entomophila. Filled black circles above genes indicate genes that show similarity to phage structural genes. Black circles above genes of S. entomophila indicate genes that have been shown to be essential for the antifeeding effect of S. entomophila. S. entom: Serratia entomophila A1MO2 AND4, Vibrio campbellii AND4; MED217, Leeuwenhoekiella blandensis MED217; HTCC2594, Erythrobacter litoralis HTCC2594; NB231, Nitrococcus mobilis Nb-231 (two loci); NB311A, Nitrobacter sp. Nb311A; HTCC2654, Rhodobacterales bacterium HTCC2654. [/fig]
[table] Table 1: Secretion systems and PAIs in genome sequenced marine bacteria. [/table]
[table] Table 2: Frequencies of putative virulence genes in the Global Ocean Sampling metagenomic data set. [/table]
[table] Table S2: Comparison of the proteins in type VI secretion systems of Salmonella enterica subsp. Typhimurium (SALTY), Vibrio cholerae (V.chol.), Rhizobium leguminosarum (RHIZO) and marine bacteria (Roseobacter sp. MED193, Marinomonas sp. MED121, Vibrio sp. MED222, Reinekea sp. MED297, Oceanobacter sp. RED65, Photobacterium sp. SKA34). Standard BLASTP was made with BLASTP 2.2.14 with no filtering, expectancy: 10, word size: 3, scoring matrix: BLOSUM 62, gap costs: opening 11, extension 1. Numbers given represent '% amino acid identity/% aa similarity/length in aa of matched region'. The conserved domains in the table legends indicate domains detected upon BLASTP analyses in GenBank. Table S3. Comparison between antifeeding PAIs from Serratia entomophila (Afp1-Afp18), Photorhabdus luminescence and a number of marine bacteria (see https://research. venterinstitute.org/moore/MultiOrganism? MULTIORGANISM_SORT_ATTR=SORT_ORGANISM for sequence access). Number in parenthesis represent '% amino acid identity/% aa similarity' as found in standard BLASTP search without low-complexity filtering and BLOSUM 45 scoring matrix. Proteins were searched for PFAM and COG protein motifs. [/table]
|
The Association between VDR Gene Polymorphisms and Diabetic Retinopathy Susceptibility: A Systematic Review and Meta-Analysis
Aims. Studies on the associations of vitamin D receptor (VDR) gene polymorphisms with diabetic retinopathy (DR) susceptibility reported conflicting results. A systematic meta-analysis was undertaken to clarify this topic. Methods. A systematic search of electronic databases (PubMed, EMBASE, and CNKI) was carried out until March 31, 2016. The pooled odds ratio (OR) and 95% confidence interval (CI) were used to assess the strength of the association. Results. A total of 7 studies fulfilling the inclusion criteria were included in this meta-analysis (649 cases and 707 controls). Pooled ORs showed a significant association between FokI polymorphism and DR risk in all the four genetic models (OR = 1.612 (1.354∼1.921), 1.988 (1.481∼2.668), 1.889 (1.424∼2.505), and 2.674 (1.493∼4.790) in allelic, dominant, recessive, and additive models, resp., < 0.01), but not for TaqI or BsmI polymorphism ( > 0.05). Similar results were found in the subgroup analysis. Sensitivity analysis indicated that the results were relatively stable and reliable. Results of Begg's and Egger's tests suggested a lack of publication bias. Conclusions. Our meta-analysis demonstrated that DR was significantly associated with VDR gene FokI polymorphism. However, due to the relatively small sample size in this meta-analysis, further studies with a larger sample size should be done to confirm the findings.
# Introduction
Diabetic retinopathy (DR) is regarded as the leading cause of legal blindness in adults, characterized by increased vascular permeability, tissue ischemia, and neoangiogenesis [bib_ref] Pathogenesis of diabetic microangiopathy: an overview, Barnett [/bib_ref] [bib_ref] Diabetic retinopathy, Cheung [/bib_ref]. As one of the most prominent pathological microvascular complications in diabetes, the prevalence of DR in diabetes patients has been estimated at 34.6% and that of proliferative diabetic retinopathy (PDR) has been estimated at 7.0% [bib_ref] Global prevalence and major risk factors of diabetic retinopathy, Yau [/bib_ref] , but the frequency varies in different ethnicities.
It has been established that good diabetes control helps to prevent DR; however, the mechanisms underlying the role of hyperglycemia in DR remain unclear. A subanalysis of the Diabetes Control and Complications Trial (DCCT) showed strong retinopathy transmission in families of patients with severe DR but not in those with nonsevere DR, supporting potential involvement of genetic factors in DR. Therefore, it is important to identify the genetic susceptibility factors for DR, which would be helpful to clarify the pathogenesis of DR.
As a secosteroid hormone, vitamin D is acquired and synthesized from the diet and ultraviolet radiation. In addition to its well-known function of maintaining normal homeostasis of calcium and phosphorus, it also has potent nonclassical properties, such as anti-inflammatory, antioxidant, antiangiogenic, and antiproliferative properties [bib_ref] Vitamin D in health and disease: a literature review, Basit [/bib_ref] [bib_ref] Vitamin D in health and disease, Heaney [/bib_ref]. It has been reported that vitamin D could inhibit vascular smooth muscle cell growth in vitro through its antiproliferative action. And vitamin D deficiency has been associated with increased prevalence of retinopathy in young T1DM [bib_ref] Vitamin D deficiency is associated with retinopathy in children and adolescents with..., Kaur [/bib_ref] and T2DM [bib_ref] Vitamin D and retinopathy in adults with diabetes mellitus, Patrick [/bib_ref] patients. The active form of vitamin D acts through a specific vitamin D receptor (VDR), which is widely expressed in human tissues and organs, including the retina [bib_ref] Immuno-localization of the calcitriol receptor, calbinclin-d28k and the plasma membrane calcium pump..., Johnson [/bib_ref]. Therefore, the gene encoding VDR is regarded as a candidate gene involved in DR and has been studied in several populations.
## Biomed research international
The human VDR gene is located on chromosome 12q13.1, with at least 5 promoter regions and 8 protein-coding exons. Several polymorphisms in the VDR gene have been suggested to be involved in the development of DR. However, the results are conflicting and inconclusive. This may be attributed to the limited sample size and inadequate statistical power, which might affect their reliability. A meta-analysis is a statistical procedure of pooling the data from individual studies, increasing effective sample size, enhancing statistical power of the analysis, and producing a single estimate of an effect [bib_ref] Meta-analysis of genetic association studies, Lee [/bib_ref]. Therefore, we performed a comprehensive meta-analysis to further evaluate the association of VDR gene common polymorphisms with DR susceptibility; we focused on the polymorphisms of FokI (rs10735810), BsmI (rs1544410), ApaI (rs7975232), and TaqI (rs731236), as they had been shown to be highly polymorphic and the most studied polymorphisms.
# Methods
## Literature search.
Eligible studies were systematically searched in PubMed, EMBASE, and China National Knowledge Infrastructure (CNKI) databases up to March 31, 2016, with keywords including "diabetes OR diabetic retinopathy" and "VDR OR vitamin D receptor" and "polymorphism OR mutation OR variation OR SNP". No language restrictions were applied. Additional studies were identified by a hand search for references of original studies and review articles about the association of VDR gene polymorphisms with DR. For detailed search strategies, please see S1 (in Supplementary Material available online at http://dx.doi.org/10.1155/ 2016/5305282).
## Inclusion and exclusion
Criteria. Studies were chosen if they met the following criteria: (1) published studies; (2) evaluated association between VDR polymorphisms and DR risk; (3) a case-control or cohort study based on unrelated individuals; (4) sufficient data for examining odds ratios (ORs) with 95% confidence intervals (CIs); (5) genotype distributions of polymorphism of the control population consistent with Hardy-Weinberg equilibrium (HWE). The most recent article would be used to extract data if the authors published more than one article with the same study data. Case reports, editorials, reviews, abstracts from conferences, republished or duplicate studies, and studies with insufficient information for data extraction were excluded.
## Data extraction and quality assessment.
Two investigators (Y. Zhang and W. Xia) independently extracted data and both of their results were submitted to a third investigator (P. Lu) for verification. If there were inconsistencies, the three investigators discussed the disagreements to resolve the differences. The following information was collected: (1) the first author's name and publication year; (2) country of origin and ethnicity of samples; (3) sample size and gender ratio (male, %), duration of diabetes and glycosylated hemoglobin (HbA1c) level; (4) genotyping methods and genotype distribution.
The Cochrane recommended case-control study quality assessment tool and the Newcastle-Ottawa Scale (NOS) tools were used to evaluate the quality of the eligible studies.
## Statistical
Analysis. STATA software 12.0 (STATA Corp., College Station, TX, USA) was used for all statistical analyses. Genotype frequency was assessed by chi-square test in the control group for HWE. The strength of the association between VDR polymorphisms and DR susceptibility was assessed through calculating the pooled odds ratios (ORs) and 95% confidence intervals (CIs) of test. Four genetics models were used for analyses: allelic model, dominant model, recessive model, and additive model; and the values were corrected for multiple testing using the false discovery rate [bib_ref] Controlling the false discovery rate in behavior genetics research, Benjamini [/bib_ref]. tests and 2 statistic were used to test the heterogeneity among studies, and > 0.10 and 2 < 50% were considered to be of low heterogeneity. Sensitivity analysis was conducted by sequentially excluding each study to assess the stability of the results. Publication bias was assessed by Begg's and Egger's tests. < 0.05 was considered significant for all tests.
# Results
## Characteristics of published studies.
A total of 360 studies were retrieved. Based on titles and/or abstracts, we excluded 36 reviews (or meta-analysis, editorials) and 309 irrelevant studies. As the result in EMBASE was the same as that in PubMed, therefore, 5 studies were retrieved. The remaining 10 studies were included for full-text view. One abstract from conference was excluded. One article was excluded owing to lack of complete data (we tried to contact the author by email and had no response until we submitted our study) [bib_ref] Association between Bsm1 polymorphism in vitamin D receptor gene and diabetic retinopathy..., Hong [/bib_ref]. One article was excluded for departure from HWE in the control group [bib_ref] The relationship between vitamin D receptor gene TaqI polymorphism and diabetic retinopathy, Sun [/bib_ref]. Finally, 7 eligible studies (649 cases and 707 controls) published from 2002 to 2015 were included in this meta-analysis [bib_ref] Effects of vitamin D receptor gene polymorphism and clinical characteristics on risk..., Zhong [/bib_ref] [bib_ref] Relationship between Vitamin D receptor gene FokI polymorphism and diabetic retinopathy in..., Hou [/bib_ref] [bib_ref] Association between DNA polymorphism of human Vi tamin D receptor (VDR) gene..., Wu [/bib_ref] [bib_ref] Retinopathy and nephropathy in type 1 diabetic patients-association with polymorphysms of vitamin..., Bućan [/bib_ref] [bib_ref] Clinical risk factors and the role of VDR gene polymorphisms in diabetic..., Cyganek [/bib_ref] [bib_ref] Association between a protein polymorphism in the start codon of the vitamin..., Taverna [/bib_ref] [bib_ref] Taq I polymorphism of the vitamin D receptor and risk of severe..., Taverna [/bib_ref] , and the data was extracted. The study selection procedure was shown in [fig_ref] Figure 1: Results of the literature search strategy [/fig_ref]. Generally, the major design characteristics of all eligible studies were in accordance with the NOS tool and therefore were of relatively high quality. Reported articles about GWAS of DR were also searched.
Among the 7 studies, 6 studies focused on the association of DR risk and FokI polymorphism [bib_ref] Effects of vitamin D receptor gene polymorphism and clinical characteristics on risk..., Zhong [/bib_ref] [bib_ref] Relationship between Vitamin D receptor gene FokI polymorphism and diabetic retinopathy in..., Hou [/bib_ref] [bib_ref] Association between DNA polymorphism of human Vi tamin D receptor (VDR) gene..., Wu [/bib_ref] [bib_ref] Retinopathy and nephropathy in type 1 diabetic patients-association with polymorphysms of vitamin..., Bućan [/bib_ref] [bib_ref] Clinical risk factors and the role of VDR gene polymorphisms in diabetic..., Cyganek [/bib_ref] [bib_ref] Association between a protein polymorphism in the start codon of the vitamin..., Taverna [/bib_ref] , 3 on TaqI polymorphism [bib_ref] Retinopathy and nephropathy in type 1 diabetic patients-association with polymorphysms of vitamin..., Bućan [/bib_ref] [bib_ref] Clinical risk factors and the role of VDR gene polymorphisms in diabetic..., Cyganek [/bib_ref] [bib_ref] Taq I polymorphism of the vitamin D receptor and risk of severe..., Taverna [/bib_ref] , 3 on BsmI polymorphism [bib_ref] Effects of vitamin D receptor gene polymorphism and clinical characteristics on risk..., Zhong [/bib_ref] [bib_ref] Retinopathy and nephropathy in type 1 diabetic patients-association with polymorphysms of vitamin..., Bućan [/bib_ref] [bib_ref] Clinical risk factors and the role of VDR gene polymorphisms in diabetic..., Cyganek [/bib_ref] , and 2 on ApaI polymorphism [bib_ref] Effects of vitamin D receptor gene polymorphism and clinical characteristics on risk..., Zhong [/bib_ref] [bib_ref] Clinical risk factors and the role of VDR gene polymorphisms in diabetic..., Cyganek [/bib_ref]. All the polymorphisms were genotyped by polymerase chain reactionrestriction fragment length polymorphism (PCR-RFLP). A total of 4 studies included Caucasian populations [bib_ref] Retinopathy and nephropathy in type 1 diabetic patients-association with polymorphysms of vitamin..., Bućan [/bib_ref] [bib_ref] Clinical risk factors and the role of VDR gene polymorphisms in diabetic..., Cyganek [/bib_ref] [bib_ref] Association between a protein polymorphism in the start codon of the vitamin..., Taverna [/bib_ref] [bib_ref] Taq I polymorphism of the vitamin D receptor and risk of severe..., Taverna [/bib_ref] , and 3 included Asian populations [bib_ref] Effects of vitamin D receptor gene polymorphism and clinical characteristics on risk..., Zhong [/bib_ref] [bib_ref] Relationship between Vitamin D receptor gene FokI polymorphism and diabetic retinopathy in..., Hou [/bib_ref] [bib_ref] Association between DNA polymorphism of human Vi tamin D receptor (VDR) gene..., Wu [/bib_ref]. Four studies were conducted in type 2 diabetes patients [bib_ref] Effects of vitamin D receptor gene polymorphism and clinical characteristics on risk..., Zhong [/bib_ref] [bib_ref] Relationship between Vitamin D receptor gene FokI polymorphism and diabetic retinopathy in..., Hou [/bib_ref] [bib_ref] Association between DNA polymorphism of human Vi tamin D receptor (VDR) gene..., Wu [/bib_ref] [bib_ref] Clinical risk factors and the role of VDR gene polymorphisms in diabetic..., Cyganek [/bib_ref] ] and 3 in type 1 diabetes patients [bib_ref] Retinopathy and nephropathy in type 1 diabetic patients-association with polymorphysms of vitamin..., Bućan [/bib_ref] [bib_ref] Association between a protein polymorphism in the start codon of the vitamin..., Taverna [/bib_ref] [bib_ref] Taq I polymorphism of the vitamin D receptor and risk of severe..., Taverna [/bib_ref]. The study characteristics were displayed in [fig_ref] Table 1: Characteristics of 7 studies included in this systematic review and meta-analysis [/fig_ref] , and the genotype distributions of all studies are summarized in [fig_ref] Table 2: Genotype frequencies of VDR polymorphisms in 7 studies included in this systematic... [/fig_ref]. The distributions of the genotypes in the control populations were consistent with HWE in all of the studies ( > 0.05).
## Association of vdr gene foki polymorphism and risk of dr.
Six relevant studies with a total number of 548 cases and 608 controls were included in FokI polymorphism analysis [bib_ref] Effects of vitamin D receptor gene polymorphism and clinical characteristics on risk..., Zhong [/bib_ref] [bib_ref] Relationship between Vitamin D receptor gene FokI polymorphism and diabetic retinopathy in..., Hou [/bib_ref] [bib_ref] Association between DNA polymorphism of human Vi tamin D receptor (VDR) gene..., Wu [/bib_ref] [bib_ref] Retinopathy and nephropathy in type 1 diabetic patients-association with polymorphysms of vitamin..., Bućan [/bib_ref] [bib_ref] Clinical risk factors and the role of VDR gene polymorphisms in diabetic..., Cyganek [/bib_ref] [bib_ref] Association between a protein polymorphism in the start codon of the vitamin..., Taverna [/bib_ref] , 4 in type 2 diabetes patients and 2 in type 1 diabetes patients. The summary results of meta-analysis were shown in [fig_ref] Table 3: Meta-analysis of VDR gene FokI polymorphism and DR susceptibility [/fig_ref]. Pooled ORs showed a significant association between FokI polymorphism and DR risk in all the four genetic models (allelic, dominant, recessive, and additive models). No significant heterogeneity was found except for additive model [fig_ref] Table 3: Meta-analysis of VDR gene FokI polymorphism and DR susceptibility [/fig_ref]. Then, we conducted subgroup analysis stratified by population (Caucasian versus Asian). Overall, heterogeneity in all the four genetic models was not statistically significant either in Asian or in Caucasian populations, and the ORs and 95% CIs were therefore calculated in fixed-effect model. The results indicated that FokI polymorphism was significantly associated with an increased DR risk in all four genetic models (allelic, dominant, recessive, and additive models) in Asian populations; and no significant association was found in Caucasian populations in all the four genetic models (Table 3, [fig_ref] Figure 2: Continued. [/fig_ref].
We also conducted subgroup analysis stratified by type of diabetes. A significant association between FokI polymorphism and DR risk in all four genetic models (allelic, dominant, recessive, and additive models) was found in type 2 diabetes patients; and no significant association was found in type 1 diabetes patients in all the four genetic models [fig_ref] Table 3: Meta-analysis of VDR gene FokI polymorphism and DR susceptibility [/fig_ref].
## Association of vdr gene taqi polymorphism and risk of dr.
Three relevant studies with a total number of 205 cases and 309 controls were included in TaqI polymorphism analysis [bib_ref] Retinopathy and nephropathy in type 1 diabetic patients-association with polymorphysms of vitamin..., Bućan [/bib_ref] [bib_ref] Clinical risk factors and the role of VDR gene polymorphisms in diabetic..., Cyganek [/bib_ref] [bib_ref] Taq I polymorphism of the vitamin D receptor and risk of severe..., Taverna [/bib_ref]. Taverna et al. first reported an association between TT genotype and low risk for severe DR in French type 1 diabetes patients [bib_ref] Taq I polymorphism of the vitamin D receptor and risk of severe..., Taverna [/bib_ref]. However, no association was found in either study in type 1 diabetes patients by Bućan et al. [bib_ref] Retinopathy and nephropathy in type 1 diabetic patients-association with polymorphysms of vitamin..., Bućan [/bib_ref] or study in type 2 diabetes patients by Cyganek et al. [bib_ref] Clinical risk factors and the role of VDR gene polymorphisms in diabetic..., Cyganek [/bib_ref]. In our meta-analysis, pooled ORs and 95% CIs in four genetic models (allelic, dominant, recessive, and additive models) were 1.145 (0.879∼1.492), 1.647 (0.582∼4.662), 1.035 (0.600∼1.785), and 1.235 (0.689∼2.213), respectively. So, no significant association between TaqI polymorphism and risk of DR was suggested.
## Association of vdr gene bsmi polymorphism and risk of dr.
Three relevant studies with a total number of 198 cases and 320 controls were included in BsmI polymorphism analysis [bib_ref] Effects of vitamin D receptor gene polymorphism and clinical characteristics on risk..., Zhong [/bib_ref] [bib_ref] Retinopathy and nephropathy in type 1 diabetic patients-association with polymorphysms of vitamin..., Bućan [/bib_ref] [bib_ref] Clinical risk factors and the role of VDR gene polymorphisms in diabetic..., Cyganek [/bib_ref]. All the studies were conducted in type 2 diabetes patients. Study by Wu et al. demonstrated a significant association of BsmI genotypes with cumulative prevalence of retinopathy ( < 0.05) [bib_ref] Association between DNA polymorphism of human Vi tamin D receptor (VDR) gene..., Wu [/bib_ref]. However, no association was detected in studies by Zhong et al. [bib_ref] Effects of vitamin D receptor gene polymorphism and clinical characteristics on risk..., Zhong [/bib_ref] or Cyganek et al. [bib_ref] Clinical risk factors and the role of VDR gene polymorphisms in diabetic..., Cyganek [/bib_ref]. Our meta-analysis showed no significant association between BsmI polymorphism and risk of DR, and heterogeneity was not statistically significant. Pooled ORs and 95% CIs in four genetic models (allelic, dominant, recessive, and additive models) were 1.031 (0.775∼1.373), 1.080 (0.579∼2.017), 1.025 (0.706∼1.487), and 1.130 (0.587∼2.175), respectively.
## Association of vdr gene apai polymorphism and risk of dr.
Two relevant studies reported the association between ApaI polymorphism and risk of DR (a total of 179 cases and 292 controls). Study by Zhong et al. demonstrated that VDR gene ApaI polymorphism was not associated with DR risk in Han Chinese type 2 diabetes patients ( = 0.92) [bib_ref] Effects of vitamin D receptor gene polymorphism and clinical characteristics on risk..., Zhong [/bib_ref]. The same result was found by Cyganek et al. in Polish type 2 diabetes patients ( = 0.23) [bib_ref] Clinical risk factors and the role of VDR gene polymorphisms in diabetic..., Cyganek [/bib_ref]. Results of metaanalysis showed no significant association between ApaI polymorphism and risk of DR, and heterogeneity was not statistically significant. Pooled ORs and 95% CIs in four genetic models (allelic, dominant, recessive, and additive models) were 1.154 (0.882∼1.509), 1.101 (0.710∼1.707), 1.372 (0.868∼2.169), and 1.308 (0.743∼2.303), respectively.
# Sensitivity analysis.
In the sensitivity analysis, the influence of each study on the pooled OR was examined by repeating the meta-analysis while omitting each study, one at a time. The results indicated that the overall significance of the ORs was not altered by any single study for all the four genetic models of the FokI polymorphism [fig_ref] Table 4: Sensitivity analysis of the meta-analysis on VDR gene FokI polymorphism and DR... [/fig_ref]. This indicated that the results of the meta-analysis about VDR gene FokI polymorphism and risk of DR were relatively stable and reliable.
## Publication bias.
Potential publication bias of the metaanalysis about VDR gene FokI polymorphism and risk of DR was examined by Begg's and Egger's tests. Begg's funnel plot was symmetrical in shape, and the value of Egger's test indicated a lack of publication bias [fig_ref] Table 5: Publication bias analysis of the meta-analysis on VDR gene FokI polymorphism and... [/fig_ref] and [fig_ref] Figure 3: Forest plots for meta-analysis of VDR gene FokI polymorphism and DR risk [/fig_ref]. The results showed no evidence of obvious asymmetry for all the four genetic models.
# Discussion
In the present study, we systematically reviewed all available published studies and performed a meta-analysis to evaluate the association of VDR gene polymorphisms with DR. Seven studies were included in this meta-analysis. Pooled ORs showed a significant association between FokI polymorphism and DR susceptibility in all the four genetic models (allelic, dominant, recessive, and additive models). Sensitivity analysis further showed that the association was stable, and Begg's and Egger's tests indicated a lack of publication bias. Contrary to our meta-analysis, a previous meta-analysis reported no association between FokI polymorphism and DR [bib_ref] Associations study of vitamin D receptor gene polymorphisms with diabetic microvascular complications:..., Liu [/bib_ref]. We believed our results were more reliable and stable based on four more included studies and larger sample size, thoughtful design, and strict criterion for the included studies.
The most investigated VDR gene polymorphism was FokI polymorphism (rs10735810), a functional T-to-C substitution at exon 2. It abolished the first translation initiation site and resulted in a peptide lacking three amino acids, which influenced the transcriptional activity of VDR [bib_ref] Genetics and biology of vitamin D receptor polymorphisms, Uitterlinden [/bib_ref]. The FF genotype of FokI polymorphism had been associated with higher VDR mRNA copy numbers and increased transcriptional activity of VDR [bib_ref] Vitamin D receptor (VDR) mRNA and VDR protein levels in relation to..., Ogunkolade [/bib_ref]. So, it was presumed that potential beneficial effects of vitamin D on the retina (e.g., immunomodulatory, anti-inflammatory, antioxidant, antiangiogenic, and antiproliferative properties) were higher in patients carrying the F allele than in f-carrier patients. This association had been studied in several populations with conflicting and inconclusive results. So, we preformed this meta-analysis, and our results confirmed the significant association. The results were relatively stable and reliable. In subgroup analysis, the association was limited in Asian BsmI, ApaI, and TaqI were all located at the 3 untranslated region of the gene, which was involved in regulation of gene expression, especially through the modulation of mRNA stability [bib_ref] Vitamin D receptor gene polymorphisms in relation to Vitamin D related disease..., Uitterlinden [/bib_ref]. It was interesting to note that significant linkage disequilibrium was found among the TaqI, ApaI, and BsmI polymorphisms [bib_ref] Vitamin D receptor gene polymorphisms and type 2 diabetes: a meta-analysis, Li [/bib_ref]. Although these polymorphisms had been reported to be associated with reduced steady state VDR mRNA [bib_ref] Vitamin D receptor: no evidence for allelespecific mRNA stability in cells which..., Verbeek [/bib_ref] , no association with risk of DR was observed in this meta-analysis. One study in Korean type 2 diabetes patients found that patients with B allele (BB or Bb) of BsmI polymorphism had significant association with lower risk of DR (severe nonproliferative DR or proliferative DR; 7.4%, Begg's funnel plot with pseudo 95% confidence limits 5/68) compared with patients without B allele (bb; 17.3%, 81/469; = 0.035) [bib_ref] Association between Bsm1 polymorphism in vitamin D receptor gene and diabetic retinopathy..., Hong [/bib_ref]. However, for deficiency of complete data, this study was not included in our meta-analysis.
Some limitations existed in the current meta-analysis that must be considered. First, the conclusion was based on a relatively small number of participants. Therefore, our results might be underpowered. Second, we performed stratification only by race and type of diabetes. The subgroup analysis by type of DR was not performed, because few studies gave data about proliferative DR and nonproliferative DR. Third, our meta-analysis was based on unadjusted estimates without being adjusted for other covariates, such as age, family history, and duration of diabetes. A large, prospective clinical study that includes additional clinical data, such as type of treatment and presence of other microvascular complications, and anthropometric parameters into account is needed to confirm the importance of VDR polymorphism in the development of DR.
# Conclusions
In conclusion, our meta-analysis results indicated that there was a significant association between the VDR gene FokI polymorphism and DR susceptibility. However, due to the relatively small sample size in this meta-analysis, in order to reach a more definitive conclusion, further studies based on larger sample size and substantiation of the variations through functional studies are still needed.
[fig] Figure 1: Results of the literature search strategy. [/fig]
[fig] Figure 2: Forest plots for meta-analysis of VDR gene FokI polymorphism and DR risk. [/fig]
[fig] Figure 3: Forest plots for meta-analysis of VDR gene FokI polymorphism and DR risk. [/fig]
[table] Table 1: Characteristics of 7 studies included in this systematic review and meta-analysis. DM1: type 1 diabetes; DM2: type 2 diabetes. DR: diabetic retinopathy; NDR: diabetes without retinopathy. SDR: severe diabetic retinopathy (severe nonproliferative DR or proliferative DR); NSDR: nonsevere diabetic retinopathy (no DR or minimal nonproliferative DR). NC: normal control. [/table]
[table] Table 2: Genotype frequencies of VDR polymorphisms in 7 studies included in this systematic review and meta-analysis. [/table]
[table] Table 3: Meta-analysis of VDR gene FokI polymorphism and DR susceptibility. [/table]
[table] Table 4: Sensitivity analysis of the meta-analysis on VDR gene FokI polymorphism and DR susceptibility. Calculated with random-effect model. [/table]
[table] Table 5: Publication bias analysis of the meta-analysis on VDR gene FokI polymorphism and DR susceptibility.populations and in type 2 diabetes patients. The discordancy might be a result of the difference in genetic backgrounds between Asian and Caucasian populations (all the 3 studies were performed in Caucasian populations). [/table]
|
Fabrication of a Cranial Prosthesis Combined with an Ocular Prosthesis Using Rapid Prototyping: A Case Report
Rapid prototyping (RP) is a technique of manufacturing parts by the additive layer manufacturing technology; where, a three-dimensional (3D) model created in a computer aided design (CAD) system is sectioned into 2D profiles, which are further constructed by RP layer by layer. Its use is not limited to industrial or engineering fields and has extended to the medical field for the manufacturing of custom implants and prostheses, the study of anatomy and surgical planning. Nowadays, dentists are more frequently encountered with the individuals affected with craniofacial defects due to trauma. In such cases, the craniomaxillofacial rehabilitation is a real challenge to bring the patients back to society and promote their well-being. The conventional impression technique for facial prosthesis fabrication has the disadvantage of deforming the soft tissue and causing discomfort for the patient. Herein, we describe the fabrication of a cranial prosthesis combined with an ocular prosthesis with RP and stereolithography.
# Introduction
Facial defects secondary to the treatment of neoplasms, congenital malformations and trauma result in multiple functional and psychosocial difficulties. Prosthetic rehabilitation to restore these facial disfigurements may improve not only the level of function but also the self-esteem of patients [bib_ref] Treatment satisfaction with facial prostheses, Chang [/bib_ref]. The disfigurement associated with the loss of an eye causes significant stress, primarily adjusting to the functional disability that results. Eye is an even organ and therefore its reproduction is challenging, since the prosthesis must be similar to the natural eye as much as possible [bib_ref] The ocular impression: A review of the literature and presentation of an..., Mathews [/bib_ref]. The conventional impression techniques for manufacturing facial prosthesis have the disadvantage of deforming the soft tissues due to the tension caused by the impression material and also cause discomfort for the patient [bib_ref] Optical data acquisition for computer-assisted design of facial prostheses, Runte [/bib_ref]. Recently, Rapid prototyping (RP) system was developed as a simple method for fabricating models and prosthesis without a facial impression [bib_ref] Fabrication of an orbital prosthesis using a noncontact three-dimensional digitizer and rapid-prototyping..., Yoshioka [/bib_ref]. Using stereolithography, a RP technique, 3D computer models are converted to solid physical models, which allows better planning of the prosthesis. This technique was first used in oral and maxillofacial surgery by .
## Case report
A 28-year-old patient was referred to the Department of Prosthodontics seeking treatment for a left ocular defect. The patient revealed a history of severe facial injury sustained from a roadside accident two months back for which he had undergone primary treatment. Extraoral examination showed that the patient had a large ocular defect on the left side extending to the supraorbital region. Due to extensive soft tissue and hard tissue loss, the defect had a hollowed out appearance. The skin covering the defect was in satisfactory condition . Intraoral examination was not significant.
## Radiographic examination
Lateral cephalogram and anterior-posterior view of the skull revealed that the defect extended superiorly involving the supraorbital ridge, the frontal and parietal regions and inferiorly the infraorbital ridge and anterior one-third of the zygomatic arch . Considering the multidisciplinary approach, a pre-prosthetic surgical intervention was planned to optimize the patient's quality of life and success of the prosthetic treatment. After detailed discussion and thorough examination of the defect, it was planned to first surgically reconstruct the bony defect with heat cure acrylic resin plate. Data were extracted from the patient's computed tomographic (CT) scan to make a computer 3D model of the defect. Then, by stereolithography, a RP technique (layer-wise additive manufacturing), a solid physical model of the defect was fabricated .
Material used was Thermoplastic Build Material i.e. ABS-M30 Thermoplastic. Wax pattern (Modeling Wax, DPI, Mumbai, India) was made on the stereolithographic model and then the acrylic resin plate (DPI, Mumbai, India) was fabricated accurately . The acrylic plate was then surgically fixed with titanium screws to obtain proper contour of the defect . The patient was then recalled after one month following complete healing of the surgical site for the management of the ocular defect. Ocular examination revealed a healthy intraocular tissue bed. There was adequate depth and undercut between the upper and lower fornices for retention of the prosthesis .
## Fabrication of prosthesis
An impression of the ocular defect was made with an ocular acrylic impression tray attached to a 5 mL modified disposable syringe (Dispo Van). Light viscosity polyvinyl siloxane (Reprosil; Dentsply DeTrey GmbH, Konstanz, Germany) was injected into the socket through the modified tray syringe assembly. Sufficient material was injected to elevate the lid contour similar to the normal side and the patient was instructed to perform eye movements until the impression material sets. The impression was removed from the socket and initially poured to the level of the height of contour with type IV dental stone (KALROCK, Kalabhai Karson Pvt. Ltd., Mumbai, India). After setting of the stone, keyholes were made and the assembly was boxed and second layer was poured, to obtain a twopiece cast for the orientation of the ocular prosthesis. Wax pattern was fabricated with modeling wax (DPI, Mumbai, India). The fit was evaluated by observing the extensions into the fornices. Acrylic resin stock eye was selected with matching iris size, color, and approximate sclera shape. Peripheral and posterior surfaces were reduced and retentive grooves were made on the posterior surface .The modified stock eye was adapted to the wax pattern. A plastic sleeve was secured with sticky wax over the pupil, perpendicular to the plane of iris for aligning the prosthesis in correct relationship to the natural eye . Try-in was done to evaluate the bulk, lid contour and esthetics of the artificial eye. At this stage a wash impression was made with irreversible hydrocolloid of the stock eye to redefine the contour of the socket bed for proper retention and stability . After satisfactory adjustments, flasking and dewaxing were done. The mold was then packed with clear, heat cure acrylic resin (DPI, Mumbai, India). A tinge of pink colored resin was given at the inner and outer canthus of the eye mold to simulate the natural eye and then curing was done . After final finishing and polishing, the prosthesis was inserted [fig_ref] Figure 10: Post-operative photograph glass eye, an acrylic eye is easier to fit and... [/fig_ref].
# Discussion
Although various reconstructive and regenerative treatments have been developed, a facial prosthesis is an effective rehabilitation for the patients with congenital or acquired defects. The fabrication of a facial prosthesis requires numerous complicated techniques to obtain a morphologically satisfactory outcome [bib_ref] Fabrication of an orbital prosthesis using a noncontact three-dimensional digitizer and rapid-prototyping..., Yoshioka [/bib_ref]. A conventional impression can be uncomfortable for patients because most of the face has to be covered with impression material until it sets. Furthermore, the weight of the impression material or the patient's body posture during impression may lead to deformation and inaccuracies. Sculpting the wax prototype is also time consuming and requires professional training and skills [bib_ref] Fabrication of an orbital prosthesis using a noncontact three-dimensional digitizer and rapid-prototyping..., Yoshioka [/bib_ref]. Technologies such as 3D surface capture (3D scanning), 3D-CAD and layer additive manufacturing process (RP and manufacturing) have been investigated for maxillofacial prosthetic applications to create models for diagnosis, treatment planning and surgical simulation [bib_ref] Rapid Prototyping Technologies and their Applications in Prosthodontics, a Review of Literature, Torabi [/bib_ref]. In our case also similar method was used to fabricate solid physical model of the defect using CT scan of the patient and ABS-M30 Thermoplastic.
It not only eliminated the tedious conventional impression making procedure but also made the fabrication of the prosthesis much easier. Various other materials that can be used for the fabrication of these models are thermoplastic polymeric materials such as ABS-M30i Thermoplastic, PC Thermoplastic, PC-150 Thermoplastic, ULTEM 9085 Thermoplastic and etc. The fabrication of an artificial eye is not limited to this modern age. They have been used for centuries, with the earliest known examples found in mummies dating back to the fourth dynasty in Egypt (1613-2494 BC) [8]. Ambroise Pare a French surgeon dentist is considered to be the pioneer of modern artificial eye. In 1575, Pare fabricated artificial eye made of glass as well as porcelain. Techniques and materials were constantly improved and now the currently available prosthetic materials most commonly used are methacrylate or acrylic resins, polyurethane elastomers and silicones . In this case, a stock acrylic eye (20R/5, SMR, Ophthalmic, Mumbai, India) matching the patient's normal eye was selected to serve the purpose. Unlike As the patient had a large extensive defect involving the supraorbital ridge, frontal, parietal and infraorbital ridges along with the anterior one-third of the zygomatic arch, it was quite difficult to restore the defect. But the patient was satisfied with the outcomes of treatment. The patient was given proper instructions and recalled on regular basis.
# Conclusion
Despite remarkable advances in surgical management of maxillofacial defects, ocular and orbital defects cannot be satisfactorily managed by plastic surgery alone. Implant-retained ocular prosthesis is always a better option but may not be always feasible mainly due to economical reasons. In such cases, the conventional method of fabrication of ocular prosthesis using stock eye provides good results both functionally and esthetically. It actually rehabilitates the patient to live a normal socially acceptable life.
[fig] Figure 1, Figure 4: Pre-operative photograph Fig 2: Anterior-posterior radiograph Fig. 3: Stereolithography Wax pattern fabrication Fig. 5: Acrylic plate fixed Fig. 6: Intra-ocular tissue bed [/fig]
[fig] Figure 7: (A) Polyvinyl siloxane injected, (B) Impression obtained, (C) Modified tray-syringe assembly [/fig]
[fig] Figure 8 71, Figure 9: (A) Selected stock eye (B) Alignment of prosthesis; (C) Wash impression with irreversible hydrocolloid A B C C January 2016; Vol.13, No.1 www.jdt.tums.ac.ir Packing of eye prosthesis [/fig]
[fig] Figure 10: Post-operative photograph glass eye, an acrylic eye is easier to fit and adjust, unbreakable, inert to ocular fluids, esthetically good, longer lasting and easy to fabricate [8-10]. [/fig]
|
Spontaneous tumor lysis syndrome in a patient with advanced gastric adenocarcinoma: a case report
Tumor lysis syndrome (TLS) is an oncologic emergency that usually occurs after initial treatment of a malignant tumor. It manifests as hyperuricaemia, hyperkalaemia, hyperphosphataemia and hypocalcaemia, ultimately resulting in acute kidney failure, seizures, cardiac arrhythmias, and even death.Here, we report a very rare case of spontaneous TLS in a patient with advanced gastric adenocarcinoma who eventually succumbed to renal failure. Extra vigilance towards electrolyte imbalances should be given during initiation of therapy in cases of large gastric cancer with severe distant metastasis. Risk assessment prior to surgery, early diagnosis and comprehensive treatment strategies are vital in improving the prognosis of gastric cancer patients with TLS. Urgent hemodialysis should be implemented as soon as possible in order to prevent further renal deterioration.
# Introduction
Tumor lysis syndrome (TLS) is an oncologic emergency that usually occurs after the initial treatment of a malignant tumor. Chemotherapy often causes lysis of large quantities of tumor cells, leading to the release of intracellular contents into the blood which causes hyperuricaemia, hyperkalaemia, hyperphosphataemia and hypocalcaemia. These electrolyte and metabolic abnormalities ultimately result in symptoms such as acute kidney failure, seizures, cardiac arrhythmias, and even sudden death [bib_ref] Pathophysiology, clinical consequences, and treatment of tumor lysis syndrome, Davidson [/bib_ref] [bib_ref] Oncological emergencies: tumor lysis syndrome, Sobota [/bib_ref].
TLS commonly occurs in hematologic cancers in view of their high cell turnover, rapid proliferation rates and increased chemosensitivity [bib_ref] The tumor lysis syndrome, Howard [/bib_ref]. While the incidence of TLS has been increasingly reported in solid tumors [bib_ref] Tumor lysis syndrome following trastuzumab for breast cancer: a case report and..., Taira [/bib_ref] [bib_ref] Tumor lysis syndrome in a nonsmall cell lung cancer, Noyes [/bib_ref] [bib_ref] Spontaneous tumor lysis syndrome in renal cell carcinoma: a case report, Norberg [/bib_ref] [bib_ref] A Case of Spontaneous Tumor Lysis Syndrome in Malignant Melanoma, Takeuchi [/bib_ref] , TLS in gastric cancer is uncommon [bib_ref] An unusual presentation of tumor lysis syndrome in a patient with advanced..., Vodopivec [/bib_ref]. TLS usually occurs within a week of initiation of cytotoxic therapy and may occur spontaneously in certain circumstances [bib_ref] Acute spontaneous tumor lysis syndrome, Jasek [/bib_ref] [bib_ref] Spontaneous tumour lysis syndrome in hepatocellular carcinoma presenting with hypocalcemic tetany: An..., Agarwala [/bib_ref] [bib_ref] Spontaneous Tumor Lysis Syndrome in Small-Cell Lung Cancer: A © Translational Cancer..., Weerasinghe [/bib_ref]. Here, we report a rare case of spontaneous TLS in a patient with advanced gastric cancer.
## Case presentation
A 62-year-old gentleman was admitted in the Second Affiliated Hospital of Zhejiang University, School of Medicine, with a 20-day history of worsening weakness and poor appetite. This was associated with discomfort of the right flank, hard stools and a loss of weight of 5 kg over one month. No nausea, vomiting or fever were reported. The patient had no history of renal disease. Physical examination on admission revealed BMI of 24.2 (a height of 172 cm and a weight of 71.6 kg). Left cervical lymph nodes were not palpable. Mild tenderness was present in the epigastric region. Gastroscopy showed a huge annular irregular mass on the wall of lower two thirds of the gastric body with peripheral mucosa edema [fig_ref] Figure 1: Gastroscopy showed a huge annular irregular swelling mass on the gastric body [/fig_ref]. Biopsy confirmed the presence of a low-grade adenocarcinoma. Abdominal enhanced CT scan revealed a large-scale thickness (size of 9 cm) of the gastric wall with serosa invasion, infiltration of the left lower ureter, multiple enlarged perigastric lymph nodes (lymph nodes No. 2, 4, 5, 6), suspicious metastatic nodes in greater omentum, and a small amount of pelvic effusion [fig_ref] Figure 2: Radiological examinations [/fig_ref]. Furthermore, 18 F-fluorodeoxyglucose (FDG) PET/CT scan showed that diffuse enhancement in the stomach (SUVmax =6.72), perigastric lymph nodes (SUVmax =4.22), a blurry omentum and a small amount of pelvic effusion [fig_ref] Figure 2: Radiological examinations [/fig_ref].
Laboratory findings at admission are as follows: white blood cells: 13. Laparoscopic exploration was performed on the ninth day of admission, revealing numerous white nodules in the peritoneum, ligamentum teres hepatis, and pelvic cavity [fig_ref] Figure 3: Laparoscopy observed numerous white nodules in abdominal cavity [/fig_ref]. Intraoperative frozen section one nodule revealed metastatic low-grade adenocarcinoma. Numerous free tumor cells were observed upon microscopic inspection [fig_ref] Figure 4: Free tumor cells observed under microscopical examination of the peritoneal lavage [/fig_ref].
An intraabdominal catheter was inserted in preparation of subsequent intraperitoneal chemotherapy. Surprisingly, the patient's renal function deteriorated immediately after surgery, as evidenced by a serum creatinine level of 746 μmol/L. On the nineteenth day of admission, bilateral ureteral stents were inserted in order to alleviate ureteral obstruction. Subsequently, serum creatinine declined to 150 μmol/L. Nevertheless, this decrease was short-lived as serum creatinine began to rise gradually again. Potassium and phosphorus levels were slightly increased, while calcium levels remained low during the entire course of treatment. Inflammatory biomarkers such as white blood cells and C-reactive protein (CRP) were relatively high despite antiinfection therapy [fig_ref] Table 1: Laboratory examinations during the time of hospital [/fig_ref] [fig_ref] Figure 5: Trends of serum creatinine [/fig_ref]. The patient was diagnosed as having TLS during multi-disciplinary team (MDT) discussion of the case. Subsequently, the patient was treated with volume expansion, diuretics, sodium bicarbonate, anti-infective therapy, with the aim to improve his electrolyte imbalances. Despite these interventions, serum creatinine continued rising, reaching a peak of 1,005 μmol/L. The patient refused hemodialysis and eventually succumbed to renal failure a month after his initial surgery.
# Discussion
TLS is frequently reported during initiation of therapy against malignant tumors. It comprises of a series of metabolic abnormalities caused by massive release of intracellular contents into the blood that exceeds the ability of renal clearance. It is manifested by hyperuricaemia, hyperkalaemia, hyperphosphataemia and hypocalcaemia [bib_ref] Pathophysiology, clinical consequences, and treatment of tumor lysis syndrome, Davidson [/bib_ref] [bib_ref] Oncological emergencies: tumor lysis syndrome, Sobota [/bib_ref]. While there is currently no universal definition of TLS, the most accepted TLS classification was proposed by Cairo and Bishop, which was modified from works of Hande and Garrow [bib_ref] Acute tumor lysis syndrome in patients with high-grade non-Hodgkin's lymphoma, Hande [/bib_ref]. Based on the Cairo and Bishop
## Classification, tls is stratified into laboratory tls (l-tls)
and clinical TLS (C-TLS). L-TLS is defined as changes in any two or more measurable values, including uric acid of 8 mg/dL or greater, potassium 6 mmol/L or greater, phosphorus 2.1 mmol/L or greater, calcium 1.75 mmol/L or less, or a 25% change from baseline in any of these electrolytes. C-TLS is defined as presence of L-TLS and at least one clinical manifestation including renal insufficiency, arrhythmia, seizure or sudden death [bib_ref] Tumor lysis syndrome: risk factors, diagnosis, and management, Burns [/bib_ref]. The patient in our case suffered from hyperuricemia (8.74 mg/dL) and hyperphosphataemia (2.20 mmol/L) at baseline, and experienced a decrease in calcium at almost 25% from baseline as well as along with renal insufficiency (1,005 μmol/L). These findings fulfill the criteria of both L-TLS and C-TLS. TLS is usually associated with haematological malignancies in view of the large cell turnover, higher proliferation rates and increased chemosensitivity of these cells [bib_ref] Acute tumor lysis syndrome in patients with high-grade non-Hodgkin's lymphoma, Hande [/bib_ref] [bib_ref] Tumor Lysis Syndrome in Patients with Hematological Malignancies, Belay [/bib_ref] [bib_ref] Incidence of chemotherapy-related tumour lysis syndrome at Kenyatta National Hospital, Busakhala [/bib_ref]. Despite the increasing reports of TLS in solid tumors, the incidence of TLS in gastric adenocarcinoma is very rare [bib_ref] An unusual presentation of tumor lysis syndrome in a patient with advanced..., Vodopivec [/bib_ref]. To the best of our knowledge, there were only 5 reports of TLS in gastric adenocarcinoma [bib_ref] An unusual presentation of tumor lysis syndrome in a patient with advanced..., Vodopivec [/bib_ref] [bib_ref] Spontaneous acute tumor lysis syndrome with advanced gastric cancer, Woo [/bib_ref] [bib_ref] Spontaneous acute tumour lysis syndrome in gastric adenocarcinoma: a case report and..., Goyal [/bib_ref] [bib_ref] Tumor lysis syndrome after capecitabine plus cisplatin treatment in advanced gastric cancer, Han [/bib_ref] [bib_ref] Acute tumor lysis syndrome in the setting of advanced gastric cancer, Kobayashi [/bib_ref]. Three of them were chemotherapy-induced while another two developed spontaneous TLS. However, the patient in one of the case reports of spontaneous TLS was noted to have received chemotherapy 2 months prior to the development of TLS, while the other case report of a patient with spontaneous TLS was diagnosed at time of presentation. While the risk factors for the development of spontaneous TLS has yet to be identified, previous studies have revealed several risk factors that may be associated with spontaneous TLS in solid tumor, such as tumor extension, a large initial tumor burden, bulky tumors, extensive metastasis, extrinsic compression of the genitourinary tract by the tumor, tumor cells with high proliferative rate, and abnormal pretreatment laboratory findings such as elevated LDH, serum creatinine, and uric acid. Patient-related factors such as preexisting nephropathy, hypotension, and obstructive uropathy are also risk factors for developing TLS [bib_ref] An unusual presentation of tumor lysis syndrome in a patient with advanced..., Vodopivec [/bib_ref]. In this report, the patient was diagnosed with metastatic gastric cancer, an indication of a large tumor burden. Furthermore, this patient was found to have elevated LDH, serum creatinine, and uric acid levels at the point of presentation. Ureteral obstruction was also present in this patient. All these factors placed this patient at a high risk of developing TLS. Additionally, surgical procedures and severe infection may have further increased the risk of TLS in this patient as operative procedures have the potential to trigger tumor cell death or to stimulate tumor cell growth.
The metabolic abnormalities in TLS, if untreated, will lead to life-threatening complications such as acute kidney [bib_ref] An unusual presentation of tumor lysis syndrome in a patient with advanced..., Vodopivec [/bib_ref]. Early recognition and prevention are especially crucial for patients at high risk. Treatment of TLS focuses on correction of electrolyte disturbances and preservation of renal function. A typical TLS treatment regimen involves volume expansion, urinary alkalinisation, allopurinol, rasburicase, or even dialytic modalities if necessary [bib_ref] Tumour lysis syndrome: implications for cancer therapy, Mika [/bib_ref]. Our patient received volume expansion, urinary alkalinisation and correction of electrolyte disturbances as soon as he was diagnosed as TLS. Nevertheless, his renal function continued to deteriorate, and without the initiation of lifesaving haemodialysis, this patient eventually succumbed to his disease.
# Acknowledgments
We acknowledge Liang Han for his revision of English grammar. We acknowledge Lin Chen for providing PET-CT imaging. Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee(s) and with the Helsinki Declaration (as revised in 2013). Written informed consent was obtained from the relatives of the patient for publication of this case report and any accompanying images.
Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the noncommercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.
[fig] Figure 1: Gastroscopy showed a huge annular irregular swelling mass on the gastric body. [/fig]
[fig] Figure 2: Radiological examinations. (A,B) Abdominal enhanced CT scan revealed a large-scale thickness of gastric wall with serosa invasion (red arrow), multiple enlarged perigastric lymph nodes, suspicious metastatic nodes in greater omentum, and a small amount of pelvic effusion (red arrow); (C) 18 F-fluorodeoxyglucose (FDG) PET/CT scan showed diffuse enhancement in stomach (red arrow), perigastric lymph nodes, and a blurry omentum. [/fig]
[fig] Figure 3: Laparoscopy observed numerous white nodules in abdominal cavity. [/fig]
[fig] Figure 4: Free tumor cells observed under microscopical examination of the peritoneal lavage (H&E stain, ×400). [/fig]
[fig] Figure 5: Trends of serum creatinine (A), potassium, calcium (B), white blood cells (C) and C-reactive protein levels (D). [/fig]
[table] Table 1: Laboratory examinations during the time of hospital [/table]
|
Fetuin-B, a potential link of liver-adipose tissue cross talk during diet-induced weight loss–weight maintenance
BACKGROUND/OBJECTIVES: Numerous hepatokines are involved in inter-organ cross talk regulating tissue-specific insulin sensitivity. Adipose tissue lipolysis represents a crucial element of adipose insulin sensitivity and is substantially involved in longterm body weight regulation after dietary weight loss. Thus, we aimed to analyze the impact of the hepatokine Fetuin-B in the context of weight loss induced short-and long-term modulation of adipose insulin sensitivity. SUBJECTS/METHODS: 143 subjects (age > 18; BMI ≥ 27 kg/m 2 ) were analyzed before (T-3) and after (T0) a standardized 12-week dietary weight reduction program. Afterward, subjects were randomized to a 12-month lifestyle intervention or a control group. After 12 months (T12) no further intervention was performed until 6 months later (T18) (Maintain-Adults trial). Tissue-specific insulin sensitivity was estimated by HOMA-IR (predominantly liver), ISI Clamp (predominantly skeletal muscle), and free fatty acid suppression during hyperinsulinemic-euglycemic clamp (FFA Supp ) (predominantly adipose tissue). Fetuin-B was measured at all concomitant time points. RESULTS: Circulating Fetuin-B levels correlated significantly with estimates of obesity, hepatic steatosis as well as HOMA-IR, ISI Clamp , FFA Supp at baseline. Fetuin-B decreased during dietary weight loss (4.2 (3.5-4.9) vs. 3.8 (3.2-4.6) µg/ml; p = 2.1 × 10 −5 ). This change was associated with concomitant improvement of HOMA-IR (r = 0.222; p = 0.008) and FFA Supp (r = −0.210; p = 0.013), suggesting a particular relationship to hepatic and adipose tissue insulin sensitivity. Weight loss induced improvements of insulin resistance were almost completely preserved until months 12 and 18 and most interestingly, the short and long-term improvement of FFA Supp was partially predicted by baseline level of Fetuin-B. CONCLUSIONS: Our data suggest that Fetuin-B might be a potential mediator of liver-adipose cross talk involved in short-and long-term regulation of adipose insulin sensitivity, especially in the context of diet-induced weight changes. TRIAL REGISTRATION: ClinicalTrials.gov number: NCT00850629, https://clinicaltrials.gov/ct2/show/NCT00850629, date of registration: February 25, 2009. Nutrition and Diabetes (2021) 11:31 ; https://doi.
# Introduction
Long-term success of dietary weight loss interventions is known to be limited by frequently observed body weight regain. Recent data indicate that especially adipose tissue function might represent a crucial element in this phenomenon. Identification of molecular pathways regulating adipose function, particularly with respect to adipose insulin sensitivity, may therefore help to unveil the underlying mechanism involved in body weight regulation. Numerous cytokines produced in hepatocytes, myocytes, and adipocytes are involved in the inter-organ cross talk promoting a tissue-specific insulin efficacy. For example, the hepatokine Fetuin-A inhibits insulin receptor tyrosine kinaseand is increased in the diabetic and prediabetic state as well as in liver steatosis. It was proposed to induce myocellular insulin resistance in humansvia TLR4 mediated adipose tissue inflammation.
Fetuin-B, which shares about 22% homology at protein leveland belongs to the same family of cysteine protease inhibitors, was recently characterized as a separate hepatokine. It is apparently involved in plaque instability and vascular inflammationand increased levels were found in patients with coronary artery disease, acute coronary syndrome, and acute myocardial infarction. Very recently, elevated hepatic Fetuin-B mRNA expression or circulating Fetuin-B levels were reported in type 2 diabetesand liver steatosis. Fetuin-B seems to be particularly associated with intrahepatic lipid content, although controversial data were also reported recently. Even if functional similarities between both hepatokines were described, the physiological function seems to be different. The impact of Fetuin-B on insulin sensitivity, which was previously described for Fetuin-A, is not entirely clear. Although experimental data indicate a Fetuin-B mediated impairment of insulin action in hepatocytes as well as a decrease of insulin stimulated glucose uptake in myotubes, the glucose infusion rate during hyperinsulinemic-euglycemic clamps was not changed by Fetuin-B in mice. Peter and colleagues did not reveal significant correlations between circulating Fetuin-B and any estimate of insulin sensitivity in a cohort of comprehensively phenotyped humans. Nevertheless, a metabolic effect of Fetuin-B was recently supported by numerous data indicating a link to lipid metabolism. An association with elevated total cholesterol, LDL cholesterol, and triacylglycerol was described in somebut not in all cohorts. More specifically, knock down of Fetuin-B resulted in increased expression of fatty acid synthase, while CPT1, a key enzyme of lipid oxidation, was suppressed. These data suggest a stimulation of lipolytic pathways by Fetuin-B. However, the exact nature of this relationship remains known. In particular, a potential interaction with insulin-mediated inhibition of lipolysis is still unclear. Insulin-mediated effects on muscular glucose uptake, hepatic gluconeogenesis, and adipose lipid breakdown are strongly modified by body weight reduction. As increased adipose lipid turnover was recently shown to be crucial for sustained weight loss, such an effect might be especially relevant in the context of body weight reduction. Therefore, we were particularly interested in the relationship between Fetuin-B and insulin-mediated regulation of lipolysis (free fatty acid (FFA) suppression) during short and long-term course of diet-induced weight loss in humans.
## Materials and methods participants and exclusion criteria
The study was performed between 2010 and 2016 at the endocrine trial center of the Charité Medical School. Screening was performed to rule out abnormal thyroid function and hypercortisolism using 1 mg dexamethasone suppression tests. Moreover, any systemic disease or biochemical evidence of severe hepatic or renal dysfunction was also excluded. Individuals with recent weight changes of more than 5 kg during the last 2 months, with changes of smoking habits or diet behavior during the last 3 months were excluded. Furthermore, patients with severe chronic diseases such as instable coronary heart disease, severe renal insufficiency (eGFR < 30 ml/min), liver diseases, severe psychological diseases, severe endocrine disorders, cancer, chronic infections, or comparable chronic disorders were also excluded. Drugs modifying energy homeostasis and body weight were not allowed during this trial (with exception of thyroxin). The study protocols were approved by the Institutional Review Board of the Charité Medical School (EA1/140/12). All methods were performed in accordance with the relevant guidelines and regulations. Informed consent was obtained from all participants.
## Study design
Details of the performed weight loss-weight maintenance trial in adults (Maintain-Adult, ClinicalTrials.gov NCT00850629) were already described. In brief, we performed a 12-months randomized controlled weight maintenance intervention in 156 overweight or obese subjects (120 female and 36 male) (BMI ≥ 27 kg/m 2 ) after an initial weight loss period of 12 weeks. The major characteristics of the trial are shown in. After the initial weight loss period, we compared the effects of a 12months multimodal lifestyle intervention to maintain body weight with a control group within a randomized controlled trial.
Pre-trial weight loss phase. A structured weight reduction program (caloric restriction using a very low energy diet and nutritional counseling, physical exercises, and psychological advices) was used to achieve a weight loss of at least 8%. The detailed protocol was already reported previouslyand is given in the supplement.
Twelve-months randomized weight maintenance phase. Eligible subjects (n = 143, 112 female and 31 male) were randomized into the intervention or control group. Subjects in the control group were no longer involved in any form of counseling. A continuous counseling was performed in the intervention group for the next 12 months. This multimodal lifestyle intervention was comparable to sessions of the weight loss period. Details of the protocol and the intervention were already reportedand are described in the supplement.
Follow-up period. After 12 months all subjects (intervention and control group) underwent a free-living period of 6 months without any further active intervention.
## Randomization and masking
Randomization was performed by the study team using a stratified randomization list. Stratification considered gender and body weight at baseline (three BMI strata). Subjects were not blinded to group assignment.
## Phenotyping
A comprehensive phenotyping was performed before (T-3) and after (T0) weight loss, after 12 months of randomized weight maintenance intervention (T12), and after follow-up of 6 months (T18). All participants received a dietary recommendation of a balanced energy intake for the 3 days preceding phenotyping. The phenotyping focussed on anthropometric, hormonal, and metabolic evaluation. Following a 10-h overnight fast, all patients were investigated at the endocrine trial center of the Charité Medical School at 8.00 a.m. Waist circumference was measured three times and the mean was calculated. At 9.00 a.m. fasting blood samples were taken. Moreover, subjects also underwent a body impedance analysis using AKERN BIA 101 (SMT medical GmbH & Co. KG, Würzburg, Germany) and a hyperinsulinemic-euglycemic clamp at T-3, T0, and T12 as previously described. Blood samples were centrifuged, and plasma and serum samples were frozen immediately at −80°C.
## Outcomes
The primary outcome defined as weight regain after 18 months (absolute change of BMI from T0 to T18 (kg/m 2 )) was reported previously. Predefined secondary outcomes were the analysis of hormonal, transcriptional, and metabolic predictive markers of body weight regain, metabolic improvement, and cardiovascular risk factors. Currently, we report the data of secondary analyses including estimates of tissue-specific insulin sensitivity and Fetuin-A and B as predictors of long-term changes of adipose insulin sensitivity.
## Laboratory analyses
Capillary blood glucose was measured using the glucose oxidase method (Dr. Müller Super GL, Freital, Germany). Triglycerides, cholesterol, LDL-and HDL-cholesterol, CRP, and liver enzymes were measured by standard laboratory methods using Cobas ISE direct and c111 Analyzer (Roche Diagnostics, Mannheim, Germany). Serum insulin was measured by fluoroimmunometric assay (AutoDelfia; Perkin Elmer, Rodgau, Germany) (inter-assay CV 2.3-3.5%, intra-assay CV 1.7-2.4%). Non-esterified fatty acids (FFA) were quantified in serum using a commercially available colorimetric assay (NEFA HR2, Wako, Neuss, Germany) performed on ABX Pentra 400 (HORIBA ABX, Montpellier, France) (inter-assay CV < 5%, intraassay CV < 1%). Fetuin-B was assessed in serum samples by ELISA using a commercial ELISA kit (Human Fetuin B DuoSet ELISA; R&D Systems, Minneapolis, USA), following the manufacturer's protocol (intra-assay CV: 5.3%, inter-assay CV: 8.2%). Plasma fetuin-A was also determined by an ELISA method (BioVendor Laboratory Medicine, Brno, Czech Republic (intra-assay CV: 3.6%, inter-assay CV: 4.5%)) in accordance with the manufacturer's instructions.
## Statistics and calculations
Insulin sensitivity was assessed by dividing the average glucose infusion rate (GIR, mg glucose/min) during the steady state of the hyperinsulinemic-euglycemic clamp by the body weight (M-value). The insulin sensitivity index (ISI Clamp ), which reflects predominantly skeletal muscle insulin sensitivity, was calculated as ratio of M-value to the serum insulin concentration (I, mU/l) in this period of the clamp. HOMA-IR was calculated to assess whole-body insulin sensitivity, which is predominantly driven by hepatic insulin efficacy. Effects of insulin on lipolysis were calculated by the suppression of FFA during hyperinsulinemic-euglycemic clamp and expressed as relative changes compared to fasting FFA levels (FFA Supp ). Hepatic steatosis index (HSI) was calculated using liver enzymes, BMI, gender, and diabetes state as described by Lee and colleagues. Statistical procedures were performed using SPSS version 25.0 (SPSS Inc., Chicago, IL, USA) and SAS software, version 9.4 (SAS Institute). Data reported herein reflect secondary analyses and are based on per protocol analysis including data of all available participants at the corresponding time point. Comparisons were made via paired Student's t-test for normally distributed data and Wilcoxon test for skewed data. Correlations between variables were investigated by Pearson's correlation coefficient for normally distributed data or Spearman's rank correlation coefficient for skewed data. Data were adjusted for age and gender as described in the "Results" section. Results were considered significant, if the two-sided α was below 0.05. Data were presented as median and limits of the interquartile range (IQR: 25th-75th percentile) and plotted as raw values unless stated otherwise.
Multivariate linear regression models were performed to analyze the impact of baseline fasting Fetuin-A or Fetuin-B levels at T-3 on weight lossinduced relative changes of HOMA (ΔHOMA) and FFA Supp (ΔFFA Supp ). These models included age, gender, and concomitant decline of BMI as potential confounders. Comparable regression models were performed to analyze the impact of baseline Fetuin-B levels on long-term improvement of HOMA and FFA Supp . As the hyperinsulinemic-euglycemic clamp was only performed at T-3, T0, and T12, we have chosen these time points for calculation. Given the effect of the 12 months intervention on BMIat T12, the models regarding long-term improvement were also adjusted for treatment group and concomitant BMI changes. Finally, the additive predictive effect of both Fetuin-B on the dependent variables was assessed by likelihood ratio test comparing the models including or excluding Fetuin-B.
Time course of CRP, HSI, liver enzymes, HOMA-IR, ISI Clamp , FFA Supp , Fetuin-A, and Fetuin-B levels between T-3 and T18 was analyzed using mixed-model, repeated-measures analyses of variance, which considered the correlation between repeated observations and used all available subsequent observations for all participants with values at randomization, regardless of further assessment completion. Means were modeled as a function of study visit (T-3, T0, T12, and T18). The model included adjustment for gender, age, group assignment, and BMI at baseline. An unstructured covariance structure was used. P values were adjusted for multiple testing using Bonferroni correction.
# Results
Baseline characteristics of the participants are shown inas reported previously. Both, Fetuin-A and Fetuin-B decline with increasing age. Interestingly, Fetuin-B levels were associated with estimates of obesity, liver steatosis (HSI), and insulin resistance (high HOMA-IR, low ISI Clamp , and weaker insulin-mediated suppression of FFAs (FFA Supp )). Although some of these associations were also found for Fetuin-A, no relationship was seen to estimates of obesity and FFA Supp. Given potential effects of age and gender on both fetuins, we repeated these analyses including adjustment for age and gender, which confirmed all relationships. To support an independent association of Fetuin-B and FFA Supp , an additional adjustment for BMI, HOMA-IR, and ISI Clamp (all known to be associated with FFA Supp ) was performed. Thereby the relationship between Fetuin-B and FFA Supp was slightly attenuated (r = 0.206, p = 0.018), but was still observable.
While Fetuin-A levels did not differ between females and males. Fetuin-A as well as Fetuin-B declined during weight loss. The decrease of Fetuin-B correlates with improvement of estimates of obesity (BMI (r = 0.337; p = 4.2 × 10 −5 ), fat mass (r = 0.281; p = 0.002), waist circumference (r = 0.262; p = 0.002)), ΔHSI (r = 0.179; p = 0.034), whole body (ΔHOMA-IRas well as adipose insulin sensitivity (ΔFFA Supp. No relationship to ΔISI Clamp (r = −0.135; p = 0.115) was found. Additional adjustment for gender and age did not substantially modify these findings (BMI: r = 0.329; p = 8. In contrast, weight loss induced decline of Fetuin-A was not significantly correlated with the extent of body weight reduction (r = 0.156; p = 0.063), ΔFFA Supp (r = −0.054; p = 0.527), ΔWC (r = 0.087; p = 0.302), ΔFM (r = 0.108; p = 0.244) and ΔHSI (r = 0.128; p = 0.130). Only a weak correlation with ΔHOMA-IR and ΔISI Clampcould be detected.
After the end of the weight loss intervention, BMI increased between T0 and T18 (1.9 (1.3-2.5) kg/m 2 ),. Although this regain was accompanied by a moderate impairment of HOMA-IR, ISI Clamp , and FFA Supp , those parameters were still improved at T12 (HOMA-IR, ISI Clamp , and FFA Supp ) and T18 (HOMA-IR) compared to baseline. In contrast, elevation of Fetuin-A and Fetuin-B between T0 and T18 resulted in fetuin levels at T12 and T18 almost comparable to baseline. Accordingly, estimates of liver steatosis (transaminases, HSI) and inflammation (CRP) demonstrated a longterm improvement between baseline and T18. These data were adjusted for baseline BMI, age, gender, and randomization state. Interestingly, long-term improvement of HSI was correlated with the decline of Fetuin-A (r = 0.336; p = 4.3 × 10 −4 ) and Fetuin-B (r = 0.216; p = 0.027).
## Impact of basal fetuin-b on short-term modulation of insulin sensitivity
Based on the relationship between HOMA-IR, FFA Supp , and Fetuin-B levels at baseline and during weight loss, we wondered if baseline Fetuin-B may be also predictive for weight lossinduced improvement of these estimates of insulin resistance, independent of body weight reduction. To answer this question, we performed a linear regression analysis including age, gender, ΔBMI, and fasting Fetuin-B levels before weight loss as independent variables. While ΔHOMA-IR was not related to basal Fetuin-B levels, a higher ΔFFA Supp was associated with male gender and higher Fetuin-B levels at baseline. Although this model explained only about 7% of the ΔFFA Supp during weight loss, a comparable model without Fetuin-B explained a substantially lower variability (R 2 = 0.026; p = 0.018 for comparison between models). Furthemore, we additionally adjusted all analyses for ΔFetuin-B, as ΔFetuin-B was associated with basal Fetuin-B levels. This did not modify the results. Notably, basal Fetuin-A was not related to ΔFFA Supp
## Impact of basal fetuin-b on long-term modulation of insulin sensitivity
Given the predictive impact of Fetuin-B on short-term improvement of FFA Supp during weight loss (ΔFFA Supp ), we finally aimed to analyze, whether baseline Fetuin-B levels can also predict longterm changes of adipose insulin sensitivity. Therefore, we performed comparable linear regression models including age, gender, fasting baseline Fetuin-B levels, randomization group, and concomitant long-term change of BMI. Basal Fetuin-B was again not related to long-term changes of HOMA-IR. However, in addition to male gender and long-term weight changes, higher baseline Fetuin-B levels were associated with Δ T3T12 FFA Supp. Additional adjustment for Δ T3T12 Fetuin-B did not modify this finding. A comparable model without Fetuin-B explained a substantially lower variability (R 2 = 0.098; p = 0.002 for comparison between models) than the full model. In analogy to previous analyses, basal Fetuin-A levels were not predictive for changes of adipose insulin sensitivity .
# Discussion
Hepatic cytokines have been shown to affect different insulin target tissues. In accordance with the metabolic impact of Fetuin-A, our findings confirmed previously described associations of Fetuin-A with estimates of liver steatosis as well as the impact on whole body and myocellular insulin resistance. Fetuin-B could represent a novel player also involved in the metabolic inter-organ cross talk. In line with such an assumption, we confirmed data revealing a relationship of Fetuin-B to estimates of obesity, liver fat, whole body, and myocellular insulin resistance, even if this was not described in all cohorts. However, previous findings also demonstrate a correlation of Fetuin-B with parameters of lipid metabolism. This suggests a potential role of Fetuin-B in the regulation of adipose tissue function. As a substantial role of weight loss-induced modification of adipose tissue function on body weight maintenance was revealed recently by us, Fetuin-B might be of particular interest for long-term body weight regulation. Now we report for the first time a relationship between Fetuin-B and insulin-dependent suppression of FFAs, an estimate of adipose insulin sensitivity. Given the independent association of high Fetuin-B levels with low insulin-mediated suppression of adipose tissue lipolysis it is tempting to speculate, that Fetuin-B might represent a crucial element diminishing antilipolytic effects of insulin, e.g., during postprandial hyperinsulinemia. The observed association of Fetuin-B with fasting FFAs suggests that Fetuin-B could be also relevant in the regulation of fasting lipolysis. However, this relationship could vice versa also reflect a FFA mediated increase of Fetuin-B release, as numerous polyunsaturated fatty acids can activate farnesoid X receptor (FXR), a nuclear receptor known to induce hepatic Fetuin-B expression.
In the context of such an effect of Fetuin-B on lipolysis, elevated Fetuin-B levels in obesity and fatty liver disease might represent a counterbalancing mechanism to attenuate further lipid storage. Accordingly, a weak association of weight loss-induced short-and long-term decline of Fetuin-B and estimates of liver steatosis (HSI) was revealed in our cohort. Interestingly, this assumption is further supported by experimental data. Fetuin-B knockdown in liver cells resulted in increased lipid content and expression level of genes involved in fatty acid synthesis, whereas key enzymes of fatty acid oxidation declined.
Our study revealed that increased Fetuin-B level in obesity could be attenuated by dietary weight loss. This would result in a reduction of previously described antilipogenic properties of Fetuin-B. Actually, weight loss-induced decline of Fetuin-B was associated with an increase of insulin-mediated FFA suppression in our trial. This interaction seems to be specific for adipose tissue, as changes of myocellular insulin sensitivity were not related to Fetuin-B reduction.
Although our data mostly reflect associations, it is tempting to speculate that elevated Fetuin-B levels represent a novel mechanism supporting adipose insulin resistance found in subjects with obesity and increased liver fat. Interestingly, higher Fetuin-B levels are apparently also relevant for regulation of adipose tissue function during diet-induced weight loss-weight maintenance, as both short and long-term improvement of adipose insulin sensitivity was independently associated with higher baseline levels of Fetuin-B. Even if the cellular structures potentially underlying such an interaction of Fetuin-B and adipose insulin resistance are currently unknown, this adds relevant novel aspects to existing knowledge. Thus, further research is clearly warranted to elucidate the details and even the direction of the here observed relationship.
Parts of our findings are in contrast to recent data. Although we confirmed previous reports demonstrating a weight loss-induced decline of Fetuin-A, the decline of Fetuin-B was modest in most yet analyzed cohorts. Differences of baseline characteristics and especially the lower number of study participantsor the degree of weight lossmight explain the stronger and significant effects observed in our study. Furthermore, Peter and colleagues demonstrated a relationship of weight loss-induced changes of insulin sensitivity and modification of Fetuin-A but not Fetuin-B. However, these data were primarily based on estimates of global insulin sensitivity during oral glucose load, while we aimed to use estimates of tissue-specific insulin sensitivity. This may provide a different perspective into the interaction of Fetuin-B with insulin action and may explain the different findings. Nevertheless, the interpretation of our data is limited by some additional factors. First, as most of our data are based on associations, further data are clearly required. However, some experimental and animal findings are in accordance with our results in humans and provide potential molecular explanations. Next, several behavioral, social, and environmental factors are known to have substantial impact on long-term effects of dietary weight loss interventions. Although we aimed to standardize the dietary intake and physical activity during the group sessions, we cannot exclude that these factors may have influenced our results. Moreover, hepatic steatosis was not measured directly, even if the HSI is used in several trials as a surrogate of liver steatosis. Finally, numerous techniques are described to assess adipose tissue insulin sensitivity. Even if hyperinsulinemiceuglycemic clamp is considered as a well-established analytical approach, different insulin infusion rates (4-80 mU/m 2 , partly used stepwise)and outcome measures (suppression of plasma FFAs, suppression of plasma glycerol or glycerol and FFA rates of appearance using tracer dilution technique)were used. Interestingly, results based on low dose insulin infusion rates might be influenced by changes of lipolysis in skeletal muscle, which is rather stable during high insulin dosage. These data using tissue-based microdialysis technique indicate that a higher dose of insulin, which results in comparable insulin levels during steady state as seen in our participants, should be preferred to differentiate suppression of lipolysis in skeletal muscle and adipose tissue. Although tracer dilution technique is widely accepted to assess adipose tissue lipolysis, suppression of FFA during hyperinsulinemiceuglycemic clamp is also frequently used. Given these data, we believe that we used a valid approach, even if adipose tissue microdialysis was not performed in this study.
On the other hand, numerous strengths of the current trial should be mentioned. These include the large sample size, the long duration of the intervention, subsequent observation, and the comprehensive phenotyping including detailed assessment of myocellular and adipose tissue insulin sensitivity by hyperinsulinemic-euglycemic clamp in such a large cohort.
Taken together, we revealed a novel relationship of Fetuin-B to several markers of insulin resistance, especially estimates of adipose tissue function. If this would reflect a causal relationship, such a mechanism could counterbalance further lipid storage in obesity and fatty liver disease. Weight loss-induced changes of Fetuin-B were primarily associated with an increase of whole body and adipose insulin sensitivity. Most interestingly, basal Fetuin-B level were related to short-and long-term improvement of adipose insulin sensitivity inhibiting lipid breakdown during hyperinsulinemia. Considering previous experimental data, Fetuin-B might be therefore linked to insulin-dependent FFA metabolism in adipose tissue, which highlighted this molecule as a promising target involved in long-term regulation of adipose tissue function and body weight regulation. In any case, our prospective data support that further studies should elucidate the role of Fetuin-B in the context of organ-specific glucose and lipid metabolism.
## Data availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request. |
Transient mydriasis in toxoplasmosis retinochoroiditis
A B S T R A C TPurpose: To report a case of transient anisocoria and mydriasis in a 14 year old boy with toxoplasmosis retinochoroiditis. Observation: The patient presented with panuveitis and mydriasis which persisted for 18 days and spontaneously resolved. Conclusions and importance: Mydriasis is a rare potential neurological manifestation of toxoplasmosis retinochoroiditis. Clinicians should be aware of this rare cause of anisocoria.
# Introduction
Toxoplasma gondii is an intracellular protozoan parasite which may cause ocular and systemic manifestations in the affected host. Congenital ocular toxoplasmosis is frequently bilateral and acquired toxoplasmosis lesions are typically unilateral although bilateral presentation may occur in approximately 30% of cases.The most common finding in ocular toxoplasmosis is retinochoroiditis, a disease in which necrosis of the retinal fiber layer leaves a pigmented scar that can result in decreased vision and potential recurrence. The diagnosis of ocular toxoplasmosis is made by testing for IgG serology in addition to clinical findings. Polymerase chain reaction testing of aqueous specimens can may useful in establishing the diagnosis of undifferentiated retinitis. Intraocular inflammation typically resolves within 6 weeks in immunocompetent individuals. Neurological manifestations of both congenital and acquired toxoplasmosis include but are not limited to seizures, paresis, hemianopia, ataxia or cranial nerve palsies. [bib_ref] Longitudinal study of new eye lesions in children with toxoplasmosis who were..., Phan [/bib_ref] [bib_ref] Ophthalmic outcomes of congenital toxoplasmosis followed until adolescence, Wallon [/bib_ref] Pupil miosis is frequently observed in eyes with panuveitis if anterior chamber inflammation is present. Herein we report a case in which transient mydriasis developed in a symptomatic patient with toxoplasmosis retinochoroiditis.
## Case report
A 14 year old boy with a history of pica as a child was admitted to the neurology service with decreased vision, headache and right sided mydriasis, greatest under bright illumination conditions [fig_ref] Figure 1: External photograph demonstrating right mydriasis and anisocoria greater in bright light [/fig_ref]. No pupil constriction could be induced with light or accommodation. There was no history of prior exposure to medications with mydriatic or cycloplegic effects. Examination of old photographs did not reveal any evidence of prior anisocoria. Uncorrected visual acuity was 20/60 in the right eye without pinhole improvement. Visual motility was full and the eyes were orthotropic. Eye pain did not occur with eye movement and there were no signs of orbital inflammation. Examination of the right eye revealed intraocular pressure of 30 mmHg. Small keratic precipitates were present on the endothelium inferiorly, the cornea was otherwise normal in appearance. One plus cell was present both in the anterior chamber and vitreous. A focus of necrotizing retinitis in the temporal midperiphery, with numerous satellite lesions along both arterioles and venules, centered approximately 15mm from the optic disc [fig_ref] Figure 2: Color fundus photograph demonstrating focus of retinochoroiditis in the temporal midperiphery of... [/fig_ref]. The anterior chamber was deep and the crystalline lens was clear. The left eye was normal. Visual evoked potentials, magnetic resonance images of brain and orbits, magnetic resonance angiogram, and lumbar puncture tests were normal. A complete blood cell count, complete metabolic panel, erythrocyte sedimentation rate, antinuclear antibody, double stranded DNA, and rheumatoid factor were all normal or negative. There was serologic evidence of prior exposure to toxoplasma gondii and cytomegalovirus, the remaining of the serologies for infectious causes of necrotizing retinitis were negative [fig_ref] Table 1: Results of serologic evaluation [/fig_ref]. Because the clinical presentation was most consistent with toxoplasmosis retinochoroiditis and was not typical of cytomegalovirus retinitis he was treated with trimethoprim 160mg/sulfamethoxizole 800mg twice daily for 6 weeks, prednisolone acetate 1% four times daily and timolol maleate 0.5% once daily. No pilocarpine or other miotic, mydriatic or cycloplegic agents were instilled into either eye during the first 2 months after presentation. One week after discharge https://doi.org/10.1016/j.ajoc.2020.100700 Received 25 October 2018; Received in revised form 17 August 2019; Accepted 5 April 2020 from the hospital the intraocular pressure was normal, the anterior chamber was quiet and the headache had resolved. Mydriasis was still present at the next outpatient visit, 18 days following onset, but had completely resolved without sequelae by the time of an outpatient visit 2 months after presentation. Uncorrected visual acuity in the right eye recovered to 20/20, transillumination defects of the iris did not develop and the area of retinitis evolved into a chorioretinal scar.
# Discussion
Toxoplasmosis is a rare cause of anisocoria. There are two reports in the literature discussing the findings of anisocoria in patients with toxoplasmosis infection. In one case, a 23 year old Caucasian woman with a macular scar presumed secondary to toxoplasmic retinochoroiditis was reported to have an afferent pupillary defect and anisocoria. [bib_ref] Pupil motor perimetry, Roy [/bib_ref] In a series of 43 patients with anisocoria, 16 were reported to have active toxoplasma gondii cervical adenitis, two of which presented symptomatically with headaches and decreased visual fixation. [bib_ref] Anisocoria in active toxoplasmosis, Somoza [/bib_ref] Pupil miosis occurs when anterior uveitis results in breakdown of the blood ocular barrier with subsequent release of prostaglandins into the aqueous fluid. Pupil mydriasis may occur in anterior uveitis associated with herpetic viral ocular infection or when the intraocular pressure rises to levels high enough to cause ischemia of the iris sphincter. In this case, mydriasis persisted for 18 days despite normalization of IOP and the intraocular pressure was not elevated to levels which are known to cause ischemia of the pupil sphincter. Although we cannot exclude the possibility that the intraocular pressure may have been elevated to higher levels prior to presentation, there was no corneal edema nor did the patient report vision worse than at presentation to suggest greater elevation of the intraocular pressure. No palpable cervical adenopathy was present and the cervical lymph nodes did not appear enlarged on magnetic resonance angiography images of the neck. We speculate that diffuse involvement of the temporal fundus along with inflammation overlying the long ciliary nerve may have been causative factors in this case.
# Conclusions
Mydriasis in toxoplasmosis is a rare yet potential neurological manifestation of toxoplasmosis retinochoroiditis. Clinicians should be aware of this rare cause of anisocoria.
## Patient consent
Consent to publish this case has been obtained from the patient and legal guardians in writing. Institutional review board waiver was granted.
[fig] Figure 1: External photograph demonstrating right mydriasis and anisocoria greater in bright light. [/fig]
[fig] Figure 2: Color fundus photograph demonstrating focus of retinochoroiditis in the temporal midperiphery of right eye with multiple satellite lesions along both arterioles and venules. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.) [/fig]
[table] Table 1: Results of serologic evaluation. [/table]
|
Constructing Endophenotypes of Complex Diseases Using Non-Negative Matrix Factorization and Adjusted Rand Index
Complex diseases are typically caused by combinations of molecular disturbances that vary widely among different patients. Endophenotypes, a combination of genetic factors associated with a disease, offer a simplified approach to dissect complex trait by reducing genetic heterogeneity. Because molecular dissimilarities often exist between patients with indistinguishable disease symptoms, these unique molecular features may reflect pathogenic heterogeneity. To detect molecular dissimilarities among patients and reduce the complexity of high-dimension data, we have explored an endophenotype-identification analytical procedure that combines non-negative matrix factorization (NMF) and adjusted rand index (ARI), a measure of the similarity of two clusterings of a data set. To evaluate this procedure, we compared it with a commonly used method, principal component analysis with k-means clustering (PCA-K). A simulation study with gene expression dataset and genotype information was conducted to examine the performance of our procedure and PCA-K. The results showed that NMF mostly outperformed PCA-K. Additionally, we applied our endophenotype-identification analytical procedure to a publicly available dataset containing data derived from patients with late-onset Alzheimer's disease (LOAD). NMF distilled information associated with 1,116 transcripts into three metagenes and three molecular subtypes (MS) for patients in the LOAD dataset: MS1 (n 1~8 0), MS2 (n 2~7 3), and MS3 (n 3~2 3). ARI was then used to determine the most representative transcripts for each metagene; 123, 89, and 71 metagene-specific transcripts were identified for MS1, MS2, and MS3, respectively. These metagene-specific transcripts were identified as the endophenotypes. Our results showed that 14, 38, 0, and 28 candidate susceptibility genes listed in AlzGene database were found by all patients, MS1, MS2, and MS3, respectively. Moreover, we found that MS2 might be a normal-like subtype. Our proposed procedure provides an alternative approach to investigate the pathogenic mechanism of disease and better understand the relationship between phenotype and genotype.
# Introduction
The identification of genes that contribute to human disease is an important step toward understanding disease etiology and can facilitate the development of diagnostic tools, preventive medicine, and novel treatments. Complex diseases are caused by multiple genetic, environmental, and behavioral factors. If a disease has heterogeneous etiologies, then the detection of operable genes is difficult as one set of genes can be important for one etiology, but not another. Therefore, the identification of the genetic determinants of complex diseases is difficult. Endophenotype is an intermediate phenotype that combined genetic factors associated with a disease to reduce genetic heterogeneity [bib_ref] The endophenotype concept in psychiatry: Etymology and strategic intentions, Gottesman [/bib_ref]. This approach assumes that complex diseases can be described by sets of simple and measurable disease characteristics, with each characteristic representing a basic biological phenomenon. In the literature, synonyms for endophenotype include intermediate phenotype, biological marker, and sub-clinical trait, although each term has slightly different implications [bib_ref] Subclass Mapping: Identifying Common Subtypes in Independent Disease Data Sets, Hoshida [/bib_ref] [bib_ref] A method for predicting disease subtypes in presence of misclassification among training..., Zhang [/bib_ref] [bib_ref] Integrative Genomics Analyses Reveal Molecularly Distinct Subgroups of B-Cell Chronic Lymphocytic Leukemia..., Mosca [/bib_ref] [bib_ref] Endophenotype properties of niacin sensitivity as marker of impaired prostaglandin signalling in..., Smesny [/bib_ref]. The endophenotype approach may be useful in exploring different pathways leading to the onset of a complex disorder. For example, patients with the same diagnosis may differ greatly in the number and severity of symptoms, suggesting heterogeneity in the causal pathways [bib_ref] Molecular Phenotypes Distinguish Patients with Relatively Stable from Progressive Idiopathic Pulmonary Fibrosis..., Boon [/bib_ref] [bib_ref] Breast Cancer Molecular Subtypes in Patients With Locally Advanced Disease: Impact on..., Huber [/bib_ref] [bib_ref] Distinct molecular mechanisms underlying clinically relevant subtypes of breast cancer: gene expression..., Sorlie [/bib_ref]. Therefore, the creation of more homogeneous subgroups of patients based on their endophenotypes may facilitate our understanding of the involved biological processes.
The identification of disease subtypes is important because homogeneous groups likely reflect stronger clinical, pathological, and genetic coherence, and this may facilitate the understanding of the mechanisms underlying a disease. The molecular heterogeneity of a complex disease may suggest the existence of molecular subtypes [bib_ref] Complex disease, gender and epigenetics, Kaminsky [/bib_ref] [bib_ref] Gene Expression Profiles of Tumor Biology Provide a Novel Approach to Prognosis..., Anguiano [/bib_ref]. Genomic tools such as DNA microarrays hold great potential for the deciphering of the molecular patterns of disease and the identification of new and improved clinical markers. Gene expression profiling has been applied extensively to studies on gene function, gene regulation, cellular processes, and disease subtypes. Many human genes show natural variation in expression levels [bib_ref] Genetic analysis of genome-wide variation in human gene expression, Morley [/bib_ref] [bib_ref] Mapping determinants of human gene expression by regional and genome-wide association, Cheung [/bib_ref] , which suggests that gene expression levels may be used to establish endophenotypes to identify genes that confer disease susceptibility [bib_ref] Gene expression levels as endophenotypes in genome-wide association studies of Alzheimer disease, Zou [/bib_ref] [bib_ref] A neural network model for constructing endophenotypes of common complex diseases: an..., Lynn [/bib_ref] [bib_ref] Using endophenotypes for pathway clusters to map complex disease genes, Pan [/bib_ref].
In general, gene expression datasets contain thousands of genes derived from a relatively small number of samples. The gene expression data can be represented by a matrix A (A m|n ) of m transcripts in n samples. As such, standard statistical methods are not appropriate for analyzing gene expression data. Unsupervised clustering methods represent an alternative approach for exploring molecular dissimilarities among patients. To date, several methods have been applied for dimension reduction such as principal component analysis (PCA) [bib_ref] Principal component analysis for clustering gene expression data, Yeung [/bib_ref] , singular value decomposition [bib_ref] Singular value decomposition for genome-wide expression data processing and modeling, Alter [/bib_ref] , and independent component analysis [bib_ref] A review of independent component analysis application to microarray gene expression data, Kong [/bib_ref]. These methods capture overall gene behaviors that cluster genes based on global similarities in their expression data [bib_ref] Biclustering of gene expression data by non-smooth non-negative matrix factorization, Carmona-Saez [/bib_ref]. Recently, Lee and Seung [bib_ref] Learning the parts of objects by non-negative matrix factorization, Lee [/bib_ref] proposed non-negative matrix factorization (NMF), a matrix factorization method, A&W |H, where the elements of A, W , and H are all non-negative. NMF imposes non-negative constraints to detect local gene behaviors, in contrast with the approaches used by other linear representation clustering methods. NMF differs from PCA and singular value decomposition by enforcing the constraint that the two factors W and H must be non-negative, i.e. all elements must be equal to or greater than zero, and the factorization of matrices is generally non-unique. NMF has been applied to microarray data, protein sequence data, and data from neuroscience studies [bib_ref] Molecular Characterization of Breast Cancer with High-Resolution Oligonucleotide Comparative Genomic Hybridization Array, Andre [/bib_ref] [bib_ref] Using non-negative matrix factorization for single-trial analysis of fMRI data, Lohmann [/bib_ref] [bib_ref] Gene expression profiling of advanced-stage serous ovarian cancers distinguishes novel subclasses and..., Yoshihara [/bib_ref].
NMF creates a small number of gene subspaces from all of the genes in a genome and summarizes the sample gene expression patterns in each of the gene subspaces [bib_ref] Metagenes and molecular pattern discovery using matrix factorization, Brunet [/bib_ref]. These gene expression patterns are then used to cluster samples into distinct tumor types and subtypes. NMF is superior to both hierarchical clustering and self-organizing mapping in subtype discovery. A previous study has compared the performance of dimension reduction using NMF and PCA with several widely studied, related cancer microarray datasets and has shown that NMF outperformed PCA in reducing data dimensionality [bib_ref] Reducing microarray data via nonnegative matrix factorization for visualization and clustering analysis, Liu [/bib_ref].
In the present study, with the advantages of higher clustering accuracy and superior dimension reduction, we investigated an endophenotype-identification analytical procedure that first applies NMF to explore the potential molecular dissimilarities of a complex disease based on high-throughput microarray data and then uses the adjusted rand index (ARI) [bib_ref] Learning parts-based representations of data, Ross [/bib_ref]. ARI is a measure of the similarity of two clustering groups of a data set. In our study, ARI was applied to select informative transcripts for each molecular subtype. Both NMF and PCA are methods that reduce the dimension linearly and aim to find a small set of transcripts that describe the underlying information contained in A m|n with a lower dimension, k. We also used a simulation to compare the efficiency of NMF and PCA with k-means analysis (PCA-K) for endophenotype construction. Finally, we used a publicly available dataset derived from patients with late-onset Alzheimer's disease (LOAD) [bib_ref] Genetic Control of Human Brain Transcript Expression in Alzheimer Disease, Webster [/bib_ref] to evaluate the feasibility of our proposed procedure.
# Materials and methods
Endophenotype-identification Analytical Procedure: Molecular Subtype Construction via NMF NMF is an algorithm based on decomposition by parts that can reduce the dimension of expression data from thousands of genes to a handful of gene sets [bib_ref] Metagenes and molecular pattern discovery using matrix factorization, Brunet [/bib_ref]. When applying NMF to a matrix A, the matrix A can be factored into two matrices W and H with A m|n &W m|k |H k|n , where the columns k of matrix W are called metagenes, as defined by Brunet et al. [bib_ref] Metagenes and molecular pattern discovery using matrix factorization, Brunet [/bib_ref]. [fig_ref] Figure 1: Schematic representation of factorization in NMF method [/fig_ref] showed the factorization structure [bib_ref] Metagenes and molecular pattern discovery using matrix factorization, Brunet [/bib_ref]. In this study, the entry w pi of matrix W is the coefficient of transcript p in metagene i. The entry h ij represents the expression level of metagene i in sample j. Each column of matrix H represents the metagene expression level of the corresponding patient sample. After applying NMF with an appropriate value of k, a kdimension metagene expression level will be generated for each sample, h .j~( h 1j , h 2j , Á Á Á , h kj ) T for sample j~1, Á Á Á , n. In this study, the standard NMF factorization with an algorithm adopted from Lee and Seung [bib_ref] Gene expression levels as endophenotypes in genome-wide association studies of Alzheimer disease, Zou [/bib_ref] was used to produce two non-negative matrices, W and H. The factorization process was begun by randomly initializing matrices W and H and then iteratively updating them to minimize a divergence function
[formula] D~P i, j A i, j log (A i, j =(WH) i, j ){A i, j z(WH) i, j for metagene i~1, Á Á Á , k and sample j~1, Á Á Á , n. [/formula]
To apply NMF to construct endophenotypes, the entry w pi of matrix W was regarded as the importance value of transcript p for p~1, Á Á Á , m in metagene i for i~1, Á Á Á , k. Transcripts with large w pi values contributed more toward the metagene when w pi w0 for any i value. Note that the dimension was reduced from m transcripts to k metagenes. Regarding h .j , sample j was placed in cluster i if h ij was the largest entry in h .j , as suggested by Brunet et al. [bib_ref] Metagenes and molecular pattern discovery using matrix factorization, Brunet [/bib_ref].
One key issue in NMF is choosing an appropriate k such that the samples decompose into k ''meaningful'' clusters. To evaluate how many clusters were appropriate, a widely accepted criterion, the cophenetic correlation coefficient (r cc ), which is based on a consensus matrix, was used to determine cluster number [bib_ref] Metagenes and molecular pattern discovery using matrix factorization, Brunet [/bib_ref] [bib_ref] Consensus clustering: A resampling-based method for class discovery and visualization of gene..., Monti [/bib_ref]. The r cc is a measure to evaluate the stability of the clusters from NMF. It is computed by the Pearson correlation of two distance matrices. The first matrix is the consensus matrix which measures the distance between samples. The second distance matrix is made by the linkage used in the reordering of the consensus matrix [bib_ref] Metagenes and molecular pattern discovery using matrix factorization, Brunet [/bib_ref]. The entries of the consensus matrix range from 0 to 1 and reflect the probability that samples i and j cluster together. Based on the idea of consensus clustering, r cc range from 0 to 1. A large r cc indicates high clustering stability; therefore, the number of clusters k with the largest r cc was chosen. To visualize the sample clustering stability associated with a given cluster number k, average linkage hierarchical clustering were used to reorder the columns and rows of the consensus matrix [bib_ref] Metagenes and molecular pattern discovery using matrix factorization, Brunet [/bib_ref]. For this study, the R software (http://www.r-project.rg)package ''NMF'' [bib_ref] A flexible R package for nonnegative matrix factorization, Gaujoux [/bib_ref] was used to perform NMF.
## Endophenotype-identification analytical procedure: informative transcript selection for biological interpretation of metagenes
Most entries w pi of matrix W were close to zero, meaning that metagene was determined by a relatively small number of transcripts with large importance values. In addition, redundant transcripts in W complicated the biological interpretation of metagenes and identification of molecular subtypes. To exclude redundant transcripts, we determined which transcripts were most informative for each metagene by ranking each column of matrix W . The transcript selection process was performed as follows. First, for each w .i~( w 1i , Á Á Á , w mi ) T , the entry w pi in w .i for p~1, Á Á Á , m was ranked in descending order. Given a certain value t, the top t transcripts were selected for each metagene i~1, Á Á Á , k. Then, another round of NMF was carried out, based solely on nonredundant t|k transcripts selected from k metagenes.
To choose an appropriate t, the process was repeated t times for each metagene. ARI is a frequently used cluster validation measurement that verifies the agreement between two partitions [bib_ref] Learning parts-based representations of data, Ross [/bib_ref]. In the present study, let C be the set that contained nonredundant t|k transcripts and R be the set that contained m transcripts. ARI was used to assess the agreement of the molecular subtypes between set C and set R. ARI values range from 0 to 1.
Higher index values indicate more agreement between sets. In this study, the minimum t among those with ARI $0.95 was selected as the optimal informative transcripts set. The selected transcripts were then used for pathway analyses.
MetaCore TM (GeneGo, Inc., St. Joseph, MI, USA) is a webbased computational platform designed for systems biology. It provides tools for gene list enrichment analysis, multi-experiment comparison, interactome analysis, and biological network analysis [bib_ref] Pathway mapping tools for analysis of high content data, Ekins [/bib_ref] [bib_ref] A novel method for generation of signature networks as biomarkers from complex..., Nikolsky [/bib_ref]. The biological characteristics of each molecular subtype were determined using MetaCore pathway analysis. The top t transcripts that overlapped among k metagenes were excluded in the MetaCore pathway enrichment analysis.
## Pca-k
PCA is a common method that is used to provide a simple overview of complex structure using a small number of principal components (PCs) to reduce high-dimension data. The PCs are linear combinations of original transcripts that explain the variation in the data and are orthogonal to each other. In gene expression analysis, the PCs have been referred to as metagenes, as introduced by Ma et al. [bib_ref] Principal component analysis based methods in bioinformatics studies, Ma [/bib_ref]. Briefly, to apply PCA to transcription data, the data matrix A with m transcripts and n samples was decomposed by singular value decomposition to A m|n~Um|n L n|n V T n|n , where U m|n was a score matrix, V n|n was a loading matrix, and L n|n was a matrix containing the socalled singular values. Matrices U and V were orthogonal. To find potential molecular subtypes, the score matrix was used for clustering with the standard k-means clustering method and the squared Euclidean distance measure [bib_ref] Reducing microarray data via nonnegative matrix factorization for visualization and clustering analysis, Liu [/bib_ref].
## Simulation studies
For comparing the performance between NMF and PCA-K in constructing endophenotype, we considered informative genes and non-informative genes based on the idea proposed by Fogel P. et al [bib_ref] Inferential, robust nonnegative matrix factorization analysis of microarray data, Fogel [/bib_ref] in order to mimic real microarray data. In the simulation studies, we considered three different molecular subtypes, T1, T2, and T3, each with an equal sample size of 80. The noninformative genes represented those that had no contribution in distinguishing molecular subtypes. First, for simulating noninformative genes, the gene expression level was described by the model
[formula] Y m|n~Gm|n zE m|n ,ð1Þ [/formula]
where matrix Y (the gene expression data) had size m|n with rows representing genes and columns representing samples, matrix G was the log 2 -transformed mean of the overall gene expression level using log 2 of 100 [bib_ref] Inferential, robust nonnegative matrix factorization analysis of microarray data, Fogel [/bib_ref] , matrix E was the multivariate random variable simulated from a multivariate normal distribution with a mean vector of zero. We assumed that all genes in a molecular subtype were correlated at the same level. Hence, to simultaneously simulate multiple molecular subtypes, the covariance matrix was considered as a diagonal partition consisting of the correlation matrix for each molecular subtype. The variance of each gene was set to be 0.3, and the correlation level between genes was set to moderate dependence ranging from 0.4 to 0.6 [bib_ref] A mixture model approach for the analysis of microarray gene expression data, Allison [/bib_ref]. In addition, the proportion of non-informative genes, h, was assumed at 70%, 80%, and 90% [bib_ref] Inferential, robust nonnegative matrix factorization analysis of microarray data, Fogel [/bib_ref]. Recent works [bib_ref] Systems genetics, bioinformatics and eQTL mapping, Li [/bib_ref] [bib_ref] Detection and interpretation of expression quantitative trait loci (eQTL), Michaelson [/bib_ref] [bib_ref] Data-driven assessment of eQTL mapping methods, Michaelson [/bib_ref] has demonstrated that co-regulated genes are expected to rise and fall together in expression levels. The genetic variation can be used to identify groups of genes that might share a common background. Correlated variation in groups of genes may result in common phenotypes [bib_ref] The influence of genetic variation on gene expression, Williams [/bib_ref]. To add the variation in gene expression to our simulation, we presumed that the difference in gene expression level between molecular subtypes was resulted from the variants in the DNA sequences (i.e., singlenucleotide polymorphisms [SNPs]) [bib_ref] Searching for polymorphisms that affect gene expression and mRNA processing: Example ABCB1..., Wang [/bib_ref]. Hence, for informative genes, we used a linear model to link the relationship between SNP data and gene expression as proposed by many eQTL studies such as Stranger et al. [bib_ref] Population genomics of human gene expression, Stranger [/bib_ref] and Chen et al. [bib_ref] Variations in DNA elucidate molecular networks that cause disease, Chen [/bib_ref]. We assumed a one-toone correlation between a SNP and a gene expression level, that is, the expression level from microarray is related to SNP allele frequency. Therefore, a group of SNP markers were selected by fixing the difference of minor allele frequencies (MAF) (f ) to predict a subtype. To include informative genes, equation 1 was rewritten as follows:
[formula] Y m|n~Gm|n zb|S m|n zE m|n ,ð2Þ [/formula]
where matrix G and matrix E were the same as in equation 1. Matrix S (the genotype data) was generated from a multinomial distribution with consideration of various minor allele frequency (MAF) differences among subtypes. To simulate informative genes, 160 subtype-specific overexpressed genes were generated for each of the molecular subtypes. To simulate overexpressed genes in T1, the MAF was set to 0.1+ f for subtype T1, and the MAF was set to 0.1 for subtypes T2 and T3, where f was the MAF difference between T1 and the other subtypes (T2 and T3). Similarly, the process for generating overexpressed genes was repeated for other subtypes. An additive genetic model was used, which coded copies of the minor allele as 0, 1, or 2. The coefficient b represented the magnitude of the SNP effect on gene expression level. Finally, matrix Y in equation 2 was applied to both NMF and PCA-K.
To evaluate the performance of NMF and PCA-K in constructing endophenotypes, an accuracy measure, purity [bib_ref] Sparse non-negative matrix factorizations via alternating non-negativity-constrained least squares for microarray data..., Kim [/bib_ref] , was used. Assuming l is the true number of clusters, after applying a clustering method (e.g. NMF and PCA-K), k clusters are obtained. Purity is given by the following equation [bib_ref] Sparse non-negative matrix factorizations via alternating non-negativity-constrained least squares for microarray data..., Kim [/bib_ref] :
[formula] Purity~X k q~1 n q n p(C q ), p(C q )~1 n q max 1ƒrƒl (n r q ) [/formula]
where C q is a particular cluster of size n q from a clustering method, n r q is the number of samples in cluster q that belongs to the original cluster r (1ƒrƒl) and n is the total number of samples. Hence, Purity indicates the proportion of samples in cluster q being clustered correctly in cluster r. The larger the value of purity is, the better the clustering performance. All simulation scenarios were run 1000 times. The average purity measure was calculated for each run.
## Application to load dataset
To demonstrate the feasibility of our analytic procedure in an endophenotype study, we used a publicly available dataset derived from patients with LOAD [bib_ref] Genetic Control of Human Brain Transcript Expression in Alzheimer Disease, Webster [/bib_ref]. The dataset is freely available at http://labs.med.miami.edu/myers/. LOAD is a common complex neurodegenerative disorder that occurs in people over the age of 65 years. The dataset contained data on 176 patients with a diagnosis of LOAD and on 188 patients who were neurologically normal. The cRNA data contained 8,650 transcripts that were obtained using the Illumina Human Reseq-8 Expression Bead Chip following standard operating procedures as described in Webster et al. [bib_ref] Genetic Control of Human Brain Transcript Expression in Alzheimer Disease, Webster [/bib_ref]. The DNA data contained 372,084 autosomal SNPs that were obtained using the Affymetrix Gene Chip Human Mapping 500 K Array set.
Before evaluating our proposed procedure, we preprocessed the LOAD dataset. First, transcripts were eliminated when expression values for a given transcript were missing for .30% of the subjects. This resulted in the removal of 338 transcripts. For the remaining 8,312 transcripts, log 2 -transformed values were used to reduce outlier impact and ensure normality. Many of the genes on the array were not expressed, were expressed at low levels, or were expressed at a level with no biological significance. To further reduce the effects of irrelevant or noisy variables, we selected transcripts with mean expression values in the upper 70 th percentile and with variances in the upper 50 th percentile, as suggested by Langfelder et al. [bib_ref] WGCNA: an R package for weighted correlation network analysis, Langfelder [/bib_ref]. The remaining 1,116 transcripts met all filtering criteria. Missing data from these 1,116 transcripts were replaced by the mean value of the transcripts across patients. Therefore, further analyses were constructed without any missing data.
Rare alleles are more likely to result in spurious finding due to a higher relatedness between individuals sharing rare alleles [bib_ref] Genome-wide association studies: past, present and future, Mccarthy [/bib_ref]. In addition, loci with a low MAF (,10%) have significantly lower power to detect weak genotypic risk [bib_ref] The Framingham Heart Study 100 K SNP genome-wide association study resource: overview..., Cupples [/bib_ref] [bib_ref] Genomewide association with diabetes-related traits in the Framingham heart study, Florez [/bib_ref]. Therefore, to avoid spurious findings resulting from rare genotypes, SNP markers with low MAFs were excluded; specifically, we adopted the criteria of removing SNPs with MAF ,10% [bib_ref] The effect of minor allele frequency on the likelihood of obtaining false..., Tabangin [/bib_ref] , which resulted in a total of remaining 283,475 SNP markers for further analysis. For incomplete genotype data, ''beagle'' software [bib_ref] A unified approach to genotype imputation and haplotype-phase inference for large data..., Browning [/bib_ref] was used for imputation. Due to the exploratory nature of our study, a loose threshold (p-value ,10 24 ) was used as the significance criterion in differential expression and metagene expression QTL analyses.
# Results
## Simulation study
Because the results for the simulated scenarios were similar, [fig_ref] Figure 2: Simulation results of NMF and PCA-K for various f and b at... [/fig_ref] demonstrates the average purity results for NMF and PCA-K with various values for b, f and h~70%. [fig_ref] Figure 1: Schematic representation of factorization in NMF method [/fig_ref] shows results for h~80% and h~90% with various values for b and f . As shown in [fig_ref] Figure 2: Simulation results of NMF and PCA-K for various f and b at... [/fig_ref] , the NMF method performed better than the PCA-K method under different scenarios. When the MAF difference, f , was ,0.4, the average purity of the NMF and PCA-K methods was close to 0.6 under different b values. This indicated that the classification using the NMF and PCA-K methods was not satisfactory when the dissimilarity among subtypes was not clear. With increasing values of f , the average purity of NMF and PCA-K methods was higher under different b values. The average purity of the NMF method increased dramatically at f~0:4, whereas the average purity of the PCA-K method increased sharply at f~0:6. These results showed that NMF had a better sensitivity than PCA-K for detecting divergence among subtypes. Additionally, with different h values and under b~0:5 and f~0:5, the average purity for NMF and PCA-K decreased slightly [fig_ref] Figure 3: Simulation results of NMF and PCA-K for various proportions of non-informative genes... [/fig_ref]. This finding indicated that the impact of noise was not strong.
## Results for the load dataset: molecular subtypes among patients
We applied NMF to the LOAD dataset containing 1,116 transcripts derived from 176 patients. A cophenetic correlation coefficient (r cc ) was calculated for each value of k. [fig_ref] Figure 4: Unsupervised clustering using NMF [/fig_ref] shows r cc levels corresponding to k~2 to 5. r cc peaked at k~3 (r cc~0 :9439) and fell off sharply as k increased, indicating that k~3 was the best fit for this dataset. Three clusters yielded the highest r cc values, which reduced the initial dimensionality of the 1,116 transcripts to three metagenes and grouped the 176 patients into three molecular subtypes: MS1 (n 1 = 80), MS2 (n 2 = 73), and MS3 (n 3 = 23) shown in [fig_ref] Figure 4: Unsupervised clustering using NMF [/fig_ref].
Hence, NMF decomposed the 1:116|176 LOAD transcription data matrix into matrix W and matrix H. The non-negative matrix W was 1,116|3, with each of the three columns representing a metagene. The non-negative matrix H was 3|176, with each of the three rows representing the metagene expression levels for the corresponding sample. The three metagenes captured gene expression patterns specific to three different molecular subtypes of patients. The three metagene expression levels were named MGL1, MGL2, and MGL3, respectively.
Devarajan [bib_ref] Nonnegative Matrix Factorization: An Analytical and Interpretive Tool in Computational Biology, Devarajan [/bib_ref] used metagene expression levels to determine which sample subtypes belonged to which specific metagenes. We therefore adopted the same approach. [fig_ref] Figure 5: Association of metagene expression levels with molecular subtypes [/fig_ref] shows boxplots of the MGL1-3 values for each molecular subtype. Patients grouped in MS1 had higher MGL1 values as compared with patients grouped in MS2 or MS3 [fig_ref] Figure 5: Association of metagene expression levels with molecular subtypes [/fig_ref]. Patients grouped in MS2 had the highest MGL2 values [fig_ref] Figure 5: Association of metagene expression levels with molecular subtypes [/fig_ref]. The greatest differences were seen for MGL3, with patients grouped in MS3 having the highest MGL3 values [fig_ref] Figure 5: Association of metagene expression levels with molecular subtypes [/fig_ref]. In this way, MGL1, MGL2, and MGL3 represented distinct biological characteristics of molecular subtypes MS1, MS2, and MS3, respectively.
## Results for the load dataset: informative transcript selection for molecular subtypes
The w pi values corresponding to p transcripts in the W matrix were sorted in descending order within each metagene, i (i~1,2,3). To identify the most relevant t transcripts for each metagene, eight subsets were selected, i.e. t = 50, 100, 150, 200, 250, 300, 400, and 800, respectively. Non-redundant transcript subsets for the eight values of t when k~3 were listed in [fig_ref] Table 1: Identification of the optimal transcript subset for molecular subtype representation [/fig_ref]. The non-redundant transcripts represented the union of the top t|3 transcripts for the 3 metagenes. For example, when the top 50 transcripts (t~50) were chosen for each metagene (total 150 transcripts), 137 transcripts were the union of the 150 transcripts from 3 metagenes [fig_ref] Table 1: Identification of the optimal transcript subset for molecular subtype representation [/fig_ref]. The transcript subset with t~800 contained all 1,116 transcripts and thus was not considered further. NMF was then applied seven times to obtain new molecular subtypes for the seven remaining transcript subsets. Using a molecular subtype containing all 1,116 transcripts as the reference group, ARI values were calculated to evaluate molecular subtype agreement between transcript subsets, t, and the reference group. According to our criteria, t~150 was the minimum t with an ARI value of .0.95 and therefore was selected [fig_ref] Table 1: Identification of the optimal transcript subset for molecular subtype representation [/fig_ref]. In comparison with the reference group, only one patient was misclassified with the NMF method. Therefore, based on this analysis, the top 150 transcripts for metagenes 1, 2, and 3 were used to represent the biological characteristics of molecular subtypes MS1, MS2, and MS3, respectively. To interpret the biological characteristics of each molecular subtype, overlapped transcripts (i.e., those found in more than one metagene) were excluded to ensure uniqueness, and 123, 89, and 71 metagenespecific transcripts were identified for MS1, MS2, and MS3, respectively.
## Results for the load dataset: enrichment pathway analysis via metacore
Furthermore, the metagene-specific transcripts were used to conduct a pathway analysis using the GeneGO pathway map from MetaCore. The pathways for each metagene-specific transcript with a significance of p-value ,10 23 are shown in [fig_ref] Table 2: Significant pathways for metagene-specific transcripts [/fig_ref].
## Enriched pathways in metagene1-specific transcripts
In metagene1-specific transcripts, there were three significant enriched pathways. In the Development Gastrin in differentiation of the gastric muscosa pathway, Gastrin is a peptide hormone produced primarily by G cells, endocrine cells located in the gastric antrum. Transcription of the Gastrin gene gives rise to a 0.7 kb mRNA coding for a 101 amino acids precursor, known as Progastrin. Progastrin is processed to mature amidated Gastrin (Gastrin 17). Gastrin 17 mediates its effects primarily through Cholecystokinin B receptor (CCKBR) [bib_ref] Gastrin and cancer: a review, Ferrand [/bib_ref] [bib_ref] Cholecystokinin and gastrin receptors, Dufresne [/bib_ref]. The CCKBR is a seven transmembrane G-protein coupled receptor (GPCR) that is expressed in the gastric fundus, parietal ECL and D cells. GPCR are involved in numerous key neurotransmitter systems in the brain that are disrupted in Alzheimer's disease (AD) [bib_ref] Role of gastrin peptides in carcinogenesis, Grabowska [/bib_ref]. The second enriched pathway was Cytoskeleton remodeling Neurofilaments. Cytoskeleton of most eukaryotic cells consists of three distinct, yet interconnected, filament systems: actin filaments, microtubules and intermediate filaments (IF). Neurofilaments are the principal intermediate filament type expressed by neurons. The third enriched pathway was Neurophysiological process dopamine D2 receptor transactivation of PDGFR in central nervous system (CNS). Dopamine is a major transmitter and neuromodulator in the CNS. This transmitter mediates its signaling through GPCRs. Dysregulation of dopamine receptor activity account for neuropsychotic disorders such as Parkinson's diseases, schizophrenia, and Alzheimer's disease [bib_ref] Dopamine receptors: from structure to function, Missale [/bib_ref]. Enriched Pathways in Metagene2-specific Transcripts There were eight enriched pathways in metagene2-specific transcripts. We briefly described the top 3 significant enriched pathways. The first pathway was Immune response function of MEF2 in T lymphocytes. The transcription factor Myocyte Enhancer Factor 2 (MEF2) is a family of muscle-enriched transcription factors that have an essential role in myogenesis. It has been implicated as playing a pivotal role in neuronal survival as well as in the development, differentiation, and plasticity of the CNS [bib_ref] Suppression of a MEF2-KLF6 Survival Pathway by PKA Signaling Promotes Apoptosis in..., Salma [/bib_ref] [bib_ref] Myocyte enhancer factor-2 transcription factors in neuronal differentiation and survival, Heidenreich [/bib_ref]. The second pathway was Cytoskeleton remodeling Neurofilaments. In [fig_ref] Table 2: Significant pathways for metagene-specific transcripts [/fig_ref] , we found that the metagene1-specific transcripts and metagene2-specific transcripts shared the same pathway. Howev-er, the genes in this pathway were different in metagene1-specific transcripts and metagene2-specific transcripts. This suggested that the genes in metagene1-specific transcripts had different process link to this pathway compared to the genes in metagene2-specific transcripts. The third pathway was Neurophysiological process role of CDK5 in presynaptic signaling. Cyclin-dependent kinase 5 (CDK5) is a member of the small serine/threomine cyclin-dependent kinase family with high activity in the central nervous system [bib_ref] Cdk5 deregulation in the pathogenesis of Alzheimer's disease, Cruz [/bib_ref].
## Enriched pathways in metagene3-specific transcripts
The significant enriched pathway in metagene3-specific transcripts was Development EPO-induced PI3K/AKT pathway and Ca (2+) influx. Erythropoietin (EPO) is a lineage-specific hematopoietic growth factor required for survival, proliferation and differentiation of committed erythroid progenitor cells [bib_ref] Erythropoietin, iron, and erythropoiesis, Goodnough [/bib_ref] [bib_ref] Erythropoietin, the biology of erythropoiesis and epoetin alfa -An overview, Bieber [/bib_ref]. EPO has been shown to have multiple effects on neurons [bib_ref] Erythropoietin as an antiapoptotic, tissue-protective cytokine, Ghezzi [/bib_ref].
## Application of molecular subtypes in the load dataset: differential expression analysis
To demonstrate the application of the molecular subtypes identified by our procedure, we compared discrepancies in differential expression analysis between patients with LOAD and normal subjects. We also conducted a differential expression analysis for patients with each molecular subtype and compared it to that for the normal subjects. A t-test was used to identify differentially expressed transcripts among the 1,116 transcripts, with significant thresholds of p-value ,10 24 . Our results found that 660, 887, 30, and 736 transcripts were differentially expressed for all patients and for patients grouped in MS1, MS2, and MS3, respectively, when compared with transcripts from normal subjects. Only 3 of the 660 differentially expressed transcripts identified for all patients were not included among those identified for patients grouped in the molecular subtypes. In comparison, it is worth noting that only 30 transcripts were significant for patients grouped in MS2, which suggests that MS2 might represent a normal-like subtype.
Furthermore, when comparing all patients with normal subjects, there were 660 differentially expressed transcripts. Among those transcripts, 14 were also listed in AlzGene database [bib_ref] Systematic meta-analyses of Alzheimer disease genetic association studies: the AlzGene database, Bertram [/bib_ref]. These 14 differentially expressed transcripts detected using all patients were also found from the results when comparing normal subjects with MS1 [fig_ref] Figure 6: Venn diagram of differentially expressed transcripts listed in AlzGene [/fig_ref]. In addition, 8 out of the 14 differentially expressed transcripts were intersected with differentially expressed transcripts identified when comparing normal subjects with MS3. We noted that there was no candidate susceptibility genes listed in AlzGene found as differentially expressed transcripts for MS2.
The gene encoding apolipoprotein E (APOE) has been identified as playing an important role in the development of Alzheimer's disease [bib_ref] Molecular genetics of late-onset Alzheimer's disease, Kamboh [/bib_ref] [bib_ref] Genomewide association study of Alzheimer's disease with psychotic symptoms, Hollingworth [/bib_ref] [bib_ref] A highdensity whole-genome association study reveals that APOE is the major susceptibility..., Coon [/bib_ref]. In the differentially expressed analysis, we found that APOE gene was not significantly different (p-value = 0.3641) comparing all patients with normal subjects. When patients were grouped into three subtypes, the p-values of the differential expression analyses were 1.03610 -11 , 0.00087, and 0.00182 for MS1, MS2, and MS3, respectively (comparing with normal subjects). At the significant threshold p,10 24 that we chose to use in this study, APOE gene was significant when comparing MS1 with normal subjects. A boxplot was drawn to show the expression level of APOE gene for 3 molecular subtypes, all patients and normal subjects. The median expression level of APOE gene was the highest in MS1 [fig_ref] Figure 3: Simulation results of NMF and PCA-K for various proportions of non-informative genes... [/fig_ref].
## The validation of molecular subtypes in the load dataset
To validate the differential expression analysis derived from each molecular subtype in contrast to normal subjects, the changes of effect size were evaluated by Cohen's d [bib_ref] A power primer, Cohen [/bib_ref]. The results showed that the distribution of effect size in 1,116 transcripts for patients grouped in MS2 centered at zero and variation was small, thus, MS2 had little difference from normal subjects [fig_ref] Figure 2: Simulation results of NMF and PCA-K for various f and b at... [/fig_ref]. The distribution of effect size in 1,116 transcripts for patients grouped in MS1 and for patients grouped in MS3 had larger variation compare to all patients [fig_ref] Figure 2: Simulation results of NMF and PCA-K for various f and b at... [/fig_ref]. This showed that patients grouped by NMF enhanced the effect size in the differential expression analysis.
To examine whether the existence of heterogeneous subtypes among patients was random or not, a permutation strategy was used as suggested by Allison et al. [bib_ref] Microarray data analysis: from disarray to consolidation and consensus, Allison [/bib_ref]. In the permutation analysis, we randomly regrouped patients into three molecular subtypes. The differential expression analysis was then carried out in each regrouped subtype compared to normal subjects, and the effect size was recorded. The empirical distribution of effect size Pathway enrichment analysis was conducted using metagene-specific transcripts. Significant biological pathways were detected by MeatCore at a significance level pvalue ,10 [bib_ref] Gene expression profiling of advanced-stage serous ovarian cancers distinguishes novel subclasses and..., Yoshihara [/bib_ref]. Pathways are listed in order of significance, e.g., most significant pathway are presented at the top. a Name of biological pathway selected by MetaCore. b The number of metagene-specific transcripts associated with pathway/the number of all genes associated with pathway.
c Gene ID of metagene-specific transcripts associated with pathway. d Different genes in "Cytoskeleton remodeling Neurofilaments" pathway were identified in metagene 1 and metagene 2. doi:10.1371/journal.pone.0040996.t002
for the randomly grouped patients with 1,116 transcripts was estimated. We combined the effect size across 1,000 permutations according to average quantiles of empirical distribution. The distribution of effect size for the randomly grouped patients was very similar to that observed for all patients, regardless of the proportion of patients that were grouped together [fig_ref] Figure 2: Simulation results of NMF and PCA-K for various f and b at... [/fig_ref]. Therefore, these findings indicated that the definition of molecular subtypes in the LOAD dataset presented in this study is meaningful and can improve the efficiency of the identification of differentially expressed transcripts.
## Metagene expression quantitative trait locus (qtl) mapping
To explore the genetic variants that are responsible for the molecular subtypes identified by differential transcript profiling, the genotype data were used to identify associations between genetic variants (i.e., SNPs) and metagene expression levels. Associations between SNP markers and each of the three MGLs were tested using a linear regression model in which individual MGLs were each regressed on individual SNP markers, and the significance level was set at a threshold p-value of 10 -4 . The results showed that 11, 25, and 31 SNP markers were associated with MGL1, MGL2, and MGL3, respectively, and were named MGL1-QTL, MGL2-QTL, and MGL3-QTL, respectively.
To explore the additive property of the effects of these SNP markers on molecular subtype identification, the frequencies of upregulated alleles in each MGL-QTL were calculated for MS1-3. The up-regulated allele was the one that had a positive coefficient in the regression model; in other words, the up-regulated allele enhanced the metagene expression level. The up-regulated alleles were called u-alleles. This analysis was intended to help understand the possible relationship that the SNPs regulated metagene expression levels for a specific molecular subtype. We calculated the mean allele frequency of u-alleles for each MGL-QTL. The mean allele frequency of u-alleles of 11 MGL1-QTL was higher in MS1 (0.6658) than in MS2 (0.5924) and MS3 (0.4895). Similarly, the mean allele frequency of u-alleles related to the 25 MGL2-QTL was higher in MS2 (0.4925) in comparison with MS1 (0.3797) and MS3 (0.3248); the mean allele frequency of u-alleles of 31 MGL3-QTL had higher frequency in MS3 (0.3050) than in MS1 (0.1370) and MS2 (0.1494). The results showed that the molecular subtypes with its respective MGL had higher frequencies of u-alleles in the LOAD data.
# Discussion
Patients with a common disease can display molecular heterogeneity [bib_ref] Genomic medicine and neurological disease, Boone [/bib_ref]. These molecular differences often are important because they can reflect the biological processes that underlie pathogenic diversity [bib_ref] Genomic subtypes of breast cancer identified by array-comparative genomic hybridization display distinct..., Jonsson [/bib_ref]. Molecular markers such as transcripts or protein levels can be used to identify molecular dissimilarities among patients. Identification of these specific disease subtypes could dramatically improve our understanding of disease pathology.
In our simulation study, the magnitude of the SNP effect on gene expression level (b), the MAF difference among subtypes (f ), and the proportion of non-informative genes (h) were considered to evaluate the efficiency of endophenotype construction using the NMF and PCA-K methods. Our findings demonstrated that the magnitude of the effect of MAF difference (f ) was larger than the magnitude of the SNP effect on gene expression level (b). This finding might result from our assumption that SNPs with differences in allele frequencies varied significantly between molecular subtypes. The difference might result from different patterns of expression levels among different molecular subtypes. In previous cancer microarray studies, NMF was used to examine the efficiency of detection of potential disease subgroups [bib_ref] Improving molecular cancer class discovery through sparse non-negative matrix factorization, Gao [/bib_ref] [bib_ref] Highresolution genomic profiles define distinct clinico-pathogenetic subgroups of multiple myeloma patients, Carrasco [/bib_ref]. Our simulations provide an alternative insight for the detection of molecular subtypes.
In our simulation study, the variations of gene expression levels were controlled only by the difference of MAF of some specific SNP markers among subtypes. In reality, other factors such as heterogeneity of genetic effect between different subtypes or different direction of effect between subtypes may influence gene expression levels. When these effects are considered in our method, the diversity of gene expression level will increase. According to a previous study [bib_ref] Nonnegative Matrix Factorization: An Analytical and Interpretive Tool in Computational Biology, Devarajan [/bib_ref] , NMF has good performance in class discovery when diversity of groups is high. Therefore, it is expected that NMF will still be useful in constructing endophenotypes.
In this study, we assumed that all genes in a molecular subtype were correlated at the same level in simulation study. To check this assumption, we further examined the variation of correlation for LOAD dataset. The average correlation with metagenes for each molecular subtype was calculated. The average correlation 6 standard error were 0.442760.003, 0.231260.0028, and 0.636960.0031 for MS1, MS2, and MS3, respectively. The results showed that the assumption seems to be acceptable for the simulation study.
One of the advantages of NMF is that it produces two nonnegative matrices that provide a parts-based local representation of the data. In comparison with NMF, PCA produces both positive and negative entries in metagenes (i.e., PCs) that can cancel each other out, at least in part, and this creates difficulties in the capturing of local data characteristics [bib_ref] Nonnegative Matrix Factorization: An Analytical and Interpretive Tool in Computational Biology, Devarajan [/bib_ref]. In addition, entries associated with NMF are interpreted readily as the relative contribution of genes and are well grounded in physical reality [bib_ref] Learning parts-based representations of data, Ross [/bib_ref]. Another advantage of NMF is that it can simultaneously cluster genes and patients. In this study, NMF was used to reduce data from thousands of transcripts into a small number of metagenes, k. This simplification provided a summary of patient gene expression levels from corresponding metagenes and was used to extract informative patterns from complex data. For application in complex disease, we applied NMF to the LOAD dataset and successfully explored three potential molecular subtypes, MS1, MS2, and MS3.
There have been several improvements to the standard NMF method. In our present study, we further applied non-smooth NMF (nsNMF) [bib_ref] Biclustering of gene expression data by non-smooth non-negative matrix factorization, Carmona-Saez [/bib_ref] , an extended method by making use of nonsmoothness constraints, to simulations at a few different values of MAF difference among subtypes (f ) and magnitude of the SNP effect on gene expression level (b) based on a fixed proportion of non-informative genes (h = 70%). Overall speaking, the average purity of for nsNMF was better than standard NMF [fig_ref] Figure 4: Unsupervised clustering using NMF [/fig_ref]. The results indicated that nsNMF method refined standard NMF method in endophenotype construction.
In addition to NMF, we proposed to incorporate the process of informative transcript selection in the identification of molecular subtypes. Here we used ARI to measure agreement between molecular subtypes to ensure that the utilization of the top selected transcripts,t, for each metagene captured the same molecular subtypes as when all transcripts, m, were used. Consequently, we anticipated that the correlations between metagene expression obtained from the selected and the entire set of transcripts would be high. For the LOAD dataset, the correlation coefficients for the metagene expression levels between the top 150 transcripts and the original 1,116 transcripts were 0.9844, 0.9840, and 0.9772 for MGL1, MGL2, and MGL3, respectively. As a result, the proposed process could help to exclude redundant transcripts and to maintain a similar molecular pattern as that observed with the original m transcripts.
In order to find the biological characteristics for each molecular subtype, the metagene-specific transcripts that overlapped among 3 metagenes were excluded to decrease the noise for finding the specific characteristics for each molecular subtype in pathway analysis. The removal of these genes may lose potential information of showing linkages between pathways.
To evaluate the predictive power of three metagene expressions in molecular subtypes, a multinomial logistic regression using three metagene expression levels with three molecular subtypes was constructed. The predictive power was almost 98%. The result showed that these metagene expression levels predict molecular subtypes well.
In addition to log 2 transformation to preprocess the transcripts profiling, we also used arcsinh transformation. To examine whether transformation methods changed clustering, we then applied NMF method to the data set and utilized ARI index to compare the clustering between results obtained from the two transformation methods. The ARI value was 0.9259 which indicated that the results for the two transformation methods were similar. Therefore, it seems that the impact of using different transformation methods to the clustering results is mild.
In our real data, we found that it was feasible to construct endophenotype using data mining tools with genomic data in LOAD study. To utilize the most information in LOAD study, microarray data was used to construct endophenotypes and SNP data was used to explore the correlation between endophenotypes and SNP markers for representing the underlying etiology of Alzheimer's disease. However, for some phenotypes, microarray data may not fully represent the underlying genetic etiology. For example, microarray expression level from lymphocyte drawn from peripheral blood may not well represent patients' blood pressure [bib_ref] The transcriptome in blood: Challenges and solutions for robust expression profiling, Fan [/bib_ref]. Due to differential spatial and temporal expression patterns, some genes are only expressed in specific tissues [bib_ref] Using gene expression to investigate the genetic basis of complex disorders, Nica [/bib_ref]. One of the limitations of NMF is that it can only be applied to continuous data. Other clustering methods e.g. random forest that can handle discrete data can be considered if only SNP data is available.
We have considered several thresholds of p-value for association analysis between metagene gene expression level (MGL) and SNP marker. Since this was an exploratory study, we chose to use a loose threshold p-value ,10 24 . Therefore, proper multiple testing was not fully considered.
When examining significant MGL-QTLs, there was no overlapping SNP marker among molecular subtypes. This result might be due to the exclusion of overlapping SNP markers before conducting QTL analysis. Therefore, it is still possible that there is some heterogeneity of genetic effects between different subtypes.
In addition, we found that numerous candidate susceptibility genes in the AlzGene database could not be reproduced using data from all patients. This finding may be due to the dilution effect contributed by the normal-like subtype, MS2. These results showed that the use of the entire patient dataset may fail to detect potentially important susceptibility genes that are specific to molecular subtypes of patients, as identified using our proposed procedure.
In our study, we utilized a permutation study to examine whether the existence of molecular subtypes among patients was random or not. The results showed that the identified molecular subtypes were apparent in LOAD patients. Moreover, we also applied another popular clustering method, PCA-K, to LOAD dataset and found 3 molecular subtypes. Results showed that the clustering pattern of PCA-K was similar to those of NMF with ARI = 0.98 and both of them could identify a normal-like subtype. These results indicated the existence of heterogeneous subtypes in LOAD patients.
In pathway analysis, the candidate pathways obtained from our study have not been identified for Alzheimer's disease related studies, however, we found that the three enriched pathways in metagene1-specific transcripts were related to neurotransmitter and neuronal system [bib_ref] D2-class dopamine receptor inhibition of NMDA currents in prefrontal cortical neurons is..., Beazely [/bib_ref] [bib_ref] The gastrin-releasing peptide receptor as a therapeutic target in central nervous system..., Roesler [/bib_ref]. The enriched pathways in metagene2-specific transcripts were related to neuronal system and immune system [bib_ref] Cdk5 deregulation in the pathogenesis of Alzheimer's disease, Cruz [/bib_ref] [bib_ref] Suppression of death receptor signaling in cerebellar Purkinje neurons protects neighboring granule..., Linseman [/bib_ref] and the enriched pathway in metagene3-specific transcripts reported effect on learning and memory [bib_ref] Erythropoietin inhibits calcium-induced neurotransmitter release from clonal neuronal cells, Kawakami [/bib_ref]. The results showed that each molecular subtype with its respective metagene has specific characteristics. This finding may help us to understand the pathophysiology of Alzheimer's disease.
Therefore, a consideration of the molecular differences among patients is recommended to identify the genetic causes of a complex disease. However, if one tests multiple genetic subtypes, one should adjust the statistical testing accordingly. Moreover, a potential disadvantage focusing on molecular subtypes is that statistical power may be compromised when large data set is reduced to create homogeneous subgroups. Hence, further study regarding power reduction in endophenotype using smaller data set with possible stronger genetic effect should be considered. Carefully chosen endophenotypes may reveal alternative pathophysiological disease processes. Collectively, our findings suggest that this proposed approach is an effective method for constructing complex disease endophenotypes. [fig_ref] Figure 1: Schematic representation of factorization in NMF method [/fig_ref] Results of NMF and PCA with k-means for simulation. The simulations for a range of magnitude of SNP effect (b) and proportions of non-informative genes (h). The x-axis represented the MAF differences f . The y-axis represented the average purity given by NMF (red) and PCA-K (blue). The average purity of each method was shown as mean+standard error.
## Supporting information
(TIF) [fig_ref] Figure 4: Unsupervised clustering using NMF [/fig_ref] Simulation results of NMF and nsNMF for various f and b at h~70%. The simulations for a range of MAF differences (f ) and magnitude of SNP effect (b) under the proportion of non-informative genes h~70%. The x-axis represents the MAF differences. The y-axis represents the average purity given by NMF (red) and PCA-K (blue). A-C indicated b = 0.3 (A), 0.5 (B), and 0.8 (C), respectively. The average purity of each method was shown as mean+standard error. (TIF)
[fig] Figure 1: Schematic representation of factorization in NMF method. NMF decomposes matrix A m|n into W m|k and H k|n . The columns k of matrix W are called metagenes and the entry w pi of matrix W is the coefficient of transcript p in metagene i (p~1, Á Á Á , m; i~1, Á Á Á , k). The entry h ij of matrix H represents the expression level of metagene i in sample j (j~1, Á Á Á , n). doi:10.1371/journal.pone.0040996.g001 [/fig]
[fig] Figure 2: Simulation results of NMF and PCA-K for various f and b at h~70%. The simulations for a range of MAF differences (f ) and the magnitude of SNP effect (b) under the proportion of non-informative genes h~70%. The x-axis represents the MAF differences. The y-axis represents the average purity given by NMF (red) and PCA-K (blue). The average purity of each method was shown as mean+standard error. A-C indicated b = 0.3 (A), 0.5 (B), and 0.8 (C), respectively. doi:10.1371/journal.pone.0040996.g002 [/fig]
[fig] Figure 3: Simulation results of NMF and PCA-K for various proportions of non-informative genes h. The simulations for a range of proportions of non-informative genes (h) under MAF difference f~0:5 and magnitude of SNP effect b~0:5. The x-axis represents the proportions of non-informative genes. The y-axis represents the average purity given by NMF (red) and PCA-K (blue). The average purity of each method was shown as mean+standard error. doi:10.1371/journal.pone.0040996.g003 [/fig]
[fig] Figure 4: Unsupervised clustering using NMF. (A). Cophenetic correlation coefficients associated with different numbers of clusters k. Y-axis is the cophenetic correlation coefficient (r cc ); x-axis is the number of clusters. The arrow displays the r cc at k~3. (B). Heat map of reordered consensus matrix for k~3. Dark blue indicates samples never assigned to the same cluster, and red indicates samples always assigned to the same cluster. doi:10.1371/journal.pone.0040996.g004 [/fig]
[fig] Figure 5: Association of metagene expression levels with molecular subtypes. Boxplots indicate metagene expression levels (MGL1-3) for each molecular subtype (MS1-3) in the LOAD samples. (A) MGL1 is the highest in MS1. (B) MGL2 is the highest in MS2. (C) MGL3 is the highest in MS3. doi:10.1371/journal.pone.0040996.g005 [/fig]
[fig] Figure 6: Venn diagram of differentially expressed transcripts listed in AlzGene. The Venn diagram showed the differential expressed transcripts in all patients, MS1, and MS3. These differentially expressed transcripts were listed in AlzGene database. doi:10.1371/journal.pone.0040996.g006 [/fig]
[fig] Figure S2: The validation results of molecular subtypes with LOAD data. (A) Patients grouped by NMF: the plot showed the distribution of effect size in 1116 transcripts for all patients (black) and molecular subtypes (MS1 = red, MS2 = green, and MS3 = blue). (B) All patients randomly grouped: the plot showed the empirical distribution of effect size in randomly grouped patients by average quantile across 1000 times of permutations (entire patients = black, MS1 = red, MS2 = green and MS3 = blue). (TIF) Figure S3 The gene expression level of APOE gene for all patients, control subjects and molecular subtypes. Boxplots indicate the gene expression level of APOE gene for all patients, control subjects and each molecular subtype (MS1-3). (TIF) [/fig]
[table] Table 1: Identification of the optimal transcript subset for molecular subtype representation. [/table]
[table] Table 2: Significant pathways for metagene-specific transcripts. [/table]
|
Quantifying Reversible Surface Binding via Surface-Integrated Fluorescence Correlation Spectroscopy
## Si-fcs image acquisition
Unless stated otherwise, SI-FCS analysis was performed on image sequences of 64x64 pixel (4x4 binning) and for comparable analysis resized to the camera resolution of 256x256 pixels. Image stacks were recorded for 1.5 million frames for 7 nt, 8 nt and 9 nt, and 150,000 frames for 10 nt, respectively. The exposure time was 10 ms and the camera frame rate 85 Hz for 7 nt, 8 nt, 9 nt, or 10 Hz 10 nt. The EM gain setting was varied, but does not influence the kinetics (data not shown). To exclude any effect of the EM gain, we varied this parameter systematically in an initial set of experiments, but observed no influence on the autocorrelation decay times obtained from SI-FCS measurements (data not shown).
## S3
# Si-fcs data analysis
The autocorrelation curves were computed and analyzed using a custom-written Matlab 2017a (The MathWorks, Natick, USA) software: Each image was subdivided into 7X7 ROIs, each of them covering 31X31 pixels, spaced in a grid around the center of illumination. The effect of the choice of ROI size is discussed in . The signal in each ROI was integrated, yielding 49 intensity traces, which were bleach and drift-corrected by a single exponential, and individually correlated using the multiple-τ algorithm 1 , in which we doubled the bin width after every sixteenth point in the autocorrelation curve. The obtained autocorrelation curves were fitted individually by a single exponential decay with an offset, from which the amplitude and the characteristic decay time were obtained. For samples containing two species, the autocorrelation curves were fitted by a sum of two exponentials with an offset. In the titration experiments, the biexponential fit accounted for a supposedly non-specific component appearing at high concentrations (10 nM for 10 nt, 100 nM for 9 nt).
## Monte carlo simulations of si-fcs measurements
If not mentioned otherwise, Monte Carlo Simulations were performed using a home-written MATLAB code (R2016a, MathWorks). The time step between two iterations was set to = 1 ms and the signal from 10 iterations was integrated to form one time point in the signal trace. This corresponds to a time resolution of the detector of 10 ms. For these simulations, we only considered fluctuations in signal originating from binding and unbinding events. Therefore, we initialized surf immobile binding sites at a surface and had a fraction = 1 1+ < > of them initially occupied. Moreover, we defined a probability of binding binding = 1 − − < > and unbinding unbinding = 1 − − . During each iteration, we treat occupied and unoccupied binding sites differently. For all bound sites we generated a uniformly distributed random number in the interval (0,1). If the random number was smaller than a threshold given by unbinding , the binding site was converted to an unoccupied state, otherwise it remained unchanged. The transitions from the unbound to bound state were simulated following an equivalent strategy. Each bound receptor contributed with the brightness 1 to the signal per iteration, each unoccupied binding site did not contribute to the signal.
The described code does not simulate images, but only the integrated signal over a simulated area. Although it is straightforward to add the functionality of simulating images, all simulations with varying surface density (supplementary figure S6) were performed using the previously published Picasso tool 2 .
## Fluorescently labeled complementary ssdna in solution
Labeled imager strands with the sequence 5'-CTAGATGTAT-3'-Cy3B were purchased from Eurofins Genomics.
## Buffers
For simplicity, we name the used buffers A+ and B+. Buffer A+ contains 10 mM Tris-HCl, 100 mM NaCl, 0.05 v% Tween20 and is adjusted to pH 8. Buffer B+ contains 5 mM Tris-HCl, 10 mM MgCl2, 1 mM EDTA, 0.05 v% Tween20 and is adjusted to pH 8.
## Preparation of dna origami samples
DNA origami structures were synthesized as previously described 2 . In brief, structures were folded in a one-pot reaction with 40 µl total volume containing 10 nM scaffold (M13mp18), 10 nM biotinylated staples for surface attachment, 100 nM core staples and 1 µM extended staples in 1xTE buffer supplemented with 12.5 mM MgCl2. Structures were folded by first holding for 5 min at 80 °C, then going from 65 °C to 4 °C S4 over the course of 3 hours. The assembly of the DNA origami structures was confirmed by super-resolution imaging with DNA-PAINT (see .
## Assembly of the sample chamber
Surfaces with immobilized DNA origami structures were prepared following a previously reported protocol 2 . In brief, high precision #1.5 coverslips (Paul Marienfeld GmbH, Germany) were sonicated in acetone (chemical grade, Merck KGaA, Germany) for 10 minutes and then rinsed twice with ethanol (chemical grade, Merck Millipore, Germany) and water (milli-Q, Merck Millipore, Germany) and gently dried with pressurized air. The cleaning of the coverslip was completed by putting a drop of 2-propanol on it (Uvasol, Merck KGaA, Germany) and wiping with a paper tissue (Kimtech Science, Sigma Aldrich, Germany). The same procedure was performed on microscope slides (76x26 mm², Menzel, Germany). The high precision coverslip and the microscope slide were assembled into a flow chamber by gluing them together with double-sided sticky-tape (Scotch, Conrad Electronic SE, Germany), yielding a roughly 5x22x0.08 mm 3 large chamber. In a series of volume exchanges, the flow chamber was first incubated with 20 µl of 1 mg/ml albumin, biotin-labeled bovine (Sigma-Aldrich) in buffer A+ for two minutes, washed with 40 µl buffer A+, incubated with 20 µl of 0.5 mg/ml streptavidin (Thermo Fisher Scientific) in buffer A+ for two minutes, washed with 40 µl buffer A+, washed with 40 µl buffer B+, incubated with 20 µl of 0.5 nM of the desired folded DNA origami structures, which were dissolved in buffer B+, for ten minutes, washed with 40 µl buffer B+ and finally loaded with 20 µl of imager strand in the required concentration. In a final step, the chamber was sealed using two-component epoxy glue (Toolcraft, Conrad Electronic SE, Germany). We verified the final concentration of fluorescently labeled ssDNA by confocal FCS measurements (see .
## Supplementary theoretical basis
Theoretical Model for the Autocorrelation Function of a One-Component Binding-Unbinding Reaction without Diffusion Considerable effort has been previously put into the derivation or approximation of an all-embracing correlation curve, which covers lateral 2D-diffusion, 3D diffusion and reversible binding [bib_ref] Measuring Surface Dynamics of Biomolecules by Total Internal Reflection Fluorescence with Photobleaching..., Thompson [/bib_ref] [bib_ref] Theory for Ligand Rebinding at Cell Membrane Surfaces, Lagerholm [/bib_ref] [bib_ref] Total Internal Reflection with Fluorescence Correlation Spectroscopy: Combined Surface Reaction and Solution..., Starr [/bib_ref] [bib_ref] Total Internal Reflection Fluorescence Correlation Spectroscopy: Effects of Lateral Diffusion and Surface-Generated..., Ries [/bib_ref]. To reduce the complexity, and as we are mainly interested in the measurement of binding kinetics, we pursue a simplified approach. We define the autocorrelation function , which is directly computed from acquired images, as
[formula] ( ) = 〈 (0) ( )〉 〈 〉 2(1) [/formula]
Here, is the fluorescence signal, which can be decomposed into a correlated contribution ( ) and an uncorrelated background ( ). In a typical equilibrium system, the total fluorescence signal can be expressed in terms of its mean 〈 〉 and the fluctuations ( ) around this mean:
[formula] ( ) = 〈 〉 + ( ) = 〈 〉 + 〈 〉 + ( ) + ( ) (2) [/formula]
Provided that ( ) is uncorrelated background, the computed autocorrelation function reduces to:
[formula] ( ) = 〈 (0) ( )〉 �〈 〉+〈 〉� 2(3) [/formula]
Practically, it is rather relevant to measure the correlation curve ( ) based on ( ) and normalize to the mean of ( ). Provided that the background can be measured in a separate blank control sample, ( ) can be calculated easily 7 :
[formula] ( ) = ( ) 〈 〉 2 �〈 〉−〈 〉� 2(4) [/formula]
It is worth noting that the amplitude of the autocorrelation curve is decreased by uncorrelated background, but the temporal decay is not altered. In the context of the SI-FCS measurements presented here, the uncorrelated background can be not only background noise or stray light, but also the signal contribution from freely diffusing ligand. The latter can be considered as uncorrelated background if the 3D diffusion of labeled ligand through the detection volume is occurring on a much shorter timescale than the considered binding kinetics.
To obtain an expression for ( ), we note that the fluorescence signal fluctuations have a contribution from freely diffusing and from bound ligand ~+ . As discussed above, only the latter is correlated on the considered timescale. Analogously, we note that 〈 〉~〈 〉. Following the common scheme of derivations for FCS, we assume equivalence of the time and ensemble averages and rewrite ( ) as follows:
[formula] ( ) = ∫ 3 ∫ 3 ′ Φ ( ) ( − ′ ) 〈 〉 2 (∫ d 3 )²(5) [/formula]
## S6
Here, the integrals run over the entire detection volume. To make use of equation 5, we aim to find an expression for , or the concentration correlation function
[formula] Φ ( ) = 〈 (0) ( )〉 (6) [/formula]
The fluctuations in C are governed by binding kinetics, which we assume to be a simple bimolecular reaction of the type A + B ⇌ C, where A is a ligand, freely diffusing above a surface, B is an unbound receptor, which is immobilized at a surface, and C is the bound receptor-ligand-pair (see . Under the assumption that all diffusion dynamics through a considered region of interest are equilibrated, the change of the concentration of conjugates , will be governed by a source and a sink term:
[formula] d d = −(7) [/formula]
Here, we introduced the association rate and the dissociation rate . Both rates fulfill the well-known relation to the dissociation constant and the mean concentrations:
[formula] = = 〈 〉〈 〉 〈 〉(8) [/formula]
As the total number of surface receptors = 〈 〉 + 〈 〉 = const is constant, it is evident that a decrease of receptor-ligand pairs will result in an increase of free receptor by the same magnitude: = − . Therefore, the differential equation for is easily transformed into a differential equation for Φ :
[formula] dΦ ( ) d = −( 〈 〉 + )Φ ( )(9) [/formula]
Differential equations of this kind are very well known and have the simple solution
[formula] Φ ( ) = Φ 0 − /(10) [/formula]
The obtained exponential function decays with the characteristic time constant
[formula] = ( 〈 〉 + ) −1 = � −1 + −1 � −1 [/formula]
, which is related to the dwell time = −1 and the association time = ( 〈 〉) −1 .
To obtain an expression for Φ 0 , we follow the argumentation of Thompson and colleagues 3 : First, Φ 0 needs to meet the initial condition Φ ( = 0) = Φ 0 = 〈 2 〉. This quantity is known as the variance. To find the underlying distribution, we note that for every given point in time, each surface receptor occupies one out of two states: bound to a ligand or unbound. Provided that all receptors are independent, this corresponds to a binomial distribution, which has the variance Φ 0 = (1 − ). Here we used again the total surface concentration of receptors = 〈 〉 + 〈 〉, and introduced the fraction of bound receptors
[formula] = 〈 〉 〈 〉+〈 〉 = 1 1+ 〈 〉 [/formula]
= , which can be interpreted as the success probability of the binomial distribution.
Analogously, the fraction of unoccupied receptors reads
[formula] (1 − ) = 〈 〉 〈 〉+〈 〉 = 1 1+ 〈 〉 = . Therefore, we obtain Φ 0 = 〈 〉 = 〈 〉(1 − )(11) [/formula]
And finally, after insertion into equation 5, we get an analytic expression for the autocorrelation function of reversible binding:
S7
[formula] ( ) = 1 − / = 1 1− − /(12) [/formula]
Here, we introduced the average number of bound receptors and the total number of receptors in the detection volume. Alternatively, equation 12 can be obtained as a limiting case of the advanced derivation of the full autocorrelation by Thompson and colleagues 3 . Interestingly, the amplitude 0 = lim →0 ( ) of the correlation is not only proportional to the absolute number of occupied binding sites, but also depends on the fraction of unoccupied sites. However, if ≫ , i.e. in case of low concentration of labeled ligand 〈 〉 ≪ , the number of occupied binding sites can be obtained directly as the inverse of the correlation amplitude.
## S8
# Supplementary tables
Free energy of DNA hybridization . Hybridization parameters for different DNA sequences with the target sequence 5'-CTAGATGTAT-3'. SI-FCS measurements were performed in a low concentration regime of labeled strands, such that the dissociation rate is directly estimated from the characteristic decay time of the autocorrelation curve
[formula] sequence concent ration 〈 〉[ ] overlap predicted � � measured [ ] measured [ − ] estimated - − [ − − ] 5'-TTATACATC-3'10 7〈 〉 ≪ �⎯⎯⎯� = −1 . [/formula]
For each pair of sequences, the free energy Δ of hybridization was estimated using the Nucleic Acid Package (NUPACK) [bib_ref] Analysis and Design of Nucleic Acid Systems, Zadeh [/bib_ref] with the following settings: temperature = 296.15 K, concentration of Na + 50 mM, concentration of Mg 2+ 9 mM. The calculations were performed based on the parameters provided by SantaLucia 9 , which had to be adjusted for our buffer conditions. To describe our conditions best, we used the minimum concentration of Na + compatible with ref. [bib_ref] Unified View of Polymer, Dumbbell, and Oligonucleotide DNA Nearest-Neighbor Thermodynamics, Santalucia [/bib_ref] , which compensates partially for the Tris in our buffer. The remaining Na + could be accounted for by lowering the Mg 2+ , although the relevant equivalent amount of Mg 2+ would be small [bib_ref] Predicting Stability of DNA Duplexes in Solutions Containing Magnesium and Monovalent Cations, Owczarzy [/bib_ref] [bib_ref] Oligonucleotide Melting Temperatures under PCR Conditions: Nearest-Neighbor Corrections for Mg2+, Deoxynucleotide Triphosphate,..., Von Ahsen [/bib_ref] [bib_ref] Part 1. Basic Requirements for Designing Optimal Oligonucleotide Probe Sequences, Mitsuhashi [/bib_ref]. The dissociation constant and the binding free energy Δ are linked via the well-known equation Δ = − ln 0 . Here, we introduced the gas constant R, the temperature T, and a reference constant 0 = 1 M, which has the sole purpose to ensure that the logarithm is applied to a dimensionless quantity. Consequently, after obtaining as the inverse of , the association rate is calculated as = 0 Δ . The obtained association rates can be regarded as estimates of the true rates. The estimated association rates are in line with values reported elsewhere [bib_ref] Single-Molecule Fluorescence Imaging of Interfacial DNA Hybridization Kinetics at Selective Capture Surfaces, Peterson [/bib_ref] [bib_ref] Thermodynamic Dependence of DNA/DNA and DNA/RNA Hybridization Reactions on Temperature and Ionic..., Lang [/bib_ref] [bib_ref] Single-Molecule Kinetics and Super-Resolution Microscopy by Fluorescence Imaging of Transient Binding on..., Jungmann [/bib_ref] [bib_ref] Single-Molecule Kinetics Reveal Cation-Promoted DNA Duplex Formation Through Ordering of Single-Stranded Helices, Dupuis [/bib_ref] [bib_ref] Quantitative Super-Resolution Imaging with qPAINT, Jungmann [/bib_ref]. Moreover, all estimated association rates appear to be similar, regardless of the basepair overlap. The same observation was recently reported for 9 nt and 10 nt hybridization 15 .
## S9
# Supplementary figures
Control for photobleaching . Dependence on Excitation Power. a) Bleaching is observed as an apparent reduction of the residence time and therefore also as a reduction of . The autocorrelations were averaged over several ROIs and measured with 〈 〉 = 10 nM and 9 nt base pair overlap. We identified irradiances below 0 < 0.037 kW cm 2 to be free from bleaching. For 10 nt the residence times are significantly longer and the bleaching free regime was determined to be 0 < 1.6 - 10 −3 kW cm 2 , which was achieved by reducing the excitation power and the frame rate of acquisition from 85 Hz to 10 Hz (data not shown). b) The Gaussian shape of the illumination profile may induce a spatial profile of the obtained from different ROIs. For high irradiances, the apparent diffusion time is not only globally lowered due to bleaching, but also is shortest in the center of illumination. For irradiances below 0 < 0.037 kW cm 2 , this spatial distribution becomes negligible. . Reproducibility of SI-FCS measurements. To estimate the robustness and reproducibility of this method, we repeated identical measurements on one and the same sample. When superimposing the autocorrelation curves from seven measurements on the hybridization of ssDNA with 9 nt overlap, all curves were indistinguishable . For every measurement, the standard deviation of the characteristic decay times obtained from 49 ROIs was considerably lower than 10% of the mean, demonstrating that consistent results are obtained over the entire field of view, independent of the local illumination profile. Moreover, the comparison of the average characteristic decay times from several independent measurements showed that the scatter is less than 5% of the overall average, that is 〈 〉 = 4.81 ± 0.05) , with the error being the standard deviation of the seven measurements. This series of similar measurements demonstrated the excellent accuracy of SI-FCS for the determination of binding rates. S11 Determination of kinetic rates from two measurements . Extraction of association and dissociation rates from two measurements. Based on equation 1 (main text) the knowledge of two points of the titration curve is in principle sufficient to determine the association rate , the dissociation rate and therefore also the equilibrium constant = ⁄ . a) For 9 nt, the full titration curve (left panel), (center panel) and (right panel) calculated from pairs of two measurements along the titration curve. The relative difference � / − / ,titration � / ,titration � is represented by the color of the points. For pairs of concentrations (〈 1 〉, 〈 2 〉) of differently dominated regimes ( ≪ 〈 1 〉 and > 〈 2 〉) the rates can be recovered with an error smaller than 20% (highlighted as squares). Pairs of concentrations which were different less than a factor of two were excluded from the analysis and marked as crosses. Concentration pairs leading to a relative error of more than 100% saturated the color scale and were marked as diamonds. b) Same as panel a), but the analysis was based on data sets from 10 nt hybridizations. [fig_ref] 4: Here, is the average number of particles in the detection volume and... [/fig_ref]. Effect of the measurement time. We performed Monte Carlo simulations to evaluate the effect of the duration of individual measurements. To this end, we simulated 10 signal traces, which were 10 5 times longer than the characteristic decay time, fitted the resulting autocorrelation curves and related the results to the initially set characteristic decay time. To assess the effect of the measurement time but keep the statistics comparable, we cut the initial traces into shorter traces and repeated the analysis, thereby keeping constant the total number of binding events observed. For example, to analyze the case of traces, which are 10 3 times longer than the characteristic decay time, we split each of the 10 initial traces (10 5 times longer than ,sim ) into 100 sub-traces and subsequently computed and analyzed all resulting 1000 autocorrelation curves. [fig_ref] 4: Here, is the average number of particles in the detection volume and... [/fig_ref] shows the obtained decay time ,meas as a function of the duration of individual simulations. Both parameters are normalized to the characteristic decay time ,sim set in the simulation, which we expect to be the relevant time scale when assessing the required measurement time [bib_ref] Photon Correlation Measurements at Large Lag Times: Improving Statistical Accuracy, Schätzel [/bib_ref] [bib_ref] Statistical Analysis of Fluorescence Correlation Spectroscopy: The Standard Deviation and Bias, Saffarian [/bib_ref]. To avoid any effects of poor sample statistics, we repeated every simulation 10 times and varied the number of considered binding sites from 1 to 1000. The corresponding results are shown as mean (central line) plus/minus standard deviation of the ten simulations (shaded area). No difference between the different settings is discernible and it appears that the bias from the autocorrelation (biased estimator [bib_ref] Correlation Techniques in Dynamic Light Scattering, Schätzel [/bib_ref] [bib_ref] Photon Correlation Measurements at Large Lag Times: Improving Statistical Accuracy, Schätzel [/bib_ref] converges to zero for sufficiently long measurements. Nonetheless, it should be noted, that the required measurement times for binding studies using SI-FCS can be long. When aiming for a bias smaller than 10%, one should conduct measurements at least 300-fold longer than the characteristic decay time. Therefore, slow dynamics require particularly long measurement time, which makes the use of a S13 focus stabilization system essential. Nonetheless, for a correlation curve originating from binding and unbinding the convergence happens significantly faster than for a correlation curve originating from 2D diffusion [fig_ref] 4: Here, is the average number of particles in the detection volume and... [/fig_ref]. The reason is the different shape of the correlation curves. For reversible binding, the autocorrelation curve decays relatively fast as a single exponential ( )~− , whereas the autocorrelation curve for 2D diffusion 2 ( )~�1 + � −1 has a significantly longer tail. [fig_ref] 4: Here, is the average number of particles in the detection volume and... [/fig_ref] demonstrates this effect by superimposing both autocorrelation curves with a time axis normalized to the relevant decay times and respectively.
## Reproducibility of si-fcs measurements
## Required measurement time
To confirm the simulation results, we re-analyzed the measurements for 7-10 nt in a similar way as we analyzed the long simulated traces. By this, we can superimpose the simulated dependence from [fig_ref] 4: Here, is the average number of particles in the detection volume and... [/fig_ref]. Strikingly, without any fitting involved, the simulation and the experimental results follow the same trend. Moreover, in this depiction, we see no difference between different nucleotides, which is expected as we normalize by . Nonetheless, we can conclude that the underlying dynamics for 7-10 nt hybridization are identical, and only occur on different time scales, which makes the depicted relation a universal concept. For short measurement times, the bias of the experimental data seems to be slightly larger than the simulation suggests, which we attribute to noise involved in the measurements. From [fig_ref] 4: Here, is the average number of particles in the detection volume and... [/fig_ref] we conclude that the systematic bias on the obtained characteristic decay time cannot be smaller than indicated by the simulated dependence. The bias cannot be reduced below this line, not even by an increasing number of binding events observed within the same measurement duration. The only option to reduce the systematic bias is an increased duration of the total measurement time. Furthermore, [fig_ref] 4: Here, is the average number of particles in the detection volume and... [/fig_ref] shows that an individual measurement has to be at least 300fold longer than the characteristic decay time to achieve a bias smaller than 10%. Based on this finding, we can replot [fig_ref] 4: Here, is the average number of particles in the detection volume and... [/fig_ref] without normalization of all times and superimpose a border, which corresponds to the 10% systematic bias (solid line, [fig_ref] 4: Here, is the average number of particles in the detection volume and... [/fig_ref]. All points on the right hand side of this line have a bias of less than 10%, which gives a direct and quick check for whether a particular measurement was long enough to provide the required accuracy. S14 SI-FCS measurements of reversible binding are calibration-free . SI-FCS is calibration-free and robust to defocused imaging. The calibration of the detection volume is a crucial step in confocal FCS measurements. On the other hand, camera-based SI-FCS measurements on diffusion dynamics in supported lipid bilayers have been shown to be calibration-free by variable pixel binning during post-processing of images [bib_ref] Calibration and Limits of Camera-Based Fluorescence Correlation Spectroscopy: A Supported Lipid Bilayer..., Bag [/bib_ref]. In contrast, when looking at the reversible binding of labeled freely diffusing ligand to a surface-immobilized target, SI-FCS is intrinsically calibrationfree, as it does not require any lateral spatial resolution. Here, the signal fluctuation originates from binding and unbinding events to the surface. Any changes in the lateral size of the detection volume do not qualitatively alter the underlying kinetics of the fluctuating signal. To prove this, we performed a series of measurements at different axial positions of the detection volume relative to the sample surface. This is equivalent to systematic defocusing, which varies the lateral size of the PSF. The SI-FCS experiments presented here are within reasonable limits robust to any defocusing of the sample. The situation changes when looking at lateral diffusion through a detection volume, e.g. a region of interest spanning a certain number of pixels on the camera detector. In this case, the decay of the autocorrelation curve depends on the diffusion coefficient, the size of the region of interest and the size of the PSF 6,20 . a) Autocorrelation curves and single-exponential fits obtained from the same sample but different axial positions of the specimen. The axial positions are color coded and can be inferred from panel b). b) Characteristic decay times obtained from single-exponential fits of the SI-FCS autocorrelation functions measured for different sample positions along the optical axis.
## S15
Surface density accessible to SI-FCS . Performance of SI-FCS at different receptor surface densities: Simulations. We simulated SI-FCS image series (see Supplementary Theoretical basis) using the Picasso software tool 2 . To assess the surface densities of receptors which are compatible with SI-FCS, we started off by simulating SI-FCS measurements for 24X24 pixels (160x160 nm² per pixel), with = 1 s −1 , 〈 〉 = 0.008 s −1 , and 3 to 100,000 receptors in the considered area. The simulated frame rate was 10 Hz and a total of 40,000 frames was simulated for each run, corresponding to a total measurement duration meas = 4000 s. Each simulation was performed ten times to develop a feeling for the scatter of results. -c show ten individual curves each (light grey), superimposed with their mean (blue line) and the corresponding single exponential fit (red line) from which the characteristic decay time is extracted. These curves demonstrate once more that a single exponential describes the autocorrelation curves appropriately. Second, the least from 300 binding sites on, the statistical scatter for the simulated conditions becomes small. Each individual autocorrelation curve represents on average � −1 + −1 〈 〉 −1 � meas −1 binding events (for = 300 less than 10 4 events). For all receptor densities considered, we recovered the simulated characteristic decay times accurately . With reasonable computation times (maximum number of binding sites 10 5 ), we could not find a regime where SI-FCS could not recover the simulated characteristic decay times. This is a clear advantage of SI-FCS over tracking based approaches in which the residence time is determined. Here, we are not confined to regimes where individual particles can be identified. ,000 and 30,000 binding sites. Clearly, already in the case of 3,000 binding sites, a tracking based approach would suffer from misassignments during the reconstruction of tracks, leading to overestimations of residence times. We note that the number of binding sites or the surface density of binding sites are not particularly good parameters to compare the performance of SI-FCS and tracking-based estimations of the residence time. As tracking relies on the detection of individual particles, it is rather relevant to replot in terms of the occupied binding sites per resolution disk = resolution disk . Here, we used the total density of binding sites S, and the fraction of occupied binding sites = 1 1+ < > (see Supplementary Methods). As the resolution disk we define a circle with the 1/e² value of the Gaussian-shaped PSF as radius. This equals to the assumption that two resolution limited spots can be distinguished reliably if their centers are separated by at least . As shown in , the SI-FCS approach reliably reproduces the simulated binding times not only in a regime where tracking-based approaches would perform, but also at surface densities of bound receptors which exceed the tracking regime by at least two orders of magnitude.
## S17
Measurements surface density . Performance of SI-FCS at different receptor surface densities: Experimental data a) Autocorrelation curves obtained from five individual SI-FCS measurements of the hybridization kinetics of a 9 nt overlap. The measurements differ in the origami surface density, which is intrinsically difficult to control quantitatively. As a simple approach, we incubated the surfaces for the same time with different concentrations of DNA origami (0.03, 0.1, 0.3, 1.0, and 3 nM). SI-FCS yields reasonably consistent decay times for the range of investigated surface densities, although a slight increase is observed for very high surface densities. The dashed line corresponds to the surface concentration we used in typical SI-FCS measurements . b) Representative images corresponding to the conditions described in a) For the lowest surface concentrations, particle tracking approaches could be potentially conducted, whereas for origami concentrations above 0.1 nM, tracking approaches would clearly fail, because individual events cannot be identified any longer. SI-FCS yielded smooth low-noise autocorrelation curves even in regimes that were clearly not accessible to tracking approaches. Time series were recorded without binning with a resolution of 256x256 pixel. To validate that the target concentration of ligand is indeed reached in the sample, we performed confocal FCS measurements on a commercial LSM 780 Confocor3 system (Zeiss AG, Oberkochen, Germany). As we were interested in the concentration of imager strand in solution, we positioned the confocal volume 30 µm above the origami-coated surface. In an initial measurement, the confocal volume was calibrated using Alexa546NHS (ThermoFisher) and its reported diffusion coefficient = 341 µm 2 s at 22.5°C 21 . As our experiments were carried out at 27°C, we adjusted the diffusion coefficient using the well-known relation ~( ) and an empirical expression for the temperature dependence of the viscosity of water [bib_ref] Viscosity of Liquid Water in the Range -8 C to 150 C, Kestin [/bib_ref]. We ensured that the autocorrelations had no bleaching or triplet contributions by performing identical measurements at a wide range of irradiances (data not shown). Therefore, the correlation curves could be fitted by a simple 3D diffusion model function: . shows three representative correlation curves, their corresponding fits and the residuals. As expected, the amplitude scales with the concentration. Nonetheless, the investigated concentrations do not have an effect on the diffusion of ligand itself as the correlation curves become indistinguishable when normalized by the S21 amplitude at zero lag time . For the range of measured concentrations the target concentration and the measured concentration coincide within 10% . As a byproduct, we obtained the diffusion coefficient of the labeled 10 nt ssDNA strand in the specified buffer (B+) at 27°C: = (205.2 ± 7.9) µm 2 s , which is in good agreement with a previously reported results [bib_ref] Unified Description of Electrophoresis and Diffusion for DNA and Other Polyions, Stellwagen [/bib_ref]. The presented numbers correspond to mean and standard deviation of 51 measurements, each of them at least 10 min long.
[formula] ( ) = −1 �1 + � −1 �1 + 2 � [/formula]
## S22
Effect of ROI size . Effect of the ROI size. a) Average normalized autocorrelation curves for 9 nt hybridization obtained for different ROI sizes are indistinguishable and are adequately described by a single exponential decay. To facilitate high frame rates, we employed 4x4 pixel (px) hardware binning during image acquisition. b) Boxplots of the obtained corresponding characteristic decay times are independent of the ROI size. The center lines mark the median. The box edges correspond to upper and lower quartile, and are extended by the whiskers marking 1.5 times the inter-quartile range. Data points outside the whiskers are marked as crosses. For small ROIs, the overall scatter is larger, as less events are sampled within an individual ROI. The average , however, is in agreement with larger ROI sizes. Theoretically, for too large ROIs, the fluctuations become less relevant, the amplitude of the autocorrelation curve approaches zero, and a fluctuation-based autocorrelation analysis becomes less reliable. For the investigated conditions, however, this regime is not reached. On the other hand, for large ROIs and small correlation amplitudes, the autocorrelation function becomes more sensitive to mechanical and laser excitation instabilities. In this particular measurement, small oscillations of the piezo controlling the TIRF angle position could be observed as oscillations for the 128 px ROIs, but not for smaller ROIs. The chosen ROI size of 31x31 px offered spatial resolution to investigate bleaching across the illumination profile, but did not affect the overall measured . c) The larger overall scatter is illustrated by generating a map of decay times. The computational effort and memory consumption increase quadratic with the ROI size or linearly with the number of ROIs.
In this particular example, the measured is independent of the ROI size. It should be noted, that there may be conditions, e.g. for very high or very low densities of binding events, where this is not the case.
Although we experienced that SI-FCS measurements were generally robust against the choice of the ROI, we suggest to carefully explore a range of ROI sizes during post-processing.
[fig] Figure S8, Figure S10: Custom-built TIRF microscope. See supplementary methods. S19 DNA-PAINT reconstruction of binding sites on DNA origami structures Figure S9. Super-resolution of DNA origami exposing 12 single-stranded DNA handles. A) Schematic of the rectangular DNA origami structures, exposing 12 ssDNA handles. The image was generated with the Picasso software tool. 2 b) Representative DNA-PAINT images of the rectangular origamis. S20 Measurement of the ligand concentration Direct measurement of the ligand concentration. [/fig]
[fig] 4: Here, is the average number of particles in the detection volume and the diffusion time is defined as the ratio of the square of the −2 -value of the Gaussian detection volume and the four-fold diffusion coefficient = 2 The structure parameter represents the ratio of axial to lateral extension of the Gaussian-shaped detection function. The concentrations are directly obtained from the amplitude of the correlation curves: [/fig]
|
Carboxymethyl Cellulose Acetate Butyrate: A Review of the Preparations, Properties, and Applications
Carboxymethyl cellulose acetate butyrate (CMCAB) has gained increasing importance in several fields, particularly in coating technologies and pharmaceutical research. CMCAB is synthesized by esterification of CMC sodium salt with acetic and butyric anhydrides. CMCAB mixed esters are relatively high molecular weight (MW) thermoplastic polymers with high glass transition temperatures (Tg). CMCAB ester is dispersible in water and soluble in a wide range of organic solvents, allowing varied opportunity to the solvent choice. It makes application of coatings more consistent and defect-free. Its ability to slow down the release rate of highly water-soluble compounds and to increase the dissolution of poorly soluble compounds makes CMCAB a unique and potentially valuable tool in pharmaceutical and amorphous solid dispersions (ASD) formulations.
# Introduction
Cellulose is the most abundant regenerated biopolymer in the planet, with annual production of about 5 × 10 11 metric tons. Most of the cellulose is utilized in industry as a raw material in paper production. Only about 4 from 108 million tons of annually produced pulp are used for chemical production [bib_ref] Smart' materials based on cellulose: a review of the preparations, properties, and..., Qiu [/bib_ref]. Hydroxyl groups of cellulose can be reacted to form esters or ethers of different physical and chemical properties suitable for various applications [bib_ref] Application of cellulose acetate butyrate-based membrane for osmotic drug delivery, Shanbhag [/bib_ref].
Cellulose derivatives have significant roles in industry; they represent a main source for fibers, textiles, coatings, thermoplastic films, food additives [bib_ref] Smart' materials based on cellulose: a review of the preparations, properties, and..., Qiu [/bib_ref] , and pharmaceutical technologies [bib_ref] Cellulose-medical, pharmaceutical and electronic applications, chapter 3, application of cellulose and cellulose..., Shokri [/bib_ref]. Cellulose derivatives are usually classified as two main classes, esters and ethers, according to the reactant nature. Cellulose derivatives usually contain free hydroxyl groups available for additional treatments to yield mixed esters. The mixed esters have several improved properties over all neat esters. Cellulose acetate propionate (CAP) and cellulose acetate butyrate (CAB) are the most commercially important mixed esters [bib_ref] Cellulose esters, Balser [/bib_ref]. Mixed cellulose derivatives with both ester and ether groups could be also attained. Due to their low degree of substitution and high hydrolytic stability, carboxymethyl cellulose can be further esterified with organic acid anhydrides to add either single or mixed ester groups.
Cellulose esters in coating compositions improved many properties as hardness, aluminum flake orientation, flow and leveling, redissolving resistance, clarity, and gloss while it reduced dry-to-touch time, cratering, and blocking [bib_ref] Rheological aspects of carboxymethyl cellulose acetate butyrate (CMCABTM) in waterborne coatings, Lawniczak [/bib_ref] [bib_ref] Carboxymethylcellulose acetate butyrate in water-based automotive paints, Posey-Dowty [/bib_ref] [bib_ref] Aqueous dispersions of carboxylated cellulose esters, and methods of making them, Mccreight [/bib_ref] [bib_ref] Low molecular weight carboxyalkyl cellulose esters and their use as low viscosity..., Shelton [/bib_ref]. Mixed cellulose derivatives afford the benefits of conventional cellulose esters with a moderate increase in viscosity without organic solvents addition, which reduce the volatile organic compounds (VOCs) levels in coating compositions [bib_ref] Low molecular weight carboxyalkyl cellulose esters and their use as low viscosity..., Shelton [/bib_ref]. In coating compositions cellulose esters are usually applied from organic solvent solutions. Nowadays, aqueous compositions are used in adhesives, paints, and inks instead of organic solvent-based compositions in order to reduce the amount of VOC [bib_ref] Use of carboxymethyl cellulose acetate butyrate as a precoat or size for..., Obie [/bib_ref]. CMCAB is thought to bring health and safety benefits by replacing the organic solvent-based compositions with aqueous compositions being likely to ensue or exceed the standards performance of solvent-based compositions [bib_ref] Use of carboxyalkyl cellulose esters, such as carboxymethyl cellulose acetate butyrate, to..., Obie [/bib_ref].
Amorphous solid dispersions (ASD) are smart way to improve solubility and bioavailability of drugs. CMCAB has 2 Journal of Drug Delivery recently shown potential as an ASD polymer [bib_ref] Enhanced dissolution of poorly soluble drugs from solid dispersions in carboxymethyl cellulose..., Li [/bib_ref]. CMCAB ASDs provide accelerated drug release with water-soluble drugs, while with poor water-soluble drugs it provides stable supersaturated solution, with almost zero-order release profile [bib_ref] Enhancement of naringenin solution concentration by solid dispersion in cellulose derivative matrices, Li [/bib_ref].
Cellulose derivatives containing carboxyl groups are very appropriate resources for ASD because they have excellent safety profiles, strong interactions with drug particles, the ability to adapt hydrophobicity through substituent nature and degree of substitution (DS), and high glass transition temperatures (Tg) values that promote formulation stability. High Tg preserves the medium in the glassy form even at high humidity and temperatures, regulating drug molecular motion and thus preventing crystallization during storage and transportation [bib_ref] Enhanced dissolution of poorly soluble drugs from solid dispersions in carboxymethyl cellulose..., Li [/bib_ref].
## Preparation
Similar to the cellulose acetate preparation, CMCAB is synthesized by esterification of CMC sodium salt. CMC-Na was first protonated and reactivated with acid to transform them to the free acid form, CMC-H. Then the swollen chains are esterified with acetic and butyric anhydrides. Mixed esterification by acetic and butyric acid is taking place at the same time so; to achieve a higher DS of butyrate groups a great excess of butyric acid must be used [bib_ref] Low molecular weight carboxyalkyl cellulose esters and their use as low viscosity..., Shelton [/bib_ref] [bib_ref] Use of carboxymethyl cellulose acetate butyrate as a precoat or size for..., Obie [/bib_ref] [bib_ref] Preparation and characterization of cellulose acetate/carboxymethyl cellulose acetate blend ultrafiltration membranes, Han [/bib_ref] [bib_ref] Electrospun carboxymethyl cellulose acetate butyrate (CMCAB) nanofiber for high rate lithium-ion battery, Qiu [/bib_ref]. Two commercial products were developed and marketed as a low and high MW type of this mixed ester.
Enzymatic acetylation of CMC by lipase provides a high selective method of mild reaction conditions with reducing the need for hazardous chemicals [bib_ref] Enzymatic acetylation of carboxymethyl cellulose by lipases, Wang [/bib_ref]. Heterogeneous esterification of CMC with high degree of substitution could be achieved quickly by using 4-dimethylaminopyridine or 2,2dichloro-propionyl chloride [bib_ref] Acylation of carboxymethyl cellulose accelerated by 4-dimethylaminopyridine, Klemm [/bib_ref].
Mostly the CMCAB esters preparation method is as follows [bib_ref] Low molecular weight carboxyalkyl cellulose esters and their use as low viscosity..., Shelton [/bib_ref] [bib_ref] Use of carboxymethyl cellulose acetate butyrate as a precoat or size for..., Obie [/bib_ref].
## Transformation of cmc-na to the free acid form.
The preparation of the CMCAB is accomplished by converting CMC-Na into its acid form (CMC-H) by soaking in an aqueous sulfuric acid solution (hydrochloric acid, nitric acid, or acetic acid could also be used) and then washed with water to remove the acid.
About 100 g carboxymethyl cellulose (Na salt form, 0.30-0.65 DS carboxymethyl) was added to 2000-3000 g of 2-20% aqueous sulfuric acid at 27-60 ∘ C. The slurry was stirred for about 10-120 minutes for agitation, and the acid solution was filtered free of excess liquids and washed with demineralized water to recover CMC-H.
## Activation of cmc (cmc-h).
The protonated CMC was transferred to a filtration funnel and excess water was drained to approximately 20-40% activated solids. The activate was dewatered by solvent exchange of the water first with three portions of acetic acid and then with three or four portions of butyric acid (each washing portion of 200-250 g acid to 100 g CMC) to give 40% solids of activated butyric acid wet CMC(H). In between each wash the sample was drained to approximately 18% solids.
## Esterification.
The butyric acid wet CMC-H was esterified by treatment with 31 g acetic anhydride and 253 g butyric anhydride at 0-15 ∘ C. A catalyst solution consisting of 3.44 g sulfuric acid in 3.44 g acetic acid was added slowly to the reaction mixture keeping the temperature below 30 ∘ C (perchloric acid, sulfoacetic acid, or zinc chloride could also be used as catalyst). After the end of the catalyst addition, the temperature was held at 30-35 ∘ C for 60-150 minutes. Then the temperature of the reaction mixture was raised to 45-60 ∘ C and held for 2-6 hours until the complete dissolution of the solids to give trisubstituted carboxymethyl cellulose, upon precipitation into water. The different acyl content of CMCAB was obtained by adjusting the ratio of acetic anhydride and butyric anhydride during esterification reaction.
Other preparing methods for CMCA suggest that, after removing the liquid well, the resultant was placed into a kneader and mixed with 250 g of acetic acid, 5.6 g of sulfuric acid, and 150 g of acetic anhydride. The reaction was performed at 48-50 ∘ C for 4 hours. Esterification could be attained also by mixing with 84 g of acetic acid, 316 g of methylene chloride, 210 g of acetic anhydride, and 1.5 g of sulfuric acid. These methods could be adapted for CMCAB preparation by introducing butyric acid and butyric anhydride along/instead of acetic acid and acetic anhydride.
## Hydrolysis and neutralization.
As with conventional esters, this ester is usually completely substituted for product uniformity. This product is then hydrolyzed using sulfuric acid to provide a desired partially substituted carboxymethyl cellulose ester. Hydrolysis is important to offer gel-free solutions in organic solvents and to provide improved compatibility with other resins in coatings formulations. The free hydroxyl groups obtained after hydrolysis are essential crosslinking sites in many coatings applications. For optimum thermal and hydrolytic stability of the final product, it is important to neutralize the strong acid catalyst [bib_ref] Low molecular weight carboxyalkyl cellulose esters and their use as low viscosity..., Shelton [/bib_ref].
A solution of 95 g water, 95 g acetic acid, and 2 g sulphuric acid was added dropwise to the reaction mixture over 30-45 minutes at 40-45 ∘ C. The contents were hydrolyzed by heating to 60-72 ∘ C for 2-4.5 hours. Then, the excess sulfuric acid is neutralized by addition of 7.53 g magnesium acetate tetrahydrate, dissolved in 25 g water and 25 g acetic acid.
## Precipitation and filtration.
The reaction mixture, either the fully substituted or partially hydrolyzed forms of the CMCAB, was then poured into about 20 times its volume of water, and the formed precipitate was filtered, washed well from organic acids and inorganic salts with water, and dried at 60 ∘ C under vacuum to obtain the acid form of CMCAB as a white granular powder.
An optional precipitation method is to add 10% aqueous acetic in sufficient volume to yield about 30% organic acid in the final precipitation bath, about 3,000 mL, followed by the addition of an equal amount of water to harden the precipitated particles.
## Cmcab salts. cmcab with ds cm of 0.3 is insoluble in
water. Water solubility is achieved when the carboxylic acid groups are fully neutralized into their salts form. However, partial solubility and thus stable dispersions are attained by partial neutralization of the acid groups, where ammonia or various organic amines are commonly employed to create ammonium carboxylate salts along the polymer backbone. The tertiary amine, N,N-dimethylethanolamine, is frequently used for partial neutralization. CMCAB salts promote water interactions that stabilize the dispersion, and the degree of neutralization controls the dispersion properties. Too much neutralization will promote CMCAB water solubility, potentially resulting impractically in high dispersion/solution viscosity [bib_ref] Rheological aspects of carboxymethyl cellulose acetate butyrate (CMCABTM) in waterborne coatings, Lawniczak [/bib_ref] [bib_ref] Aqueous dispersions of carboxylated cellulose esters, and methods of making them, Mccreight [/bib_ref].
Different CMCAB salt dispersions, sodium, potassium, calcium, or ammonium carboxylate may be formed by first dissolving CMCAB (acid form) into an organic solvent (acetone), and then 0.1-0.5 N neutralizing alkaline solution is added dropwise and then dried under vacuum by means of a rotary evaporator to give the corresponding salt.
As the CMCAB salt is obtained, the degrees of substitution by carboxymethyl group and by acetyl group were determined. The resulting white solid CMCAB had the following general composition: acid number 47-60, sulfur 25-100 PPM, DS of carboxymethyl groups 0.26-0.35, butyryl 1.64-2.26, acetyl 0.49-0.60, and hydroxyl 0.10-0.60.
## Properties
Carboxymethyl cellulose acetate butyrate mixed esters are thermoplastic polymers with relatively high molecular weight (MW) and high glass transition temperatures (Tg). CMCAB is relatively nonpolar; it is insoluble in water and soluble in common organic solvents as alcohols, ketones, esters, and glycol ethers [bib_ref] Zero-order release formulations using a novel cellulose ester, Posey-Dowty [/bib_ref]. By complete neutralization of the carboxylic acid groups to their salts form, water solubility occurs. CMCAB solubility is affected by many factors as carboxyl content, percent of carboxylic group neutralization, degree of substitution, substitution homogeneity, percent of acetate and butyrate groups, and the viscosity [bib_ref] Low molecular weight carboxyalkyl cellulose esters and their use as low viscosity..., Shelton [/bib_ref].
Ethyl acetate is known as a better solvent for CMCAB than acetone. CMCAB films formed from ethyl acetate solution were smooth and flat. Contact angle measurements considered CMCAB surfaces as hydrophobic due to the position of methyl groups. CMCAB surface energy is close to that of collagen [bib_ref] Solution behavior and surface properties of carboxymethyl cellulose acetate butyrate, Amim [/bib_ref].
Rheological properties viscosity, thixotropy loop, and dynamic viscoelasticity of CMCAB water dispersions (WD) were conducted. The viscosity of WD increases as the ratio of solvent/water decreased. Significant increase in viscosity was noticed by neutralization until a 50% degree and then it increases slowly. WD is a positive thixotropy liquid and is viscous at low concentration and neutralization degree. WD begins to form instable gel structure as concentration and neutralization degree increased [bib_ref] Rheological properties of carboxymethyl cellulose acetate butyrate water dispersions, Lv [/bib_ref].
## Applications
Carboxymethyl cellulose acetate butyrate combines two known commercial cellulose derivatives, carboxymethyl cellulose (CMC) and cellulose acetate butyrate (CAB). CMCAB is specifically designed for use in the waterborne coating industry. CMCAB makes coatings application more consistent and defect free; it provides to waterborne systems improved flow and leveling combined with sag resistance, reduces dry time, controls viscosity, has excellent metal flake control, and also facilitates the dispersion of hydrophobic species into water-based coating systems [bib_ref] Recent developments for carboxymethyl cellulose acetate butyrate in waterborne coatings, Eggers [/bib_ref]. CMCAB attains these properties through high Tg, near-Newtonian rheology, and a sharp viscosity/solids relationship [bib_ref] Rheological aspects of carboxymethyl cellulose acetate butyrate (CMCABTM) in waterborne coatings, Lawniczak [/bib_ref].
The aqueous pigment dispersions could be prepared by adding an organic solvent solution of the CMCAB ester to a metallic dispersion and neutralizing, partially or fully, the ester by base addition (ammonia or amine) and then mixing in water. The resulting mixture may then be added directly to a dispersion containing only aluminum flake and an organic solvent. The esters also have specific use as wetting agents in high solids coatings. Low molecular weight carboxyalkyl cellulose esters are useful in coating and ink compositions as low viscosity binder resins and rheology modifiers [bib_ref] Low molecular weight carboxyalkyl cellulose esters and their use as low viscosity..., Shelton [/bib_ref].
A stable aqueous coating dispersion is provided containing up to 50%, based on the total solids weight, cellulose mixed ester with organic solvent, water, and a suitable amine neutralized acrylic resin [bib_ref] Aqueous coatings composition containing cellulose mixed ester and amine neutralized acrylic resin..., Walker [/bib_ref]. Aqueous dispersion of carboxylated cellulose esters in water that is suitable for use in a variety of waterborne coating compositions while having a relatively low VOC content was suggested [bib_ref] Aqueous dispersions of carboxylated cellulose esters, and methods of making them, Mccreight [/bib_ref].
Low molecular weight CMCAB with a maximum degree of substitution as coating additive provides high solids and low viscosity coating compositions, without the drawbacks usually found with typical low molecular weight cellulose esters such as the increase in organic solvent ratio to reserve the required viscosity. They have a marginal effect on solution and spray viscosities of high solids coatings. It displays superior compatibility when mixed with other coating resins, thus yielding clear films compared to conventional cellulose esters [bib_ref] Low molecular weight carboxyalkyl cellulose esters and their use as low viscosity..., Shelton [/bib_ref].
CMCAB provides enhanced aluminum flake orientation and improved hardness. They can provide a high gloss protective coating for several substrates, especially metal and wood. The solids percent can be increased relative to organic solvent, consequently reducing the VOC of the formulation [bib_ref] Low molecular weight carboxyalkyl cellulose esters and their use as low viscosity..., Shelton [/bib_ref].
When carboxymethyl cellulose esters of higher acids, including CMCAB, are treated with ammonia or an amine they are readily dispersed in waterborne formulations, containing metallic pigments such as aluminum flakes and mica. It improves rheological properties by a remarkable increase in viscosity with a minor increase in ester concentration.
To prevent sagging and dripping in waterborne coatings, rapid solids increases due to solvent evaporation are important. Due to its exceptional balance of hydrophilicity from the carboxylate groups and hydrophobicity from its acetate groups, CMCAB affords a good substrate wetting with wonderful flow and leveling and excellent redissolving resistance and decreased defects in waterborne paints as cratering, sagging, and picture framing [bib_ref] Rheological aspects of carboxymethyl cellulose acetate butyrate (CMCABTM) in waterborne coatings, Lawniczak [/bib_ref]. Also CMCAB in water dispersions keeps metallic flakes orientation in suspension considerably longer compared with polyurethane thickeners [bib_ref] Carboxymethylcellulose acetate butyrate in water-based automotive paints, Posey-Dowty [/bib_ref]. CMCAB has high compatibility with common plasticizers used in coating applications so clear films were produced compared to conventional cellulose esters [bib_ref] Use of carboxymethyl cellulose acetate butyrate as a precoat or size for..., Obie [/bib_ref].
High solids and water dispersions of CMCAB mixed esters are used as renewable wood adhesives. Aqueous CMCAB size composition improves the holdout of a coated, cellulosic fiber board [bib_ref] Use of carboxymethyl cellulose acetate butyrate as a precoat or size for..., Obie [/bib_ref]. Holdout is the ability of a coating to stay at or near the substrate surface, not to penetrate that substrate. As a result of enhanced holdout the coated surface appears smoother, uniform with increased gloss [bib_ref] Use of carboxymethyl cellulose acetate butyrate as a precoat or size for..., Obie [/bib_ref]. CMCAB in a wood stain formulation provides good adhesion under an overcoat [bib_ref] Low molecular weight carboxyalkyl cellulose esters and their use as low viscosity..., Shelton [/bib_ref].
As a wood adhesive, CMCAB dispersions will require higher solids contents up to 40%, such that the resulting films would create an adhesive layer with mechanical integrity. Adhesive viscosity, film formation, penetration, joint-performance, and fracture toughness are dependent on MW, the percent solids, percent neutralization, and the organic solvent components. CMCAB interacts strongly with wood and provides a potential use as wood adhesive.
Although petroleum glassy polymers have a huge market, there are few studies on glassy polymers from biomass [bib_ref] Transparent polymer blends composed of cellulose acetate propionate and poly(epichlorohydrin), Yamaguchi [/bib_ref]. Amorphous glassy polymers, like CMCAB, may produce strong films. The glass transition temperature (Tg) is the temperature at which amorphous polymers soften and attain mobility like viscous liquid [bib_ref] Characterization of aqueous polymer dispersions, Wiese [/bib_ref]. Polymers of higher Tg values produced frequently brittle films and need plasticizers to improve their properties [bib_ref] Plasticizers, Wadey [/bib_ref]. Usually esters with smaller substituents, such as CA, have higher Tgs, roughly 180 ∘ C, and hence require more plasticizer than mixed esters with larger groups like CMCAB with a Tg between 135 and 141 ∘ C [bib_ref] Zero-order release formulations using a novel cellulose ester, Posey-Dowty [/bib_ref] [bib_ref] Thin films of carbohydrate based surfactants and carboxymethyl cellulose acetate butyrate mixtures:..., Amim [/bib_ref]. CMCAB, with its relatively lower Tg than many cellulose esters, is an easily compressed matrix for pharmaceutical actives. When used in proper amounts, it is possible to produce a CMCAB/drug tablet for a one-time dose to last up to 24 h. CMCAB matrix provides pH controlled release with no need for an enteric coating of the tablet [bib_ref] Zero-order release formulations using a novel cellulose ester, Posey-Dowty [/bib_ref].
CMCAB has good miscibility with various pharmaceutical actives. It forms amorphous matrices with poorly soluble drugs and enhances its water solubility and bioavailability [bib_ref] Zero-order release formulations using a novel cellulose ester, Posey-Dowty [/bib_ref]. CMCAB in drugs formulation provides slow or zeroorder pH controlled release, stable solid blends, and better drug solubility and stabilization [bib_ref] Enhanced dissolution of poorly soluble drugs from solid dispersions in carboxymethyl cellulose..., Li [/bib_ref]. CMCAB was compressed with drugs into tablets to get the pH-responsive drug delivery systems [bib_ref] Zero-order release formulations using a novel cellulose ester, Posey-Dowty [/bib_ref].
Amorphous solid dispersions (ASD) are a smart technique to increase the drugs water solubility for oral bioavailability. Carboxylated cellulose derivatives are very appropriate composition for ASD due to their unique properties of no toxicity and high Tg and they provide strong hydrogen bonding with the drug to create pH controlled drug release. CMCAB is functional ASD to avoid drugs crystallization, while controlling drug release to the pH of the small intestine [bib_ref] Thin films of carbohydrate based surfactants and carboxymethyl cellulose acetate butyrate mixtures:..., Amim [/bib_ref].
Carboxylated cellulose derivatives were evaluated for their ability to form an ASD with ellagic acid (bioactive natural flavonoid) [bib_ref] Stability and solubility enhancement of ellagic acid in cellulose ester solid dispersions, Li [/bib_ref] , quercetin (dietary flavonoid abundant in foods) [bib_ref] Solid dispersion of quercetin in cellulose derivative matrices influences both solubility and..., Li [/bib_ref] , curcumin (hydrophobic polyphenol) [bib_ref] Both solubility and chemical stability of curcumin are enhanced by solid dispersion..., Li [/bib_ref] , and naringenin (bioactive flavonoid) [bib_ref] Enhancement of naringenin solution concentration by solid dispersion in cellulose derivative matrices, Li [/bib_ref] in order to improve their solubility in aqueous solution. These compounds and cellulose esters were readily blended by spray-drying to produce amorphous solid dispersions even at very high concentrations. ASD is an interesting approach to enhance solubility, bioavailability, effectiveness, simplicity, benign nature, and quite slow release from CMCAB dispersions.
Poly(3-hydroxybutyrate) (PHB) gel formation blends with CMCAB were prepared. The gel formation procedure provided new ways to prepare immiscible blends with the advantage of using benign solvents [bib_ref] Preparation of poly (3-hydroxybutyrate)/carboxymethyl cellulose acetate butyrate blends using gel formation, Gomes [/bib_ref].
Nanofibrous CMCAB mats with excellent superhydrophobicity were prepared by electrospinning [bib_ref] Superhydrophobicity of CMCAB fibrous mats produced by electrospinning, Xie [/bib_ref]. Spincoated films were prepared from CMCAB solution in tetrahydrofuran (THF). THF vapor allowed the cellulose esters chains mobility, causing considerable changes in the film morphology. Smooth and homogeneous porous CMCAB films were detected after only 3 min of exposure to THF vapor [bib_ref] Effect of humidity and solvent vapor phase on cellulose esters films, Blachechen [/bib_ref].
CMCAB is expected to have the same or superior benefit provided by carboxymethyl cellulose acetate (CMCA). CMCA, in membranes for reverse osmosis, give an elastic film that was tougher and of better desalination characteristics than the standard cellulose acetate membrane [bib_ref] Reverse osmosis studies on desalination membranes formed from chemically modified cellulose acetate, El-Taraboulsi [/bib_ref]. Blending of CMCA/cellulose acetate (CA) resulted in novel blend membranes with enhanced features such as lower contact angle. Morphology, permeability, hydrophilicity, and antifouling properties of the prepared CA/CMCA blend membranes enhanced considerably by the combination of CMCA and CA [bib_ref] Preparation and characterization of cellulose acetate/carboxymethyl cellulose acetate blend ultrafiltration membranes, Han [/bib_ref].
Acetylated cellulose ether provides advancing properties and bleach activation to the heavy duty detergent composition. The acetylated ether acts as a bleach activator by reacting with a bleaching agent, such as sodium perborate monohydrate, to generate peracetic acid.
Carbohydrate based surfactants are largely used in microemulsions and in drugs solubilization. Thin films of mixtures containing CMCAB and carbohydrate based surfactant, sorbitanmonopalmitate or poly(oxyethylene) sorbitanmonopalmitate, were spin-coated onto silicon wafers [bib_ref] Thin films of carbohydrate based surfactants and carboxymethyl cellulose acetate butyrate mixtures:..., Amim [/bib_ref]. CMCAB films were deposited from ethyl acetate solutions onto silicon wafers or aminoterminated surfaces and used for lectins adsorption [bib_ref] Effect of amino-terminated substrates onto surface properties of cellulose esters and their..., Amim [/bib_ref].
Nanometer fiber composite material was obtained from lithium iron phosphate (LiFePO4, LFP)/CMCAB by electrospinning. Under the protection of inert gas, modified LFP/carbon nanofibers (CNF) were achieved by carbonization in 600 ∘ C. Cellulose materials that were applied to lithium battery by electrospinning can improve battery performance [bib_ref] Electrospun carboxymethyl cellulose acetate butyrate (CMCAB) nanofiber for high rate lithium-ion battery, Qiu [/bib_ref]. CMCAB had a structure of multicarbon functional groups, which provide abundant carbon resources and attain a dense conductive carbon structure by modification. This property established the CMCAB application in the electrochemistry [bib_ref] Electrospun carboxymethyl cellulose acetate butyrate (CMCAB) nanofiber for high rate lithium-ion battery, Qiu [/bib_ref].
Conflict of InterestsThe authors declare that there is no conflict of interests regarding the publication of this paper. |
Genomic Characterization of Two Novel RCA Phages Reveals New Insights into the Diversity and Evolution of Marine Viruses
Viruses are the most abundant living entities in marine ecosystems, playing critical roles in altering the structure and function of microbial communities and driving ocean biogeochemistry. Phages that infect Roseobacter clade-affiliated (RCA) cluster strains are an important component of marine viral communities. Here, we characterize the genome sequences of two new RCA phages, CRP-9 and CRP-13, which infect RCA strain FZCC0023. Genomic analysis reveals that CRP-9 and CRP-13 represent a novel evolutionary lineage of marine phages. They both have a DNA replication module most similar to those in Cobavirus group phages. In contrast, their morphogenesis and packaging modules are distinct from those in cobaviruses but homologous to those in HMO-2011-type phages. The genomic architecture of CRP-9 and CRP-13 suggests a genomic recombination event between distinct phage groups. Metagenomic data sets were examined for metagenome-assembled viral genomes (MAVGs) with similar recombinant genome architectures. Fifteen CRP-9-type MAVGs were identified from marine viromes. Additionally, 158 MAVGs were identified containing HMO-2011-type morphogenesis and packaging modules with other types of DNA replication genes, providing more evidence that recombination between different phage groups is a major driver of phage evolution. Altogether, this study significantly expands the understanding of diversity and evolution of marine roseophages. Meanwhile, the analysis of these novel RCA phages and MAVGs highlights the critical role of recombination in shaping phage diversity. These phage sequences are valuable resources for inferring the evolutionary connection of distinct phage groups. IMPORTANCE Diversity and evolution of phages that infect the relatively slow-growing but dominant Roseobacter lineages are largely unknown. In this study, RCA phages CRP-9 and CRP-13 have been isolated on a Roseobacter RCA strain and shown to have a unique genomic architecture, which appears to be the result of a recombination event. CRP-9 and CRP-13 have a DNA replication module most similar to those in Cobavirus group phages and morphogenesis and packaging modules most similar to those in HMO-2011type phages. HMO-2011-type morphogenesis and packaging modules are found in combination with distinct types of DNA replication genes, suggesting compatibility with various DNA replication modules. Altogether, this study contributes toward a better understanding of marine viral diversity and evolution.
they have successfully obtained a large amount of genetic information of double-stranded DNA (dsDNA) metagenome-assembled viral genomes (MAVGs). Despite the enormous number of available viral genomes retrieved from metagenomic studies, the majority of MAVGs have limited similarity to phage isolates, and most open reading frames (ORFs) remained unannotated. In addition, the evolutionary history and functional capacity of marine viruses are still poorly understood.
Marine viral communities are mostly composed of phages that infect bacteria. Phages are known to have pervasively mosaic genomes, resulting from complex evolutionary processes during a long period of time. Current knowledge of the genetic diversity of phages suggests that genomic evolution of bacteriophages is mainly driven by extensive lateral gene transfer via genetic recombination and fast mutation. Pairwise genome comparison of over 2,000 phage genomes suggested that phages evolved with low or high gene flux modes, depending on the host, lifestyle, and genomes of phages. The most influential theory regarding lateral gene transfer is the theory of modular evolution, which proposes that evolution proceeds mainly at the level of functionally interchangeable units (modules). Phage evolution should thus be considered acting on functional modules. In the case of marine bacteriophages, successful (survivable) interchange of genomic modules must entail the continued integrity of the phage function, positive function execution, and functional compatibility. The sequences of phage genomes can provide useful information about how the phages have evolved. Therefore, the increasing number of MAVGs is important to understand phage evolution. In recent decades, phages infecting major marine bacteria groups have become a hot spot in marine virology due to the ecological and functional importance of their hosts. Among major bacteria groups, the Roseobacter group dominates coastal waters and the polar oceans. The Roseobacter group has diverse pathways for the metabolism of a variety of organic compounds. Among diverse Roseobacter lineages, RCA, CHAB-I-5, SAG-O19, and NAC11-7 dominate, but they are difficult to isolate and cultivate in the lab. As the largest lineage of the Roseobacter group, the RCA cluster comprises 35% of all planktonic bacteria in the southern ocean and up to 20% to 40% of all roseobacters in temperate and polar oceans. To date, more than 40 roseophages have been reported. However, due to difficulties in host cultivation, researchers have only begun to explore the diversity and ecology of phages infecting dominant Roseobacter lineages. In 2019, Zhang et al. first reported phages infecting RCA strains. These RCA phages are genetically and evolutionarily diverse, belonging to four distinct phage groups. Furthermore, phage groups represented by RCA phages have been shown to exhibit wide distribution patterns and high relative abundance compared to other known roseophage groups.
Here, we report two novel phages that infect Roseobacter RCA strain FZCC0023. Interestingly, the genomes of CRP-9 and CRP-13 seem to contain genomic modules of distinct origins. Genes associated with viral replication are most similar to those in the Cobavirus genomes. Conversely, the morphogenesis and packing-related genes in CRP-9 and CRP-13 are most similar to those in HMO-2011-type phages. Moreover, many similar recombinant MAVGs were identified from marine viromic databases. Our results not only expand the understanding of the roseophages but also have broader implications for understanding the evolution of marine phages.
# Results and discussion
Morphology of CRP-9 and CRP-13. Negative staining electron microscopy reveals that both CRP-9 and CRP-13 have short tails and isometric heads of similar size in diameter , about 70 nm, which are larger than that of the cobavirusesbut similar to that of HMO-2011-type RCA phage CRP-1.
General genomic characteristics of CRP-9 and CRP-13. The genomic size of CRP-9 is 56,157 bp with 73 predicted ORFs and a tRNA-Arg (TCT) gene. CRP-13 is similar to CRP-9 with a genome size of 55,015 bp, encoding 74 ORFs and a tRNA-Arg (TCT) gene. The G1C content of CRP-9 and CRP-13 is 41.6% and 44.2%, respectively, lower than in their host FZCC0023 (53.8%). Overall, CRP-9 and CRP-13 are closely related, sharing approximately half their genes (39 ORFs with 35% to 85% amino acid identity) and have a highly similar genome arrangement . Fifty ORFs in CRP-9 and 49 ORFs in CRP-13 have recognizable homologs in NCBI RefSeq; only 19 ORFs in CRP-9 and 17 ORFs in CRP-13 can be assigned to putative biological functions based on sequence homology. The ORFs with known function in CRP-9 and CRP-13 are mostly involved in the essential functions of the phage life cycle, including DNA replication and metabolism, phage morphogenesis, DNA packaging, and cell lysis .
Modules present in CRP-9 and CRP-13 genomes. The CRP-9 and CRP-13 genomes each consists of a DNA replication module, a DNA metabolism module, a morphogenesis module, and a DNA packaging module . In the DNA replication and metabolism modules, ORFs encoding thymidylate synthase, DNA primase, DNA polymerase, DNA methylase, endonuclease, and ribonucleotide reductase (RNR) were identified. Most genes in this region have similarity with their counterparts in the Cobavirus genomes and are arranged in the same order with cobaviruses , suggesting that CRP-9 and CRP-13 share a similar DNA replication strategy with cobaviruses. Among these genes, two DNA replication genes in CRP-9 and CRP-13, DNA primase and DNA polymerase genes, are most similar to those in cobaviruses (28.0% to 60.2% amino acid identity). DNA polymerase phylogeny reveals that CRP-9 and CRP-13 are clustered with cobaviruses, locating on the same branch with CRP-5, CRP-4, and ICBM2 . Although the DNA replication modules in CRP-9 and CRP-13 show conservation with cobaviruses, another two modules in CRP-9 and CRP-13 display no significant similarity to cobaviruses except for a few homologous ORFs. CRP-9 and CRP-13 genes in the morphogenesis module are most similar to those in HMO-2011-type phage genomes (35% to 99% amino acid identity) . The essential proteins encoded by this module include tail structure protein, tail fiber protein, major capsid protein, and portal protein. The terminase large unit and small subunit (TerL and TerS) in their DNA packaging modules are also most similar to those of HMO-2011-type phages (39% to 93% amino acid identity) . We also observed that some DNA metabolism genes in CRP-9 and CRP-13 also share identity with those in HMO-2011-type phage genomes.
CRP-9 and CRP-13 can be classified roughly at the genus level based on the phage genus definition criterion (namely, sharing .40% of genes). Here, we refer to this group as the CRP-9-type phage group, characterized as harboring Cobavirus-type DNA replication genes and HMO-2011-type morphogenesis and packaging modules.
The HMO-2011-type and Cobavirus groups are two dominant phage groups in the ocean, exhibiting broad distribution patterns and high relative abundance. To date, all cobaviruses were isolated from marine roseobacters. The HMO-2011-type group currently has four members; one infects the SAR116 bacterium IMCC1322, and the other three infect RCA roseobacters. The genomes of CRP-9 and CRP-13 seem to be the combination of HMO-2011-type and cobaviruses genomes. This novel module combination in the CRP-9 and CRP-13 genomes is of great interest. As the modular theory proposed, functional modules can be shuffled between phage genomes by recombination, resulting in new combinations of modules and thus create potentially novel and viable phages. Phage genomic modules can be exchanged with other phages by recombination at specific locations between the modules. It has been suggested that groups of exchanged genes must function together. It is currently difficult to track the evolutionary history of this phage type or to identify the boundaries where recombination has occurred, because these phages share limited similarity with other types of phages at the nucleoide level. Lateral gene transfer between phages mainly occurs during coinfection, that is, when more than one phage infects the same host. Coinfection has been identified in some marine bacteria genomic data. Such recombination may have occurred between different phages that coexisted within the same host or between a resident prophage and an infecting phage. It is noteworthy that all HMO-2011-type phage isolates encode an integrase gene, implying the ability to establish a lysogenic life cycle.
AMGs identified in CRP-9 and CRP-13. Auxiliary metabolic genes (AMGs) are a class of phage-encoded metabolic genes speculated to have functions similar to those of their respective host metabolic genes, regulating host metabolism and improving phage adaptability. AMGs are usually classified by function into class I and class II. AMGs encoding class I proteins are present in the Kyoto Encyclopedia of Genes and Genomes (KEGG) metabolic pathways, while AMGs encoding class II proteins are not present in KEGG metabolic pathways. AMGs are frequently identified in marine roseophage genomes. In roseophage genomes, the most prevalent AMGs are those involved in metabolism and nucleotide synthesis. This study identified several AMGs in CRP-9 and CRP-13, including the phosphate starvation-inducible protein gene (phoH), the nucleoside triphosphate pyrophosphohydrolase gene (mazG), and the glutaredoxin gene. The phoH genes detected in numerous marine phages have previously been described with diverse function, which includes participating in phospholipid metabolism, RNA modification, and fatty acid beta-oxidation. The identification of phoH from CRP-9 may suggest that CRP-9 is more successful during infection when the host is undergoing low-phosphate conditions.
The mazG gene identified in CRP-9 is a highly conserved gene prevalent in bacterial genomes and has also been identified from many marine phage genomes, including all known RCA phages. It has been suggested that MazG protein is important for maintaining the starved host metabolism and therefore benefit the propagation of infecting phages. In Escherichia coli, MazG has been reported to interfere with the FIG 2 Unrooted maximum-likelihood phylogenetic tree of phage DNA polymerases constructed with conserved polymerase domains. Black and white circles indicate nodes with 90% to 100% and 70% to 89% bootstrap support, respectively. Roseophages, cyanophages, and pelagiphages are shown in red, green, and blue, respectively. Roseophages isolated in this study are marked with red asterisks. Scale bar represents amino acid substitutions per site. function of the MazEF toxin-antitoxin system by regulating the cellular level of (p)ppGpp. However, a recent study showed that a cyanophage MazG has no binding or hydrolysis activity against alarmone (p)ppGpp but has high hydrolytic activity toward dGTP and dCTP, and it was spectated to play a role in hydrolyzing high G1C host genome for phage replication. The G1C content of CRP-9 is significantly lower than its host. It is possible that MazG protein in CRP-9 has similar hydrolytic activity to facilitate phage genome replication.
Prevalence of HMO-2011-type morphogenesis and packaging modules in MAVGs. We next sought to investigate the prevalence of the HMO-2011-type morphogenesis and packaging modules in marine MAVGs. A total of 318 MAVGs were retrieved from the Global Ocean Virome 2.0 (GOV 2.0). Genes encoding major capsid, tail protein, portal protein, and terminase in these MAVGs all share high amino acid identity with HMO-2011-type phages, organized in the same order, which suggests close evolutionary relationships. In addition, there are very few insertions and deletions within the HMO-2011-type morphogenesis and packaging modules. Among these 318 MAVGs, 145 contain an HMO-2011-type DNA replication module, thus belonging to the HMO-2011-type phage group. The remaining 173 MAVGs contain DNA replication genes with distinct evolutionary origins (Data Set S1). These results confirm that the HMO-2011-type morphogenesis and packaging modules are present in various phage groups in combination with distinct DNA replication genes. We used the DNA polymerase, capsid, and TerL sequences of these 173 MAVGs for phylogenetic analyses, B and C). The capsid and TerL phylogenies also show that all 173 MAVGs cluster with HMO-2011-type phages.
A conserved DNA polymerase domain (PF00476)-based phylogenetic analysis separated the 173 MAVGs into four major groups (I to IV), distinct from the HMO-2011-type group. Among these 173 MAVGs, 15 are associated with CRP-9 and CRP-13 in the DNA polymerase tree, group I). Genomic comparison shows that all these 15 MAVGs are largely syntenic to CRP-9 and CRP-13 throughout the whole-genome region. Therefore, all 15 MAVGs belong to the CRP-9-type phage group.
Twelve MAVGs are shown to cluster with RCA phage CRP-6 in the DNA polymerase tree. CRP-6 represents a novel phage type, sharing limited homology with other known phage isolates. Genomic comparison shows that these 12 MAVGs and CRP-6 have a very similar DNA replication module but are distinct in morphogenesis and packaging modules. All 12 MAVGs contain a set of DNA replication genes most similar to CRP-6, including DNA primase, DNA polymerase, and a few other associated genes of unknown function. We noticed that the majority, 10 of these 12 MAVGs, contain an integrase gene homologous to that in HMO-2011-type phages (36.5% to 42.7% amino acid identity). In contrast, an integrase gene was not identified in other groups. In HMO-2011-type phages, integrase gene is located upstream of DNA replication genes. In other MAVGs, the integrase genes may have been lost during recombination, or their ancestor did not contain the integrase gene.
The remaining 146 MAVGs, we found, are more closely related to HTVC103P-type pelagiphages in the DNA polymerase tree and can be further divided into two groups, group III and IV). As in CRP-9 and CRP-13, a remarkable feature of the HTVC103P-type group is that these pelagiphages all harbor a set of morphogenesis and packaging genes similar to their counterparts in HMO-2011-type genomes. Genomic comparison reveals remarkable synteny among group III MAVGs and HTVC103P-type pelagiphages. They share 33.3% to 97.7% of their ORFs with HTVC103P-type pelagiphages and have conserved DNA replication, morphogenesis, and packaging modules with HTVC103P-type pelagiphages. Thus, they all fall within the HTVC103P-type phage group and may infect SAR11 bacteria. In addition, the G1C content of group III MAVGs ranged from 30.8% to 34.0%, similar to the G1C content of all known pelagiphages. There is an obvious but more distant relationship between group IV MAVGs and HTVC103P-type pelagiphages. Compared with group III genomes, they share a relatively lower percentage of their ORFs (20.9 to 65.9%) with HTVC103P-type pelagiphages and have a wider range of G1C content (31.9 to 44.4%). Due to the lack of cultured representatives in this group, the hosts of group IV remain to be further explored.
The analysis of these MAVGs reveals that similar HMO-2011-type morphogenesis and packaging modules can exist in combination with various DNA replication modules. In each case, we found the DNA polymerase gene location closely associated with the helicase gene, and each DNA polymerase-helicase combination was exclusive to each group. In phage genomes, DNA polymerase reacts with DNA helicase in the DNA replication process. Clearly, a particular DNA polymerase-helicase combination is vital for a functional DNA replication module. Due to the intimate interactions of DNA replication genes, some phages that undergo DNA replication module replacement survive to infect a host.
The novel module combination of CRP-9, CRP-13, and these MAVGs supports the modular nature of phage genomes. Similar observations were previously reported for several phage genomes. Collectively, these results indicate that genetic recombination plays a major role in the diversification of phage genomes, producing novel combinations of genomic modules. Many phage types have arisen from recombination between two phage ancestors. Phage genomes generally have a higher frequency of recombination to mutation. The recombination rate is also highly variable across the genome regions. The horizontal exchange of sequences among phage genomes is thought to be rampant between phage genomes and randomly distributed over the genomes. Only a minor fraction of sequence exchange, without interrupting module function, can produce functional and competitive recombinants. Based on our observation, it can be speculated that DNA replication modules were transferred between phage genomes, which resulted in new combinations of functional modules and production of new phage types. The identification of these MAVGs suggests that HMO-2011-type morphogenesis and packaging modules are functionally compatible with various DNA replication modules. Existing phages with these types of combinations have survived natural selection. We searched for MAVGs containing the HMO-2011-type DNA replication module and the Cobavirus-like morphogenesis and packaging modules. A total of 8 corresponding MAVGs were successfully retrieved from the GOV 2.0 database. These 8 MAVGs all contain an HMO-2011-type DNA replication module with DNA polymerases having a conserved DnaJ domain (box in. The discovery of this MAVG genotype further supports the idea that recombination plays a critical role in generating novel phage genotypes.
Due to the prominent role of recombination in the diversification of bacteriophage genomes, the assessment of phylogenetic relationships among different phage groups is challenging. Many phage genome sections display different evolutionary histories. In these cases, classification of these phages is challenging, phylogenetic analysis based on a single hallmark gene can be problematic, and network analysis based on gene content may require caution. In some studies, some "hybrid" phage gene modules have a high level of nucleotide identity with other types of phage gene modules through whole-genome nucleotide comparisons, suggesting that the exchange events occurred relatively recently. However, in this study, it was difficult to reconstruct the modular recombination history of these phages; the precise boundaries of these "modules" have remained elusive using current genomic data due to CRP-9-type genomes lacking nucleotide-level similarity with other types of phages.
Conclusion. In this study, a novel type of marine RCA phages has been identified, providing more insights into the diversity and evolution of marine RCA phages. They both possess a Cobavirus-type DNA replication module, suggesting that they use the same DNA replication strategy with cobaviruses. However, their morphogenesis and packaging modules share common ancestry with HMO-2011-type phages. This is the first case of Cobavirus-type DNA replication genes being found outside the Cobavirus group. These two RCA phages and recombinant forms of MAVGs confirm that the recombination of modules among phages is an important evolutionary process that shapes the structure of marine phages. The global marine phage population may harbor a much higher diversity of genomic architectures than has been previously recognized.
# Materials and methods
Host strain and growth condition. Roseobacter RCA strain FZCC0023 was isolated from coastal water of Pingtan Island, China in 2017 by dilution to extinction method. The detailed phylogenetic information on FZCC0023 has been described in an earlier study. FZCC0023 was grown at 23°C in modified natural seawater-based medium amended with 1 mM FeCl 3 , excess vitamins, 1 mM NH 4 Cl, 100 mM KH 2 PO 4 , and mixed carbon source.
Source waters and isolation of CRP-9 and CRP-13. A seawater sample was collected from the coast of Pattaya Beach, Thailand (latitude, 12°569N; longitude, 100°539E) in March 2018. The other seawater sample was collected from the German Bight, North Sea (latitude, 53°569N; longitude, 7°489E) in March 2019, by the RV Heincke during the cruise HE526 (58). Water samples were filtered through a 0.1-mm syringe filter (Pall Gelman Laboratory, USA) and stored at 4°C until use. The phages were isolated according to the procedure described in previous reports. Briefly, 0.1 mm filtered seawater was added to exponentially growing FZCC0023 cultures. The cultures were incubated at 23°C until cell lysis was detected by using a Guava easyCyte flow cytometer (Merck Millipore, Billerica, MA). The presence of phage particles was also confirmed by epifluorescence microscopy. To obtain the pure phage clone, the extinction dilution procedure (59) was repeated three times. The purity of the RCA phages was examined by genome sequencing.
Transmission electron microscopy. CRP-9 and CRP-13 lysates were filtered through 0.1-mm-poresize filters and then concentrated by Amicon Ultra centrifugal filters (30 kDa; Merck Millipore). Next, CRP-9 and CRP-13 were further concentrated by an ultracentrifuge (Beckman Coulter, USA) at 50,000 Â g for 2 h. The concentrated phage samples were adsorbed onto a copper grid in the dark and stained with 2% uranyl acetate for 2 min. A Hitachi transmission electron microscope was used to observe the samples at 80 kV.
Phage DNA extraction. The preparation and concentration of phage lysates were carried out as described by Briefly, 250 ml of host culture was infected with each phage at a phage-tohost ratio of approximately 3:1. After host lysis, the phage lysates were filtered through 0.1-mm-pore-size filters and then further concentrated by Amicon Ultra centrifugal filters (30 kDa; Merck Millipore). Phage DNA was extracted using a formamide treatment, phenol-chloroform extraction method. Subsequently, the phage genomes were sequenced with paired-end sequencing by Mega Genomics (Beijing, China) using the HiSeq 2500 sequencing system (Illumina). The Illumina raw reads obtained were quality controlled, adaptor trimmed, and assembled using CLC Genomic Workbench 11.0.1 software (Qiagen, Hilden, Germany) with default settings.
Genome annotation. The open reading frames (ORFs) of CRP-9 and CRP-13 were predicted by GeneMark. The translated ORFS were analyzed by BLASTP (E value , 1E-3, coverage $ 50%, amino acid identity $ 25%) query against the NCBI nonredundant protein sequences (nr) and NCBI reference proteins (refseq_protein) databases. Proteins were also used to search the PFAM (https://pfam.xfam.org) (62) and HHpred (https://toolkit.tuebingen.mpg.de/tools/hhpred) databases (63) to identify protein families and distant protein homologs, respectively. tRNA gene identification was performed with tRNAscan-SE.
Retrieval of MAVGs. A total of 515,588 MAVGs from the following metagenomic data sets were downloaded for analysis: (i) the Med-DCM fosmid library (5), (ii) the Global Ocean Viromes (7), and (iii) the Global Ocean Virome 2.0 (GOV 2.0). The following search strategy was used to retrieve the MAVGs. First, HMO-2011-type capsid and TerL sequences (including sequences from HMO-2011-type phages, CRP-9, CRP-13, and HTVC103P-type pelagiphages) were searched against genomes of MAVGs using TBLASTN (E value , 1E-3, coverage $ 80%), resulting in 1,242 MAVGs that contain both HMO-2011-type capsid and TerL genes. After this initial search, a search was conducted to identify DNA polymerase genes in the MAVGs. Among the 1,242 MAVGs, 318 MAVGs had a DNA polymerase domain (PF00476) and were retained for further analysis. Next, 145 MAVGs containing an HMO-2011-type DNA polymerase were excluded from further analysis. The remaining 173
MAVGs all contain the HMO-2011-type morphogenesis and packing modules, but their DNA replication modules are distinct from HMO-2011-type phages (Data set S1). All 173 MAVGs were recovered from GOV 2.0, and no MAVGs were obtained from the Med-DCM fosmids or GOV data sets. The above processes were applied to retrieve MAVGs containing HMO-2011-type morphogenesis and packing modules plus other types of DNA replication modules.
We also searched for MAVGs containing the HMO-2011-type replication genes and the Cobavirus-type morphogenesis and packing modules. In brief, Cobavirus-like capsid and TerL sequences (including P12053L, CRP-4, CRP-5, vB_LenP_ICBM1, and vB_LenP_ICBM2) were searched against genomes of MAVGs by BLASTP (E value , 1E-3, coverage $ 80%), resulting in 438 MAVGs. A DNA polymerase domain was identified from 144 of these 438 MAVGs. Next, HMO-2011-type DNA polymerase sequences were searched against 144 MAVGs genomes by TBLASTN (E value , 1E-3, coverage $ 70%). A conserved DnaJ domain containing two repeats of the CXXCXGXG motif was used as the criterion to identify HMO-2011-type DNA polymerases.
Phylogenetic analysis. Multiple alignments were computed using MUSCLE softwareand edited with Gblocks. ModelFinder (67) was used to evaluate optimal amino acid substitution models for alignments. Phylogenetic trees were constructed using the IQTREE v2.0.3 with 1,000 ultrafast bootstrap replicates.
Data availability. The genome sequences of CRP-9 and CRP-13 have been deposited in the GenBank database under accession numbers MW514246 and MW514247.
# Supplemental material
Supplemental material is available online only. SUPPLEMENTAL FILE 1, XLSX file, 0.02 MB. |
The experience of loneliness among young people with depression: a qualitative meta-synthesis of the literature
Background: Young people have a higher prevalence of loneliness than other age groups, and they are also at risk of depression. Quantitative studies describe a bidirectional association between loneliness and depression, but there is limited understanding of how these influence each other. Little is known about the experience of loneliness among young people with depression. Qualitative approaches may help understand the relationship between loneliness and depression among young people, and how to intervene to improve outcomes. We aimed to conduct a meta-synthesis to understand the complex inter-relationship between loneliness and depression among young depressed people by synthesising evidence from a systematic review of qualitative studies.Methods:We conducted a meta-synthesis of qualitative studies capturing experiences of loneliness among young people with depression. We systematically searched six electronic databases for selected search terms, critically appraised eligible studies, and analysed the data from included studies using the approach of thematic synthesis. We used feedback from an inter-disciplinary research workshop to improve reflexivity. Results: Our inclusion criteria identified fourteen studies. Our analysis identified four themes: (1) social withdrawal due to poor mental health, (2) non-disclosure of depression contributing to social distance, (3) the desire to connect, and (4) paradoxes of loneliness and depression. These themes illustrated a range of pathways between depression and loneliness, and a sense of how these might be mutually reinforcing. Our findings suggest that where depressed individuals engage in certain behaviours (withdrawing; not confiding) for a range of reasons, this can lead to feelings of loneliness, an awareness of which worsens their mood, thus perpetuating their depression. Conclusions: Young people with depression experience loneliness as an insurmountable distance between themselves and others. Our findings identified non-disclosure of depression, and the debilitating nature of the depressive symptomatology, as factors perpetuating a vicious cycle of loneliness and depression. They suggest that approaches to tackling the problem might include helping young people communicate about their depression to trusted friends and educating their social networks in how to support them. The wider research literature suggests that cognitive interventions may have a role in shifting maladaptive cognitions about their social world.
# Background
Loneliness is defined as a negative emotional state that arises when there is a perceived discrepancy between desired and actual social relationships. The adverse effects of loneliness on mental, and physical healthare now well established. Most work describing the association between loneliness and mental illness has focussed on depression. Depression is the third leading cause of disease worldwide, based on World Health Organisation (WHO) rankings, and its incidence appears to be increasing internationally [bib_ref] Predictors of depressive symptoms in college students: a systematic review and meta-analysis..., Liu [/bib_ref]. Onset of major depression extends from mid-adolescence to the mid-40s, but almost 40% experience their first episode of depression before age 20 years, with peaks in prevalence in the second and third decades of life. Cross-sectional work shows that people with depression are ten times more likely to feel lonely than the general population [bib_ref] Feelings of loneliness among adults with mental disorder, Meltzer [/bib_ref]. Longitudinal studies demonstrate that loneliness not only increases the risk of becoming depressed [bib_ref] Incidence and outcome of depressive symptoms in nursing home patients in the..., Smalbrugge [/bib_ref] [bib_ref] Secular changes in the relation between social factors and depression: a study..., Sjöberg [/bib_ref] [bib_ref] Loneliness, health, and longevity, Stessman [/bib_ref] , and worsens depressive symptoms amongst those who are already depressed [bib_ref] Associations between loneliness and perceived social support and outcomes of mental health..., Wang [/bib_ref] , but also that loneliness and depression influence each other reciprocally [bib_ref] Loneliness as a specific risk factor for depressive symptoms: cross-sectional and longitudinal..., Cacioppo [/bib_ref]. This means that people who are lonely are more likely to be become depressed, but also that their depression reinforces their loneliness. The mechanisms underlying this complex inter-relationship between loneliness and depression are unclear and need further investigation, particularly in young people.
Population-based surveys describe a U-shaped age distribution of loneliness, with high rates of loneliness among young people and among the elderly [bib_ref] Where are all the lonely people?" a population-based study of high-risk groups..., Lasgaard [/bib_ref]. However, the majority of epidemiological work on the health impacts of loneliness have been conducted in older age groups . Extrapolating the findings of studies in older age groups to younger people is problematic given that experiences of loneliness vary in different demographic and cultural groups [bib_ref] Loneliness of the marginalized, Rokach [/bib_ref]. The social context of loneliness is also very different in young people to later stages of life. Additionally, the experience of depression is also likely to vary by age, with symptoms of irritability and interpersonal difficulties being particularly prominent among adolescents [bib_ref] Characteristics of adolescent depression, Crowe [/bib_ref].
The few studies that have been done in young people suggest that social isolation in childhood predicts loneliness in young adulthood [bib_ref] Lonely young adults in modern Britain: findings from an epidemiological cohort study, Matthews [/bib_ref] and that chronic peerrelated loneliness in childhood predicts adolescent depression [bib_ref] Loneliness in children and adolescents with chronic physical conditions: a meta-analysis, Maes [/bib_ref] [bib_ref] Childhood loneliness as a predictor of adolescent depressive symptoms: an 8-year longitudinal..., Qualter [/bib_ref] [bib_ref] Unraveling the role of loneliness in depression: the relationship between daily life..., Van Winkel [/bib_ref]. Given the evidence from studies of older age groups, loneliness is likely to be a factor affecting quality of life and prognosis among young people with depression. It is also likely to compound the barriers described by young people in accessing formal or informal help, namely the stigma of mental illness and a reluctance to talk about feelings or emotions [bib_ref] Identifying barriers to mental health help-seeking among young adults in the UK:..., Salaheddin [/bib_ref]. Adolescence and young adulthood is the greatest risk period for the emergence of depression and also one in which loneliness might be most stigmatising given strong social pressure to appear connected [bib_ref] Advancing our understanding of loneliness and mental health problems in young people, Pitman [/bib_ref]. In view of the high prevalence of loneliness amongst young people, and the lack of research focussed on this age group, it is important to gain a better understanding of the experience of loneliness among young people with depression, as well as its causes and consequences, to tailor the design of acceptable age-appropriate treatments [bib_ref] Associations between loneliness and perceived social support and outcomes of mental health..., Wang [/bib_ref].
Loneliness is a subjective construct related to the concepts of social isolation [bib_ref] Social isolation in mental health: a conceptual and methodological review, Wang [/bib_ref] , alienation [bib_ref] Agency of depressed adolescents: embodiment and social representations, De Mol [/bib_ref] , social connectedness [bib_ref] Conceptual framework for social connectedness in mental disorders: systematic review and narrative..., Hare-Duke [/bib_ref] , lack of belonging [bib_ref] The need to belong: desire for interpersonal attachments as a fundamental human..., Baumeister [/bib_ref] and social capital [bib_ref] Social isolation in mental health: a conceptual and methodological review, Wang [/bib_ref]. Loneliness is distinct from social isolation, which is an objective measure of the absence of relationships with other people. Loneliness is also distinct from solitude in that loneliness is an unpleasant experience, whereas solitude implies a desire to be alone and is not necessarily a negative experience. Quantitative work shows that loneliness and social isolation are moderately correlated and both are associated with depression [bib_ref] Social isolation, loneliness and depression in young adulthood: a behavioural genetic analysis, Matthews [/bib_ref] [bib_ref] Social isolation, loneliness and their relationships with depressive symptoms: A population-based study, Ge [/bib_ref]. Behavioural genetic analysis finds that young people who are lonely are often depressed, partly because the same genes influence loneliness and depression [bib_ref] Social isolation, loneliness and depression in young adulthood: a behavioural genetic analysis, Matthews [/bib_ref]. Environmental factors are also important; lonely young adults are more likely to have been bullied and socially isolated as children [bib_ref] Lonely young adults in modern Britain: findings from an epidemiological cohort study, Matthews [/bib_ref]. The distinction between loneliness and social isolation is important because socially isolated young adults do not necessarily feel lonely [bib_ref] Social isolation, loneliness and depression in young adulthood: a behavioural genetic analysis, Matthews [/bib_ref] and young adults who feel lonely do not necessarily spend less time with others in comparison to their less lonely peers.
Available evidence suggests that different age groups experience loneliness differently. Comparison of the social networks of young and middle-aged adults show that young adults reported twice as many days feeling lonely and isolated than late middle-age adults, despite having larger networks [bib_ref] Loneliness and social isolation among young and late middle-age adults: associations with..., Child [/bib_ref]. Interview data from an English community sample show that children and young people aged 10 to 24 years describe loneliness as a sense of exclusion, disconnection from others and unhappiness with relationships. Children as young as 5 years understand a concept of loneliness, a sadness associated with this, and how it motivates them to make contact with others [bib_ref] Childhood loneliness as a predictor of adolescent depressive symptoms: an 8-year longitudinal..., Qualter [/bib_ref].
By understanding how loneliness and depression influence each other in young people, there is potential for improving depressive symptoms and depressive outcomes through well-developed and appropriately targeted interventions focused on loneliness. The aim of this meta-synthesis was therefore to summarise qualitative research describing the experience of loneliness and depression among young people, to provide insights into the relationships and pathways between them.
# Methods
## Design
Meta-synthesis is a research method that uses rigorous qualitative methods to synthesize existing qualitative studies, with the aim of constructing greater meaning through an overarching interpretation based on the qualitative studies included [bib_ref] Methods for the synthesis of qualitative research: a critical review, Barnett-Page [/bib_ref] [bib_ref] Understanding qualitative Metasynthesis: issues and opportunities in early childhood intervention research, Erwin [/bib_ref] [bib_ref] Metasynthesis: An Original Method to Synthesize Qualitative Literature in Psychiatry, Lachal [/bib_ref] [bib_ref] Qualitative metasynthesis: issues and techniques, Sandelowski [/bib_ref]. Thematic synthesis is influenced by the meta-ethnography process and grounded theory [bib_ref] Metasynthesis: An Original Method to Synthesize Qualitative Literature in Psychiatry, Lachal [/bib_ref] and involves conceptual coding of data to construct an encompassing model providing insights to the phenomenon studied. The approach identifies patterns across qualitative data and aims to enrich the understanding of a topic, creating new theoretical insights as well as serving as a tool to develop suitable interventions [bib_ref] Conducting a meta-ethnography of qualitative literature: lessons learnt, Atkins [/bib_ref]. It entails an iterative cyclical process comparing and contrasting themes between different studies and attempting to encompass the data using a set of themes that are relevant within each study, constructed as a hierarchical tree structure [bib_ref] Metasynthesis: An Original Method to Synthesize Qualitative Literature in Psychiatry, Lachal [/bib_ref].
For this study we applied an accepted six step method for conducting a meta-synthesis [bib_ref] Metasynthesis: An Original Method to Synthesize Qualitative Literature in Psychiatry, Lachal [/bib_ref] , consisting of: 1) defining the research question and selection criteria, 2) using those criteria to select studies, 3) undertaking a quality assessment, 4) extracting and presenting formal data, 5) conducting data analysis and 6) reporting the synthesis.
## Protocol and search strategy
Before commencing, we registered our meta-synthesis protocol with PROSPERO, the international prospective register for systematic reviews [bib_ref] International Prospective Register of Systematic Reviews, Prospero [/bib_ref]. We conducted our search using six electronic databases (MEDLINE, PsycINFO, CINAHL, Scopus, ProQuest and Web of Science) from database inception to 21 March 2019. These databases were chosen to capture studies from a range of research disciplines, including medicine, psychology, sociology and anthropology. We developed search terms as a team (see Appendix 1), including terms capturing depression and mental illhealth, loneliness and social isolation, and qualitative research. Given the conceptual overlap between loneliness and other constructs such as perceived social support, our search terms also included several other words capturing these concepts [bib_ref] Associations between loneliness and perceived social support and outcomes of mental health..., Wang [/bib_ref]. We also conducted a search of the Ethos British Library database to find any relevant PhD dissertations and hand searched the reference lists of any eligible studies to reduce the chance of missing important studies.
## Selection: inclusion & exclusion criteria
We screened titles and abstracts of identified studies for eligibility, followed by full text review where indicated, using the following inclusion criteria to identify studies that: a) used a qualitative research design such as semistructured interviews or focus groups. b) sampled adolescents and/or young adults aged 11 to 30 years with a depressive disorder. We chose the age range 11 to 30 years to cover WHO definitions of adolescents (aged 10-19 years), youth (aged 15-24 years), teenagers (aged 15-19 years), and young adults (aged 20-24 years), with a wider margin at the upper limit in order to ensure that we did not exclude studies including young adults as a proportion of those sampled or studies where young people reflected back on their recent adolescence. c) explored how young people with depression experience loneliness, both in relation to a current depressive episode and/or in reflecting back on past episodes d) included participants with self-reported depression or depression diagnosed by a health professional, regardless of severity of depression or treatment received. We included studies that involved participants with depression, with or without a comorbid anxiety or personality disorder. We also included studies in which the depressive episode was in the context of a diagnosis of bipolar disorder.
Studies were excluded if they: a) sampled participants above the age of 30 only, or used a mixed sample of age groups above and below 30 b) sampled participants without a history of depression. c) sampled participants with: a co-morbid chronic physical disability (e.g. rheumatoid arthritis); any co-morbid mental illness other than an anxiety disorder or personality disorder; or a co-morbid neurocognitive disorder (e.g. Alzheimer's disease). This was to avoid capturing the experience of depression in the context of co-morbid conditions beyond common mental disorders. d) presented data mentioning loneliness fleetingly or not at all. For example, a study with data on one participant saying: 'I feel lonely because I'm depressed' would not be deemed sufficient in detail to convey anything meaningful about the experience of loneliness. e) presented data describing solely the objective presence or absence of social support, rather than subjective feelings about perceived social support, social isolation, or social network size.
f) used a quantitative research design. g) were not written in English or Dutch.
## Data screening and extraction
One researcher (LA) conducted the search, removed duplicates, and screened the titles and abstracts of all studies for relevance, before assessing the full text of identified studies for eligibility. Three researchers (MB, EP and AP) were each randomly assigned 10% of these studies for full text assessment to ascertain agreement over inclusion/exclusion, meeting regularly as a group to discuss eligibility.
## Quality appraisal
One researcher (LA) appraised all eligible studies for quality using the Critical Appraisal Skills Programme (CASP), a 10-item quality assessment tool for qualitative research, discussing this with the wider team. Studies were appraised on these items grouped under three categories; validity (clarity of research aims, appropriateness of qualitative methodology, research design, recruitment strategy, and data collection, appropriate consideration of researcher reflexivity), results (ethical considerations, appropriateness of data analysis, clarity of findings stated), and utility (the value of the research). Study characteristics and appraisal criteria were summarised in a proforma [fig_ref] Table 1 Table: presenting characteristics and quality appraisal of included studies Table presenting characteristics and... [/fig_ref]. We chose not to exclude studies based on our assessment of low quality. Instead, our synthesis of findings took into account our CASP-based judgements on the quality of included studies, as suggested in methodological guidance [bib_ref] Metasynthesis: An Original Method to Synthesize Qualitative Literature in Psychiatry, Lachal [/bib_ref] [bib_ref] Conducting a meta-ethnography of qualitative literature: lessons learnt, Atkins [/bib_ref] , and can therefore be interpreted in this context.
# Data analysis
For each included study, one researcher (LA) identified any text relating to loneliness in the results section (whether quotes or interpretation) and imported relevant passages into a qualitative data analysis software packageto facilitate the process of thematic synthesis. Three researchers (AP, MB, EP) independently assessed a set of studies each to identify which passages to import and compare judgements on which data to include or exclude.
Having established our final database of extracted qualitative data, one researcher (LA) then coded the full dataset, and three researchers (AP, MB, EP) independently coded data from two randomly-allocated studies each. All four researchers then compared their coding to develop an initial coding framework. This was then refined through an iterative process, to develop a taxonomy of analytical themes.
## External validity
To improve external validity and reduce researcher subjectivity, we presented the findings of this draft thematic framework at an inter-disciplinary research workshop held in London on 26th June 2019. This formed part of the research activities of the United Kingdom Research and Innovation (UKRI) Loneliness and Social Isolation in Mental Health Research Network (see Acknowledgements). The 58 participants included health and social care practitioners, university and voluntary sector researchers, policy makers, lived experience researchers, and mental health service users. Following an oral presentation of findings, one researcher (LA) led two independent 45-min sessions with 8 attendees in each group to discuss the coding framework in more detail. The comments made were used to revise the coding framework and improve reflexivity.
## Reflexivity
Our multidisciplinary research team reduced the dominance of one perspective. This was important, as a metasynthesis is an overarching interpretation from the joint analysis of primary studies [bib_ref] Methods for the synthesis of qualitative research: a critical review, Barnett-Page [/bib_ref] , with a high risk of subjectivity and personal bias. LA is a social scientist and medical student with an interest in the links between mental health and loneliness, AP and SJ are psychiatrists and academics with an interest in sociology and social psychology, EP has a research background in experimental psychology and biological anthropology, with experience of having worked in the voluntary sector, while MB is a mental health occupational therapist and academic. This team approach, and the use of a multidisciplinary research workshop to discuss findings, reduced the focus on loneliness from a medical perspective by including insights from multiple disciplines.
# Results
## Description of included studies
Our search identified 9188 studies, which was reduced to 6540 after deleting 2648 duplicates (see [fig_ref] Figure 1: Flow diagram for included studies [/fig_ref]. Following screening of titles/abstracts we excluded 6351 studies for irrelevance. Following full text review of the remaining 188 studies we excluded 177 based on our exclusion criteria, included eleven studies and added three more based on hand searching the references from the selected studies, identifying 14 eligible studies [bib_ref] Agency of depressed adolescents: embodiment and social representations, De Mol [/bib_ref] [bib_ref] How African American adolescents manage depression: being with others, Al-Khattab [/bib_ref] [bib_ref] Concerns and hopes among adolescents attending adolescent psychiatric outpatient clinics, Anttila [/bib_ref] [bib_ref] HS adolescent depression: a Metasynthesis, Dundon [/bib_ref] [bib_ref] HS the experience of major depression: Adolescents' perspectives, Farmer [/bib_ref] [bib_ref] What's love got to do with it? The relational nature of depressive..., Granek [/bib_ref] [bib_ref] HS a qualitative exploration of depression in emerging adulthood: disorder, development, and..., Kuwabara [/bib_ref] [bib_ref] Depression in adolescence: from qualitative research to measurement, Lachal [/bib_ref] [bib_ref] The experience of young people with depression: a qualitative study, Mccann [/bib_ref] [bib_ref] Adolescents coping with mood disorder: a grounded theory study, Meadus [/bib_ref] [bib_ref] Beyond a diagnosis: the experience of depression among clinically-referred adolescents, Midgley [/bib_ref] [bib_ref] The experience of depression: a qualitative study of adolescents with depression entering..., Weitkamp [/bib_ref] [bib_ref] Living in the shadow of fear: adolescents' lived experience of depression, Woodgate [/bib_ref] , which we included in this meta-synthesis. We achieved 100% agreement on study eligibility between four authors. Characteristics of each study are shown in [fig_ref] Table 1 Table: presenting characteristics and quality appraisal of included studies Table presenting characteristics and... [/fig_ref] , including an assessment of study quality using CASP criteria.
The total number of participants was 388, with sample sizes in each study ranging from 5 to 107. Participants' ages ranged from 11 to 30 years, and roughly three. One study was a meta-synthesis of six qualitative studies [bib_ref] HS adolescent depression: a Metasynthesis, Dundon [/bib_ref] , which included one identified in our own search [bib_ref] HS the experience of major depression: Adolescents' perspectives, Farmer [/bib_ref]. We decided to include this meta-synthesis as a unified whole rather than disaggregating its included studies because a number of those studies were unpublished theses that were unavailable to us. All studies involved individuals with a history of depression, whilst two studies included participants with depressive episodes in the context of bipolar disorder. In those two studies an unknown proportion had bipolar in one, whilst two out of nine participants had bipolar in the other [bib_ref] Adolescents coping with mood disorder: a grounded theory study, Meadus [/bib_ref]. Most studies (n = 12) collected interview data, while two studies used free text from written self-reports. Included studies used a range of qualitative analytic methods: thematic analysis, interpretative phenomenological analysis, grounded theory, discourse analysis, framework analysis, hermeneutic phenomenology, content analysis, and comparative method analysis.
## Thematic synthesis
Our thematic synthesis of 14 eligible studies identified four analytic themes: (1) social withdrawal due to poor mental health, (2) non-disclosure of depression contributing to social distance (with four sub-themes), (3) the desire to connect, and (4) paradoxes of loneliness and depression. Quotes given in italics denote primary data.
## Theme 1: social withdrawal due to poor mental health
A key theme we identified related to the debilitating nature of depressive symptoms, which made it very hard for some young people to engage with others. Nearly all studies (n = 13) described the experience of depression as causing those individuals to withdraw from others, relating this to difficulties in being around others due to low motivation and low energy.
"There would be days that I just couldn't get out of bed. I didn't want to face people. I didn't want to look at anybody, I just wanted to stay there and I guess just sulk by myself, and I just didn't have any energy." (Female in her 10s, USA sample) [bib_ref] HS adolescent depression: a Metasynthesis, Dundon [/bib_ref]. Some individuals described feeling better when isolating themselves from peers, because being around others was so emotionally draining. "I come home it's just kind of like a relief", explained Lana (teenager, UK sample), who had been bullied at school for reasons unspecified [bib_ref] Beyond a diagnosis: the experience of depression among clinically-referred adolescents, Midgley [/bib_ref]. Some participants avoided others by spending time in their rooms or going for walks alone. One female participant in her teens from the USA explained, "I just kind of wanted to be by myself." [bib_ref] How African American adolescents manage depression: being with others, Al-Khattab [/bib_ref] Participants described having stopped participating in activities they had previously enjoyed or not feeling able to fully engage in such activities. A female in her teens from the USA, who had taken an active role in the performing arts since the age of 2 years, explained.
"I was in show choir and throughout that year I just didn't really enjoy it. I was fine with standing in the back, which really wasn't like me. My wanting to be in the back just wasn't normal." [bib_ref] How African American adolescents manage depression: being with others, Al-Khattab [/bib_ref] Low self-esteem seemed to affect some young people sampled, who felt that their depression had worsened their insecurities, leading them to withdraw socially. The depression had apparently eroded their belief that anyone could find them likeable, resulting in them withdrawing to avoid other people.
"I become even more withdrawn than I normally am, and it's based on the insecurity, and it came up the unlikeability thing again, that I'm not likable inherently so what's the use of pretending that I am because eventually they are going to find out." (Sarah, teenager; Canadian sample) [bib_ref] What's love got to do with it? The relational nature of depressive..., Granek [/bib_ref].
Participants also spoke of an inability to feel affection from others: "When you're depressed you feel like you don't have anybody." (Tina, teenager, USA sample) [bib_ref] HS the experience of major depression: Adolescents' perspectives, Farmer [/bib_ref]. The syndrome of depression set young people apart from their peers and made them feel different. This change was noted by others, even if they did not necessarily recognise it as depression, and this could lead to others' withdrawal. The sense of rejection was apparent in young people who coped by isolating themselves, thus compounding their sense of differences between them and others.
People just started drifting away, like they were asking, "What's wrong with you?" I wanted to ask them, "Why don't you talk to me anymore?" I felt they were saying "You're different now!" I just began to hide away a lot and I would say, "I just want to be alone". (female teenager, USA sample).
## Theme 2: non-disclosure of depression and social distance
The second theme, emerging from 12 studies, was more explicitly related to feelings of loneliness. As young people dealing with depression were hesitant about disclosing their depressed feelings to people in their social networks, they avoided being open about their true selves. This sense of otherness through concealment enhanced participants' feelings of loneliness. Some individuals described being very aware of putting up a façade and of making extensive efforts to maintain this front to avoid talking about their mental health issues.
"I would put on a smile for my parents and my siblings. Whenever somebody would leave and I knew I was going to be alone, they would ask me, "Are you going to be alright?" And I would say "Yes, of course," because I didn't want them to know what I was dealing with. But, it was a living hell. I put up a really good façade for them, like all cheery and happy, nothing's wrong." (Female in her 20s; USA sample) [bib_ref] How African American adolescents manage depression: being with others, Al-Khattab [/bib_ref].
A range of reasons were given for the non-disclosure of depressed mood, summarised in the four sub-themes below.
## Subtheme 2.1 fear of being judged
Young people in the included studies commonly expressed fear of being judged negatively if they identified themselves as suffering from depression, or of being perceived as unbearable or embarrassing if they vented their feelings. The negative consequences they feared included social exclusion and isolation, as borne out by their experiences:
"I cannot talk about my sadness, in fact, I don't dare to talk about it, because then you are considered as a weak person. I see that some people feel pity for me, but they don't talk to me, they prefer to run away because they are afraid and do not know how to react to someone who is sad." (Female, teenager, Belgian sample) [bib_ref] Agency of depressed adolescents: embodiment and social representations, De Mol [/bib_ref].
"If I could talk to them [friends] I would, but I just didn't feel like I could talk to them. They would keep on going, 'You're weird' or something." (Sandra, teenager, USA sample) [bib_ref] HS the experience of major depression: Adolescents' perspectives, Farmer [/bib_ref].
## Subtheme 2.2 preserving friendships
Another reason for not disclosing their depression was that the young people sampled clearly valued their friendships and wanted to preserve existing networks. They feared losing these connections if they shared their feelings of depression. There was also a fear of burdening others, in that by not disclosing their depressive thoughts they hoped to minimise the negative impact of their depression on others. Many adolescents had experienced negative changes or the ending of friendships as a consequence of mental health problems and this reinforced their reluctance to reveal their feelings to friends.
"I'm afraid that friends and significant others can't see me the same way as before or something might change between us if I told them all my troubles. I don't want to bother anybody with my worries." (Unknown gender, 15-17 years old; Finnish sample) [bib_ref] Concerns and hopes among adolescents attending adolescent psychiatric outpatient clinics, Anttila [/bib_ref].
## Subtheme 2.3 difficulty explaining oneself
Beyond deliberate efforts to avoid talking about their feelings, young people also found it hard to explain why they felt depressed. Pressure to explain themselves arose from members of their peer group, who struggled to comprehend their experiences, expressing this through intolerance. Their own inability to formulate or articulate an explanation frustrated young people with depression and had the effect of widening the gulf between them and others.
"When you feel bad, you need to have an external explanation for why you have these feelings, because the fact that you feel bad must be caused by something. Participants stated that they often received the question: 'Why are you feeling so bad?' Adolescents shared that they cannot give a constructive answer because they do not know why they have these feelings. They could not give explanations because there were no specific causes for them. Due to the inability to provide a real explanation regarding the causes, their feelings and depression are not recognized by others." [bib_ref] Agency of depressed adolescents: embodiment and social representations, De Mol [/bib_ref] Subtheme 2.4 perceived futility of explaining oneself Experiencing depression engendered a realisation of being different from one's peers. This gave rise to the belief that others would not understand one's situation and that there was therefore no point in discussing it. Young people with depression indicated that they feared others were likely to trivialise, dismiss or ignore their depressive symptoms. Again, their previous negative experiences of others failing to understand them taught some young people not to disclose their feelings. The lack of any incentive or opportunity to confide and feel understood made young people feel very lonely.
"Having others reach out, however, was not always beneficial. Some participants, especially females, did not feel comfortable opening up to those who reached out to them. These participants did not believe the other person would understand what they were going through, believed their problems were 'no one else's business' or doubted the person's motives for reaching out." [bib_ref] How African American adolescents manage depression: being with others, Al-Khattab [/bib_ref] "Despite the fact that all the individuals in this sample acknowledged social support as an important part of their daily lives, the belief that others cannot understand their experiences often caused individuals to feel alone." [bib_ref] HS a qualitative exploration of depression in emerging adulthood: disorder, development, and..., Kuwabara [/bib_ref] Theme 3: the desire to connect Despite young people reporting a tendency to disengage from certain social interactions, they still expressed a desire for connection and a desire to feel 'normal'.
"At the same time, the adolescents hoped to have more friends and to be included in their peer group. In addition, they wished to have a good time with the friends and to have somebody to talk to about their problems and feelings." [bib_ref] Concerns and hopes among adolescents attending adolescent psychiatric outpatient clinics, Anttila [/bib_ref] "Most individuals have a strong need to connect and have positive relationships with others especially middle school students."In this sub-theme we identified a conflict with the experiences described in sub-themes 2.3 and 2.4 above, in that although some individuals expressed a wish to talk about their issues, they also experienced difficulties in doing so. Such barriers included a fear of the consequences, particularly the threat of rejection (and perhaps stigmatisation) from peers. To address this, some preferred to share their problems with people who they knew had faced the same mental health issues in preference to their wider peer group, amongst whom it was not always clear who had experienced depression themselves.
Shadow clearly had the wish to disclose to someone, which he expressed in a wish for some kind of group therapy to meet people where he could actually speak about his problems: "And maybe, that you can talk about it in a group that you can say: "I am [Shadow], I have this and that problem. What do you think, what is your impression, what is your problem?" . .. Because I can't possibly walk into my classroom and say: "you know what happened to me?" Well, I could, but. .." (Male, teenager; German sample) [bib_ref] The experience of depression: a qualitative study of adolescents with depression entering..., Weitkamp [/bib_ref].
## Theme 4: paradoxes of loneliness and depression
This theme described a number of paradoxes or vicious cycles that were apparent in various forms across a number of studies. Whilst some young people talked about a need or a tendency to withdraw socially, this came with an awareness that such avoidance could create or worsen feelings of loneliness.
"During their depressive experiences, participants felt a distinct separateness from others and often chose solitude over being with others even when feeling lonely" [bib_ref] HS the experience of major depression: Adolescents' perspectives, Farmer [/bib_ref].
"Being around people was, was always a bad thing for me. I constantly felt the need to be alone. .. and I always felt like interacting with other people was difficult for me. .. Ya, that was confusing because I felt lonely but I didn't feel like being around anyone at the same time" (Jeff, in his 20s, Canadian sample) [bib_ref] What's love got to do with it? The relational nature of depressive..., Granek [/bib_ref].
Some young people described their friends showing a form of understanding by not asking too many questions, but then feeling cut off because of an apparent absence of overt concern.
Sometimes when some of my friends are … .. ok with ignoring me, with not asking about it, I feel like kind of I know it's ridiculous, but unloved. (Female, teenager, UK sample) [bib_ref] Beyond a diagnosis: the experience of depression among clinically-referred adolescents, Midgley [/bib_ref].
Another trap that some young people described was a vicious cycle of loneliness and depression, was the suggestion that the manner in which they processed feelings about loneliness reinforced their depression.
"They were unable to initiate or sustain relationships because of feelings of severe discomfort around people. They described a cycle of feeling lonely, often as a result of their breakups, and then feeling depressed about the loneliness, causing a self-fulfilling prophecy by further alienating and self-isolating themselves from others." [bib_ref] What's love got to do with it? The relational nature of depressive..., Granek [/bib_ref] A fear of stigma was also mentioned as a reason for withdrawing from others, but this came at the price of increasing loneliness. Sometimes a yearning to connect with others coexisted with an inability to be with them. However, where they withdrew from others, young people were prevented from getting support from others, thus increasing their sense of alienation from friends.
"While some disclosed their depression to friends, others withdrew, fearful of the perceived stigma and loss of status from being labelled as having mental illness …. However, retreating from others contributed to their loneliness and isolation." [bib_ref] The experience of young people with depression: a qualitative study, Mccann [/bib_ref] The difficult choice that some young people faced, was between withdrawing socially to hide their depression and then feeling excluded, or remaining superficially socially engaged but living behind a façade in not disclosing their depression. In the latter case, the strain of concealing their low mood could create a sense of greater alienation from their peers.
"This process of social isolation was characterized by ambivalent feelings. Participants explained that on the one hand they feel the necessity to share their emotions with others, but on the other hand they felt it was impossible to do this. Consequently, they felt caught up in a vicious circle which made them feel alienated from themselves and from of their social world." [bib_ref] Agency of depressed adolescents: embodiment and social representations, De Mol [/bib_ref].
No suggestions were made by participants as to how to break such vicious cycles, but a note of optimism was sounded in relation to recovery from depression. During an episode of depression young people characteristically described the experience of yearning for a connection with others, feelings of being very different from others, and a perception that their problems were incomprehensible to their peers; all of which meant that when unwell, gaining a sense of connectedness was out of their reach. However, on recovering from an episode of depression, those who reflected back on those unwell periods had better insight into such traps and were able to see how a lifting of their symptoms removed many of the barriers to connecting with others.
# Discussion
## Main findings
This meta-synthesis of fourteen qualitative studies describing the experience of loneliness among young people with depression identified four main themes, conveying the social consequences of both loneliness and depression, and illustrating a range of pathways between them and a sense of how these could be mutually reinforcing. Young people described the symptoms of their depression as leading to social withdrawal. Although they did not name them as such, the symptoms they described match those featuring in diagnostic criteria for depression (low energy, anhedonia, avolition, low self-esteem). The first theme suggests that although debilitating, these symptoms could lead to social isolation but not necessarily feelings of loneliness. Indeed, some young people gained a sense of relief when not having to engage in their usual social roles although this pattern of social withdrawal risked increasing the probability of feeling lonely over the longer term. Our second theme was more explicitly related to loneliness, capturing how non-disclosure of depression made it hard for young people to feel connected because it distanced them from others. Contrary to the experiences of those who preferred to avoid others even at the cost of feeling lonely, data coded under our third theme illustrated a longing to be among others. Our fourth theme described a set of paradoxes faced by young depressed people, including that of yearning for connection, or believing it might help with low mood, yet being unable to tolerate being around others. This theme also described a selfperpetuating cycle of loneliness and depression, providing insights into the cognitive processes underlying this bidirectional relationship. Our data suggest that where depressed individuals engage in certain behaviours (withdrawing and not confiding) this can lead to feelings of loneliness, an awareness of which worsens their mood, thus perpetuating their depression. Again, this process of self-isolation and social alienation builds a sense of one pathway in an apparent bidirectional relationship.
We did not observe any gender or age patterning of themes, but it was hard to rule these out without access to the primary data. Ranging in age from 11 to 30 years, many older subjects reflected back on their previous experiences, but it was not always clear what age they were referring to. We also did not identify any mention of difficulties accessing social contacts or in meeting people, in contrast to the issue of sparse social networks being described as a problem for older age groups [bib_ref] Loneliness and depression in the elderly: the role of social network, Domènech-Abella [/bib_ref] [bib_ref] Loneliness, social networks, and health: a cross-sectional study in three countries, Rico-Uribe [/bib_ref]. In our data psychological aspects of depression were more prominent than social aspects of depression. Although some participants mentioned social anhedonia there was little mention of impaired social communication (for example impaired emotion recognition) or of impaired social perception (for example reduced empathy), as might otherwise be seen in depression [bib_ref] Social functioning in major depressive disorder, Kupferberg [/bib_ref]. However, participants may have lacked an awareness of this and we lacked collateral accounts to triangulate their own. We also did not identify overt descriptions of perceived stigma, although this was implied in an avoidance of disclosing depression to peers. Our themes built up a theoretical framework focussed on a dominant pathway of depression leading to loneliness (and loneliness worsening depression) than on loneliness having predated (and/ or contributed) to the onset of depression. However, the nature of our study meant that it was impossible to probe individuals' thoughts about predisposing factors and this warrants further interview work.
## Findings in the context of other studies
We believe this to be the first meta-synthesis of qualitative studies describing the experience of loneliness in young people with depression. Quantitative testing of causal models of adolescent depression suggest that loneliness and low self-esteem increase the probability of depression, with low self-esteem having an indirect effect on depression via loneliness [bib_ref] A causal model of adolescent depression, Brage [/bib_ref]. Our data implied that loneliness was not an antecedent of depression so much as a consequence of it, although none of the included studies probed this. Whilst we did not identify any gender patterning, US survey data suggest that loneliness was one of the most common self-reported features of depression among female adolescents with depression but far less apparent among depressed boys [bib_ref] Characteristics of adolescent depression, Crowe [/bib_ref].
Our finding that some young people distance themselves from others through a fear of being rejected is consistent with the interpersonal hostility theory of loneliness. This theory posits that loneliness generates negative social cognitions through which others can be perceived as threatening, competitive and unwelcoming. This leads to a self-fulfilling prophecy in which social contact is put on hold through fear of negative evaluation. This process of interpersonal hostility leads to more isolation, which increases the likelihood of feeling lonely, creating a self-reinforcing loneliness cycle accompanied by feelings of low self-esteem [bib_ref] Loneliness matters: a theoretical and empirical review of consequences and mechanisms, Hawkley [/bib_ref].
Empirical studies exploring the responses of young people towards peers suffering from mental illness demonstrate clear stigmatising attitudes towards people with mental health problems and a preference to avoid them [bib_ref] Influences on young people's stigmatising attitudes towards peers with mental disorders: national..., Jorm [/bib_ref]. This would confirm what young people with depression fear: being judged and avoided. Their preference for non-disclosure is therefore unsurprising. Such work also shows that young people's attitudes towards peers with mental health problems are influenced by their parents' attitudes and by their previous exposure to people with mental health problems [bib_ref] Influences on young people's stigmatising attitudes towards peers with mental disorders: national..., Jorm [/bib_ref]. This suggests that there is scope to modify attitudes and create more accepting environments for young people with depression.
Adolescence is a time characterized by hormonal, physical, psychological and social change [bib_ref] Development of the social brain in adolescence, Blakemore [/bib_ref] , during which identity formation, role transition, independence and creating relationships are of critical importance. During this period adolescents become less dependent on parental attachment as peer relationships become more important [bib_ref] In my mind I was alone: suicidal Adolescents' perceptions of attachment relationships, Bostik [/bib_ref] , but this can mean that the effects of social exclusion by peers are felt more acutely [bib_ref] In my mind I was alone: suicidal Adolescents' perceptions of attachment relationships, Bostik [/bib_ref]. Attachment theory suggests that psychiatric symptoms (such as depression) and feelings of loneliness arise when there is an absence of opportunity to make affectional bonds, or when bonds once made are repeatedly disrupted. It is also theorised that young people with attachment disorders come to define themselves as outsiders and through this they risk becoming chronically lonely. It is possible that early experience of unreliable and unresponsive attachment figures can lead to insecure attachment perceptions, including a lack of trust, low self-esteem and difficulties with affect regulation and intimacy [bib_ref] In my mind I was alone: suicidal Adolescents' perceptions of attachment relationships, Bostik [/bib_ref]. This is supported by evidence that lonely young adults are more likely to have been bullied and socially isolated as children [bib_ref] Lonely young adults in modern Britain: findings from an epidemiological cohort study, Matthews [/bib_ref] and that securely attached adolescents report more emotionally close friendships and greater social acceptance by peers than insecurely attached peers [bib_ref] In my mind I was alone: suicidal Adolescents' perceptions of attachment relationships, Bostik [/bib_ref]. Such work suggests that early parenting interventions to improve attachment have the potential to prevent loneliness and depression, although this requires formal testing.
# Strengths and limitations
This meta-synthesis used a robust systematic search strategy to identify studies collecting qualitative accounts of loneliness in young people with depression across a range of countries. We followed received guidelines on conducting a meta-synthesis and used an interdisciplinary team approach in conducting our analysis. We addressed threats to validity by presenting preliminary results at a workshop attended by academics, voluntary sector practitioners, and people with lived experience, requesting feedback with which to revise our thematic framework.
The nature of our research question meant that research subjects were hard to reach and it is possible that those willing to participate in included studies are not representative of the wider population of depressed young people. The predominance of female participants from high-income countries, also limits generalisability. Whilst our research question related specifically to the experience of loneliness in the context of depression, our search strategy did not restrict eligibility to studies that had similar aims. Instead, by including studies with the broader remit of understanding the experiences of depression among young people we were able to identify instances where loneliness was mentioned in this context. Whilst this meant that included studies did not necessarily probe the experience of loneliness in depression, restricting the richness and variety in data relevant to our research question, we were able to identify and describe this where it was mentioned. Any meta-synthesis analyses the findings of previously analysed data and is therefore a subjective interpretation of an interpretation [bib_ref] Understanding qualitative Metasynthesis: issues and opportunities in early childhood intervention research, Erwin [/bib_ref]. This could potentially undermine the integrity of individual studies by ignoring the context, leading to superficial interpretations rather than a deeper understanding [bib_ref] Qualitative metasynthesis: issues and techniques, Sandelowski [/bib_ref]. This was a particular threat in our meta-synthesis given the range of analytic approaches used in the constituent fourteen studies. A further limitation of the meta-synthesis approach is that such interpretations are made by a group of researchers with their own perspectives, which might plausibly differ from those of another team. However, we addressed this through our consideration of reflexivity in discussions of the multidisciplinary research team. To address some of these limitations and gain a more in-depth understanding of the pathways between loneliness and depression in young adults, it would be important to conduct qualitative interview studies with a specific focus on these links.
## Clinical and policy implications
Our study provides valuable insights for clinicians, teachers, parents, peers and researchers into the social challenges faced by young people with depression, helping them understand how feelings of loneliness might arise in those who feel depressed and how they might compound depressive symptoms, impeding recovery. This is consistent with the evidence that poor subjective social support is associated with poorer recovery from depression [bib_ref] Associations between loneliness and perceived social support and outcomes of mental health..., Wang [/bib_ref].
It was common for research participants to feel that others did not understand them and the distress associated with this was very apparent. Lay dissemination of the findings of this study might help young depressed people feel more understood by those in their social networks and suggest ways in which others can support them appropriately. There is evidence that social media campaigns focussed on mental health awareness and stigma reduction can promote help-seeking for mental illness, but this is thought to be attributable to wider societal awareness in creating environments where it is acceptable to disclose mental illness [bib_ref] Youth mental health services utilization rates after a large-scale social media campaign:..., Booth [/bib_ref]. Creating a culture in which it is more acceptable to disclose mental ill-health has the potential to interrupt the vicious cycle of social withdrawal giving rise to loneliness and thereby worsening mood. It may be useful for clinicians to explain to young people with depression that it is common for young patients not to disclose mental health difficulties to their peers, but that this might potentially reduce their sense of connectedness. Clinicians might be able to educate them about the impact of loneliness on the prognosis of their depression and explore appropriate ways to address this.
Early intervention in loneliness would appear to be critical to prevent lonely young adults from being trapped in loneliness as they grow older [bib_ref] Lonely young adults in modern Britain: findings from an epidemiological cohort study, Matthews [/bib_ref] and to prevent it from limiting psychosocial functioning and diminishing quality of life at an important stage of social and emotional development. However, whilst this study provides an understanding of the links between loneliness and depression in social terms, it does not detail the psychological processes involved in those pathways. A better understanding of the psychological factors that engender and perpetuate loneliness in young people with depression would help identify those that are modifiable and contribute to developing effective interventions. Combining insights from cognitive neuroscience and social psychology would be helpful in the foundations for this interventional work.
# Future research
We have mentioned the need to conduct interview studies directly probing the experience of loneliness in young people with depression, with balanced gender representation. Given the lack of ethnic diversity in our sample and cultural dimensions of social connections, we also need studies conducted in different ethnic groups. Such work might probe, for example, the more extreme phenomenon of hikikomori in Japan where young people with mental health problems engage in extreme selfisolation [bib_ref] Youth social withdrawal behavior (hikikomori): a systematic review of qualitative and quantitative..., Li [/bib_ref]. We have also mentioned the need to develop effective interventions. In the wider literature on approaches to addressing loneliness among people with mental health problems, changing cognitions to shift maladaptive cognitions is viewed as most promising, but lacks a robust evidence base [bib_ref] A life less lonely: the state of the art in interventions to..., Mann [/bib_ref]. Amongst trials of interventions to improve subjective and/or objective social isolation for people with mental health problems, again the most promising interventions include cognitive modification for subjective social isolation, as well as interventions with mixed strategies and supported socialisation for objective social isolation [bib_ref] The effectiveness of interventions for reducing subjective and objective social isolation among..., Ma [/bib_ref]. More research is needed to develop and assess the acceptability of interventions that address cognitions about social engagement among young people with mental health problems, before trialling them rigorously.
# Conclusions
Our meta-synthesis of fourteen qualitative studies capturing experiences of loneliness among young people with depression identified four themes revealing the challenges faced by these individuals in their social networks. They described how the symptoms of their depression hampered social engagement, leading to social withdrawal. A preference not to disclose their mental health problems, for a variety of reasons, compounded the perception of differences between depressed young people and their peers. Many longed for a connection with others, but could not tolerate the experience of being with others. Some participants described a self-reinforcing cycle of loneliness and depression, from which it was hard to see a way out. Although participants did not suggest how to intervene to break this cycle, our findings suggest that supporting young people with depression to find appropriate ways to disclose their problems has potential to promote a sense of feeling understood by their peers. The published literature also suggests a role for cognitive interventions to shift maladaptive cognitions about their social world. More widely, a change in societal attitudes towards young people with mental illness would also help promote a sense of feeling accepted and socially connected.
# Appendix 1
## Search strategy
To reduce the risk of missing relevant studies our initial search protocol included a term to capture a diagnosis of personality disorder. However, as per our inclusion/ exclusion criteria, only studies relating to young people with depression were included.
MeSH terms: Mood disorder / Affective disorder / Personality disorder / Anxiety disorder / (and related relevant MeSH depending on database) OR Depress* OR Personality disorder** OR Borderline personality disorder or emotionallyunstable personality disorder or histrionic personality disorder or narcissistic personality disorder OR Avoidant personality disorder or dependent personality disorder or obsessive/compulsive personality disorder) AND Loneliness loneliness [MeSH] OR loneliness OR lonely (social* adj/2 isolation network* support contact* relation* capital distance alienat* detach* tie* participation activ* engage* connect* disconnect* cohes* embedded* interact*) AND Qualitative research "Qualitative research" Interviews as Topic "Qualitative adj2 (stud* OR method*)" "Lived experience" "Mixed?method*" "Thematic analysis" "semi-structured" or semistructured or unstructured (guide*) adj2 (interview* or discussion*)) Focus groups ethnograph* fieldwork group discussion Patient perspective
[fig] Figure 1: Flow diagram for included studies [/fig]
|
Probing helicity and the topological origins of helicity via non-local Hanbury-Brown and Twiss correlations
Quantum Hall edge modes are chiral while quantum spin Hall edge modes are helical. However, unlike chiral edge modes which always occur in topological systems, quasi-helical edge modes may arise in a trivial insulator too. These trivial quasi-helical edge modes are not topologically protected and therefore need to be distinguished from helical edge modes arising due to topological reasons. Earlier conductance measurements were used to identify these helical states, in this work we report on the advantage of using the non local shot noise as a probe for the helical nature of these states as also their topological or otherwise origin and compare them with chiral quantum Hall states. We see that in similar set-ups affected by same degree of disorder and inelastic scattering, non local shot noise "HBT" correlations can be positive for helical edge modes but are always negative for the chiral quantum Hall edge modes. Further, while trivial quasi-helical edge modes exhibit negative non-local"HBT" charge correlations, topological helical edge modes can show positive non-local "HBT" charge correlation. We also study the non-local spin correlations and Fano factor for clues as regards both the distinction between chirality/helicity as well as the topological/trivial dichotomy for helical edge modes.In presence of magnetic field and at low temperatures, chiral quantum Hall (QH) edge modes appear in a 2DEG 1, 2 . In such 2D QH bars, edge modes flow in a manner (shown inFig. 1(a), left panel) such that at the top edge electrons only move in one direction to the right. At the other, i.e., bottom edge electrons flow to the left in exactly opposite direction. So if a electron in the top edge has to change its direction, it has to scatter to the opposite edge (bottom), according to the chiral traffic rule 3 . At low temperatures and in samples, e.g., Mercury Telluride/ Cadmium Telluride (HgTe/CdTe) heterostructure's 3 with strong spin-orbit coupling quantum spin Hall (QSH) edge modes appear. Herein spin of the electron is locked to its momentum. If at the top edge of the sample spin up electrons are moving in one direction (say, right) then spin down electrons are moving in the opposite direction (left). And at the bottom edge vice versa. Thus a new traffic rule comes into effect-helical traffic rule and these edge modes are therefore helical 3 , seeFig. 1(a) (middle panel). To scatter, an electron into the opposite direction its spin has to flip. This is prohibited by time reversal(TR) symmetry as QSH samples obey TR symmetry in contradistinction to QH samples which don't. However, it is not always that the origin of helical edge modes in QSH samples is topological, recently there have been cases 4 where spin-momentum locked quasi-helical edge modes appear but these are not topologically protected. It has to be pointed out that the spin momentum locking among quasi-helical edge modes does not survive non-magnetic disorder and intra-edge backscattering comes into effect. These are termed trivial quasi-helical edge modes and are shown inFig. 1(a) (right panel).The reason it is necessary to probe helicity is because the QSH state is a new state of matter-it is a 2D topologically ordered phase in absence of magnetic field. The QSH edge modes due to the potent spin-orbit interaction, realize a 1D metal wherein spin is tied to the direction of motion. This unique state of matter has to be experimentally and rigorously probed such that its existence is beyond doubt. Secondly, this confusion regarding the origin of helical edge states whether its really topological and therefore protected from disorder and inelastic scattering in the sample or its due to some trivial reason and thus of non-topological origin is a current topic of interest as evidenced by the recent works in this field 4, 5 . Further, as the QSH edge modes can be used in low power
information processing due to their robustness against disorder, it is very important to identify these helical edge modes and their topological origins.
There are different methods for distinguishing between helicity and chirality. The usual way to probe the existence of chiral/helical edge modes is via conductance measurements in multi terminal transport experiments [bib_ref] Nonlocal transport in the quantum spin hall state, Roth [/bib_ref] [bib_ref] Spin polarization of the quantum spin Hall edge states, Brune [/bib_ref]. Lets consider an elementary set-up as in(left panel)-a two terminal conductor with probes 1 and 2 as source and sink. If a third probe is added in between probe 1 and 2 as a voltage probe(left panel)) then for QH sample one edge mode enters probe 3 from source and another edge mode goes out from it. So to maintain net current zero at probe 2 its potential is adjusted to potential of the source, and the two terminal conductance of the sample remain the same as before (without voltage probe 2). We can understand this from Landauer-Buttiker formalism, ⇒V 2 = V 1 /2, where G 21 = G 23 = G and V 3 = 0. So its potential is adjusted to half of the potential of the source [bib_ref] Edge-state physics without magnetic fields, Buttiker [/bib_ref]. In QSH (Helical-Topological) samples the conductance is reduced by adding a voltage probe and is e h . Measuring the conductance with the inclusion of a voltage probe one can differentiate between the topological helical and chiral edge modes. Now what about trivial quasi-helical edge modes these are not topologically protected and therefore these are susceptible to intra-edge scattering. At the top edge the electronic edge mode with spin-up (shown in red) has a finite probability of spin-flip scattering and reversing its path, the same thing happens for the spin-down electron (shown in blue). The small arrows in between the quasi-helical (trivial) edge modes indicates this process. The three terminal conductance for the quasi-helical trivial case then is − f [bib_ref] Absence of backscattering in the quantum hall effect in multiprobe conductors, Buttiker [/bib_ref] e h , where f is the probability of intra-edge scattering a measure of the vulnerability of trivial quasi-helical edge modes to disorder and inelastic scattering. In clean samples where the probability of intra-edge scattering (f) is expected to be small, relying on conductance measurement alone may not be wise. Therefore in this work we focus attention on the noise in particular the non-local HBT correlations.
Herein, we have assumed that the trivial quasi-helical edge modes in absence of any disorder are similar to the topological helical edge modes and are spin-momentum locked. In other words-helical, up-spin edge modes at same edge have exactly opposite momentum to down-spin edge modes. There is evidence that trivial quasi-helical edge modes have spin-momentum locking 4 but it's not fool-proof. In case there is no spin-momentum locking in the trivial phase then this case resembles a ballistic scattering state. The two terminal ballistic conductance will be double for the no spin-momentum locked trivial phase as compared to that with spin-momentum locking. We show via a simple calculation in section 4.3.1 the validity of our assertion. However, the ballistic phase unlike the spin-momentum locked phase will be susceptible to not just spin flip scattering but also back scattering in presence of sample disorder. This may complicate the situation and again one has to take recourse to the non-local noise to resolve this situation. Thus, the topological or otherwise origin of helical edge modes needs further probe via the nonlocal HBT correlations.
The subject of this work on distinguishing topological chiral and helical edge modes and determining whether the origin of the helical edge modes is topological or not via non-local HBT correlations has not been dealt with in any previous work. However, some works have looked at other aspects of topological helical and chiral edge modes. In two of our recent works [bib_ref] Are quantum spin hall edge modes more resilient to disorder, sample geometry..., Mani [/bib_ref] [bib_ref] Fragility of non-local edge mode transport in the quantum spin hall state, Mani [/bib_ref] we also have explored the distinct attributes of chiral QH and helical QSH topological edge modes. Further, to the aforesaid works, few more papers [bib_ref] Probing the helical edge states of a topological insualtor by cooper-pair injection, Adroguer [/bib_ref] [bib_ref] Electrical manipulation and measurement of spin properties of quantum spin hall edge..., Vayrynen [/bib_ref] [bib_ref] Corner junction as a probe of helical edge states, Hou [/bib_ref] [bib_ref] Spin-polarized scanning-tunneling probe for helical luttinger liquids, Das [/bib_ref] have explored the topic of helical vs. chiral edge modes using superconductors [bib_ref] Probing the helical edge states of a topological insualtor by cooper-pair injection, Adroguer [/bib_ref] , with polarized STM tips [bib_ref] Spin-polarized scanning-tunneling probe for helical luttinger liquids, Das [/bib_ref] , with corner junctions [bib_ref] Corner junction as a probe of helical edge states, Hou [/bib_ref] and finally exploiting the Rashba coupling [bib_ref] Electrical manipulation and measurement of spin properties of quantum spin hall edge..., Vayrynen [/bib_ref]. All these works while relying on different systems have a common conductance measurement which acts as the arbiter of helicity. Since in quantum spin Hall systems spin is locked to momentum, relying on just conductance measurements is risky, wherein detecting degree of spin polarization in samples exposed to disorder and spin-flip scattering will be tricky. In this paper we aim to use the Hanbury-Brown and Twiss or shot-noise correlations to probe the presence of helical edge modes and determine its origins whether topological or not. Non-local shot noise correlations on the other hand are seen to use the disorder and/or inelastic scattering present as a resource in being better able to differentiate between chirality and helicity and also between trivial and topological edge modes.
The theoretical examination of noise in QSH systems has mostly focused on the effect of electron-electron (e-e) interactions on the current-current correlations within a helical Luttinger liquid model describing the QSH state. Further these studies are in presence of a QPC in a QSH bar, as in refs There are also few papers on current-current correlation studied via the scattering matrix approach or other than helical Luttinger liquid approach, see refs 19 and 20. In ref. 21 differential noise is studied in presence of magnetic moment which is strongly dependent on frequency of the current. In our work zero frequency non-local correlations are studied for QSH systems as regards distinguishing chiral versus helical (topological) and quasi-helical (trivial) phases. The focus of the aforesaid references is not on identifying the topological origins of helical edge modes neither on the distinction between chiral and helical edge modes as is the case in our work.
However, apart from noise, various research groups around the world have made intriguing attempts at inferring helicity in edge mode transport QSH systems via the conductance.
A very recent proposal concerns a π shift seen in the conductance measurement of a QSH system 22 , this work relies on quantum point contacts(QPC's) which due to the Klein effect for Dirac states will be difficult to achieve experimentally in QSH systems. This method also has an inherent weakness in that such a π shift is only observed when backscattering is absent. This implies presence of disorder will trip this method up rendering it un-fructuous. Another interesting proposal aims to use a Hong-Ou-Mandel interferometer [bib_ref] Electronic hong-ou-mandel interferometry in two-dimensional topological insulators, Ferraro [/bib_ref] with QSH/QH edge modes. This work uses noise and proposes to use the dip in noise at zero power as a probe. However, this dip is shown as function of the time delay between two sources in the interferometer and its magnitude is compared for chiral and helical cases. These dips are affected by number of edge modes making the clear cut differentiation difficult. Further, no comment is made on the presence of disorder and inelastic scattering. Another work which includes disorder [bib_ref] Disorder and magnetic-field-induced breakdown of helical edge conduction in an inverted electron-hole..., Pikulin [/bib_ref] and tries to distinguish between chiral and helical edge modes via a quantization of the conductance measurement obviates the weakness of refs 22 and 23 but has an inherent weakness in that-with disorder the quantization vanishes. An interesting proposal which also uses the noise correlations [bib_ref] z 2 peak of noise correlations in a quantum spin hall insulator, Edge [/bib_ref] to distinguish between chiral and helical edge modes in presence of disorder purports to be better than ref. 24 but then it again would be difficult to experimentally realize with current technology because of its reliance on QPC's. Another related work concerns the amount of net spin tunneling between edge states and this can be also used as an arbiter for helicity [bib_ref] Tunneling between edge states in a quantum spin hall system, Strom [/bib_ref] , however herein too effects of disorder and inelastic scattering are not dealt with, finally a related work suggests the use of noise [bib_ref] Nonequilibrium noise correlations in a point contact of helical edge states, Lee [/bib_ref] and uses a four terminal QPC to probe the helicity versus chirality dilemma, however herein too the dependence on QPC's will hamper any experimental realization. Further, the distinction between chiral and helical cases is via a difference in magnitude of the noise while a better arbiter is the sign which we will focus on in this work and will aim to surmount the challenges in the above proposals. On the question of topological helical vs. trivial quasi-helical edge modes there have been a couple of experimental papers [bib_ref] Edge transport in the trivial phase of inas/gasb, Nichele [/bib_ref] which have shown that quasi-helical edge modes do exist in trivial insulator but only a single theoretical work has dealt with this problem. In ref. [bib_ref] Phase diagram of two interacting helical states, Santos [/bib_ref] , the authors propose a method to distinguish between the two which relies on the addition of two non-magnetic impurities in an other wise clean QSH sample. The occurrence of localized zero modes identifies the topological origin of the helical edge modes. Notwithstanding the complexity of the method this approach also will be hard to fashion experimentally since detecting zero modes is a non-trivial task.
Further, while local shot noise correlations have been calculated in some recent works with QSH samples 19 to our knowledge this is the first work wherein both the non-local charge and spin shot noise correlations have been used as a probe of helicity and its topological origins and also to discriminate between chiral and helical edge modes. In 1950, R Hanbury Brown and R Twiss found out the diameter of radio stars via a intensity-intensity correlation experiment [bib_ref] Entangled hanbury brown twiss effects with edge states, Buttiker [/bib_ref] [bib_ref] A new type of interferometer for use in radio astronomy, Hanbury Brown [/bib_ref]. The fermionic analog of this famous experiment was realized in refs 29 and 30 for a 2DEG in the chiral QH regime. The correlations were shown to be completely anti-correlated meaning fermions in obedience to Paulli principle exclude each other. These correlations also go by the name of shot noise which measures the correlations between fluctuations of the current [bib_ref] Shot noise in mesoscopic conductors, Blanter [/bib_ref]. In this work we find that measuring the non-local HBT noise correlations between two disordered probes in presence of inelastic scattering via a voltage probe one can differentiate between the chirality or helicity of the edge modes as well as the topological or trivial origin of the Helical edge modes. We observed that in the mentioned condition while the cross correlations are always negative for chiral QH case, it can be positive or negative depending on the disorder in the topological Helical QSH case. So getting a positive nonlocal correlation in these conditions identifies topological helical edge modes. Further, the shot noise correlations can be of two types for Helical case: one the charge and the other spin. We find in this paper that while there is no distinction between charge and spin noise correlations for topological helical edge modes, they are completely different for trivial quasi-helical edge modes enabling an effective discrimination between the topological or trivial origins of these edge modes.
The rest of the paper is organized as follows, first we focus on the chiral QH case and calculate the non-local "HBT" correlations for two disordered probes in our set-up, then proceed to case of all disordered probes. Next we add inelastic scattering to our set-up with two disordered probes and finally ending with all disordered probes with inelastic scattering. We see in all these cases non-local HBT correlations are always negative. We particularly also focus on a well known theoretical work 32 and its subsequent experimental implementation [bib_ref] Positive cross-correlations in a normal-conducting fermionic beam splitter, Oberholzer [/bib_ref] and show that due to the presence of quantum point contacts(QPC) in these studies, one get positive non-local correlations for chiral QH samples. In our work we deliberately remove QPC's since our focus is on obtaining positive correlations in helical QSH samples, where due to Dirac nature of the edge states experimental implementation of QPC's is difficult.
We next focus on the topological helical QSH case, herein we distinguish between the non-local charge and spin correlations. Similar to the chiral QH case we calculate the the non-local "HBT" correlations for two disordered probes in the QSH set-up, then discuss the case of all disordered probes. Like the QH case we next add inelastic scattering to the QSH set-up with two disordered probes and finally to all disordered probes with inelastic scattering. We see that the non-local charge correlations can be positive in presence of inelastic scattering. Next we focus on the question of distinguishing topological from trivial quasi-helical edge modes. In case of topological helical edge modes charge and spin HBT correlations are identical however these two differ for trivial quasi-helical case. The non-local spin HBT correlations turns completely positive for trivial quasi-helical case but the non-local charge correlations turn absolutely negative. We end the paper by focusing on the Fano factor, summarize our results in two tables, finally concluding it with a perspective on applications to other materials.
## Shot noise in quantum hall set-up
Following the scattering matrix approach one can calculate the noise in the sample from the scattering matrix of the system as shown below ref. 32:
[formula] ∫ ∑ = − γλ γλ α λγ β γ λ αβ { } S e h dE Tr A A f f 2 (1 ) (1) 2 wherein δ δ = − γλ α αγ αλ αγ αλ † A [/formula]
ss is defined by the scattering matrix element s ij and i j , are the indexes of the probes in the sample. In this work we detect the non-local correlations via probes {α, β → 4, 3 respectively. f γ is the Fermi-Dirac distribution at probe γ. Further, we are at zero temperature always, so f γ can take values 1 and 0 only.
Quantum Hall set-up with two disordered probes. The two probe disordered case for QH is depicted in [fig_ref] Figure 2: Figure 2 [/fig_ref]. The scattering matrix relating the incoming wave to the outgoing one is given as follows: (for < < E eV 0 1 ) at zero temperature, where E is the electronic energy and f i = Fermi-Dirac distribution at contact i which basically depends on the potential of that contact. From Eq. (1) one can calculate the nonlocal correlation between probes 3 and 4 as: Quantum Hall set-up with all disordered probes. The all probe disorder case for QH is depicted in [fig_ref] Figure 2: Figure 2 [/fig_ref]. The scattering matrix relating the incoming to the outgoing wave is given as follows:
[formula] = = − − b b b b s a a a a s r t t r t r r t , with 0 0 0 0 0 0 0 0 ,(2)∫ = − + − + − = − + − + − = + + = † † † † † † S e h dE A A f f A A f f A A f f e h e V V A A e V V A A e V V A= − − − − − − − − − − − − − − − − s a r [/formula]
rr r t t r r t t r tt t t r rr r t t r r t t r t t r tt r r r r t t r r t t r r t t r tt r r r r 1 , herein the term = − a rr r r 1 1 2 3 4 in the denominator arises because of the multiple reflections from the disordered probes [bib_ref] Are quantum spin hall edge modes more resilient to disorder, sample geometry..., Mani [/bib_ref]. This matrix satisfies the unitarity relation = = † † s s ss I. Here too the potentials are identical to two
[formula] probe disorder case-= V V 1 , = = = V V V 0 2 3 4 . Thus, = f 1 1 and = = = f f f 0 2 3 4 (for < < E eV 0 1 ) at zero temperature. [/formula]
We can calculate the nonlocal HBT correlation from Eq. (1) as shown below: wherein we have used the unitarity or conservation of probability condition
[formula] ∫ = − + − + − = − + − + − = + + = − † † † † † † S e h dE A A f f A A f f A A f f e h e V V A A e V V A A e V V A+ = + = r t R T 1 i i i i 2 2 [/formula]
. Here the correlation depends on the disorder at probes 2 and 3 which explains why the correlation for two disordered probe (contacts 1 and 3) case is zero. The nonlocal HBT correlation is negative which is the property of the Fermi-Dirac distribution which directly relates to the antisymmetric wave function of electrons. The QH setup, probes 3 and 4 are detectors kept at zero potential, a voltage is applied only to probe 1, (a) QH bar with two disordered probes (potential at probe 2 = 0), a i 's and b i 's with i = 1-4 are the incoming and outgoing waves at the probes, (b) QH bar with all disordered probes, (c) QH bar with two disordered probes and inelastic scattering-probe 2 (curvy box) is a voltage probe (with current into it I 2 = 0), (d) QH bar with all disordered probes and inelastic scattering-probe 2 (curvy box) is a voltage probe (with current into it I 2 = 0). To avoid clutter the waves are only shown in (a). (b-d) Have exactly similar waves to and from the probes, these aren't shown explicitly.
Scientific REPORTS | 7: 6954 | DOI:10.1038/s41598-017-06820-w Quantum Hall set-up with disordered probes and inelastic scattering. We have introduced inelastic scattering via voltage probe 2. The average current 〈I 2 〉 through this probe is zero. Electrons coming from the probes 1 and 3 are equilibrated to a new potential at probe 2, and their phase is randomized there. The current through any probe is defined by-
[formula] ∫ = ∑ + δ α β αβ β α I d E G f I e 1 [/formula]
, herein the second term is due to the intrinsic fluctuation and given by Eq. (1) and the conductance matrix
[formula] δ = − αβ α αβ α β αβ † G N T r s s ( [ ]) e h 2 [/formula]
with N α = No. of edge modes at contact α. We need to fix the fluctuating part of the current at probe 2, ΔI 2 = 0. This condition ΔI 2 = 0 affects the fluctuation of current at other probes 32 as follows:
[formula] ∫ ∫ ∫ ∑ ∑ ∑ δ δ µ µ δ = +∆ α + = + ∆ − + = ∆ β − + = ∆ α α α α β αβ β α β αβ β α β αβ β β α α β α α α - I I I I e dE G f I e dE G f I e dE G f f I I f e G I I [/formula]
, with the average current in contact 1 1 ,
[formula] 1 ( ) , is average of the Fermi Dirac distribution function in contact , 1 ( ) ,(4) [/formula]
2 2 2 μ 2 , μ 2 being chemical potential and average chemical potential at contact 2.
Putting α = 2, we get: putting this in Eq. (4) we get:
[formula] μ μ μ μ μ μ Δ = − +δ = − + δ δ = − − I e G I e G I I G e 1 ( ) , 0 1 ( ) ,1 ( ) 2 2Δ = δ − δ α α α I I G G I (5) 2 22 2 [/formula]
wherein the first term is due to the intrinsic part of the fluctuation and the second term is due to the voltage fluctuation at probe 2. Thus the nonlocal HBT correlation due to the inelastic scattering is written as: In our case α = 4 and β = 3.
[formula] δ δ = 〈Δ Δ 〉 = 〈 δ − δ − δ 〉 = 〈 δ δ − δ δ − δ + δ δ 〉 = − − + αβ α β α α β β α β α β β α α β αβ α β β α α β S I I I G G I I G G I I I G G I I G G I I G G G I I S G G S G G S G G G S ( ) ( ) ( )(6) [/formula]
Quantum Hall set-up with two disordered probes and inelastic scattering. Two contacts are considered to be disordered for this case. The scattering matrix relating the incoming to the outgoing wave is given as in Eq.. Here,
[formula] we have considered source = V V 1 , and = = V V 0 3 4 [/formula]
are detectors. As contact 2 is the voltage probe, putting
[formula] = − + − = − + − = − I G V V G V V e h T V V R V V e h V TV ( ) ( ) 2 [ ( ) ( )]2 ( ) 2 2I 2 = 0, we get-V 2 = T 1 V 1 . [/formula]
The Fermi-Dirac distribution functions at zero temperature in the probes are as follows-
[formula] = f 1 1 , f 3 = 0, f 4 = 0 (for < < E eV 0 1 ), = f 1 2 (for < < E eV 0 2 ) and = f 0 2 (for < < eV E eV 2 1 ) [/formula]
as probes 3 and 4 are used as detectors and are at zero voltage. Following Eq. (1) the non-local charge correlation between probes 3 and 4 is: . Following Eq. (6) we get the non-local correlation in presence of inelastic scattering as: If there are multiple no. of edge modes then the correlation is just multiplied by the no. of edge modes and it remains always negative irrespective of the disorder or inelastic scattering for QH case.
[formula] Scientific REPORTS | 7: 6954 | DOI:10.1038/s41598-017-06820-w ∫ = − + − + − + − + − = − + − + − + − + − = + + + + = − † † † † † † † † † † S e h dE A A f f A A f f A A f f A A f f A A f f e h e V V A A e V V A A e V V A A e V V A A eV V A A e h eV [/formula]
## S s s s s s s s s s s s t s s s s s s s s
[formula] e h eV TT R 2 [ ( 1 ) (1 ) ( 1 ) (1 ) ( 1 )] 2 [ ( ) ( ) ( ) ( ) ( ) ] 2 [ ( )] 2 [= − − + = − S S G G S G G S G G G S e h eV T T R 2 [ ] [/formula]
Quantum Hall set-up with all disordered probes and inelastic scattering. All contacts are considered to be disordered for this case. The scattering matrix relating the incoming to the outgoing wave is given as in Eq. (3). In the set up as shown in [fig_ref] Figure 2: Figure 2 [/fig_ref] , only one mode is shown. We have considered = V V 1 , and = = V V 0 . As contact 2 is the voltage probe, from Landauer-Buttiker formalism putting
[formula] I 2 = 0 gives = − V T V R R R 2 1 1 1 1 3 4 [/formula]
. The Fermi-Dirac distribution functions are in the zero temperature limit given as follows- (1) and (6) one can calculate the non-local correlation. in presence of inelastic scattering as-
[formula] f 1 = 1, = f 0 3 , = f 0 4 (for < < E eV 0 1 ), = f 1 2 (for < < E eV 0 2 ) and = f 0 2 (for < < eV E eV 2 1 ). From Eqs= − − + + − − + − − − − + − . S TT T R R R R a R T R R R RR R R R R R R T a R R R R R R T R R R R a RR T R R R R T (1 ) [ (1 )((1 ) (1 ) 4 ) 2 ( 2 ) (1(1 ) [/formula]
)] We plot the shot noise in presence of inelastic scattering obtained from the above equation in [fig_ref] Figure 3: Non-local correlation in quantum Hall case S in 43 vs R 3... [/fig_ref]. In [fig_ref] Figure 3: Non-local correlation in quantum Hall case S in 43 vs R 3... [/fig_ref] (dashed line) we see as the disorder at probes 1, 2 and 4 increases the non-local correlation almost vanishes for low levels of disorder at probe 3. This is because the probe with larger disorder behaves as closed for the electron, meaning electron almost cannot transmit into the probe. So it is more probable for the electron to follow a path through the probe with less disorder. This makes the electron behavior deterministic (particle like behavior) rather than probabilistic (wave like behavior), which reduces the noise correlation (almost to zero). As disorder at probe 3 increases electron path becomes more probabilistic and negative correlations appear. One can clearly conclude that probes with same or close to the same disorder will show maximum stochastic nature in the system and will show maximum negative correlation, which is shown in [fig_ref] Figure 3: Non-local correlation in quantum Hall case S in 43 vs R 3... [/fig_ref]. 2( / ) [bib_ref] Absence of backscattering in the quantum hall effect in multiprobe conductors, Buttiker [/bib_ref] in [fig_ref] Table 2: Topological helical vs [/fig_ref] is completely negative as we see in our case too. The different transmittances for different edge modes arising from a particular contact is the reason why there is a positive correlation. The experimental realization of this set-up in ref. [bib_ref] Positive cross-correlations in a normal-conducting fermionic beam splitter, Oberholzer [/bib_ref] requires QPC's in order to generate different transmittances for different edge modes which for chiral QH samples maybe alright but is quite difficult for helical QSH samples, since in the latter due to Dirac nature of edge states(Klein effect) its extremely difficult to tune their transmittances via a QPC. In this context the set-up we have which does not rely on QPC's but as we will see in next section generates positive correlations for helical edge modes becomes much more relevant for experimental implementation. Generating positive non-local correlations is the first step to generating entangled currents, which will have important applications in quantum information processing tasks.
## Shot noise in quantum spin hall set-up
As the edge modes in QSH are spin polarized, so the non-local correlation can be calculated separately for charge as well as spin. The charge shot noise formula is given as follows-
[formula] = + + + αβ αβ ↑↑ αβ ↑↓ αβ ↓↑ αβ ↓↓ S S S S S (7) ch [/formula]
The above expression can be easily derived by extending the formalism of section II to spin with the spin shot noise formula given as- Quantum spin Hall set-up with two disordered probes. The two probe disorder case for QSH is depicted in [fig_ref] Figure 5: Four terminal Quantum Spin Hall bar showing QSH edge modes [/fig_ref]. The scattering matrix relating the incoming wave to the outgoing one is given as follows: for ( < < E eV 0 1 ). From Eq. (9) one can calculate the nonlocal HBT correlation as: Quantum spin Hall set-up with all disordered probes. The case represented in [fig_ref] Figure 5: Four terminal Quantum Spin Hall bar showing QSH edge modes [/fig_ref] , depicts all disordered probe case. The scattering matrix relating the incoming to the outgoing wave is given as follows: (for < < E eV 0 1 ). From Eqs (7) and (9) Thus the nonlocal charge correlation depends on the disorder at probes 2, 3 and 4 which explains why the correlation is zero for two disordered probes case (disorder at probe 1 and 3). This correlation is always negative irrespective of the magnitude of disorder. Eq. 12 can be simplified by separating the individual spin components as follows:
[formula] ∑ α σ δ δ δ δ δ ′ ′ = ′ ′ − ′ ′ γγ = = − − − − ↑ ↓ ↑ ↓ ↑ ↓ ↑ ↓ ↑ ↓ ↑ ↓ ↑ ↓ ↑ ↓ b b b b b b b b s a a a∫ ∑ ∑ ∑ = ′ ↑ ′ ↑ − + ′ ↑ ′ ↑ − + ′ ↑ ′ ↑ − = − ′ ↑ ′ ↑ + − ′ ↑ ′ ↑ + − ′ ↑ ′ ↑ = ′ ′ + ′ ′ + ′ ′ = ↑↑= − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − s a r rr r t [/formula]
## T r r t t r t t r rr r t t t t r t t r r t t r rr r t t r r t t r t t r r r rr r t t t t r t t r t t r r r r t t r r t t r t t r r r r r r t t t t r r t t r t t r r r r t t t t r t t r r r r r r
[formula] 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 ,(10) [/formula]
## Quantum spin hall set-up with disorder and inelastic scattering.
[formula] S S G G S G G S G G G S S S S S G G S S S S G G S S S S G G G S S S S , ( ) ( ) ( ) ( )(11) [/formula]
Scientific REPORTS | 7: 6954 | DOI:10.1038/s41598-017-06820-w In the above equation
[formula] = − − + + − − + + − − + + − − + = + + + αβ −S S G G S G G S G G G S S G G S G G S G G G S S G G S G G S G G G S S G G S G G S G G G S S S S S(12)= + + + ↑↑ ↑↓ ↓↑ ↓↓ G G G G G kl ch kl kl kl [/formula]
kl is the conductance summed over all the spin indices's, for example ↑↓ G kl represents the probability that a down spin electron is transmitted as a spin up electron. The non-local spin correlations is particular to QSH case and can be similarly, as above, written as
[formula] = − − + αβ − αβ ↑↑ αβ ↑↓ αβ ↓↑ αβ ↓↓ S S S S S (13) sp in in in in in , , , , with = ↑ ↓ αβ S i j , , , [/formula]
ij defined as in Eq. 9. We proceed now by calculating the non-local charge and spin correlations in the next sub-section and beyond.
Quantum spin Hall set-up with two disordered probes and inelastic scattering. For the case of two disordered probes in QSH case, depicted in [fig_ref] Figure 5: Four terminal Quantum Spin Hall bar showing QSH edge modes [/fig_ref]. The scattering matrix relating the incoming to the outgoing wave is given as Eq. (21), following Eq. (13) we first calculate ↑↑ S 43 as follows- Similarly from Eqs (10)
[formula] ∫ ∫ ∑ ∑ ∑ ∑ ∑ = ′ ′ ↑ ′ ′ ↑ − ′ = ′ ′ ′ ′ − ′ = ′ ′ − + ′ ′ − + ′ ′ − + ′ ′ − + ′ ′ − = + + + − + = − ↑↑1 , = f 0 3 , = f 0 4 (for < < E eV 0 1 ), = f 1 2 (for < < E eV 0 2 ) and = f 0 2 (for < < eV E eV 2 1 ) [/formula]
. So from Eqs (7), [bib_ref] Edge-state physics without magnetic fields, Buttiker [/bib_ref] and (13), we calculate the non-local charge correlation in presence of inelastic scattering as: as shown in . From one can conclude that small values of T 1 (large R 1 ) and larger values of T 3 (small R 3 ) help in generating positive cross correlation. In QSH case, inelastic scattering in presence of disorder induces a positive cross correlation in the system which is unexpected for electrons as they are fermions, they should show a negative cross correlation, which is the basis of the famous HBT experiment [bib_ref] Entangled hanbury brown twiss effects with edge states, Buttiker [/bib_ref]. We can understand this in this way that QSH edge modes are spin polarized and there are spin up electrons, which after getting out of the probe 2 (voltage probe which redistributes the current) follow one edge of the Hall bar and directly reach the contact 3 (a detector), at the same time spin down electrons after getting out from same contact 2 follow the other edge of the Hall bar reaching contact 4 (another detector) via contact 1-and these two electrons can be correlated positively. Since these two electrons are traveling via two completely different paths and as different probes are disordered with varying degrees of disorder these two paths will have different transmission probabilities. But in QH case (as discussed in section II) if there are two edge modes, they do not travel via different paths (one cannot separate the paths taken by the two edge modes from voltage probe to detector) and therefore even if probes are affected with varying degrees of disorder the transmission probabilities of two edge modes will be identical. That's why positive non-local correlation is not observed in the QH set-ups as in section II even in presence of disorder and inelastic scattering.
[formula] = − − + − S e h eV TT R TT R R R 2 2 ( ) 4 (14) [/formula]
Quantum spin Hall set-up with all disordered probes and inelastic scattering. The set-up for this case is depicted in [fig_ref] Figure 5: Four terminal Quantum Spin Hall bar showing QSH edge modes [/fig_ref]. The scattering matrix relating the incoming wave to the outgoing one is given as Eq. (10). Here we have considered = V V 1 , and = = V V 0 . As contact 2 is the voltage probe, substituting = I 0
[formula] 2 gives = − V T V R R R 2 2(1 ) 1 1 1 3 4 [/formula]
. Now the Fermi-Dirac distribution functions as usual at zero temperature and with probes 3 and 4 as detectors are as
[formula] follows-= f 1 1 , = f 0 3 , = f 0 4 (for < < E eV 0 1 ), = f 1 2 (for < < E eV 0 2 ) and = f 0 2 (for < < eV E eV 2 1 [/formula]
). Thus, from Eqs (7), [bib_ref] Edge-state physics without magnetic fields, Buttiker [/bib_ref] and (14) one can calculate the non-local correlation − S ch in 43 , as the expressions are large we will analyze them in . Positive non-local correlations are obtained in this case similar to the two probe case discussed above, while inelastic scattering and the fact that up and down spin edge modes have different transmittances through the sample (due to the difference in their paths) is critical to getting positive correlations, the effect of disorder on these positive correlations is more ambiguous. Some disorder is of course necessary to have noise but other than that there is no clear cut influence of increasing/decreasing disorder on the positive correlations so obtained.
Till now we have only considered topological Helical QSH edge modes. Now we ask the question what happens to the positive correlations so obtained if we are not sure of their topological origin. This question has become relevant recently with some papers [bib_ref] Edge transport in the trivial phase of inas/gasb, Nichele [/bib_ref] showing that in a trivial insulator quasi-helical edge modes can also occur, of course they are without any topological protection. In the next section we address this question.
## Topological vs. trivial quasi-helical qsh edge modes
We consider trivial quasi-helical QSH edge modes as shown in . In ref. 4 the difference between trivial and topological QSH modes is determined from the non-local Resistance. Herein we show the non-local noise (both spin as well as charge) can be very effective in determining the topological origins of QSH edge modes. Since these trivial quasi-helical edge modes are not topologically protected there is a finite probability 'f ' that with disorder and inelastic scattering they will scatter from the other mode and change their direction and spin. We denote by parameter 'f '-the probability for an electron with a particular spin orientation in a trivial QSH edge mode to change its direction and spin via intra edge scattering. This intra-edge scattering is shown in by small arrows connecting two oppositely moving edge modes. Thus, 'linking' up and down spin modes due to the possibility of backscattering because of sample disorder/inelastic scattering. However, we note that in both cases trivial quasi-helical as well as topological helical QSH edge modes[see [fig_ref] Figure 5: Four terminal Quantum Spin Hall bar showing QSH edge modes [/fig_ref] ], spin-momentum locking is preserved in absence of any non-magnetic disorder. An up-spin electron is backscattered as a down-spin electron moving in exactly opposite direction.
Trivial QSH set-up with two disordered probes and inelastic scattering. The case represented in , depicts two disordered probes with inelastic scattering in a trivial QSH set-up. The scattering matrix relating the incoming to the outgoing wave is given as follows: gives-
[formula] = + − − − − + − − − − − − + − + − − − − − − + − − − − + − − −− − − + − − − − − +(15)1 ( 1 ) (1 ) (1 ) (1 ) ( 1 ) 1 1 1 (1 ) (1 ) (1 ) (1 ) 1 1 1 (1 ) ( 1 ) (1 ) 1 (1 )(1 ) 1 (1 ) 1 ( 1 )= − + − + − + + V R f T Rf V R f f R f f R R f (1 ) (1 ) 2 ( 1 ) (1 ) 2 (16) 2 3 2 2 1 1 1 1 2 3 2 1 2 3 2 3 [/formula]
At zero temperature, the Fermi-Dirac distribution functions are as follows:
[formula] = f 1 1 , = f 0 3 , = f 0 4 (for < < E eV 0 1 ), = f 1 2 (for < < E eV0 [/formula]
2 ) and = f 0 2 (for < < eV E eV . From Eqs (7), [bib_ref] Edge-state physics without magnetic fields, Buttiker [/bib_ref] , [bib_ref] Corner junction as a probe of helical edge states, Hou [/bib_ref] and (15) , as the expressions are large we will analyze them via plots . As intra-edge scattering probability f increases to 0.25, the correlation can be positive or negative depending on the disorder at probe 1, and as f increases to 0.5 one can see that non-local charge correlation becomes completely negative irrespective of the disorder. Strong spin flip scattering completely destroys the positive correlation effect induced in the non-local fluctuation by the inelastic scattering in the trivial QSH sample. One can also calculate the non-local spin shot noise correlation from Eqs (8), [bib_ref] Are quantum spin hall edge modes more resilient to disorder, sample geometry..., Mani [/bib_ref] , [bib_ref] Corner junction as a probe of helical edge states, Hou [/bib_ref] and [bib_ref] Spin-polarized scanning-tunneling probe for helical luttinger liquids, Das [/bib_ref]. This is plotted in . In this case we see opposite behavior to the non-local charge correlation shown in . The non-local spin correlation turns completely positive with increased intra-edge scattering. Of course the non local charge and spin correlations are identical for QH case as well as for topological QSH samples. The nonlocal HBT spin correlation can thus be a good detector of trivial QSH edge modes.
Fano Factor. The Fano factor, like the coefficient of variation, is a measure of the dispersion of the probability distribution of noise. It is basically the signal to noise ratio, named after Ugo Fano. Surprisingly, the noise is usually smaller than a Poisson distribution noise (in which the variance is equal to the mean value, and F = 1 for Poisson distributions) and it is called sub-poissonian noise (F < 1). If noise is greater than Poisson distribution then it is called super-poissonian noise. The Fano factor is defined by-= , where V 2 is defined as in Eq. 16. We compare the charge Fano factors in the topological QH and QSH cases in [fig_ref] Figure 8: The effect of disorder on charge Fano factors [/fig_ref] while the charge and spin Fano factors in the trivial QSH case in [fig_ref] Figure 8: The effect of disorder on charge Fano factors [/fig_ref]. The charge Fano factor for topological QSH case changes sign while for QH case doesn't as a function of disorder. Further, in case of QSH we have two different Fano factors corresponding to charge and spin. The spin Fano factor is super-Poissonian regardless of whether the edge modes are topological or trivial while the charge Fano factor is sub-Poissonian. Thus the spin Fano factor can also be a good arbiter of the presence or absence of topological helical edge modes. In [fig_ref] Figure 8: The effect of disorder on charge Fano factors [/fig_ref] the charge Fano factors are plotted as function of disorder for increasing intra edge scattering (f), for the topological case while Fano factor changes sign as function of disorder as intra edge scattering increases, i.e., edge modes are in trivial regime, the charge Fano factors turn more and more negative thus one can conclude that for trivial quasi-helical edge modes charge Fano factors will be negative. In [fig_ref] Figure 8: The effect of disorder on charge Fano factors [/fig_ref] we plot the spin Fano factor, although there is no sign change but entering the trivial regime the spin Fano factor increases in magnitude for increasing intra-edge scattering f. The charge shot noise measured in our case is sub-Poissionian, and the charge Fano factor is well below 1/3, which is in agreement with the experimental work of ref. 35 on QSH systems. We summarize the main results on distinction between chiral and helical edge modes and second between topological and trivial origins of QSH edge modes in two Tables 1 and 2. However, our story is not yet complete, the trivial phase we have assumed to have spin-momentum locked edge modes in absence of non-magnetic disorder but although there is overwhelming evidence regarding this it is not cent percent guaranteed [bib_ref] Edge transport in the trivial phase of inas/gasb, Nichele [/bib_ref]. In the next sub-section we address this what-if regarding the trivial phase.
What if the trivial quasi-helical phase does not have any spin-momentum locking, even in absence of non-magnetic disorder? Although its quite probable that the trivial phase in case of QSH systems is dominated by transport via quasi-helical spin-momentum locked edge modes in absence of non-magnetic and for the spin-momentum not locked trivial ballistic phase the quantized resistance for various case in the same H shaped bar shown in In the above equations we have considered no spin flip scattering (f = 0) as well as absence of any backscattering (f ' = 0). In of ref. 4 they get the quantized resistance as mentioned in Eq. 18, which implies that the edge modes are spin-momentum locked. Ref. 4 also mentions that "scanning probe techniques demonstrate the existence of edge channels also in the inverted regime, with similarities to those measured in the trivial regime", and "Furthermore, the edge channel resistance per unit length is very close to earlier reports of helical edge channels". From [fig_ref] Table 4: Topological Helical vs [/fig_ref] , we see that the 2-terminal conductances for topological and trivial phase without spin-momentum locking are different, but 2-terminal conductance for topological and trivial phases with spin-momentum locking are same (for f = 0, as the sample length is smaller than the spin-flip scattering length). This is understandable since in our systems there are no magnetic impurities so no possibility of time-reversal symmetry being broken. Thus the topological helical case wont be susceptible to spin-flip scattering due to sample disorder but trivial quasi-helical edge modes would be. Further trivial ballistic modes would be susceptible to both spin-flip as well as backscattering, since these are no longer edge modes. A simple calculation for the 2 Terminal conductance reveals that for the trivial quasi-helical phase it is
[formula] − f (1 ) e h 2 2 [/formula]
while for the trivial ballistic phase it is
[formula] − − ′ f f (2 ) e h 2 2 [/formula]
. The ballistic conductance is almost double in case of no backscattering (f′ = 0). But the situation is complicated if f ′ ≠ 0. In the latter case, one again takes recourse to the noise to determine exactly how transport is occurring in the trivial phase as shown below.
Noise correlations to the rescue again. To identify unambiguously whether the trivial phase is quasi-helical or not we consider the set-ups shown in.
The set-up for the trivial quasi-helical case is depicted inwhile for the trivial ballistic case is depicted in. The scattering matrix relating the incoming edge modes to the outgoing modes for trivial quasi-helical case is given as in Eq. 20. . Charge and spin conductance in Trivial quasi-helical and Trivial ballistic phases. f is the probability of spin-flip scattering, and f ′ is the backscattering probability which is non-zero only for trivial ballistic phase. Left column has edge modes while write column has ballistic modes and therefore for f ′ = 0, R 2T for ballistic case does not reduce to that of edge state. Putting these values in Eqs (12) and (14) for the trivial quasi-helical case with set-up as shown in.
The scattering matrix relating the incoming wave to the outgoing one for trivial ballistic case is given as in Eq. 21. Here we consider = V V 1 , and = = V V 0 . As contact 2 is the voltage probe, substituting I 2 = 0 gives = V V/3
and Helical edge modes while in [fig_ref] Table 2: Topological helical vs [/fig_ref] we bring out the differences between trivial and topological QSH edge modes. To end, we would like to point out that although in our work we have exclusively focused on chiral and helical edge modes and their topological origins our detection technique (Non-local HBT correlations) can be a very effective tool to probe helicity and its origin in Weyl semi metals 36 too.
[fig] .: V 1 = V 2 The total conductance of the QH sample (Chiral-Topological) The conductance of the QH sample does not change with addition of an extra voltage probe But in QSH sample (topological-helical) (Fig 1(b)(middle panel)) one edge mode enters the voltage probe from source and two edge modes come out of the voltage probe The current through voltage probe- [/fig]
[fig] Figure 1: (a) Chiral vs Helical (Topological) vs quasi-Helical (Trivial), (b) 3-terminal QH, QSH(topological) and QSH(trivial) bar. [/fig]
[fig] Figure 2: Figure 2. The QH setup, probes 3 and 4 are detectors kept at zero potential, a voltage is applied only to probe 1, (a) QH bar with two disordered probes (potential at probe 2 = 0), a i 's and b i 's with i = 1-4 are the incoming and outgoing waves at the probes, (b) QH bar with all disordered probes, (c) QH bar with two disordered probes and inelastic scattering-probe 2 (curvy box) is a voltage probe (with current into it I 2 = 0), (d) QH bar with all disordered probes and inelastic scattering-probe 2 (curvy box) is a voltage probe (with current into it I 2 = 0). To avoid clutter the waves are only shown in (a). (b-d) Have exactly similar waves to and from the probes, these aren't shown explicitly. [/fig]
[fig] Figure 3: Non-local correlation in quantum Hall case S in 43 vs R 3 for all disordered probes with inelastic Scientific REPORTS | 7: 6954 | DOI:10.1038/s41598-017-06820-w Why is shot noise in our quantum Hall set-up always negative but in Texier, et al.,/Oberholzer, et al., set-ups it can be positive? Our set-up is different to that of Texier, et al./Oberholzer, et al., set-ups. They considered a constriction/QPC in their sample which can back scatter electrons thereby generating noise within the system. In our case disorder is relegated to the probe/contact and thus we don't have any back scattering within the sample in contrast to Texier, et al.,/Oberholzer, et al., set-ups as shown inFig. 4. Further, in their set-ups they consider two edge modes with different transmission probabilities-one which is completely transmitted while the other is partially transmitted. However, in our case we have identical transmission probabilities for different edge modes arising from a particular contact. Also getting a positive cross correlation in their set-up depends on the no. of edge modes in the sample but in our set-up the results are independent on the no. of edge modes32,33 .The shot noise result (with inelastic scattering) derived in ref. edge modes with different transmissions.But if two edge modes have same transmission (lets say the two edge modes are partially transmitted with identical transmittances) then the shot noise result is = [/fig]
[fig] Figure 4: The Texier, et al.,/Oberholzer, et al., set-up as in refs 32 and 33 to detect positive non-local HBT correlations in a quantum Hall set up. Here, probe 2 is a voltage probe ( kept at zero voltage. Note that by using constrictions inside the sample and having edge modes transmitting with different probabilities one can engineer positive non-local correlations. However, in the setups we have in this work positive non-local correlation in quantum Hall regime are impossible. Scientific REPORTS | 7: 6954 | DOI:10.1038/s41598-017-06820-wOne can clearly see that the equations for charge and spin shot noise differ by a minus sign in front of the opposite spin correlations. This has important consequences since in presence of finite spin-flip scattering the charge and spin shot noise behave in a dis-similar manner unlike the case in absence of spin-flip wherein these are identical. [/fig]
[fig] Figure 5: Four terminal Quantum Spin Hall bar showing QSH edge modes. These edge modes differ from their QH counterparts since these are spin polarized and helical, probes 3 and 4 are detectors kept at zero potential. (a) Two disordered probes and (b) All probes disordered. = − R T 1 i i represents the reflection probability of edge modes from and into contact i with the strength of disorder in contact i ranging from < < R 0 1 i . (c) Two disordered contacts with inelastic scattering. (d) All disordered probes with inelastic scattering, (probe 2 is a voltage probe in both (c) and (d) with I 2 =0).In (a) the incoming and outgoing waves into the probes are explicitly shown, these are not repeated in (b-d) to avoid clutter.with r i and t i being the reflection and transmission amplitudes at contact i. There are four probes in the case described above and the samples have two edge modes on each side-one for spin up and the other for spin down going in the opposite directions (spin-momentum locked), the scattering matrix s is thus a [/fig]
[fig] Figure 6, Figure 7: S 43 vs. Disorder. (a) Non-local correlation (S 43 ) vs.T 1 and T 3 for two probe disorder with inelastic scattering for QSH (Positive cross correlation), (b) Non-local correlation S 43 vs R 1 for two probe disorder with inelastic scattering for QSH with parameters = . Non-local correlation S 43 vs R 4 for all probe disorder with inelastic scattering for QSH with parameters R 1 = 0.9, = . (a) QSH sample with trivial quasi-helical edge modes. There are two disordered probes with inelastic scattering included via voltage probe 2, small arrowheads indicate intra edge scattering. The effect of such intraedge scattering on positive non-local charge (b) and spin (c) correlations. Non-local charge (b) ( − S ch in 43 vs R 1 ) and spin (c) correlations (S sp in 43 , vs R 1 ) in a trivial quasi-helical QSH sample with two disordered probes ( = . scattering. Note the exactly opposite behavior to the nonlocal charge correlations. The intra edge scattering parameter: = f 0(topological) (red) and = . f 0 1 (blue), = . f 0 2 (pink), = . f 0 3 (black), = . f 0 4 (brown), = . f 0 5 (purple) in (b,c). [/fig]
[fig] Figure 8: The effect of disorder on charge Fano factors (a) Topological QH versus Topological QSH cases, and the effect of intra-edge scattering on Fano factors in (b) for charge and spin Fano factors in trivial QSH phase. The charge Fano factor (c) and spin Fano factor (d) in the trivial phase (..) are completely distinct from topological ( = f 0) QSH phase. (a) Non-local charge Fano factors in topological helical QSH and topological chiral QH cases F ch 43 vs R 2 for all disordered probes ( inelastic scattering. Intraedge scattering probability: = f 0. Note the sub-poissonian behavior in both cases for the charge Fano factor. (b) Non-local charge Fano factor F ch 43 and spin Fano factor in trivial quasi-helical QSH sample F sp 43 vs f (intraedge scattering probability) for two disordered probes ( inelastic scattering. Note the super Poissonian behavior of the spin Fano factor as compared to the charge Fano factor. (c) Non-local charge Fano factors for topological ( = f 0) and trivial ( ≠ f 0) QSH edge modes. F ch 43 vs R 3 for two disordered probes ( = . inelastic scattering as function of R 3 . (d) Non-local spin Fano factors for topological ( = f 0) and trivial ( ≠ f 0) QSH edge modes. F sp 43 vs R 3 for two disordered probes ( = . inelastic scattering as function of R 3 . [/fig]
[fig] Figure 9: (a) Spin-momentum locked trivial quasi-helical edge modes, (b) Trivial ballistic modes without spinmomentum locking. For case (b), since spin-momentum locking is broken, it is more appropriate to address these modes as ballistic although for representative comparison with the spin-momentum locking case they are shown similar to the edge modes in (a), in actuality they are anything but, the two arrows one pointing left and another pointing right on the same mode indicate backscattering while the dashed arrows linking up and down spin modes indicate spin-flip scattering. [/fig]
[table] Table 2: Topological helical vs. Trivial quasi-helical edge modes via non-local HBT correlations. [/table]
[table] Table 4: Topological Helical vs. Trivial quasi-helical vs. Trivial Ballistic phase via 2T resistance. REPORTS | 7: 6954 | DOI:10.1038/s41598-017-06820-w [/table]
|
Energy, Waves, and Forces in Bilateral Fracture of the Femoral Necks: Two Case Presentations and Updated Critical Review
# Introduction
Bilateral femoral neck fracture has long been subject to investigation, owing to its rare occurrence and the multiple causes and mechanisms supposedly responsible for its incidence. The primary cause is usually considered external shock; however, aside from high energy shocks (e.g., falling and impact with hard, rigid surfaces, traffic accidents, etc.) the underlying causes of femoral neck frailty, and how the fractures are produced on both sides, have yet to be understood. A study spanning approximately 80 years of investigations does not reveal thousands of cases, and therefore, these types of fractures are valued as rare. It is also difficult to precisely determine whether men or women are more prone to this condition, or if patients of a certain age can be found more frequently in the studies available to date, because data present both sexes and a range of ages, including children. [bib_ref] Simultaneous bilateral fractures of femoral neck in children: Mechanism of injury, Upadhyay [/bib_ref] [bib_ref] Femoral neck fractures in children: A review, Palocaren [/bib_ref] [bib_ref] Atraumatic bilateral fracture of the femoral neck in young male patient with..., Yoon [/bib_ref] [bib_ref] Low-velocity simultaneous bilateral femoral neck fracture following long-term antiepileptic therapy: A case..., Sadiq [/bib_ref]. Osteoporosis in older females is claimed as a criterion, but it is not self-consistent [bib_ref] Transient osteoporosis of pregnancy complicated by a pathologic subcapital hip fracture, Cohen [/bib_ref] [bib_ref] Transient regional osteoporosis: A rare cause of foot and ankle pain, Chowdhury [/bib_ref] [bib_ref] Isolated acetabular osteoporosis in TOH in pregnancy: A case report, Debnath [/bib_ref] [bib_ref] A case report: Pregnancy-induced severe osteoporosis with eight vertebral fractures, Ofluoglu [/bib_ref] [bib_ref] Transient osteoporosis of the knee in pregnancy, Lloyd [/bib_ref] [bib_ref] Case reports: Transient osteoporosis of the hip: An atypical case, Ma [/bib_ref] [bib_ref] Transient osteoporosis of the hip during pregnancy: A case report, Pai [/bib_ref] [bib_ref] Bilateral Femoral Neck Fracture in a Postpartum Woman: Beware of the Risk..., Sahan [/bib_ref] [bib_ref] Successful treatment for bilateral femoral neck insufficiency fractures: A rare lesion case..., Tan [/bib_ref] [bib_ref] Insufficiency fractures. Geriatr, O'connor [/bib_ref] [bib_ref] Bilateral simultaneous neck femur fracture following domestic fall in an elderly patient:..., Vijayvargiya [/bib_ref]. Underlying conditions, such as bone metabolism disorder, chronic steroid use, renal osteodystrophy, and hypocalcemia, are the main causes of bone frailty, and not of the fracture itself. Convulsions in epileptic patients, or electroshock treatment [bib_ref] Dw Powell Simultaneous Bilateral Fractures of the Neck of the Femur, Powell [/bib_ref] , have resulted in bilateral femoral neck fracture as mentioned in ref. [bib_ref] Dw Powell Simultaneous Bilateral Fractures of the Neck of the Femur, Powell [/bib_ref] , emphasizing the role of muscle contractions. A careful analysis of causes, mechanisms, and effects would allow us to separate the aspects and criteria issued from the problem into (a) underlying causes of bone frailty and (b) mechanisms that have produced bilateral femoral neck fractures. The latter can include high energy trauma [bib_ref] Simultaneous bilateral fractures of the femoral neck caused by high energy: A..., Gao [/bib_ref]. However, a thorough investigation of the possible mechanisms leading to bilateral fracture of the femoral neck should bring to the fore questions related to wave propagation through various segments of the body [bib_ref] Dw Powell Simultaneous Bilateral Fractures of the Neck of the Femur, Powell [/bib_ref] [bib_ref] Simultaneous bilateral fractures of the femoral neck caused by high energy: A..., Gao [/bib_ref] [bib_ref] Hypovitaminosis D Among Patients Admitted With Hip Fracture to a Level-1 Trauma..., Farouk [/bib_ref] [bib_ref] Absence of osteonecrosis of the femoral head following, the high degree of..., Fujita [/bib_ref] [bib_ref] Ten-year follow-up study of missed, simultaneous, bilateral femoral-neck fractures treated by bipolar..., Madho [/bib_ref] [bib_ref] Fractures due to hypocalcemic convulsion, Gur [/bib_ref] [bib_ref] Displaced stress fracture of the femoral neck in an active amenorrhoeic adolescent, Haddad [/bib_ref] [bib_ref] Hipocalcemic bilateral fracture of the femoral neck during a convulsion, Taylor [/bib_ref] [bib_ref] Stress fracture of the femoral neck, Devas [/bib_ref] [bib_ref] Fatigue fractures of the lower extremities: One of bilateral cases of bilateral..., Rengman [/bib_ref].
In order to understand the mechanisms behind bilateral femoral neck fractures, it is important to first consider the anatomy of the femur. From the internal structures, two trabecular columns (one vertical and one horizontal) are most obvious and are best visualized clinically on antero-posterior radiographs of the hip joint, in which it is seen that they pass deeply through the femoral head, terminating close to the subchondral bone (e.g., [fig_ref] Figure 1: Case 1 [/fig_ref]. When fine cracks of the femoral neck are produced, they are less likely to be visible on X-ray radiographs; therefore, from the very beginning, an MRI combined with a CT can be more helpful. In this work, [fig_ref] Figure 1: Case 1 [/fig_ref] shows the CT, and,b, shows the MRI of the preconditions in Case 1. This can lead to an initial misdiagnosis of bilateral femoral neck fracture and to doubtful interpretation of simultaneity in these types of fractures. The horizontal trabeculae, as they traverse the femoral head, are seen to cross the vertical trabeculae at approximately right angles. The highest stress loads within the femur act on the femoral neck. These are external forces and are enhanced by bone frailty.
Diagnostics 2022, 12, x FOR PEER REVIEW 2 of 10 studies available to date, because data present both sexes and a range of ages, including children. [bib_ref] Simultaneous bilateral fractures of femoral neck in children: Mechanism of injury, Upadhyay [/bib_ref] [bib_ref] Femoral neck fractures in children: A review, Palocaren [/bib_ref] [bib_ref] Atraumatic bilateral fracture of the femoral neck in young male patient with..., Yoon [/bib_ref] [bib_ref] Low-velocity simultaneous bilateral femoral neck fracture following long-term antiepileptic therapy: A case..., Sadiq [/bib_ref]. Osteoporosis in older females is claimed as a criterion, but it is not selfconsistent [bib_ref] Transient osteoporosis of pregnancy complicated by a pathologic subcapital hip fracture, Cohen [/bib_ref] [bib_ref] Transient regional osteoporosis: A rare cause of foot and ankle pain, Chowdhury [/bib_ref] [bib_ref] Isolated acetabular osteoporosis in TOH in pregnancy: A case report, Debnath [/bib_ref] [bib_ref] A case report: Pregnancy-induced severe osteoporosis with eight vertebral fractures, Ofluoglu [/bib_ref] [bib_ref] Transient osteoporosis of the knee in pregnancy, Lloyd [/bib_ref] [bib_ref] Case reports: Transient osteoporosis of the hip: An atypical case, Ma [/bib_ref] [bib_ref] Transient osteoporosis of the hip during pregnancy: A case report, Pai [/bib_ref] [bib_ref] Bilateral Femoral Neck Fracture in a Postpartum Woman: Beware of the Risk..., Sahan [/bib_ref] [bib_ref] Successful treatment for bilateral femoral neck insufficiency fractures: A rare lesion case..., Tan [/bib_ref] [bib_ref] Insufficiency fractures. Geriatr, O'connor [/bib_ref] [bib_ref] Bilateral simultaneous neck femur fracture following domestic fall in an elderly patient:..., Vijayvargiya [/bib_ref].
Underlying conditions, such as bone metabolism disorder, chronic steroid use, renal osteodystrophy, and hypocalcemia, are the main causes of bone frailty, and not of the fracture itself. Convulsions in epileptic patients, or electroshock treatment [bib_ref] Dw Powell Simultaneous Bilateral Fractures of the Neck of the Femur, Powell [/bib_ref] , have resulted in bilateral femoral neck fracture as mentioned in ref. [bib_ref] Dw Powell Simultaneous Bilateral Fractures of the Neck of the Femur, Powell [/bib_ref] , emphasizing the role of muscle contractions. A careful analysis of causes, mechanisms, and effects would allow us to separate the aspects and criteria issued from the problem into (a) underlying causes of bone frailty and (b) mechanisms that have produced bilateral femoral neck fractures. The latter can include high energy trauma [bib_ref] Simultaneous bilateral fractures of the femoral neck caused by high energy: A..., Gao [/bib_ref]. However, a thorough investigation of the possible mechanisms leading to bilateral fracture of the femoral neck should bring to the fore questions related to wave propagation through various segments of the body [bib_ref] Dw Powell Simultaneous Bilateral Fractures of the Neck of the Femur, Powell [/bib_ref] [bib_ref] Simultaneous bilateral fractures of the femoral neck caused by high energy: A..., Gao [/bib_ref] [bib_ref] Hypovitaminosis D Among Patients Admitted With Hip Fracture to a Level-1 Trauma..., Farouk [/bib_ref] [bib_ref] Absence of osteonecrosis of the femoral head following, the high degree of..., Fujita [/bib_ref] [bib_ref] Ten-year follow-up study of missed, simultaneous, bilateral femoral-neck fractures treated by bipolar..., Madho [/bib_ref] [bib_ref] Fractures due to hypocalcemic convulsion, Gur [/bib_ref] [bib_ref] Displaced stress fracture of the femoral neck in an active amenorrhoeic adolescent, Haddad [/bib_ref] [bib_ref] Hipocalcemic bilateral fracture of the femoral neck during a convulsion, Taylor [/bib_ref] [bib_ref] Stress fracture of the femoral neck, Devas [/bib_ref] [bib_ref] Fatigue fractures of the lower extremities: One of bilateral cases of bilateral..., Rengman [/bib_ref].
In order to understand the mechanisms behind bilateral femoral neck fractures, it is important to first consider the anatomy of the femur. From the internal structures, two trabecular columns (one vertical and one horizontal) are most obvious and are best visualized clinically on antero-posterior radiographs of the hip joint, in which it is seen that they pass deeply through the femoral head, terminating close to the subchondral bone (e.g., [fig_ref] Figure 1: Case 1 [/fig_ref]. When fine cracks of the femoral neck are produced, they are less likely to be visible on X-ray radiographs; therefore, from the very beginning, an MRI combined with a CT can be more helpful. In this work, [fig_ref] Figure 1: Case 1 [/fig_ref] shows the CT, and,b, shows the MRI of the preconditions in Case 1. This can lead to an initial misdiagnosis of bilateral femoral neck fracture and to doubtful interpretation of simultaneity in these types of fractures. The horizontal trabeculae, as they traverse the femoral head, are seen to cross the vertical trabeculae at approximately right angles. The highest stress loads within the femur act on the femoral neck. These are external forces and are enhanced by bone frailty.
## Case presentation
Case 1. A woman aged 68 was admitted to our clinic with great walking difficulty and strong pain in both hip joints. Her recent history showed three weeks of tough sciatica in the R side and a consequent cessation of any exercise, work outs, and normal walking, thus leading to weakness of the muscles. She had fallen in her flat by stepping on a cylindrical cable and, in an attempt to reduce the height of falling, she performed a forced squat [fig_ref] Figure 3: Simulation of the fall that led to the fracture in Case 1 [/fig_ref]. The X-ray radiographs performed immediately at admission showed a fracture of the R femoral neck, but no sign of cracks in the L side. Additional information from CT, MRI, and US scans [bib_ref] Clinically Relevant Prenatal Ultra-sound Diagnosis of Umbilical Cord Pathology, Bohîltea [/bib_ref] [bib_ref] The Strategy against Iatrogenic Prematurity Due to True Umbilical Knot: From Prenatal..., Bohiltea [/bib_ref] showed coralliform stones in her L kidney, whereas the R side was clean. Ultrasound investigation has shown an incipient degenerative process, but not to such an extent to produce an enhanced frailty of the bones. The US also confirmed, supported by the MRI and CT scans, that the stones in the left kidney were only delaying its normal function and had no major impact on her health. The patient had normal blood pressure, normal blood chemistry values, and a satisfactory general state. The L hip fracture was not observable at the time of admission but only during the surgery on the R hip. The second surgery operation, i.e., on the L hip, was performed after seven days. Noncement Taperloc Complete prostheses (Zimmer Biomet, Medical Technologies International (MTI), 206 D, 13 September Road, Bucharest, Romania) were applied to each hip. Loading started five days after the second surgery operation. The patient used a walker while assisted by a kinesiotherapist.
## Case presentation
Case 1. A woman aged 68 was admitted to our clinic with great walking difficulty and strong pain in both hip joints. Her recent history showed three weeks of tough sciatica in the R side and a consequent cessation of any exercise, work outs, and normal walking, thus leading to weakness of the muscles. She had fallen in her flat by stepping on a cylindrical cable and, in an attempt to reduce the height of falling, she performed a forced squat [fig_ref] Figure 3: Simulation of the fall that led to the fracture in Case 1 [/fig_ref]. The X-ray radiographs performed immediately at admission showed a fracture of the R femoral neck, but no sign of cracks in the L side. Additional information from CT, MRI, and US scans [bib_ref] Clinically Relevant Prenatal Ultra-sound Diagnosis of Umbilical Cord Pathology, Bohîltea [/bib_ref] [bib_ref] The Strategy against Iatrogenic Prematurity Due to True Umbilical Knot: From Prenatal..., Bohiltea [/bib_ref] showed coralliform stones in her L kidney, whereas the R side was clean. Ultrasound investigation has shown an incipient degenerative process, but not to such an extent to produce an enhanced frailty of the bones. The US also confirmed, supported by the MRI and CT scans, that the stones in the left kidney were only delaying its normal function and had no major impact on her health. The patient had normal blood pressure, normal blood chemistry values, and a satisfactory general state. The L hip fracture was not observable at the time of admission but only during the surgery on the R hip. The second surgery operation, i.e., on the L hip, was performed after seven days. Non-cement Taperloc Complete prostheses (Zimmer Biomet, Medical Technologies International (MTI), 206 D, 13 September Road, Bucharest, Romania) were applied to each hip. Loading started five days after the second surgery operation. The patient used a walker while assisted by a kinesiotherapist.
Forces acting on the femoral necks are both external and internal. The external forces are generated as a reaction to the ground-hill pad impact and are transmitted through the ankle toward the femoral neck. [fig_ref] Figure 3: Simulation of the fall that led to the fracture in Case 1 [/fig_ref]. Normal movements and body balance are achieved by forces generated by muscles acting on bones. These are known as internal forces. At rest, externally and internally generated forces are in equilibrium. These two kinds of forces yield bending and, respectively, torsion moments upon the femoral neck, with the bending moments being a product of vertical loading during daily activities that create compression at the inferior zone of the femoral neck and tension at the superior one Forces acting on the femoral necks are both external and internal. The external forces are generated as a reaction to the ground-hill pad impact and are transmitted through the ankle toward the femoral neck. [fig_ref] Figure 3: Simulation of the fall that led to the fracture in Case 1 [/fig_ref]. Normal movements and body balance are achieved by forces generated by muscles acting on bones. These are known as internal forces. At rest, externally and internally generated forces are in equilibrium. These two kinds of forces yield bending and, respectively, torsion moments upon the femoral neck, with the bending moments being a product of vertical loading during daily activities that create compression at the inferior zone of the femoral neck and tension at the superior one [fig_ref] Figure 3: Simulation of the fall that led to the fracture in Case 1 [/fig_ref]. The vertically oriented trabeculae, located in the inferior zone of the femoral neck, uptake and transfer the vertical load.
Previously published studies show that hip fractures in patients dependent on dialysis are 4.4 times greater than in the general population [bib_ref] Simultaneous bilateral femoral neck fractures in a dialysis-dependent patient: Case report and..., Zhu [/bib_ref] [bib_ref] Spontaneous bilateral femoral neck fractures in a young adult with chronic renal..., Karapinar [/bib_ref] , with an incidence of 29.3/1000 people per year [bib_ref] Increased risk of hip fracture among patients with end-stage renal disease, Alem [/bib_ref]. This information relates to unilateral hip fractures, as bilateral ones are still very rare [bib_ref] Temporal Trends in the Incidence, Treatment, and Outcomes of Hip Fracture in..., Nair [/bib_ref].
Case 2. A 37-year-old woman, with type I diabetes and dialysis dependency, was admitted to our clinic with a bilateral femoral neck fracture caused by a free-fall domestic accident. The impact with a flat, hard surface had produced the fractures. X-ray radiographs [fig_ref] Figure 4: Case 2 [/fig_ref] showed the cracks at the beginning of investigations. Hematology and biochemistry tests were acceptable, considering her complicated health condition and comorbidities. During surgery the patient suffered a second fracture in the right hip in the segment above her knee and required better fixation with cables. She had been cleared for weight bearing after two months. However, she returned only one month later with a new crack in the femur. Surgery was successful [fig_ref] Figure 4: Case 2 [/fig_ref]. Loading has been foreseen at the end of the following four months.
In this case, a high energy shock caused the initial fractures, but it must be taken into account that the patient's bones were dramatically weakened and frail due to diabetes and kidney failure. As one underlying cause of this patient's bilateral femoral neck fracture, Previously published studies show that hip fractures in patients dependent on dialysis are 4.4 times greater than in the general population [bib_ref] Simultaneous bilateral femoral neck fractures in a dialysis-dependent patient: Case report and..., Zhu [/bib_ref] [bib_ref] Spontaneous bilateral femoral neck fractures in a young adult with chronic renal..., Karapinar [/bib_ref] , with an incidence of 29.3/1000 people per year [bib_ref] Increased risk of hip fracture among patients with end-stage renal disease, Alem [/bib_ref]. This information relates to unilateral hip fractures, as bilateral ones are still very rare [bib_ref] Temporal Trends in the Incidence, Treatment, and Outcomes of Hip Fracture in..., Nair [/bib_ref].
Case 2. A 37-year-old woman, with type I diabetes and dialysis dependency, was admitted to our clinic with a bilateral femoral neck fracture caused by a free-fall domestic accident. The impact with a flat, hard surface had produced the fractures. X-ray radiographs [fig_ref] Figure 4: Case 2 [/fig_ref] showed the cracks at the beginning of investigations. Hematology and biochemistry tests were acceptable, considering her complicated health condition and comorbidities. During surgery the patient suffered a second fracture in the right hip in the segment above her knee and required better fixation with cables. She had been cleared for weight bearing after two months. However, she returned only one month later with a new crack in the femur. Surgery was successful [fig_ref] Figure 4: Case 2 [/fig_ref]. Loading has been foreseen at the end of the following four months.
In this case, a high energy shock caused the initial fractures, but it must be taken into account that the patient's bones were dramatically weakened and frail due to diabetes and kidney failure. As one underlying cause of this patient's bilateral femoral neck fracture, perhaps the major one, we assumed renal osteodystrophy, a consequence of her chronic kidney disease.
External forces generate shock by the waves that propagate from the hill pad/ground interface to the femoral neck, passing through three joints (ankle, knee, and hip), and two big bones (tibia and femur). The femur is the largest bone in the human body.
The highest stress loads within the femur are sustained by the femoral necks. As compared with the reviewed literature on bilateral femoral neck fractures, the two cases reported here show that kidney disfunction/malfunction can influence bone frailty. In Case 1, the fracture is of the stress kind, i.e., caused by repetitive loading of the femoral neck that leads to either compression side (inferior-medial neck) or tension side (superior-lateral neck) stress fractures. In [fig_ref] Figure 5: Case 1 [/fig_ref] , it is shown that the bilateral fracture occurred on the tension side.
perhaps the major one, we assumed renal osteodystrophy, a consequence of her chronic kidney disease.
External forces generate shock by the waves that propagate from the hill pad/ground interface to the femoral neck, passing through three joints (ankle, knee, and hip), and two big bones (tibia and femur). The femur is the largest bone in the human body. The highest stress loads within the femur are sustained by the femoral necks. As compared with the reviewed literature on bilateral femoral neck fractures, the two cases reported here show that kidney disfunction/malfunction can influence bone frailty. In Case 1, the fracture is of the stress kind, i.e., caused by repetitive loading of the femoral neck that leads to either compression side (inferior-medial neck) or tension side (superiorlateral neck) stress fractures. In [fig_ref] Figure 5: Case 1 [/fig_ref] , it is shown that the bilateral fracture occurred on the tension side.
In Case 1, a slight weakness of the muscles at walking can be presumed given her history of sciatica on the R side three weeks prior to the fracture. She had fallen following a slip on a cable, which means it was likely a relatively high velocity fall. It is therefore clear that using appropriate imaging tools, such as CT, MRI, and US, in addition to X-ray radiographs, can support a more precise diagnosis. Furthermore, as thrombosis is known as the main cause of death in hip fractures in general, Doppler US is extremely useful for monitoring blood circulation in the lower part of the limbs.
# Discussion
There are two factors that should be taken into account when analyzing wave prop- In Case 1, a slight weakness of the muscles at walking can be presumed given her history of sciatica on the R side three weeks prior to the fracture. She had fallen following a slip on a cable, which means it was likely a relatively high velocity fall.
It is therefore clear that using appropriate imaging tools, such as CT, MRI, and US, in addition to X-ray radiographs, can support a more precise diagnosis. Furthermore, as thrombosis is known as the main cause of death in hip fractures in general, Doppler US is extremely useful for monitoring blood circulation in the lower part of the limbs.
# Discussion
There are two factors that should be taken into account when analyzing wave propagation in falls: energy (E) and momentum (P). Therefore, almost all published works on the topic of bilateral femoral neck fractures discuss high energy or low energy impact. At the interface between the human body and the surface of impact, the shock wave and the stress wave act to satisfy energy conservation. During falling, the energy of the body is dominantly kinetic, and it is a function of the body mass and its velocity: E = 0.5m*v 2 . A complete description of energy for a free-falling body should be described by E = m*g*h + 0.5m*v 2 , where g is the acceleration of gravity and h is the height, from which the free fall happens. Momentum is also a function of mass m[kg] and velocity: P = m*v. Force is derived from momentum, F = m*a, where a is the acceleration in movement, and is the derivative in time of the velocity. These simple equations are provided to show that forces and velocities play important roles in the mechanics of the human body falling short distances, as domestic accidents often lead to bilateral femoral neck fractures. This approach was adopted in Case 2. It has been shown by other authors [bib_ref] Temporal Trends in the Incidence, Treatment, and Outcomes of Hip Fracture in..., Nair [/bib_ref] that when v > 3 m/s the probability of a bilateral femoral neck fracture occurring is the highest. Domestic accidents can be caused by various factors, such as mistaken shifting of weight, slipping, etc. The velocity in the latter is higher than in the former [bib_ref] Temporal Trends in the Incidence, Treatment, and Outcomes of Hip Fracture in..., Nair [/bib_ref] , and in agreement with the energy and momentum conservation highlighted above, body mass plays an important role. Thus, the higher the initial position, the more energetic the impact. Therefore, attempts to reduce the height of falling (e.g., from standing to kneeling) would diminish the energy dissipation at the impact point. This is the approach that matches Case 1 (see also [fig_ref] Figure 3: Simulation of the fall that led to the fracture in Case 1 [/fig_ref]. The entire dissipated energy is of the vibration kind, i.e., we start talking about waves and their propagation to determine whether it is a question of shock waves or stress waves [bib_ref] Temporal Trends in the Incidence, Treatment, and Outcomes of Hip Fracture in..., Nair [/bib_ref] [bib_ref] Sideways fall-induced impact force and its effect on hip fracture risk: A..., Sarvi [/bib_ref] [bib_ref] Human-Structure Dynamic Interaction during Short-Distance Free Falls, Shahabpoor [/bib_ref] [bib_ref] Identification of Falling Human Body Dynamics, Nam [/bib_ref].
An updated review on literature concerning bilateral femoral neck fractures would begin with the reports of H.D W. Powell [bib_ref] Dw Powell Simultaneous Bilateral Fractures of the Neck of the Femur, Powell [/bib_ref] , which present a collection of cases (see [fig_ref] Table II: [18], pp [/fig_ref] in (ref. [bib_ref] Dw Powell Simultaneous Bilateral Fractures of the Neck of the Femur, Powell [/bib_ref] , pp. 238-243)), most of them being patients in a psychiatric hospital. , pp. 238-240) shows listed cases that have not previously been reported. Most of these patients had been treated with electroconvulsive therapy, leading to strong contractions causing fracturing of femoral necks and even thoracal vertebrae. There is no question of falling, and the entire dissipating energy belongs to the stress waves in response to the shock waves induced by the therapy, leading to contractions. Statistics on sex and age show male predominance (12 men and eight women), aged between 26 and 71. It is difficult, then, to claim that women and older individuals were more exposed to bilateral femoral neck fractures. However, the collection of cases in , pp. 238-240) touches quite restrictively on the subject as all patients had been diagnosed with psychiatric conditions, and had suffered long stays in hospital. There was no mention of malignancy, osteodystrophy, or other metabolic conditions that could have influenced the frailty of bones. [bib_ref] Displaced stress fracture of the femoral neck in an active amenorrhoeic adolescent, Haddad [/bib_ref] [bib_ref] Non-traumatic, bilateral subcapital femoral fractures postpartum, Yassin [/bib_ref] , metabolic diseases affecting the bones, and deficiencies in vitamin D or calcium [bib_ref] Hypovitaminosis D Among Patients Admitted With Hip Fracture to a Level-1 Trauma..., Farouk [/bib_ref] [bib_ref] Spontaneous bilateral femoral neck fractures in a young male adult: A case..., Arisumi [/bib_ref] , were not investigated. Since the 1960s, medical research has evolved tremendously within all branches, including in orthopedics, traumatology, metabolic diseases, and others. Though some authors tend to claim that older females have a predisposition to bilateral femoral neck fractures, recent case reports show that children with a vitamin D deficiency [bib_ref] Simultaneous bilateral fractures of femoral neck in children: Mechanism of injury, Upadhyay [/bib_ref] [bib_ref] Femoral neck fractures in children: A review, Palocaren [/bib_ref] , were subject to this kind of fracture, as well as a young male who had been deprived of sunlight for several years [bib_ref] Spontaneous bilateral femoral neck fractures in a young male adult: A case..., Arisumi [/bib_ref]. Atraumatic bilateral fracture of the femoral neck in young male patients with suspected osteomalacia [bib_ref] Atraumatic bilateral fracture of the femoral neck in young male patient with..., Yoon [/bib_ref] and low-velocity simultaneous bilateral femoral neck fracture following long-term antiepileptic therapy [bib_ref] Low-velocity simultaneous bilateral femoral neck fracture following long-term antiepileptic therapy: A case..., Sadiq [/bib_ref] have also been reported recently, but are very rare indeed.
Another underlying cause of bone fragility causing bilateral femoral neck fracture is osteoporosis, encountered both in young, pregnant female patients, [bib_ref] Low-velocity simultaneous bilateral femoral neck fracture following long-term antiepileptic therapy: A case..., Sadiq [/bib_ref] [bib_ref] Transient osteoporosis of pregnancy complicated by a pathologic subcapital hip fracture, Cohen [/bib_ref] [bib_ref] Transient regional osteoporosis: A rare cause of foot and ankle pain, Chowdhury [/bib_ref] [bib_ref] Isolated acetabular osteoporosis in TOH in pregnancy: A case report, Debnath [/bib_ref] [bib_ref] A case report: Pregnancy-induced severe osteoporosis with eight vertebral fractures, Ofluoglu [/bib_ref] [bib_ref] Transient osteoporosis of the knee in pregnancy, Lloyd [/bib_ref] [bib_ref] Case reports: Transient osteoporosis of the hip: An atypical case, Ma [/bib_ref] [bib_ref] Transient osteoporosis of the hip during pregnancy: A case report, Pai [/bib_ref] [bib_ref] Bilateral Femoral Neck Fracture in a Postpartum Woman: Beware of the Risk..., Sahan [/bib_ref] [bib_ref] Sideways fall-induced impact force and its effect on hip fracture risk: A..., Sarvi [/bib_ref] , young adult females with amenorrhea [bib_ref] Displaced stress fracture of the femoral neck in an active amenorrhoeic adolescent, Haddad [/bib_ref] , or postpartum [bib_ref] Non-traumatic, bilateral subcapital femoral fractures postpartum, Yassin [/bib_ref] , and also in aged patients, both women and men [bib_ref] Insufficiency fractures. Geriatr, O'connor [/bib_ref] [bib_ref] Bilateral simultaneous neck femur fracture following domestic fall in an elderly patient:..., Vijayvargiya [/bib_ref] [bib_ref] Bilateral femoral neck fractures in an adult male following minimal trauma after..., Sood [/bib_ref] [bib_ref] Bilateral Fracture of Femoral Neck in Korea: A Case Report, Park [/bib_ref] [bib_ref] An unexpected cause of bilateral femoral neck fractures, Boog [/bib_ref]. Supposedly, metastatic conditions, as in prostate carcinoma described in ref. [bib_ref] An unexpected cause of bilateral femoral neck fractures, Boog [/bib_ref] , can lead to osteoporotic fractures in men, although they are three times less prone to osteoporotic fractures than women [bib_ref] An unexpected cause of bilateral femoral neck fractures, Boog [/bib_ref] [bib_ref] Osteoporosis in men, Khosla [/bib_ref]. Hormone suppression following, corticosteroid treatments, for example, can lead to loss of bone density in men. Among the metabolic conditions found in patients with bilateral femoral neck fractures, the works reviewed in this study mostly emphasize hypocalcemia [bib_ref] Absence of osteonecrosis of the femoral head following, the high degree of..., Fujita [/bib_ref] [bib_ref] Ten-year follow-up study of missed, simultaneous, bilateral femoral-neck fractures treated by bipolar..., Madho [/bib_ref] [bib_ref] Fractures due to hypocalcemic convulsion, Gur [/bib_ref] [bib_ref] Hipocalcemic bilateral fracture of the femoral neck during a convulsion, Taylor [/bib_ref] and renal failure, with dialysis where it was the case [bib_ref] Ten-year follow-up study of missed, simultaneous, bilateral femoral-neck fractures treated by bipolar..., Madho [/bib_ref] [bib_ref] Simultaneous bilateral femoral neck fractures in a dialysis-dependent patient: Case report and..., Zhu [/bib_ref] [bib_ref] Spontaneous bilateral femoral neck fractures in a young adult with chronic renal..., Karapinar [/bib_ref] [bib_ref] Increased risk of hip fracture among patients with end-stage renal disease, Alem [/bib_ref] [bib_ref] Temporal Trends in the Incidence, Treatment, and Outcomes of Hip Fracture in..., Nair [/bib_ref]. Case 2, which we presented here, adds to this category of patients.
# Conclusions
An attempt to analyze the causes, mechanisms, and effects of bilateral femoral neck fractures has been made in this study. The approach considers energy, waves, and forces that may lead to the assumption that shock waves are "the action upon the bone structure" and stress waves are its "reaction", in terms of Newton's laws.
This work includes two recent case presentations within the frame of an updated critical review. It allows us to separate the aspects and criteria of the problem into (a) underlying causes of bone fragility and (b) mechanisms that produce bilateral femoral neck fractures.
Case 1 is a typical case of war consequences on the biomechanics of the hill pad-tibiafemur-femoral neck system, where compression of the neck occurred due to a stress wave rather than a shock wave. This can be proven by the absence of a second fracture from the images first acquired, the only evidence being pain and walking difficulty. Case 2 shows that metabolic diseases can dramatically enhance the frequency of bilateral femoral neck fractures. Case 1 adds to the deductions of the first comprehensive case report in the 1960s, but without supporting the conclusion that men are more prone to stress wave-induced bilateral neck fractures than women, owing to their long, lean muscles. Still, bilateral fractures of the femoral neck are rare and these factors remain difficult to assess, as they have previously occurred at the same intensity at the same time. Fine cracks can grow and become visible after a period of time. Additional analyses, e.g., CT, MRI, and US scans, in support of X-ray radiographs, are extremely useful for a more precise early diagnosis, as well as for understanding pre-existing conditions related to a patient's history. Although Case 1 has stones in her L kidney, as shown by US, MRI, and CT scans, this cannot be considered a comorbidity; as a metabolic disease, no evidence of malfunction arose from the biochemistry except from a slight delay in excretion from the L kidney. Therefore, the only common feature of the two cases is the bilateral fracture of the neck of the femur caused by falling of some kind.
"Spontaneous" bilateral femoral neck fracture is not the most suitable term, because it is even not clear when "simultaneous" is entirely appropriate.
Based on the review, a standard treatment protocol should comprise of:
(a) a clinical and paraclinical examination; history of the fracture as told and remembered by the patient; history of comorbidities and their respective treatments; and continuity of medication in cases such as diabetes, dialysis, heart disease, and others of the chronic kind. (b) imaging and spectral techniques in support of primary X-ray radiographs, to get a more precise picture of the case and diagnostic. (c) surgery and appropriate manipulation and fixation of the bones, using the most suitable devices (screws, cables, cement/non-cement prostheses, etc.).
[fig] Figure 1: Case 1. (a) X-ray radiograph with inset showing the trabeculae of the femoral neck; (b) CT before fracture. [/fig]
[fig] Figure 2: (a,b) MRI images of the spine for Case 1 following three weeks of sciatica and one week before the bilateral fracture. The inflammation and the discal prolapse associated with lumbar scoliosis remarked could have led to instant pain and caused the fall. The femoral heads are not shown in this image. [/fig]
[fig] Figure 3: Simulation of the fall that led to the fracture in Case 1. [/fig]
[fig] Figure 4: Case 2. (a) The bilateral femoral neck fractures are well defined. The patient suffers from type I diabetes and is also dependent on dialysis. (b) The R fracture is reduced with a Taperloc prosthesis, but the femur cracked during insertion of the prosthesis and supplementary cables were required to fix the bone (c). (d) shows the final result, i.e., complete reduction in the fracture in the R femur one month after the first surgery, when the patient returned to the clinic with a new fracture on the same side. The strongest sources for such an enhanced frailty of the bone are presumably her metabolic diseases (see also refs.[5,[10][11][12][13][14][20][21][22][23]). [/fig]
[fig] Figure 5: Case 1: X-ray radiographs. (a) The R femoral neck fracture is visible initially, but not the L one. (b) R femoral neck fracture reduced with prosthesis. The L crack is revealed in surgery. (c) The L femoral neck fracture is fixed with prosthesis seven days post R-side surgery. [/fig]
[table] Table II: [18], pp. 241-242) summarizes cases of accidents reported by other authors. It is notable that renal abscesses, pulmonary tuberculosis, diabetes, and severe rheumatoid arthritis are mentioned in connection with diseased femoral necks, but without detail.Table III ([18], p. 243) contains cases of therapeutically induced convulsions, with predominance in male patients. The comprehensive work published by H.D.W. Powell in 1960 was focused on femoral neck fractures occurring mainly through contractions following electroconvulsive therapy. Hormone activity [/table]
|
Detection of Cryptosporidium parvum in Environmental Soil and Vegetables
# Introduction
Cryptosporidium parvum is a protozoan parasite that causes cryptosporidiosis which is characterized by enteritis of watery diarrhea. Being a zoonotic disease, it can be transmitted between humans and animals. Numerous cryptosporidial enteritis outbreaks have been reported worldwide [bib_ref] A massive outbreak in Milwaukee of cryptosporidium infection transmitted through the public..., Mac Kenzie [/bib_ref] [bib_ref] Waterborne cryptosporidiosis: current status, Smith [/bib_ref]. In Korea, the first waterborne outbreak of cryptosporidiosis was reported in 2013 [bib_ref] Epidemiological characteristics of the first water-borne outbreak of cryptosporidiosis in Seoul, Moon [/bib_ref]. Both humans and animals may be exposed to C. parvum through consumption of contaminated water and food as well as by direct contact with contaminated soil and infected hosts [bib_ref] Cryptosporidium: a water-borne zoonotic parasite, Fayer [/bib_ref]. C. parvum is released into the environment through the fecal matter of infected hosts in the form of oocysts and it remains potentially infectious for several months depending upon the environmental conditions and stresses [bib_ref] Cryptosporidium: a water-borne zoonotic parasite, Fayer [/bib_ref]. Application of manure and irrigation to agricultural fields with untreated effluent may also contribute to the prevalence of C. parvum oocysts in the farm environment [bib_ref] A twophase separation method for recovery of Cryptosporidium oocysts from soil samples, Zilberman [/bib_ref].
Foodborne outbreaks are often not characterized as well as waterborne outbreaks [bib_ref] Transport of Cryptosporidium parvum oocysts in soil columns following applications of raw..., Petersen [/bib_ref]. In the last decade, only 15 of 71 worldwide Cryptosporidium-linked outbreaks appeared to be correlated to transmission through food [bib_ref] Global distribution, public health and clinical impact of the protozoan pathogen cryptosporidium...., Putignani [/bib_ref]. Contaminated irrigation water appears to constitute a major route of contamination of fresh products [bib_ref] Occurrence of Cryptosporidium and Giardia in irrigation water and its impact on..., Chaidez [/bib_ref]. In addition, soil contaminated by manure could also be a source for contamination. There has been no report of foodborne outbreak of cryptosporidiosis in Korea until date. Additionally, little information is available regarding the contamination of soil or fresh produce by Cryptosporidium sp. in Korea. Therefore, in this study, we evaluated the extent of contamination by Cryptosporidium of both farm soil and vegetables to determine the source and route of Cryptosporidium infection.
# Materials and methods
## Sample collection and pretreatment
Soil samples were collected from 7 different localities along the western part of the Korean peninsula in January 2012 [fig_ref] Figure 1: Locations of the surveyed areas [/fig_ref]. Soil (20 g) taken from each locality was mixed with 500 mL of filtered (0.22 µm) distilled water and centrifuged at 2,000 × g for 20 min in an ultracentrifuge (Sorvall ® Plus, Thermo Scientific, Waltham, MA, USA) after sieving through a gauze. The pellet was transferred to a 50-mL tube, washed in distilled water, and centrifuged again at 2,000 × g for 20 min. The supernatant was discarded and 40 mL of distilled water was added to the pellet and mixed well. Only 1 mL of the mixed solution was taken and transferred into a microcentrifuge tube and centrifuged at 2,000 × g for 10 min in a microcentrifuge (5415R, Eppendorf, Hamburg, Germany). The pellet was collected and stored at 4°C until the DNA was purified.
Domestic food products were purchased from a local grocery market in Seoul in June 2012. Fifty grams of fresh leafy vegetables (including spinach, wild sesame leaves, and winter-grown cabbage), fruits (including blueberries, wild berries, and cherry tomatoes), and one each of root vegetables (potatoes and carrots) were put into zipper bags (Cleanwrap Inc., Gimhae, Gyeongsangnam-do, Korea). Next, 2-3 volumes of 0.1% liquinox TM (Alconox Inc., White plains, NY, USA) in 0.01 M phosphate-buffered saline (PBS; pH 7.4) was added to the bag; the bags were sealed tightly and placed on an orbital shaker (Hoefer, Holliston, MA, USA) for 15 min. The sample bags were then manually shaken repeatedly in an upside-down manner. The wash solution from the bag was collected in a bottle and centrifuged at 5,000 × g for 20 min in an ultracentrifuge (Thermo Scientific). The supernatant was discarded. The food samples in the zipper bags were rewashed with 0.01 M PBS for 15 min, and the wash solution was centrifuged as before. After centrifugation, the pellets were transferred to 50-mL tubes and centrifuged at 1,000 × g for 20 min in a centrifuge (Union 32R plus, Hanil Science Industrial, Incheon, Korea). Finally, the pellets were collected in microcentrifuge tubes. Cryptosporidium oocysts were isolated by using the Dynabeads ® anti-Cryptosporidium kit (Invitrogen, Grand Island, NY, USA) according to the manufacturer's instructions. DNA was extracted from the soil and food samples by using the QIA-quick stool mini kit (Qiagen Inc., Valencia, CA, USA) (10).
## Real-time polymerase chain reaction
The extracted DNA was used as a template in real-time polymerase chain reaction (qPCR). To obtain Cryptosporidium DNA as a positive control, C. parvum oocysts (KKU isolates) were isolated from laboratory mice (C57 BL6/J) as described by Lee et al. [bib_ref] CP2 gene as a useful viability marker for Cryptosporidium parvum, Lee [/bib_ref]. C. hominis DNA obtained from human feces [bib_ref] Cryptosporidium hominis infection diagnosed by real-time PCR-RFLP, Cheun [/bib_ref] was used as the positive control. The primers and probes used in qPCR were designed based on the gene sequences of C. parvum SWI2/SNF2 ATPase, a Rad16 ortholog gene (Rad16, Gen-Bank no. XM_625623.1; [fig_ref] Table 1: Primers and probes used for qPCR [/fig_ref]. Ab solute quantitative qPCR reactions were performed as per the method described by Lee et al. [bib_ref] CP2 gene as a useful viability marker for Cryptosporidium parvum, Lee [/bib_ref]. The results were analyzed using the LightCycler ® software (version 4.05; Roche). DNase/RNase-free water was used instead of the template DNA as a negative control. The plasmid DNA standard for qPCR was prepared as described previously using Rad 16 as the target gene [bib_ref] CP2 gene as a useful viability marker for Cryptosporidium parvum, Lee [/bib_ref].
Sequences of qPCR products were confirmed using the ABI 7700 Sequence Detector and SDS software (version 1.6.3; Applied Biosystems, Foster City, CA, USA) at the Cosmo Sequ ence Facility Service (Seoul, Korea). The gene sequences were aligned using Clone ManagerSuite 7 (Sci-Ed Software, Cary, NC, USA). To differentiate Cryptosporidium sp. by using the restriction enzyme fragmentation method, the qPCR products were digested with TaqI restriction enzyme (Takara Bio Inc., Shiga, Japan) at 65°C for 2 hr, and the DNA fragments were analyzed on 2% agarose gel.
# Results
Of the 34 farm soil samples examined, Cryptosporidium sp. was detected in 11 soil samples [fig_ref] Table 2: Positive rates of Cryptosporidium detection in soil samples according to the sample... [/fig_ref] ; 32.4% positive). The number of oocysts detected from the soil samples ranged 809-3,710/ In Hongseong-gun, Chungcheongnam-do, 6 of the 7 soil samples were positive for Cryptosporidium sp., which was the highest positive rate (85.7%) among the areas examined (except for Gyeonggi-do, where only one sample was examined). In the two locations in Jeollabuk-do, Cryptosporidium was not detected. Agarose gel electrophoresis demonstrated the presence of a 242-bp product in all Cryptosporidium-positive samples . The result of TaqI restriction enzyme digestion showed that all of the positive samples fragmented into 117-and 125-bp bands . The C. parvum DNA used as a positive control showed fragmented bands with same pattern as that of the Cryptosporidium-positive soil samples. C. hominis DNA, obtained from human diarrheal stool [bib_ref] Cryptosporidium hominis infection diagnosed by real-time PCR-RFLP, Cheun [/bib_ref] , was not fragmented by the TaqI enzyme . DNA sequence alignment data of the PCR products showed 98%-100% homology with those of C. parvum [fig_ref] Figure 4: Alignment of the qPCR product sequences belonging to the positive soil [/fig_ref] Of the 24 food samples examined, Cryptosporidium sp. was detected from 3 food samples of carrot, winter-grown cabbage, and blueberries (12.5% positive) [fig_ref] Table 3: Positive rates of Cryptosporidium detection in food samples according to the sample... [/fig_ref]. Cryptosporidium sp. was detected in only 1 of 3 samples for each kind of food examined [fig_ref] Table 3: Positive rates of Cryptosporidium detection in food samples according to the sample... [/fig_ref]. The number of oocysts detected from the food samples was 42-110/g of the sample. Agarose gel electrophoresis revealed a 242-bp product from all Cryptosporidium-positive samples , and TaqI restriction enzyme digest revealed that all positive samples had fragmented into 117-and 125-bp bands . These results indicate that the food samples were contaminated with C. parvum. DNA sequence alignment data of the PCR products also supported that the Cryptosporidium sp. detected from the food samples was C. parvum [fig_ref] Figure 4: Alignment of the qPCR product sequences belonging to the positive soil [/fig_ref].
# Discussion
Cryptosporidiosis is one of the most prevalent waterborne parasitic infections (1). Cryptosporidium is a zoonotic pathogen, and the sources of zoonotic contamination are usually feces-contaminated soil and water. In addition, food can mediate human infection by contamination from irrigation water and soil. The recent increase in the demand, global sourcing, and rapid transport of food products, especially of soft fruit and salad vegetables, enhances both the likelihood of surface contamination and the survival of pathogenic parasites in transmissive stages [bib_ref] Emerging parasite zoonoses associated with water and food, Slifko [/bib_ref].
Recently, Dado et al. [bib_ref] Detection of zoonotic intestinal parasites in public parks of Spain. Potential epidemiological..., Dado [/bib_ref] reported that 6% of examined park soils in Madrid, Spain were contaminated with Cryptosporidium sp. and they indicated that children are mainly affected by accidental ingestion of contaminated soil. Farm soil presents an additional source of infection due to microbial pathogen contamination from agricultural runoff contaminated with livestock feces. In the present study, C. parvum was detected in 11 of the 34 (32%) farm soils examined. Animal husbandry in the vicinity of the sampling sites could be an important factor influencing the results of this study, as Cryptosporidium-positive farm areas were located near areas where animal husbandry was practiced. In addition, animal husbandry was not practiced around the farms of Jeollabuk-do, where no oocysts were detected. Cryptosporidium infection rates of cattle, goats, pigs, and livestock in some rural areas of Korea have been reported to range from 10% to 94% in 2004 and 2006 [bib_ref] Infection status of pigs with Cryptosporidium parvum. Korean, Yu [/bib_ref] [bib_ref] Prevalence of cryptosporidiosis among the villagers and domestic animals in several rural..., Yu [/bib_ref] [bib_ref] Genotype analysis of Cryptosporidium spp. prevalent in a rural village in Hwasun-gun,..., Park [/bib_ref] , suggesting that the source of soil contamination with Cryptosporidium oocysts could be livestock feces. Livestock studies on Cryptosporidium infection were performed in different locations in the present study, such as Jeollanam-do and Chungcheongbuk-do. In only one locality (Hongseong-gun, Chungcheongnam-do), it was reported that 1% of pigs was infected with Cryptosporidium in 2004 [bib_ref] Infection status of pigs with Cryptosporidium parvum. Korean, Yu [/bib_ref] , and where 85.7% of soil samples were contaminated in 2012. This difference could be due to the difference in the soil collected from the sample site, indicating several changes in the agricultural environment in the last 10 yr in the same area.
There have been reports on foodborne cryptosporidiosis from the USA, UK, and Nordic countries [bib_ref] Foodborne cryptosporidiosis: is there really more in Nordic countries?, Robertson [/bib_ref]. Cryptosporidium oocysts have been detected in several sources of fruits, salad vegetables, oysters, mussels, water spinach, and zucchini [bib_ref] Cryptosporidium and Giardia as foodborne zoonoses, Smith [/bib_ref] [bib_ref] Faecal and protozoan parasite contamination of water spinach (Ipomoea aquatica) cultivated in..., Vuong [/bib_ref] [bib_ref] Cryptosporidium and Giardia in commercial and non-commercial oysters (Crassostrea gigas) and water..., Schets [/bib_ref] [bib_ref] Cryptosporidium oocysts and giardia cysts on salad products irrigated with contaminated water, Amorós [/bib_ref] [bib_ref] Surface and subsurface irrigation with effluents of different qualities and presence of..., Armon [/bib_ref]. Vegetables such as cilantro leaves, lettuce, carrots, and mung bean sprouts sold in the markets of Costa Rica and Norway have also been reported to be contaminated with Cryptosporidium oocysts [bib_ref] Occurrence of parasites on fruits and vegetables in Norway, Robertson [/bib_ref] [bib_ref] Presence of various pathogenic microorganisms in fresh vegetables in Costa Rica, Monge [/bib_ref]. In the present study, we identified Cryptosporidium sp. from carrot, winter-grown cabbage, and blueberries, and confirmed the species to be C. parvum by using a molecular genetics method. Most of the earlier studies were unable to identify the infectious Cryptosporidium sp. because of the limitations of the conventional detection systems and use of acidfast staining or fluorescent antibody labeling techniques for confirmatory identification. Although the viability and infectivity of the detected oocysts could not be determined in this study, it is necessary to develop a washing manual for food safety practices. A previous report showed that the infective dose (ID50) for Cryptosporidium was as low as 9 oocysts [bib_ref] Virulence of three distinct Cryptosporidium parvum isolates for healthy adults, Okhuysen [/bib_ref].
Presently, there are no commonly agreed-upon guidelines to assess the methods used to detect parasitic protozoa on farm products [bib_ref] Use of a common laboratory glassware detergent improves recovery of Cryptosporidium parvum..., Shields [/bib_ref]. Cook et al. [bib_ref] Towards standard methods for the detection of Cryptosporidium parvum on lettuce and..., Cook [/bib_ref] [bib_ref] Towards standard methods for the detection of Cryptosporidium parvum on lettuce and..., Cook [/bib_ref] conducted extensive studies by examining a number of wash solutions and washing techniques for vegetables. The use of Alconox ® for washing food samples has been reported to significantly increase the recovery efficiencies [bib_ref] Use of a common laboratory glassware detergent improves recovery of Cryptosporidium parvum..., Shields [/bib_ref]. Therefore, in the present study, we combined the methods described by Cook et al. [bib_ref] Towards standard methods for the detection of Cryptosporidium parvum on lettuce and..., Cook [/bib_ref] and Shields et al. [bib_ref] Use of a common laboratory glassware detergent improves recovery of Cryptosporidium parvum..., Shields [/bib_ref] for washing the vegetables.
This study has a limitation of smaller sample size in terms of the numbers of food samples and sample sites for adequately representing the overall status of soil and food contamination in Korea. Nevertheless, this study provides valuable information about soil contamination by Cryptosporidium in Korea for the first time as well as supports the hypothesis that contaminated farm soil could be a source of contamination of farm products. Although there have been no reports of foodborne outbreaks of cryptosporidiosis in Korea yet, the results of this study indicate the potential risk of such an occurrence. Therefore, it is recommended to conduct epidemiological study on a large scale with sufficient soil and food samples on Cryptosporidium contamination in the future.
# Disclosure
[fig] Figure 1: Locations of the surveyed areas. (a) Hwaseong-si, (b) Seosan-si, (c) Hongsunggun, (d) Boryeong-si, (e) Seocheon-gun, (f) Gunsan-si, (g) Buan-gun. [/fig]
[fig] Figure 2, Figure 2, Figure 2: Agarose gel electrophoresis of the qPCR products from soil samples. qPCR was performed to detect Cryptosporidium parvum SWI2/SNF2 ATPase, a Rad16 ortholog gene (GenBank No. XM_625623). (A) qPCR products of soil samples (242 bp). (B) DNA fragments generated from TaqI enzyme digests of the qPCR products. L, 100-bp ladder; ① soils from Hwaseong-si, ② Seosan-si, ③-⑧ Hongseong-gun, ⑨-⑩ Boryeong-si, ⑪ Seocheon-si; Cp, C. parvum DNA; Ch, C. hominis DNA. Agarose gel electrophoresis of the real-time polymerase chain reaction (PCR) products from soil samples. Agarose gel electrophoresis of the real-time polymerase chain reaction (PCR) products from soil samples. [/fig]
[fig] Figure 3, Figure 3: Agarose gel electrophoresis of the qPCR products from food samples. qPCR was performed to detect the C. parvum SWI2/SNF2 ATPase, Rad16 ortholog gene (GenBank No. XM_625623). (A) qPCR products of food samples (242 bp). (B) DNA fragments generated by TaqI enzyme digestion of the qPCR products. L, 100-bp ladder; 1, carrot; 2, winter-grown cabbage; 3, blueberry; Cp, C. parvum DNA. Agarose gel electrophoresis of the real-time polymerase chain reaction (PCR) products from food samples. [/fig]
[fig] Figure 4: Alignment of the qPCR product sequences belonging to the positive soil (A) and food (B) samples generated by using the Clone Manager 7. qPCR was performed to detect C. parvum SWI2/SNF2 ATPase, Rad16 ortholog gene (GenBank No. XM_625623). [/fig]
[table] Table 1: Primers and probes used for qPCR. Target gene was C. parvum SWI2/SNF2 ATPase, Rad16 ortholog gene (GenBank No. XM_625623) ions of surveyed areas. ⓐ Hwaseong-si, ⓑ Seosan-si, ⓒ Hongsung-gun, ⓓ Boryeong-si, ⓔ Seocheonan-si, ⓖ Buan-gun. [/table]
[table] Table 2: Positive rates of Cryptosporidium detection in soil samples according to the sample collection site [/table]
[table] Table 3: Positive rates of Cryptosporidium detection in food samples according to the sample collection sites [/table]
|
The distribution and relative ecological roles of autotrophic and heterotrophic diazotrophs in the McMurdo Dry Valleys, Antarctica
The McMurdo Dry Valleys (MDV) in Antarctica harbor a diverse assemblage of mat-forming diazotrophic cyanobacteria that play a key role in nitrogen cycling. Prior research showed that heterotrophic diazotrophs also make a substantial contribution to nitrogen fixation in MDV. The goals of this study were to survey autotrophic and heterotrophic diazotrophs across the MDV to investigate factors that regulate the distribution and relative ecological roles of each group. Results indicated that diazotrophs were present only in samples with mats, suggesting a metabolic coupling between autotrophic and heterotrophic diazotrophs. Analysis of 16S rRNA and nifH gene sequences also showed that diazotrophs were significantly correlated to the broader bacterial community, while co-occurrence network analysis revealed potential interspecific interactions. Consistent with previous studies, heterotrophic diazotrophs in MDV were diverse, but largely limited to lakes and their outlet streams, or other environments protected from desiccation. Despite the limited distribution, heterotrophic diazotrophs may make a substantial contribution to the nitrogen budget of MDV due to larger surface area and longer residence times of lakes. This work contributes to our understanding of key drivers of bacterial
# Introduction
The biological reduction of dinitrogen (N 2 ) to ammonium by nitrogen-fixing diazotrophs is the primary pathway by which bioavailable nitrogen enters the Earth's biosphere. Diazotrophic species include a diverse group of autotrophic and heterotrophic bacteria found free-living in terrestrial and aquatic environments, as well as in symbiotic association with plants, insects, lichen and phytoplankton (reviewed by. The contribution of nitrogen supplied by diazotrophs is of particular importance in N-limited environments such as deserts (reviewed by. In the desert southwest of the United States, for example, N-fixing cyanobacteria contribute substantial amounts of nitrogen (up to 51 mg N 2 m −2 h −1 ) to the benthic zone of desert streams, while the N-fixing activity of heterotrophic diazotrophs is associated with perennial shrubs in Tehuacán Desert, Mexico, and is thought to support the fertility of soil biota under the shrubs.
One of the most extreme environments on Earth is the McMurdo Dry Valleys (MDV) desert system of Antarctica. The MDV comprise the largest ice-free region on the continent (reviewed by, receiving only 50 to 100 mm of precipitation annually. Soils in the MDV are characterized by <2% water content, along with low organic matter, high salt content and high pH;. In spite of the extreme conditions, the application of cultureindependent methods has revealed a higher than expected level of microbial diversity and biomass in the MDV. Similar to desert ecosystems in temperate regions, ephemeral meltwater streams generated by glacial runoff constitute microbial hotspots of productivity and biomass in the MDV , but differ in the high level of illumination and ultraviolet irradiation received by MDV streams during the austral summer.
Microbial mats, composed largely of filamentous photosynthetic cyanobacteria and often dominated by diazotrophic Nostoc spp., play a key role in biogeochemical cycling within the MDV., for example, demonstrated the contribution of fixed nitrogen derived from diazotrophic cyanobacteria to downstream microbial communities in MDV streams. In addition to providing pond and stream ecosystems with biologically accessible carbon (C) and N during the austral summer, it has been hypothesized that aeolian transport of cyanobacteria mat detritus is a major source of nutrients to the oligotrophic arid soils of the MDV .
The highly conserved nifH gene, which encodes the Feprotein reductase component of the nitrogenase enzyme complex, is a commonly used marker in molecular investigations of diazotroph diversity (see reviews by. This gene shares a rough concordance with rRNA gene sequence phylogeny across several prokaryotic phyla, allowing for resolution of taxonomic identity from nifH sequences. NifH gene sequences have been used as a proxy for diazotroph diversity in a wide range of aquatic (e.g.and terrestrial environments (e.g.. While research on diazotroph diversity and activity in Antarctica has largely focused on cyanobacteria species (e.g., the relative importance of heterotrophic diazotrophs has also been noted. NifH sequence analysis of hypolithic communities from the MDV, for example, revealed that all putative diazotrophs were Proteobacteria, in spite of the dominance of non-diazotrophic cyanobacteria species in the assemblage. In another recent study, nifH sequence libraries from sediments collected in Miers Valley in the MDV bydemonstrated the presence of highly diverse assemblages of heterotrophic diazotrophs along with cyanobacteria. Results ofshowed that heterotrophic N-fixing species contributed up to 50% of total nitrogen fixed, indicating that heterotrophic diazotrophs may constitute a significant source of fixed N to the MDV ecosystem.
The relative ecological roles of autotrophs and heterotrophs within the MDV may have important implications for understanding C and N cycling in this system; although both contribute N to the system, cyanobacteria are a source of fixed carbon while heterotrophs are a sink. The contribution of each group of microbes is likely dynamic, as light and temperature regimes change seasonally or with anticipated warming due to climate variability. Ecosystem-level disturbance may also lead to the loss of key species and instability of diazotroph cooccurrence patterns, affecting nutrient cycling and ecological networks of microorganisms in the MDV that benefit from their association with diazotrophs.
The overall goals of the current study were to investigate the diversity, composition and relative distributions of heterotrophic and autotrophic N-fixing bacterial assemblages in the MDV, as well as potential relationships between diazotrophic bacteria (based on nifH sequence data) and the broader microbial community (based on 16S rRNA gene sequence data). These relationships were investigated using multivariate statistical methods, both at the whole community level and for subsets of the community. Positive correlations in relative abundance were then inferred from co-occurrence network analysis to identify potential interspecific relationships between diazotrophs and the other members of the microbial community.
# Materials and methods
Sixty-nine surface sediment samples were collected from 23 sites in the MDV over two field seasons. With some exceptions noted below, each site was sampled along a transect, with samples collected from submerged sediments at edges of ponds/lakes or streams (designated by site number.1), in the wetted region adjacent to water bodies (designated by Water samples for pH were collected from the overlying water in the submerged regions and from pore water in submerged and wetted regions, and filter-sterilized through 0.2 μm polycarbonate (PC) filters. Samples were stored cold in polypropylene centrifuge tubes, and the pH was later measured with a Hanna Instruments model 2213 pH/ORP meter (Hanna Instruments, Smithfield, RI).
Three replicate 1 cm 3 sediment cores were collected for chlorophyll-a determination. Cored samples were transferred into Whirl-Pak bags in the field and later transferred to 50 mL conical centrifuge tubes and stored frozen in the dark until analysis. Chlorophyll-a was extracted over 24 h at −20 - C using 90% acetone, and then measured on a Turner 10AU fluorometer and corrected for phaeophytin.
Water samples were collected from surface water where available and from pore water at submerged and wetted sites for dissolved nutrient as well as dissolved organic and inorganic carbon analyses. Pore water was obtained by driving a 50 mL conical centrifuge tube, modified with several drill holes, into the sediments and allowing pore water to fill the tube. Water was extracted from the tube by inserting acid-washed silicone tubing into the centrifuge tube, fitted with a 50 cc barrel syringe that allowed collection of the water. Pore water was first collected in 250 mL high-density polyethylene bottles and transported back to camp for further processing. Pore water was filtered through 25 mm Whatman GF/F filters and subsampled for nutrient concentrations: nitrate + nitrite (NO 3 + NO 2 ), orthophosphate (PO 4 ), silicate [Si(OH) 4 ] and ammonium (NH 4 ), and dissolved organic carbon (DOC). All subsamples were stored frozen.
Nutrient concentrations were measured on a Bran and Luebbe AutoAnalyzer II with MT-19 manifold chemistry module for NO 3 and NO 2 , PO 4 (Bran and Luebbe Method G-175-96) and Si(OH) 4 (Bran and Luebbe Method G-177-96) . NH 4 was analyzed according to. Pore water subsamples for DOC were acidified prior to analysis using a Shimadzu TOC-V high-temperature combustion instrument. Other soil characteristics including % moisture, conductivity and elemental analysis were determined as previously described by.
## Dna extraction
Composite samples were subsampled for DNA extraction using the Mo Bio Powersoil DNA Isolation Kit (Mo Bio Laboratories, Carlsbad, CA). Replicate samples were extracted from sample 14.1 (designated 14.1B and 14.1C) collected in Garwood Valley to evaluate representativeness of subsamples. Extracted DNA was stored on dry ice in the field until return to Crary Lab at McMurdo Station, after which samples were stored at −80 - C.
## Nifh amplification
Samples were amplified by polymerase chain reaction (PCR) using primers targeting conserved regions in the nifH gene to identify communities harboring diazotrophs. PCR reac- were incubated at 94 - C for 2 min, and amplified for 10 cycles of 94 - C for 45 s, 60 - C for 45 s, decreasing 0.5 - C per cycle, and 72 - C for 1.5 min, followed by 25 cycles of 94 - C for 45 s, 51 - C for 45 s and 72 - C for 1.5 min.
Samples from the submerged and wetted regions at Sites 1-3 and 7-12, collected in Taylor Valley in 2014, were then amplified in nested PCR reactions for nifH amplicon sequencing by Ion Torrent (Thermo Fisher Scientific). A nested PCR reaction was necessary to minimize amplification of chlorophyllide reductase gene sequences, Group V nitrogenase homologs. The initial PCR reaction consisted of 2 μL template DNA (5-10 ng), 1× polymerase buffer, 0.2 mM dNTPs, 2.5 mM Purified PCR products were then precipitated with ethanol and sequenced using an Ion Torrent PGM DNA sequencer (Ther-moFisher) at the University of Waikato DNA Sequencing Facility (Hamilton, New Zealand). The resulting sequencing reads were processed using mothur v.1.35.1to discard all reads shorter than 300 bp (the anticipated amplicon size was around 395 bp), all reads longer than the 97.5th percentile and all reads with homopolymers more than the 97.5th percentile. All reads with mismatches to the forward PCR primer or with expected error rates higher than 1% (-fastq maxee rate 0.01) were discarded using USEARCH, and all reads were truncated to 270 bp. The resulting reads were de-replicated, singleton reads found only once across all samples were discarded and the remaining reads were clustered at 95% identity.
# Phylogenetic analysis
The translated amino acid sequences of 46 nifH operational taxonomic units (OTUs) representing sequences that were present at greater than 1% of the nifH library were subjected to phylogenetic analysis. Amino acid sequences were aligned in Geneious v.10.2.2 (http://www.geneious.com/) along with nifH sequences for relevant taxa and closely related environmental isolates in GenBank. The alignment was imported into MEGA v.7for phylogenetic analysis using the neighbor-joining method. The evolutionary distances were computed using the p-distance method and the rate variation among sites was modeled with a gamma distribution.
## 16s rrna gene sequence analysis
DNA was submitted to MR DNA Laboratory (Shallowater, TX) for 16S rRNA gene sequencing. DNA was quality-checked and the V1-V3 region of the 16S rRNA gene was amplified using Eubacterial primers ill27Fmod (AGRGTTTGATCMTGGCTCAG;and ill519Rmod (5 -GTNTTACNGCGGCKGCTG-3 ), modified with barcode sequences for paired-end sequence analysis (2 × 300) on the Illumina MiSeq platform (Illumina, Inc., San Diego, CA). Paired-end reads exceeding the Illumina Q30 quality standard were joined and chimeras removed before further analysis using the QIIME v.1.9.1pipeline. Sequences were clustered into OTUs at 97% similarity and taxonomy assigned using the pick open reference otus.py script. Chloroplast sequences and 16S rRNA OTUs containing <10 sequences across all 69 samples were removed. Three groups of OTU libraries were then generated for analysis: (i) No chlor containing all remaining 16S rRNA OTUs, (ii) No cyano OTUs in which cyanobacteria sequences were removed and (iii) Cyano only OTUs that contained only cyanobacteria sequences. The No chlor OTU libraries were rarefied to a sampling depth of 33 531 and diversity analysis was conducted on rarefied libraries using the core diversity analyses.py script in QIIME v.1.9.1.
# Multivariate statistical analysis
Multivariate statistical analysis was conducted in PRIMER v.7. OTU libraries were standardized and square root transformed before computing Bray-Curtis dissimilarity matrices. Cluster analysis was used to generate dendrograms showing similarities of nifH OTU abundances representing greater than 1% of the nifH library and 16S rRNA gene OTUs. Shade plots were then constructed for visualization of OTU abundances of the nifH library and class-level distribution of taxa based on 16S rRNA gene. The Spearman rank correlation method in the RELATE function in PRIMER was used to test the null hypothesis that there was no correlation between multivariate patterns generated by the No cyano and Cyano only groups for the 16S rRNA gene or nifH OTU abundance data. The analysis of similarity (ANOSIM) function in PRIMER was applied to 16S rRNA gene OTU abundance data to test the null hypotheses that (i) there were no differences in assemblages for samples collected in submerged versus wetted versus dry regions, (ii) there were no differences in assemblages for samples collected from different valleys, (iii) there were no differences in assemblages for samples collected in areas with cryptogamic mats compared to areas with no visible mat present and (iv) there were no differences in assemblages for samples that were positive for nifH compared to those that were negative for nifH. Permutation tests randomly resampled data 999 times.
Environmental data were fourth root transformed and normalized. Principal component analysis (PCA) was conducted using PRIMER v.7to identify environmental variables in submerged and wetted regions that contributed to the variability between sites. The Biota-Environment STepwise (BEST) matching function in PRIMER was used to identify environmental variables that were correlated to changes in microbial community structure based on both 16S rRNA gene and nifH OTU abundances for submerged and wetted regions.
Null hypotheses to test correlations between a set of environmental variables and the assemblage were then examined with BEST using permutation tests that randomly resampled data 99 times.
## Network inference
Correlations between taxon relative abundances in the 16S rRNA gene library and nifH sequences were investigated using CoNet. The 16S rRNA gene library included samples from sites that were sequenced for nifH and for which taxonomy was assigned using the pick closed reference otus.py program in QIIME at 97% similarity based on Greengenes 13 5 16S rRNA gene database. 16S rRNA OTUs with fewer than 100 counts across all samples were excluded from analysis, leaving 598 taxa. The nifH library included the 46 most highly abundant sequences as described above. Sequences that were present at only one site were excluded from analysis. A distribution of pairwise scores was calculated for each of the following similarity measures: Bray-Curtis dissimilarity, Kullbacker-Leibler dissimilarity and Pearson and Spearman correlations. Initial thresholds were selected so that each of these measures contributed 1000 positive edges and 1000 negative edges to the network. The initial network was then used to generate 1000 renormalized permutations and 1000 bootstrap scores and a P-value calculated for each similarity measure. P-values were merged taking into account any correlations between similarity measures . Multiple-test corrections were conducted using Benjamini-Hochberg's methodsuch that any edges with merged P-values below 0.05 were retained. The inferred network including only positive correlations was visualized in Cytoscape v. 3.6.1.
# Results
## Environmental conditions
Environmental context is provided for submerged, wetted and dry soil habitats through measurements of soil moisture, conductivity, chlorophyll-a and inorganic and organic nutrients summarized in(Supporting Information). Not surprisingly, soil moisture was highest in soils from submerged sites (13-75%), followed by wetted soils (13-57%) and finally 'dry' soil habitats (2-18%). Chlorophyll-a concentrations were similar for submerged and wetted habitats (range 0.0023-1.32 ug/g), and roughly an order of magnitude higher than found in dry soils. Pore water pH and nutrients were collected at submerged and wetted habitats. pH was more variable at submerged habitats (5.90-7.43) compared to wetted soil habitats (6.16-7.41), with average pore water pH slightly acidic in both habitats. At submerged habitat sites, pore water NO x , PO 4 , SiO 2 and DOC were roughly 2-fold higher than the overlying water. Comparisons between pore water nutrients at submerged versus wetted sites suggest that wetted sites were elevated (∼2-fold) for NO x and SiO 2 but more similar to submerged sites for PO 4 . Average NH 4 was >10-fold higher in wetted pore water compared to submerged soil habitats, although much of this difference was due to a single site (Site 16, data not shown). PCA analysis of environmental attributes for submerged and wetted samples showed that 47.3% of the variability between samples could be represented on PC1 and PC2, Supporting Information). Pore water pH, conductivity and NO x concentrations contributed the most to variability between samples. and this study may be expected to result in differential amplification of nifH. In addition, nifH sequences reported inwere isolated from Miers Valley, Antarctica, while those identified in this study were all from Taylor Valley. In spite of the difference in primers and distance between Miers Valley and Taylor Valley, the nifH libraries sequenced in this study included eleven OTUs that shared 99% or greater similarity with nifH OTUs detected in , with four OTUs identical between the two libraries.
Of the 102 nifH OTUs, 31 were most closely related to cyanobacteria nifH sequences and represented 0.26% (sample 1.1) to 99.96% (sample 12.2) of the nifH libraries. Phylogenetic analysis of the 46 most highly abundant amino acid sequences revealed the presence of a diverse community of heterotrophic and autotrophic diazotrophs. Two OTUs had <95% similarity to nifH amino acid sequences from known diazotrophs and are labeled as 'unclassified' . The . Phylogenetic analysis of deduced heterotrophic (A) and cyanobacterial (B) nifH amino acid sequences. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (100 replicates) are shown next to the branches. The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree. The evolutionary distances were computed using the p-distance method and are in the units of the number of amino acid differences per site. The rate variation among sites was modeled with a gamma distribution (shape parameter = 1). All positions with <75% site coverage were eliminated. That is, fewer than 25% alignment gaps, missing data and ambiguous bases were allowed at any position. There were a total of 89 positions in the final dataset. Evolutionary analyses were conducted in MEGA7remaining nifH sequences clustered with heterotrophic bacteria within the Bacteroidetes and Proteobacteria .
ANOSIM analysis of nifH sequence distribution revealed no significant differences in submerged versus wetted assemblages. Cluster analysis of samples based on nifH OTUsrevealed a topology loosely correlated to sample location. The submerged sample from Spaulding Pond (sample 1.1), for example, was distinct from stream-collected samples. The nifH assemblages at Sites 9 and 12 from Green Creek and the base of Commonwealth Glacier on the north side of Taylor Valley were more similar to each other than they were to samples collected from the south side of the valley near Howard and Crescent Glaciers (Sites 2, 3, 7 and 11). The nifH assemblage in sample 2.1, located just downstream of Spalding Pond, was most closely related to sample 1.2, at the edge of the pond. Conversely, the nifH assemblage in samples collected at the wetted region of Site 2 (sample 2.2) was most similar to samples collected at Explorer's Cove (Site 10), possibly related to the higher diversity of Cyanobacteria nifH OTUs from these samples.
The nifH sequence assemblages were not significantly correlated to environmental variables (P = 0.31). When evaluated separately, however, there was a significant correlation between cyanobacteria nifH assemblages and environmental variables (Rho = 0.398; P = 0.03), with pore water pH, pore water magnesium, pore water potassium and pore water DOC concentrations contributing most to the relationship. There was no significant correlation between the heterotrophic nifH assemblage and environmental variables.
## 16s rrna gene sequence analysis
Altogether, 69 libraries of 16S rRNA gene amplicon sequences from 23 sites were generated. After removal of sequences identified as chloroplast 16S rRNA gene, the number of remaining sequences ranged from 33531 to 167085 (median = 97430;per sample. The number of observed 16S rRNA gene OTUs generated with a 97% similarity cutoff ranged from 1192 to 8140 per sample. The number of 16S rRNA OTUs in submerged and wetted regions was significantly greater than the number of OTUs in dry regions; P < 0.05). After rarefaction to 33531, Chao1 diversity was greatest in wetted regions and was significantly higher in both submerged and wetted regions compared to dry regions ( P < 0.05;. Few 16S rRNA gene OTUs were shared between all libraries and, Supporting Information). Samples collected from submerged regions shared 134 OTUs, with 34 OTUs unique to those sites, including species within Cyanobacteria, Acidobacteria, Actinobacteria, Bacteroidetes, Chloroflexi, Nitrospirae, Proteobacteria and Verrucomicrobia phyla. Wetted samples shared 164 OTUs, of which 58 were unique to wetted libraries. Unique 16S rRNA gene OTUs in this region included species from the same phyla as submerged regions, as well as species within the Planctomycetes phylum. Samples from dry regions of transects shared the fewest OTUs (77), with 22 OTUs unique to this region, including species within Acidobacteria, Actinobacteria, Chloroflexi, Gemmatimonadetes and Deinococcus-Thermus phyla.
# Multivariate statistical analysis
ANOSIM analysis of 16S rRNA gene library revealed a significant difference in bacterial assemblages based on location within a transect (submerged, wetted and dry; R = 0.341; P < 0.05), and between sites with visible mats versus samples with no visible mat (R = 0.470 and R = 0.362, respectively; P < 0.05). Bacterial communities collected from Taylor Valley versus Garwood Valley were also significantly different (R = 0.289; P < 0.05).
Cluster analysis of 16S rRNA gene libraries revealed further relationships in community structure loosely based on sample type (submerged, wetted or dry), location and field season, see Clusters I-XII). Notably, bacterial assemblages from samples that were also positive for nifH clustered together, while samples that were negative for nifH formed distinct clusters independent of position within the transect. Samples collected from submerged and wetted regions of transects that were negative for nifH (Sites 4, 5, 6, 13 and 17; Clusters III, X, XI and XII), for example, were more highly correlated to each other and/or to samples collected from dry regions than to samples that were positive for nifH. Samples collected from ponds or lakes (Sites 1, 16, 21 and 23) formed two distinct clusters (Clusters I and II). Samples collected in Victoria Valley (Sites 5 and 6) formed Cluster III, while Clusters VI and VII collected on the south side of Taylor Valley (Sites 2, 3, 7 and 11) were distinct from bacterial assemblages collected on the north side of the Valley (Sites 9 and 12; Cluster VIII).
A shade plot depicting the relative abundance of selected bacterial classes revealed the basis for patterns observed in the cluster diagram. Bacterial communities from dry samples generally had a higher relative abundance of Actinobacteria, Acidobacteria and Gemmatamonidetes, while samples collected from submerged or wetted regions generally had higher relative abundances of Cyanobacteria and/or Proteobacteria, as well as specific classes of bacteria within the Chloroflexi, Planctomycetes and Bacteroidetes. Samples collected from dry regions during the 2015 field season had higher relative abundances of Cyanobacteria, Planctomycetes and Verrucomicrobia compared to samples collected in 2014. In addition, Cluster I with samples collected from submerged and wetted regions of lakes or ponds had a high relative abundance of Bacteroidia and Anaerolineae compared to samples from streams or dry regions.
ANOSIM analysis of the heterotrophic community alone (No cyanos) or the cyanobacterial community alone (Cyanos only) revealed significant differences in bacterial assemblages based on location within the transect (R = 0.352 and 0.166, respectively; P < 0.05). When evaluated by region within a transect, the heterotrophic bacterial population was highly and significantly correlated to the cyanobacterial population (Rho = 0.736; P < 0.05). Furthermore, ANOSIM analysis of the 16S rRNA gene sequence libraries revealed significant differences between assemblages collected from submerged and wetted regions that were positive for nifH versus those that were negative (R = 0.434; P = 0.001).
BEST analysis of submerged and wetted samples indicated no significant relationship between bacterial assemblages and environmental variables. Analysis of specific subsets of bacterial assemblages (i.e. based on year, location within a transect, Only cyano and No cyanos groups) also demonstrated no significant correlation with environmental variables measured here.
## Comparisons between nifh and 16s rrna gene community structure
The community structure of nifH OTUs was highly and significantly correlated to the bacterial assemblages based on 16S rRNA gene sequences (Rho = 0.614; P = 0.01). This relationship was primarily driven by nifH OTUs within the cyanobacteria clade. When divided into cyanobacteria and heterotrophic nifH libraries, the structure of cyanobacteria nifH library was significantly correlated to community structure of the entire 16S rRNA gene library (Rho = 0.403; P = 0.001), as well as to the Only cyanos (Rho = 0.474; P = 0.001) and the No cyanos (Rho = 0.391; P = 0.001) 16S rRNA gene libraries. There was no significant correlation between heterotrophic nifH and 16S rRNA gene assemblages.
# Network analysis
Network analysis identified 41 nifH OTUs with significant positive correlations to 351 16S rRNA gene OTUs in 18 samples ; , Supporting Information). Of these, 22 heterotrophic nifH OTUs and 19 cyanobacteria nifH OTUs were positively correlated to 136 and 220 16S rRNA gene OTUs, respectively , Supporting Information). Cyanobacterial nifH OTUs represented in the network were distributed in up to 18 samples (indicated as counts; , Supporting Information), and were significantly correlated with as few as 3 and up to 44 16S rRNA gene OTU abundances (indicated as degrees; , Supporting Information). Heterotrophic nifH OTUs were significantly correlated with 2 to 39 16S rRNA gene OTUs. It might be expected that sample counts were inversely related to the degree. That is, samples with restricted distribution may be randomly correlated to a larger number of 16S rRNA gene sequences, but there was no correlation between the count and degree for nifH OTUs (R 2 = 0.0003).
Five groups or clusters within the network were labeled A-E on , based on shared 16S rRNA gene nodes (triangles) and/or the relative positions of heterotrophic versus cyanobacteria nifH nodes (circles) in the network. Cluster A included both heterotrophic and cyanobacteria nifH OTU nodes, but there were few 16S rRNA gene OTUs shared between the autotrophic and heterotrophic diazotrophs in this cluster. Cyanobacterial nifH OTUs in Cluster A were broadly distributed with generally higher relative abundances at Lower Delta Stream (Site 8) and at Green Creek and the base of Commonwealth Glacier (Sites 9 and 12) on the north side of Taylor Valley. The 16S rRNA gene OTUs that correlated with cyanobacterial nifH OTUs in Cluster A were primarily within the Bacteroidetes, Cyanobacteria (Nostocaceae) and Proteobacteria .
Two heterotrophic nifH nodes in Cluster A shown onrepresented OTUs (R22 5253 and R22 50498) from sulfate-reducing Deltaproteobacteria Group I within the Desulfovibrio and Desulfobulbus genera; , Supporting Information), found at Spaulding Pond (Site 1), upper Delta Stream (Site 2) and Explorer's Cove (Site 10). Cluster A also included the only Alphaproteobacteria nifH node in the network (OTU Rn04 89), found primarily at Explorer's Cove (Site 10) and the sites near Howard Glacier. The 16S rRNA gene OTUs that correlated with heterotrophic nifH OTUs in Cluster A were dominated by Bacteroidetes and Alphaproteobacteria , Supporting Information). Several Chlorobi and Planctomycetes 16S rRNA gene OTUs were also significantly correlated with heterotrophic nifH OTUs in Cluster A. Notably absent were cyanobacterial 16S rRNA gene OTUs correlated to heterotrophic nifH OTUs in this cluster, although the heterotrophic nifH nodes shared several 16S rRNA gene nodes with cyanobacterial nifH nodes in Cluster B , Panels 2 and 3).
Cluster B shown in included only cyanobacterial nifH OTUs , Supporting Information). These OTUs were broadly distributed with generally lower relative abundance at Lower Delta Stream (Site 8), and on the north side of Taylor Valley at Green Creek (Site 9) and the base of Commonwealth Glacier (Site 12). 16S rRNA gene OTUs that correlated with cyanobacterial nifH OTUs in Cluster B were dominated by Proteobacteria. Cyanobacterial nifH OTUs with higher abundance at the base of Howard Glacier (Sites 3 and 7) and Crescent Stream (Site 11) were significantly correlated with several Acidobacterial 16S rRNA gene OTUs, while those that were found in higher abundance in Upper Delta Stream (Site 2) and Explorer's Cove (Site 10) were correlated with Actinobacterial 16S rRNA gene OTUs.
Cluster C included a single cyanobacteria nifH node found primarily at Green Creek (Site 9) and Crescent Stream (Site 11), and was positively correlated with a number of Acidobacteria 16S rRNA gene OTUs within the photosynthetic Chloracidobacteria genus. Notably, no Actinobacteria 16S rRNA gene OTUs were significantly correlated with this nifH node.
Cluster D shown inincluded a large number of heterotrophic nifH nodes within the Betaproteobacteria, Gammaproteobacteria and Group 2 Deltaproteobacteria, as well as a number of Bacteroidetes nifH and an unclassified nifH node, representing sequences that were broadly distributed across sites. There were no cyanobacterial nifH nodes in this cluster. NifH OTUs in this cluster were positively correlated with 16S rRNA gene OTUs within the Bacteroidetes and Proteobacteria, primarily Betaproteobacteria. There were no Acidobacteria 16S rRNA gene nodes and few Cyanobacteria 16S rRNA gene nodes in Cluster D. Those cyanobacterial 16S rRNA gene OTUs that were positively correlated with heterotrophic nifH nodes in this Cluster were phylogenetically associated with Leptolyngbya, Pseudoanabaena and Phormidium genera.
Cluster E included heterotrophic nifH OTUs within the Gammaproteobacteria (R21 2604) and Betaproteobacteria (R21 4204), as well as an unclassified nifH (R21 3689) from Spaulding Pond (Site 1) and from the base of Howard Glacier . Network analysis illustrating significant correlations between nifH and 16S rRNA OTU abundance for 18 sites included in the nifH library. Node color indicates phylum except for Proteobacteria, which are identified to the class level. nifH (circles) and 16S rRNA OTUs (triangles) are connected by edges, where the edge thickness is proportional to the significance of the correlation (merged P-value) and the sizes of the nifH nodes are proportional to the number of edges connecting them. Panel 1: all nodes; Panel 2: Cyanobacteria nifH and associated 16S rRNA nodes only; Panel 3: heterotrophic nifH and associated 16S rRNA nodes only. A, B, C, D and E are for orientation and refer to groups of nodes described in greater detail in the text.
(Site 3). None of the 16S rRNA gene nodes were shared with other clusters. Acidobacteria, Actinobacteria, Chloroflexi and Gemmatimonadetes 16S rRNA OTUs dominated this cluster, while cyanobacterial 16S rRNA gene OTUs were not present.
# Discussion
Nitrogen input to MDV sediments is typically low, driven by atmospheric deposition of nitrate or through biological nitrogen fixation of nitrogen gas to ammonia (reviewed by Van Goethem and Cowan 2019). Research efforts have primarily focused on Nfixation by autotrophic cyanobacteria, primarily Nostoc spp., and the significance of cyanobacteria-derived N to the nutrient budget and ecology of microbial communities in MDV ecosystems; reviewed by Van Goethem and Cowan 2019). A study by, however, demonstrated the sizeable contribution of heterotrophic bacteria to N-fixation in sediment samples collected from ponds and outlet streams in Miers Valley, Antarctica, suggesting that nitrogen input through biological N-fixation to this desert environment may be substantially greater than expected based on estimates from cyanobacteria biomass alone. The current study extends this work to examine the distribution of heterotrophic diazotrophs across several ecosystems (submerged or wetted regions of streams, ponds or lakes, and cyanobacteria-or bryophyte-dominated mats versus no visible mat) of the MDV and to investigate relationships between diazotrophs and the broader microbial community.
Detection of the nifH gene by PCR amplification was restricted to samples with visible cyanobacteria or cryptogamic mats. Previous studies suggested that cyanobacteria are primary colonizers of sediments and soils (reviewed by, including polar desert sediments that have been newly exposed during deglaciationor in response to an induced wetting event. As 'ecosystem engineers', cyanobacteria may facilitate recruitment and colonization of heterotrophic bacteria by improving soil structure and providing a labile pool of carbon and nitrogen to oligotrophic environments. Here, samples collected from ephemeral streams produced during summer months from glacial runoff likely represent early stages of community development. These sites were dominated by cyanobacteria nifH sequences, contributing up to 99.96% of sequences obtained within a site.
Cyanobacteria nifH OTUs were exclusively assigned to the Nostocales . The absence of nifH sequences from Oscillatoriales, including Lyngbya, Phormidium and Plectonema spp., was surprising given the presence and dominance of this species complex at several sites. Greater than 24% of the 16S rRNA gene OTUs from the surface sample collected at the base of Howard Glacier (sample 3.1), for example, were identified as Phormidium. The absence of nitrogenase from Phormidium was noted in a previous study of nifH sequence diversity and nitrogenase activity in meltwater ponds of the McMurdo Ice Shelf, where Oscillatoriales dominated the cyanobacteria assemblage. These results suggest that Phormidium present in samples collected in the MDV is a non-N-fixing species or strain (e.g., or has accumulated mutations in the nifH gene that impact amplification with the primers used in this and other studies.
Cyanobacteria nifH OTU RM 252 was the only OTU detected at all sites in Taylor Valley that were positive for nifH. This OTU was closely related to Nostoc sp. cyanobiont of the lichen, Peltigera membranacea and shared 100% identity with OTU43 inas well as with a nifH sequence isolated from the Tibetan Plateau (GenBank Accession No. KR819439.1). Other putative cyanobionts detected in this investigation included nifH OTUs RM 5554 and R22 140654 based on phylogenetic analysis .demonstrated that the most abundant Peltigera cyanobionts collected from Antarctic soils were also the most abundant phylotypes in substrate from the same collection site, suggesting that the substrate included a high abundance of lichen propagules or fragments, and/or that free-living Nostoc sp. was selected for lichenization. Both RM 252 and RM 5554 nifH OTUs were broadly distributed in samples collected in the study described here, and dominated the nifH library at some sites. The broad distribution and high relative abundance of these nifH OTUs suggest that putative cyanobionts may be key constituents of diazotroph assemblages in the MDV while also contributing to the colonization success and diversity of lichens in MDV.
Other nitrogen-fixing cyanobacteria were more limited in their distribution and may have been constrained by environmental filtering.showed that soil pH was the strongest driver of diazotrophic community structure across desert and non-desert environments in Mongolia, China, while others (e.g.also observed correlations in pH or conductivity with the relative abundance and diversity of microbial taxa in the MDV, Antarctica. In results presented here, the diazotroph community structure was significantly correlated to pore water pH. OTU RM 4765, for example, was abundant in nifH libraries of samples collected on the north side of Taylor Valley at the base of Commonwealth Glacier (Site 12) and from Green Creek (Site 9), where it made up >50% of the nifH library in the sample 9.1. These sites were characterized by the highest pH and lowest conductivity levels of samples with positive amplification of nifH. Phylogenetic analysis of OTU RM 4765 shows a close relationship to nifH from Anabaena cylindrica, a freshwater cyanobacteria species that is primarily found at neutral to basic pH and has a low tolerance to salinity. Likewise, cyanobacteria nifH OTU R23 18734 was detected only at the base of Howard Glacier, where the lowest pore water pH (5.9) and highest PO 4 concentrations were measured. While conductivity was not significantly correlated to the diazotroph community structure, cyanobacteria nifH OTU R23 38659, most closely related to the euryhaline genus, Rivularia, was found primarily in Lower Delta Stream (Site 8) and Explorer's Cove (Site 10) with high conductivity levels ranging from 400 to >1600 μS/cm.
Although cyanobacteria nifH dominated at most sites, a larger number of heterotrophic nifH sequences were retrieved during this investigation. Of the 102 nifH OTUs, 71 were phylogenetically related to heterotrophic species. Heterotrophic nifH gene sequences were detected in samples collected from sites with visible mat, suggesting that heterotrophic members of the diazotroph community were tightly coupled with photosynthetic mat-forming species. Cyanobacteria provide structure and produce extracellular polymeric substances (EPS) that help to stabilize sediments and reduce desiccation, providing a favorable environment for colonization of heterotrophic species. Nitrogen fixation is also energetically costly, and heterotrophic species may rely on cyanobacteria as an external C source. , for example, demonstrated a 10-fold increase in activity and a doubling of heterotrophic diazotrophs after addition of EPS to water sample incubations. In addition to supplying carbon to the heterotrophic community, the anaerobic environment associated with cyanobacteria-produced EPS may provide protection for the nitrogenase complex of heterotrophic species .
The abundances of heterotrophic nifH OTUs relative to cyanobacteria were greatest at a permanent water body, Spaulding Pond (Site 1), as well as in the outflow stream, Upper Delta Stream (Site 2), where they contributed 99.7% (Site 1) and 50.8% (Site 2) of nifH sequences at those sites. The diazotroph assemblage at Spaulding Pond may be representative of a more established community, where development of a stable anaerobic environment served to structure the heterotrophic community. Several heterotrophic nifH OTUs at this site included sequences closely related to nifH homologs from obligate anaerobes within the Deltaproteobacteria Geobacter sp. (e.g., several of which share 99% identity with nifH OTUs reported in. Results also provide evidence that the heterotrophic diazotroph community at Upper Delta Stream was seeded by outflow from Spaulding Pond, as many of the shared sequences in wetted and submerged samples of Sites 1 and 2, as well as Site 3 (inlet stream to Spaulding Pond from base of Howard Glacier), but rarely occurred at other sites.
Submerged sediments at Explorer's Cove (Site 10) also included a diverse assemblage of heterotrophic nifH sequences. This site was in close proximity to McMurdo Sound and was characterized by a thick layer of moss, representing a late successional biocrust stage. The increased complexity of this mat environment may have supported the formation of anoxic microniches that allowed for recruitment over time of anaerobic diazotrophs, similar to Site 1 at Spaulding Pond. Indeed, diazotrophic anaerobes classified as Bacteroidetes together with Deltaproteobacteria Groups I and II made up the largest proportion of heterotrophic diazotrophs at Site 10, while aerobic species within the Beta and Gammaproteobacteria clades were not detected among heterotrophic diazotrophs at this site. NifH OTU R22 65896, phylogenetically related to nitrogenase in the Purple Non-Sulfur Bacteria group, was detected only at this site. Purple Non-Sulfur Bacteria are facultative or obligate phototrophs that typically occupy anaerobic environments. This OTU was most similar to sequences from Rhodopseudomonas palustris, a common member of marine coastal sediment microbial communities and widely known for its metabolic versatility (reviewed by. As a member of the Alphaproteobacteria, its phylogenetic position in supports the hypothesis that the nifH gene in this species was acquired through lateral gene transfer.
The diversity of heterotrophic diazotrophs in stable water features of the MDV supports results of, which noted the large contribution of this physiological group to N-fixation in samples collected from lakes and outlet streams of Miers Valley. Although many of the nifH sequences reported inare shared in this study , there was a notable absence of nifH sequences from Bacteroides in. Prior investigationsidentified a high level of intervalley heterogeneity in bacterial taxa from MDV lakes, and specifically Bacteroides, suggesting potential endemism within the bacterial community. This is further supported here by the lack of Bacteroides 16S rRNA gene from samples collected at Lake Miers (sample 21.1;as well as Victoria Valley (Sites 5 and 6). In contrast, diazotrophs within the Bacteroides genus and other member of the Bacteriodales were broadly distributed in Taylor Valleyand dominated nifH OTUs from submerged samples at Spaulding Pond. This may have important consequences with respect to nutrient cycling in MDV ecosystems, as Bacteroides are capable of metabolizing cellulosic biomass, providing a conduit of accessible carbon and fixed nitrogen to other members of the community.
Analysis of the microbial community structure based on 16S rRNA gene profiles showed that the heterotrophic community was highly correlated to cyanobacterial community structure, providing further evidence that bacterial community structure and diversity in the MDV are controlled in part by the presence and relative abundance of cyanobacteria. Similar to nifH OTU structure, 16S rRNA gene OTUs representing the broader microbial community appeared to be loosely structured in the MDV according to location, with communities isolated from ponds and lakes forming distinct clusters, and samples collected on the north side of Taylor Valley different from those on the south side. Although the community structure based on 16S rRNA gene was not correlated to environmental variables, there was a significant correlation between 16S rRNA gene and nifH community structure. Likewise, samples that were negative for nifH by PCR were more closely related to each other regardless of location within a transect (dry versus submerged or wetted), suggesting that the presence of diazotrophs may have played a larger role in structuring the bacterial community than the presence of water.
Network analysis revealed additional interspecific relationships between diazotrophs and members of the larger microbial community. Networks involving cyanobacteria members of the diazotroph community in Clusters A, B and C showed significant correlations to other members of the community, ranging from a minimum of 2 (for R22 116588) to a maximum of 44 (for R22 53758) first neighbors (median = 9 first neighbors; , Supporting Information). The ubiquitous putative lichen cyanobiont represented by nifH OTU 252 ; , Supporting Information) was correlated to only three 16S rRNA gene OTUs in the network, in spite of its broad distribution in the MDV. Network analysis presented inalso resulted in few correlations between the symbiotic diazotroph, Bradyrhizobium, and other members of the microbial community in a grassland ecosystem. This may be due to the physical isolation of symbiotic species, which would not be subject to the same biotic and abiotic environmental conditions that free-living species experience. While the distribution of heterotrophic diazotrophs was limited to sites with visible mat, results of co-occurrence network analysis presented here also indicated that heterotrophic diazotrophs participate in potential syntrophic partnerships with other members of the heterotrophic community, independent of cyanobacteria diazotroph abundance. The 17 heterotrophic nifH OTUs comprising Cluster D ; , Supporting Information), for example, were significantly correlated with up to 35 other species dominated by Alphaproteobacteria, Betaproteobacteria and Bacteroidia. Interspecific relationships shared among multiple heterotrophic diazotroph and non-diazotrophic species may expand the functional capacity of this group beyond the contribution of reduced nitrogen, and contribute to functional redundancy of the bacterial community. In addition, significant interspecific interactions may also serve to stabilize relationships in late successional communities through eco-evolutionary feedbacks that occur over varying time scalesor in response to environmental change.
The ecological roles of heterotrophic diazotrophs relative to autotrophic diazotrophs in the MDV likely represent a dynamic relationship. As noted above, results indicate that cyanobacteria are primary colonizers of ephemeral streams, while the relative abundance of heterotrophic diazotroph OTUs and their diversity in MDV ecosystems increased at sites with established biomass (such as Site 10) or a more stable environment (such as Site 1). Although seasonal variations in diazotroph communities were not examined here, variations in light and temperature that promote growth and dominance of autotrophic cyanobacteria during the Austral summer may also limit autotrophs during late summer and into fall, when heterotrophic diazotrophs would play a larger role in the ecology of the system. Anticipated climate change conditions resulting in elevated temperatures have been shown to increase overall nifH gene abundance in Antarctic soils. However, the increased glacial runoff and release of stored carbonover an extended warm season with diminishing light levels may tip the balance between heterotrophic and autotrophic species to favor growth of heterotrophic diazotrophs.
Overall, results of this study provide a better understanding of ecological roles of autotrophic and heterotrophic diazotrophs and factors regulating their distribution in the MDV. As ecosystem engineers, cyanobacteria exert some control over the structure of the microbial community, including heterotrophic diazotrophs, while interspecific relationships with N-fixing heterotrophic species that develop over time likely extend the functional capacity and stability of the community at some sites. Consistent with, results of this study indicate that heterotrophic diazotrophs in the MDV are diverse, but largely limited to lakes or ponds and their outlet streams, as well as other stable environments that offer protection from desiccation such as the mossy environment at Explorer's Cove (Site 10). Despite the limited distribution, however, contributions by heterotrophic diazotrophs to the N-budget of MDV ecosystems could be substantial due to the larger surface area and long residence times of lakes compared to ephemeral streams (e.g.. Results of this investigation will help guide further work evaluating the relative ecological roles of N-fixing heterotrophic versus autotrophic species and their contributions to microbial commotuunity dynamics in the MDV ecosystem as temperature and light regimes change over the course of a season or with anticipated changes in climate.
## Supplementary data
Supplementary data are available at FEMSEC online.
# Acknowledgments
## Nutrient analysis was completed by edmund antell and sarah
Blaser at the Wilkerson/Dugdale research laboratory at San Francisco State University. Mapwas created with assistance from Michael Wethington, Polar Geospatial Center. We would like to thank the staff of the United States Antarctic Program and Antarctica New Zealand for exceptional logistical support during each field season. We are also indebted to Antarctica New Zealand and members of the NZTABS team for their assistance during the entire season that made this work possible. |
rs4246215 is targeted by hsa-miR1236 to regulate FEN1 expression but is not associated with Fuchs’ endothelial corneal dystrophy
Fuchs' Endothelial Corneal Dystrophy (FECD) is a genetically complex disorder that affects individuals above 40 years of age; molecular pathogenesis of its associated genes is poorly understood. This study aims at assessing the association of flap endonuclease 1 (FEN1) polymorphisms, c.-69G>A (rs174538) and c.4150G>T (rs4246215) with FECD. Comet assay analysis reaffirmed that endogenous DNA damage was greater in FECD individuals. However, genetic analysis in 79 FECD patients and 234 unrelated control individuals prove that both the FEN1 polymorphisms, c.-69G>A (rs174538) and c.4150G>T (rs4246215), failed to show any genetic association with the FECD disease phenotype. In silico analysis and luciferase reporter assay identified 'G' allele of the 3'UTR located FEN1 polymorphism c.4150G>T as the target for binding of hsa-miR-1236-3p. This study indicates that although FEN1 polymorphisms, c.-69G>A (rs174538) and c.4150G>T (rs4246215) are not genetically associated with FECD, its transcript regulation reported in other diseases such as lung cancer which are genetically associated by rs4246215 could be mediated through miRNA, hsa-miR-1236-3p. Fig 2. FECD lucocytes show increased endogenous DNA damage. (A) Representative images of comets from control and FECD peripheral leucocytes are shown. (B) Ln-OTM values of the comets are plotted to compare the DNA damage in FECD cases (n = 28) against controls (n = 24). Error bars indicate standard error (SE) in both directions. https://doi.org/10.1371/journal.pone.0204278.g002 miRNA targets rs4246215 to regulate FEN1 expression PLOS ONE | https://doi.
# Introduction
Fuchs' endothelial corneal dystrophy (FECD) is an autosomal dominant, bilateral, age related disorder, known since 1910 after its published detailed description by Ernst Fuchs. [bib_ref] Dystrophia epithelialis corneae, Fuchs [/bib_ref] Disease manifestation involves decreased endothelial cell density and edematous cornea, which progressively consequent into decreased visual acuity. FECD is a genetically complex disorder with an increased predilection for females (2.5:1) and affects 3.9% of individuals over forty years of age in USA, from which it derives the name late-onset FECD. [bib_ref] Prevalence and risk factors for cornea guttata in the Reykjavik Eye Study, Zoega [/bib_ref] [bib_ref] Corneal endothelial dystrophy. A study of 64 families, Krachmer [/bib_ref] On the contrary, a rarer variant of this disease known as early-onset FECD is also detected in individuals at their first decade and harboring autosomal dominant mutations in COL8A2 gene. [bib_ref] Missense mutations in COL8A2, the gene encoding the alpha2 chain of type..., Biswas [/bib_ref] Corneal transplantation is considered as the sole alternative for restoring vision in FECD affected individuals. [bib_ref] Fuchs endothelial corneal dystrophy: current perspectives, Vedana [/bib_ref] Over the past few years only handful of studies on FECD have been carried out in India that reported aboutof five-six years are due to FECD cases solely; [bib_ref] Frequency, distribution, and outcome of keratoplasty for corneal dystrophies at a tertiary..., Pandrowala [/bib_ref] [bib_ref] Endothelial Keratoplasty: A Review of Indications at a Tertiary Eye Care Centre..., Mohamed [/bib_ref] indicating a prominent contribution of FECD in the total endothelial dystrophy cases in our country. Linkage analysis and genome wide association studies have identified a plethora of genes and their genetic variations as risk factors associated with late-onset FECD (LO-FECD). [bib_ref] Genome-wide association study identifies three novel loci in Fuchs endothelial corneal dystrophy, Afshari [/bib_ref] One of the disease pathomechanism involves increased oxidative stress and resultant apoptosis of corneal endothelial cells. [bib_ref] UV-A Irradiation Activates Nrf2-Regulated Antioxidant Defense and Induces p53/Caspase3-Dependent Apoptosis in Corneal..., Liu [/bib_ref] In a recent study, one of the two polymorphisms, c.-69G>A (rs174538) and c.4150G>T (rs4246215), studied in Flap endonuclease 1 (FEN1) gene was genetically associated with LO-FECD in Polish population.As a genome stabilization factor, FEN1 is one of the component in DNA damage repair mechanism during long-patch base-excision repair (BER).It has been established through previous studies that changes in FEN1 gene expression can increase the susceptibility of DNA damage during oxidative injury. [bib_ref] Base excision repair of oxidative DNA damage activated by XPG protein, Klungland [/bib_ref] [bib_ref] Methylation of FEN1 suppresses nearby phosphorylation and facilitates PCNA binding, Guo [/bib_ref] The current study aims at assessing the genetic association of the two FEN1 polymorphisms 69G>A and c.4150G>T in Indian population. It also aims at investigating the functional role of the associated variant by evaluating its involvement in DNA damage and having a regulatory role as a miRNA target site.
# Materials and methods
## Participants and genetic analysis
A population of 234 controls and 79 patients, of Indian origin, were recruited at L. V. Prasad Eye Institute (LVPEI) at Bhubaneswar, India after acquiring written consent from the participants. This study was approved by the Ethics review board from Institutional Ethics Committee (IEC) and Institutional Biosafety Committee (IBSC), National Institute of Science Education and Research (NISER) and Institutional Ethics Committee (IEC), LV Prasad Eye Institute (LVPEI). Recruitment of study participants was done as per the previously published reportand in accordance with the Declaration of Helsinki. Genomic DNA extracted from peripheral blood leucocytes of these individuals was used to genotype FEN1 polymorphisms, rs174538 using 5' CTCTCGCCCTTAGAAATCGC 3' and 5' GGCAACCAGTCCCTCCAG 3' and rs4246215 using 5' TATGTCAGGCTCAAACCAC 3' and 5' CAGCCAGTAA TCAGTCACAA 3' by Sanger sequencing [fig_ref] Fig 1: Representative sequence chromatographs of FEN1 polymorphisms [/fig_ref].Comet assay. Endogenous DNA damage of the peripheral blood mononuclear cell (PBMC) samples was quantified by performing alkaline comet assay with the frozen blood leucocytes from the participants following previously published literatureand expressed as the natural log of Olive tail moment (Ln-OTM) of the comets after analysing individual fluorescent comet images on CASPLab v1.2.3beta2 comet analysis software (University of Wroclaw, Institute of Theoretical Physics, Poland).
## In silico prediction of mirna sites in 3'utr of fen1
In order to analyse the putative miRNA binding sites at the 3'UTR of FEN1, the most frequently used computational algorithms such as miRWalk 2.0 (http://zmf.umm.uni-heidelberg. de/apps/zmf/mirwalk2/index.html), miRanda (http://www.microrna.org/microrna/home.do); TargetScan (http://www.targetscan.org); PicTar (http://www.pictar.org), RNA22 (https://cm. jefferson.edu/rna22/)32, FindTar3 (http://bio.sz.tsinghua.edu.cn) and SegalLab (https://genie. weizmann.ac.il/index.html) and SNPinfo database (http://snpinfo.niehs.nih.gov/) were used.
## Luciferase reporter assay
Target site for miRNA binding was assessed in vitro by reporter assays using pMIR-report luciferase vector (Invitrogen) containing CMV promoter and pGL4.74 Renilla vector as reporter control plasmid. Target oligos (67bp) for mIR-1236 as identified through bioinformatics analysis on the 3'UTR of FEN1 gene , were synthesized and annealed in 1X annealing buffer (30mM HEPES pH7.4, 100mM potassium acetate and 2mM magnesium acetate) at 90˚C for 5 mins, followed by an hour incubation at 37˚C. The annealed oligos were cloned into XhoI-NotI double digested pMIR-report vector at 3'UTR of the luciferase gene. The resultant pMIR-1236 clones (1μg) along with hsa-miR-1236-3p mimic (10pmoles, Invitrogen) and pGL4.74 vector (5ng) were transiently co-transfected into HEK293 at 80% confluency using Lipofectamine 2000 (Invitrogen). 24hr post-transfection, cells were harvested and the reporter activity of the transfected cells were measured using Dual-luciferase reporter assay system (Promega). Reporter activity for each assay group were measured in Varioskan Flash spectral scanning multimode reader (Thermo Scientific, USA). After normalization with Renilla reporter activity as transfection control, values obtained for each group were plotted as percent relative luciferase activity in comparison to empty pMIR vector.
## Relative transcript expression analysis
RNA from peripheral blood leucocytes was extracted (Nucleospin RNA blood kit, Mackerey-Nagel) and converted to cDNA using 3:1 (v/v) random hexamers and anchored oligo (dT) reverse transcription primers (Verso cDNA conversion kit, ThermoFisher Scientific). Expression of FEN1 transcript in control and FECD samples was analyzed in triplicates, taking GAPDH expression as internal control, on ABI 7500 Real time-PCR system, using SYBRgreen method. Primers used for qRT-PCR are, FEN1 5' GGAGAGCGAGCTTAGGACCG 3' and 5'CAACACAGAGGAGGGATGACTGG 3' and GAPDH 5' GAAGTCAGGTGGAGCGA GG 3' and 5' GCCCAATACGACCAAATCAGAG 3'. Transcript expression of FEN1 in c.4150TT homozygotes was compared against c.4150GG individuals. . Oligomers designed for miRNA target analysis. For cloning into pMIR-report vector, these oligomers include restriction enzyme overhangs (denoted in lowercase), miRNA target site (in BOLD) and position of c.4150G>T SNP (G/T).
## Oligo name sequence (5' -3')
WT1236/G tcgagTGAAAGTGATAGATAGCAACAAGTTTTGGAGAAGAGAGAGGGAGATAAAAGGGGGAGACAgc
[formula] WT1236/T tcgagTGAAAGTGATAGATAGCAACAAGTTTTGGAGAAGAGAGAGGGAGATAAAAGGGGGATACAgc MT1236/G tcgagTGAAAGTGATAGATAGCAACAAGTTTTGGATTTTTTTGAGGGAGATAAAAGGGGGAGACAgc [/formula]
# Statistical analysis
To estimate the genetic power for the enrolled cases and control groups post hoc, G Ã Power v3.1.9.2 statistical power analysis software (University of Düsseldorf, Germany) was used.
With the effect size of 0.5 (intermediate effect) and alpha error at 0.05, the genetic power of the sample size was estimated to be at 90.73%. The difference in age and gender distribution across cases and controls were computed by carrying out Student t-test and Fisher's exact test respectively. SPSS 23.0 statistical software for MacOS (IBM SPSS, Inc., Chicago, IL, USA) was used for assessing the genotyped polymorphisms for their association with FECD by employing Chisquared test and logistic regression with age, gender and endothelial counts as covariates. Genotypic variables were coded as 0, 1 and 2 where the homozygous risk alleles in each case were weighed as 2. Haplotype frequency and association analysis, linkage disequilibrium (LD) estimation, LD plot generation and tests for deviance from Hardy Weinberg Equilibrium (HWE) was performed on Haploview4.2 (Broad Institute, Cambridge, MA, USA). Association of polymorphisms below the threshold of 5% was considered significant for this study.
Unpaired T-test with Welch's correction was performed to analyse the differences between the LnOTM values of comets from each sample types and differences in normalized FEN1 expression fold changes between the genotypes.
# Results
## Fecd samples show significant dna damage
Whole blood from 28 FECD patients and 24 individuals with healthy cornea were selected for comet assay considering the limited supply of post-operative corneal endothelial tissues from the study participants and genomic homogeneity between PBMCs and corneal cells in accordance with previously reported studies. [bib_ref] DNA damage and repair in Fuchs endothelial corneal dystrophy, Czarny [/bib_ref] Representative images of the comets segregated on the basis of affected status can be seen in Endogenous DNA damage, expressed as Ln-OTM, was significantly higher in FECD cases (P value = 0.005, n = 28) as compared to their unrelated controls (n = 24, .
## Fen1 polymorphisms are not associated with fecd risk in indian cohort
Based on the comet data and previously reported genetic association study of FEN1 with FECD in Polish population, we aimed to assess the genetic involvement of FEN1 gene in conferring insufficient DNA repair and genotyped two polymorphisms present in its regulatory untranslated regions, c.-69G>A (rs174538; 5'UTR) and c.4150G>T (rs4246215; 3'UTR) in 79 FECD cases and 234 control individuals in Indian population. About 68% of the cases were female populated; thereby substantiating the fact that FECD is a female-biased disease reported worldwide. However, only 43% females were present in the control group. This generated statistically significant difference in gender distribution between the two groups (P value = 0.001). For genetic association tests, age-gender corrected samples were used. The mean age of the cumulative study population is 63.2 years. Considering the essential age-sensitive bracket for FECD ranging from 40-85 years, individuals ranging from 43-80 years for patients and 42-85 years for control groups have been included in this study to keep the age distribution unbiased (P value = 0.09; [fig_ref] Table 2: Demographics and clinical characteristics of the study population [/fig_ref]. Mean endothelial cell count was significantly low in FECD patients (1476 ± 284.3, P value<0.001) suggesting that these individuals suffer from progressive stage of endotheliopathy.
Distribution of alleles in both the subject groups were under Hardy Weinberg equilibrium (HWE). Genotypic and allelic frequencies for FEN1 c.-69G>A and FEN1 c.4150G>T are as detailed in . With a genetic power of the current study at 90%, both the polymorphisms failed to project a genetic association with FECD. Neither of the minor alleles (rs174538: A, rs4246215: T) differed significantly between the two groups. Various association tests for the model of inheritance like dominant, recessive, and additive also confirmed the same; thereby eliminating FEN1 polymorphisms, c.-69G>A and c.4150G>T as genetically associated with FECD.
## Fen1 polymorphism c.4150t acts as a target for hsa-mir-1236-3p
Although FEN1 is not genetically associated with FECD, we simultaneously explored the possibility of any functional trait conferred by its 3'UTR variant, rs4246215 by scanning for putative miRNA binding sites on or around this SNP using computational algorithms like miRWalk, miRanda, TargetScan, PicTar2, RNA22, FindTar, and Segal Lab based on their alignment, energy, and mirsvr scores . We identified about 274 miRNAs from TargetScan and nine miRNAs from MiRanda from the initial screening. Out of these, eight candidates were common between minimum of five algorithm outputs and only hsa-miR-1236-3p spans the SNP of interest. The binding site for hsa-miR-1236-3p includes 'G' allele from the polymorphic site and a seed region, GAAGAGA situated 19 bases upstream of it.
To validate this finding, we performed luciferase assay by transfecting HEK293 cells with constructs comprising of 60bp region from 3'UTR of FEN1 gene, harboring target sites for hsa-miR1236-3p. Each of the three types of pMIR-constructs, WT1236/G, WT1236/T and MT1236/G, were individually co-transfected along with excess of hsa-miR1236-3p mimic into HEK293 cells. Regulatory effect was assessed by measuring the percent-luciferase activity in these cells. In presence of 'G' allele, we observed a significantly reduced reporter activity (26.6%, P = 0.008) than the cells with 'T' allele or mutant constructs. This suggests that both the miRNA target sequence (GAAGAGA) and 'G' allele at c.4150G>T polymorphic site is required for hsa-miR-1236-3p mediated downregulation of luciferase activity [fig_ref] Fig 3: miRNA mediated reporter activity on 3'UTR regions of FEN1 gene [/fig_ref]. This result indicates that rs4246215 can act as a functional variant when associated with a disease by regulating the expression FEN1 gene. Allele specific FEN1 regulation was further validated through qRT-PCR with comparison between genotype specific blood samples. We observed a 2.5-fold over-expression of FEN1 in c.4150TT homozygotes when compared with c.4150GG genotype individuals [fig_ref] Fig 4: FEN1 transcript expression in PBMCs of c [/fig_ref]. This corroborates with the luciferase data and indicates that miRNA binding near rs4246215 suppresses the FEN1 expression.
# Discussion
The current study highlights two important features: (1) it accurately negates the previous study that associates FEN1 with FECD and finds experimental error as a reason for this discrepancy and (2) discovers a novel miRNA candidate that targets the 3'UTR of FEN1 gene. Our comet assay analysis on peripheral leucocytes reaffirmed that FECD individuals have high endogenous DNA damage than controls; however, reported polymorphisms of FEN1 may not be genetically involved for this phenotype. In contrast to the erroneous Polish report, both of the FEN1 polymorphisms, c.-69G>A and c.4150G>T, did not show association with FECD in Indian population. Further analysis of these polymorphisms revealed a putative miRNA binding site on the 3'UTR, which could regulate its expression. Deregulated FEN1 expression was associated with rs4246215 in lung cancer tissues. This study indicates that although FEN1 . Genetic association of polymorphisms in FEN1 gene with FECD. Allelic and genotypic distribution of SNPs, rs174538 and rs4246215 across patients and controls are tabulated. Polymorphism rs4246215 shows recessive mode of inheritance with the disease. polymorphisms, c.-69G>A (rs174538) and c.4150G>T (rs4246215) are not genetically associated with FECD, its transcript regulation in FECD and other diseases such as lung cancer could be mediated through miRNA. Upon performing a BLAST analysis on the primer pair utilized in the Polish study, it revealed that in addition to the intended 3'UTR of FEN1, an unintended region of FEN1 pseudogene on chromosome 1 which shares 98% sequence homology with the FEN1 gene on chromosome 11 would amplify. This error was confirmed when RFLP and Sanger sequencing of the same sample produced different genotypes. This makes the Polish data grossly erroneous miRNA targets rs4246215 to regulate FEN1 expression and should be interpreted with caution. Our re-designed primer pair for rs4246215 correctly amplified this region and confirmed its non-association with FECD in Indian population.
## Snp
The Polish report by, was the first to report a genetic association of FEN1 polymorphisms with FECD. Although the current study identified that their results might be erroneous, but it cannot disprove the possibility that FEN1 polymorphisms may be associated with FECD in Polish population. As elaborated by Myles et al., disease-associated SNPs differ between human populations across the globe owing to the differences in their variations in risk allele frequencies. [bib_ref] Worldwide population differentiation at disease-associated SNPs, Myles [/bib_ref] This provides the impetus to perform region-specific genetic association tests to identify the disease-associated SNPs. A re-investigation is therefore warranted to correctly understand the role of FEN1 gene in Polish individuals affected with FECD.
Although these FEN1 polymorphisms are not genetically associated with FECD, researchers have associated c.4150G>T (rs4246215) with deregulated FEN1 transcript expression in lungand breast cancer tissues [bib_ref] Association of functional FEN1 genetic variants and haplotypes and breast cancer risk, Lv [/bib_ref]. Genetic association analysis in patients diagnosed with hepatocellular, esophageal, gastric, colorectal, and glial cell cancers have reported that the homozygous c.4150T allele limits cancer progression [bib_ref] Functional FEN1 genetic variants and haplotypes are associated with glioma risk, Chen [/bib_ref] [bib_ref] Functional FEN1 genetic variants contribute to risk of hepatocellular carcinoma, esophageal cancer,..., Liu [/bib_ref] [bib_ref] Flap endonuclease 1 polymorphisms (rs174538 and rs4246215) contribute to an increased cancer..., Ren [/bib_ref]. Lung tissues of c.4150TT homozygotes showed reduced DNA damage and FEN1 transcripts as compared to c.4150GG. Researchers hypothesized that increased FEN1 mRNA expression in lungand gastrointestinal [bib_ref] Functional FEN1 genetic variants contribute to risk of hepatocellular carcinoma, esophageal cancer,..., Liu [/bib_ref] cancers in c.4150TT homozygotes could be due to unsuccessful binding of miRNA. Through bioinformatics analysis, we identified that hsa-miR-1236-3p could be a potential regulator of FEN1 transcription. Reporter assay and genotype-specific expression analysis of FEN1 gene validated that hsa-miR-1236-3p specifically targets the major allele "G" at c.4150G>T SNP position. This report provides a better understanding of designing therapeutic targets to allow FEN1 expression in affected tissues.
The current study disproves the genetic association of FEN1 polymorphisms with FECD and cautions the readers to appropriately interpret the association results reported in Polish population. It also identifies a novel miRNA candidate that might target one of the 3'UTR SNP and regulate FEN1 expression.
[fig] Fig 1: Representative sequence chromatographs of FEN1 polymorphisms (A) c.-69G>A and B) c.4150G>T. https://doi.org/10.1371/journal.pone.0204278.g001 [/fig]
[fig] Fig 3: miRNA mediated reporter activity on 3'UTR regions of FEN1 gene. Normalized luciferase activity in HEK293 cells transfected with hsa-miR-1236-3p mimic is shown to demonstrate the effect of allelic changes on luciferase activity. Trasnfected cells carrying contructs with 'G' allele exhibit reduced luciferase activity (26.6 ± 3.2) in presence of hsa-miR1236-3p as compared to those with 'T' allele (96.1 ± 2.5) or scrambled hsa-miR-1236-3p binding sequence with G allele at SNP position (MT1236/G; 101 ± 6.9). Luciferase acticity of cells transfected with empty pMIR-report vector (EV) were taken as control (100 ± 2.88). Error bars indicate standard error and à indicate P value = 0.008.https://doi.org/10.1371/journal.pone.0204278.g003 [/fig]
[fig] Fig 4: FEN1 transcript expression in PBMCs of c.4150TT homozygotes was 2.5-fold upregulated than c.4150GG individuals. https://doi.org/10.1371/journal.pone.0204278.g004 [/fig]
[table] Table 2: Demographics and clinical characteristics of the study population. [/table]
|
A Review on the Use of Impedimetric Sensors for the Inspection of Food Quality
This paper exhibits a thorough review of the use of impedimetric sensors for the analysis of food quality. It helps to understand the contribution of some of the major types of impedimetric sensors that are used for this application. The deployment of impedimetric sensing prototypes has been advantageous due to their wide linear range of responses, detection of the target analyte at low concentrations, good stability, high accuracy and high reproducibility in the results. The choice of these sensors was classified on the basis of structure and the conductive material used to develop them. The first category included the use of nanomaterials such as graphene and metallic nanowires used to form the sensing devices. Different forms of graphene nanoparticles, such as nano-hybrids, nanosheets, and nano-powders, have been largely used to sense biomolecules in the micro-molar range. The use of conductive materials such as gold, copper, tungsten and tin to develop nanowire-based prototypes for the inspection of food quality has also been shown. The second category was based on conventional electromechanical circuits such as electronic noses and other smart systems. Within this sector, the standardized systems, such as electronic noses, and LC circuit -based systems have been explained. Finally, some of the challenges posed by the existing sensors have been listed out, along with an estimate of the increase in the number of sensors employed to assess food quality.
# Introduction
The intervention of sensors in the application world has helped in developing the quality of life of human beings on a daily basis, starting with the popularization of semiconducting sensors, around three decades ago, which have been implemented primarily for industrial and environmental applications. Initially, the prototypes based on silicon substrates were usedfor sensing purposes. These are non-flexible or rigid sensors that are formed using conventional Microelectromechanical Systems (MEMS)-based techniques. Non-flexible materials with moderate or high brittle nature are used as raw materials to develop these sensors. These materials are processed using lithographic techniques, where the structure and dimensions have been considerably reduced to form microelectronics. The non-flexile sensors have a high versatility to be implemented for certain industrialand environmentalapplications. They have also been used in other forms, where a particular quality of this material had been used for quantum mechanical studies. For example, silicon nitrate solutions have been prepared to determine the strong quantum squeezing for the detection of low-count photonsand scan probe microscopies. Silicon has been typically used to form silicon nitrate solutions having Young's modulus, Poisson's ratio, and material density of 2.5 × 10 11 Pa, 0.23, and 3.1 × 10 3 kg/m 3 , respectively. These solutions were subsequently used as thin films to coat metals and probes for detecting photons.
Although these sensors are very popular, there are some limitations associated with them, which leads to subsequent limitations to their applications. Some of those limitations are high input power, degradation in their sensitivity with time, highly brittle nature, relatively shorter life cycle, and the generation of toxic products during its fabrication . This led scientists to opt for alternative options where the raw materials are flexible in nature. The flexible materials have been processed to develop sensors that had enhanced electrical, mechanical and thermal characteristics. Unlike silicon-based sensors, these sensors have the capability to be bent to a certain degree, depending upon the mechanical attributes of the electrodes and substrates of the prototypes. A wide range of techniques, including soft-lithography and printing processes, have been used to fabricate these sensors. A comparison between the pros and cons of the flexible and non-flexible sensors is given in . It is seen from the table that both types of sensors have certain advantages and disadvantages associated with them. So, a review of the exploitation of both these types of sensors for determining the food quality has been presented in the paper. . Comparative study of the pros and cons of flexible and non-flexible sensors.
## Types of sensors pros cons
Non-flexible sensors In terms of sensing principles, some of the primary ones which have been focused upon by the researchers are impedimetricand electricalsensing. Most of the standardized sensors working in the microelectronics industry operate on either of these sensing mechanisms. The choice of a particular working principle of a prototype mainly depends on the application of the sensor. For example, when the sensors are used to detect very low applied force, the changes in the capacitive values would be determined. Among these options, impedimetric sensing has been largely preferred, as it helps to obtain a large amount of information related to the target analyte. This type of sensing has certain pros and cons related to it.
Although there are some disadvantages, such as short shelf life, a high probability of crosssensitivity and high sensitivity towards the change in ambient temperature, they offer a lot more advantages in terms of static sensor characteristics. Some of them are low detection limit (LOD), good stability and reproducibility in the responses, the ability to respond to a wide linear range, low power requirement, portability good resolution, high repeatability and high accuracy. Among the popular usages of these sensors for industrial applications, the detection in the food quality holds the pivotal spot in the current era. In today's world, more advanced food-producing techniques have been developed in comparison to earlier times. Today, in the era of consumerism, food products are developed on a very large scale, and in a very quick time. This makes it necessary to avoid adulteration of the products anyway. Moreover, the high demand for a different kind of food material brings about a plausible option for them to be stored in the local stores and supermarkets in a large quantity and for a long time. The storage of food and beverages for a long time also, however, deteriorates their quality, which, if consumed even in a small amount, may drastically affect the health of a person.
The use of sensors in this area is currently one of the practical ways of achieving delicate balance due to its capability to operate in harsh conditions, perform chemical fingerprinting of the specified species, ability to deal with the biomolecules present in the food matrix, and a wider range of detection. Farahi et al.had very neatly described the critical issues related to food and water safety that are being addressed and resolved by these sensors. Advanced receptor-based and non-receptor-based sensors are being engineered to obtain prototypes with better performance in terms of sensitivity and efficiency. Some of the pathogens that are being detected in food and water are shown in. These pathogens are being dealt with integrated sensing systems that operate on lap-on-a-chip systems, where the sensors are based on biological and chemical-based receptors. For example, certain bio-catalysis models such as the "lock-and-key model" have been largely preferredfor the selective detection of biomolecules present in food, due to certain attributes, such as its simplicity and high specificity on the enzyme. developed on a very large scale, and in a very quick time. This makes it necessary to avoid adulteration of the products anyway. Moreover, the high demand for a different kind of food material brings about a plausible option for them to be stored in the local stores and supermarkets in a large quantity and for a long time. The storage of food and beverages for a long time also, however, deteriorates their quality, which, if consumed even in a small amount, may drastically affect the health of a person. The use of sensors in this area is currently one of the practical ways of achieving delicate balance due to its capability to operate in harsh conditions, perform chemical fingerprinting of the specified species, ability to deal with the biomolecules present in the food matrix, and a wider range of detection. Farahi et al.had very neatly described the critical issues related to food and water safety that are being addressed and resolved by these sensors. Advanced receptor-based and nonreceptor-based sensors are being engineered to obtain prototypes with better performance in terms of sensitivity and efficiency. Some of the pathogens that are being detected in food and water are shown in. These pathogens are being dealt with integrated sensing systems that operate on lap-on-a-chip systems, where the sensors are based on biological and chemical-based receptors. For example, certain bio-catalysis models such as the "lock-and-key model" have been largely preferredfor the selective detection of biomolecules present in food, due to certain attributes, such as its simplicity and high specificity on the enzyme..
For the food products developed at home, ubiquitous sensing prototypes need to be developed, which can be employed to test the condition of the material on a regular basis. The quality of the food products processed on an industrial scale is highly vulnerable, as an aberration, even on a very small scale, can create health-related problems on a large scale. On the large-scale production of food in the industries, sensors are classified on the basis of the nature of the product. Each of the different sensors has specific nutrient content associated with it. The measurement of this content is extremely important to determine the overall amount of intake in the body of a person. The anomaly caused due to prolonged exposure to this content may create a permanent misbalance, thus affecting the For the food products developed at home, ubiquitous sensing prototypes need to be developed, which can be employed to test the condition of the material on a regular basis. The quality of the food products processed on an industrial scale is highly vulnerable, as an aberration, even on a very small scale, can create health-related problems on a large scale. On the large-scale production of food in the industries, sensors are classified on the basis of the nature of the product. Each of the different sensors has specific nutrient content associated with it. The measurement of this content is extremely important to determine the overall amount of intake in the body of a person. The anomaly caused due to prolonged exposure to this content may create a permanent misbalance, thus affecting the health of the person.
Researchers have been trying to develop sensorsthat can determine the exact composition of the food and beverages produced in the industries. The sensors have been developed to detect the complex matrix of food samples in order to trace micro-scaled amounts of hazardous agents. These sensors have been fabricated to achieve a high-throughput with low-cost fabrication, accuracy, and high sensitivity for food testing. The prototypes have been developed using singular as well as a combination of materials, depending upon their characteristics. In some cases, the fusion of two or more materials helps to increase the sensitivity and selectivity of the prototypes. This is because, with the formation of nanocomposites, the individual properties of the conductive fillers and polymeric matrix get amalgamated for further enhancement of the generated compound. For example, due to the exceptional properties of graphene, its mixing in a polymer matrix results in nanocomposites showing a substantial enhancement in their multifunctional aspects. The output material becomes stronger with simple processing.
In some cases, metallic nanoparticles are also instructed alongside graphene to facilitate an improvement in the inter-particle contact in the host matrix. This also helps in increasing the homogeneity in the developed nanocomposites. Like the utilization of ceramic reinforced graphene composites has been largely done in the chemical and biological fields due to their high stability, robustness, inertness towards other chemicals and recyclable nature. The range of polymeric materials that have been mixed with the nanofillers also varies depending upon their requirement. One of the cases can be seen in, where a conductive polymer such as poly polystyrene sulfonate is mixed with palladium and graphene to generate highly sensitive, wearable thin-film strain sensors. One of the common raw materials for sensors that have been popularized for food-testing is carbon allotropes, due to their simple structure, high portability, and the ability to deal with certain disadvantages, such as a lack of robustness, low sensitivity, and an incapability to sense the quality of multiple biomolecules present in the food products. Different kinds of carbon materials, such as Carbon Nanotubes (CNTs), graphene, and graphene oxide, have been employedto form the sensors for food testing, but none of the review papers have explicitly categorized the sensors for food-testing based on their operating principle and processed material. This paper showcases the use of some of the sensors based on their physio-chemical characteristics, such as size, dimensions and their capability to induce high selectivity and sensitivity towards the chosen molecule.
The range of impedimetric sensors has been categorized on the basis of the raw materials used to develop them.showcases the summary of a few of the selected studies done with the use of carbon-based nanomaterials on the determination of food quality, based on essential parameters. It can be seen from the table that even though each of these prototypes has been formed using different materials, the choice of the processed material has been specifically based on its high selectivity. Even though some of these works have achieved high sensitivity with the fabricated prototypes, all these sensors have been labeled for increasing their specificity towards the molecule under consideration. For example, the detection sensitivity of the graphene oxide-based sensors detecting kanamycinwas as low as 100 fM, but these sensors were labeled with Flourescein amidite (FAM)-labelled kanamycin-binding aptamer (KBA) to increase their detection capability towards a series of tested kanamycin concentrations ranging between 1 pm and 1 µM. It is a time-consuming process as well as it increases complications in the system. Moreover, the labeling of the sensors fixes the concentrations of the tested kanamycin samples. With label-free graphene-oxide sensors, the approach would be much simpler, and it would obtain more direct information with the hitting and validating technique. On the contrary to the detection of kanamycin, a very high sensitivity of 260.75 mA/mM was obtained for the detection of glucosewith the label-free field-effect transistors. These devices have been able to detect glucose at very low concentrations of 0 to 30 mM. These sensors should be encouraged for rapid detection and high stability.
The sections of the paper are organized in the following manner. Section one highlights the necessity of the detection of food quality, along with the significance of sensors for this application. Section two mentions the different types of impedimetric sensors that have been fabricated and utilized for the inspection of food quality. This section has been sub-categorized based on the conductive materials that are used to form the sensors. The two sub-classes that are predominantly used for detection of food quality are nanomaterial-based sensors and conventional electronic sensors. Among the first category, explanations related to different forms of graphene and metallic nanoparticle-based sensors have been done. For conventional sensors, electronic nose and other embedded circuitries have been elucidated. Section three shows some of the challenges currently faced by the existing sensors in this application, in addition to some of the novel ideas that can be implemented in this application. Finally, the conclusion is given in the last section of the paper.
## Type of impedimetric sensors utilized for the inspection of food quality
Among the sensors that have been developed and implemented to determine the quality of food, some of the major research work has been mentioned here. MEMS and nanoelectromechanical (NEMS) and other printing techniques have been used to fabricate the sensors. In terms of raw materials, crystalline and amorphous elements have been used to form the electrodes and substrates of the sensors. Certain elements such as graphene nanosheets and metallic nanowires have also been chosento form the conductive part due to their excellent electrical, mechanical, and thermal properties. Various synthetic polymers have also been used to counteract some of the disadvantages of the MEMS-and NEMS-based sensors, such as the high cost of fabrication due to the requirement of cleanroom facilities, time-consuming processes, and the use of hazardous chemicals. These polymers also add certain advantages, such as stretchability, transparency, and flexibility. These polymers include three types, namely elastomers, thermosets, and thermoplastics. Each of these categories has its own characteristics, the uses of which are based on the type of sensors that are developed for the detection of food quality. Some of the synthetic polymers that come under these categories are polydimethylsiloxane (PDMS), polyimide (PI), epoxy-based SU-8, polypropylene (PP), polystyrene (PS), poly (methyl methacrylate) (PMMA), polyethylene terephthalate (PET) and cyclic olefin copolymers (COC). Among them, the elastomeric polymers are the ones that serve the best for the fabrication of wearable devices, whereas the thermoplastics are favorable for the fabrication of disposable sensing devices.
The third category of substrates covers the ones developed from cellulose, which are inexpensive, porous, hydrophilic, transparent and have a biodegradable nature. The sensors formed with these substrates are highly applicable for point-of-care devices, due to their specific biodegradable coatings. Cellophane-based substrates are other materials that are sub-categorized under this sector. They are mainly developed from wood, cotton, and other sources of cellulose. They are readily used for food packaging industries, and as disposable sensors. The final category is called hybrid material, which combines two or more of the three types mentioned above. As shown in, different carbon allotrope-based materials have been preferred a lot as conducting material to form the sensors. Apart from carbon material, the deployment of electronic noses is a very standardized technique for food quality inspection, which consists of an array of sensors. The type of sensors may include metal-oxide-semiconductor field-effect transistor (MOSFET), metal oxides (MOX), or other commercial sensors, depending on the specific operation of the nose.
The change in the quality of food is dependent on the growth of different kinds of microorganisms, which makes the food inedible. The sealed and loose packets containing food may be a site for bacterial growth, as a result of leakage and exposure to moisture. The resulting decrease in food quality creates problems for consumption, as well as a commercial loss because of the perishing of the product. The shelf life of food products constantly changes depending upon the senescence, microbial spoilage, and chemical deterioration, which ultimately leads to rancidity and growth of pathogens. Some of the techniques to preserve the quality of food include storage at a low temperature, controlled atmospheric packaging, pasteurizing, and deliberately increasing the acidity content of the food. The requirement of constant monitoring of the food quality makes it necessary to develop integrated sensing prototypes.
## Impedimetric sensors based on graphene and other nanomaterials
The use of impedimetric sensors to determine the quality of food and beverages has been a very popular research topic in recent years. Among them, nanomaterials have been largely preferred due to their enhanced electromechanical characteristics, high surface-to-volume ratio, and high specificity. The utilization of a lot of nanomaterials, including graphene and other forms, has been explained in the next two sub-sections. These materials have used to form impedimetric sensors for rapid, efficient and economical sensing. Various type of nanomaterials has been consideredto form sensors with high sensitivity and selectivity. Some of the common ones are graphene, gold, and mono-and bimetallic nanoparticles and metallic nanocomposites. Researchers have been working on designing and fabricating biosensors with resources having enhanced electromechanical and thermal characteristics. For example, Avramescu et al.have depicted the use of screen-printed graphite-based sensors operating as amperometric detectors. The sensors have been used as disposable prototypes formed with NAD+-dependent dehydrogenases, tyrosinase, and genetically modified acetylcholinesterases. Some of the elements that have been detected using these sensors are D-lactic acid, L-lactic food, and acetaldehyde. Real-time samples, such as wine, were tested for validating the performances of these prototypes. Below shown are some of the examples related to the use of graphene and other nanomaterials for the detection of food quality. Each of the devices used the conductive material in a certain form that would assist them to achieve the highest selectivity and sensitivity towards the target molecule.
## Graphene-based impedimetric sensors
Graphene, due to its unique electrical properties, has largely been used for the quality assurance and safety of food. Some of the attributes of graphene that have been essential for the detection of food quality are high adsorption capability due to high surface area, high electrical conductivity, and high catalytic activity. The use of graphene to form nano-hybrids has also been shown by Chen et al., where the primary aim was to detect the presence of oxalic acid (OA) in human bodies. Since the excess amount of OA, as obtained from consumed fruits and vegetables, creates a complex with the available calcium ions in the body, it is required to control their concentration to avoid certain organs like stomach and kidneys from getting affected. The graphene nanosheets were mixed with platinum nanoparticles to form the hybrid for impedimetric detection of OA. The sensing phenomena were based on cyclic voltammetry (CV) and differential pulse voltammetry (DPV) techniques. The electrochemical oxidation of oxalic acid took place on the platinum nanoparticle-loaded graphene nanosheet-modified electrodes, as a result of which, well-defined peaks were obtained. A notable decrease in the overpotential was also obtained, in comparison to those of the bare or graphene nanosheet-modified electrode. The range and LOD of OA were from 0.1 to 50 mM, and 0.1 mM, respectively.
Similar to platinum nanoparticles, palladium was also used alongside graphene to form conductive hybrids for sensing purposes. This hybrid was placed on film-coated glassy carbon electrodes for the detection of hydrogen peroxide, present in foods, pharmaceutical, industrial, and environmental analysis. Voltammetric and amperometric techniques were employed to conduct the impedimetric experiments with the palladium nanoparticle-loaded graphene nanosheet-modified electrodes. The modified glassy electrodes demonstrated direct and mediatorless responses to the target analyte at low potentials. The LOD for these experiments was 0.05 µM, whereas the measured range lay between 0.1 and 1000 µM. Some of the colorants, such as sunset yellow (SY) and tartrazine (TT), present in aromatic dyes, are very harmful to human beings if consumed in high amounts. These colorants are responsible for causing certain anomalies, such as migraines, allergies, anxiety, diarrhea, and even cancer.
Another work where graphene-based materials have been used for quality testing was shown by Gan et al.. They proposed a novel idea where the graphene layer was wrapped using phosphotungstic acid (PTA) to form selective material for detection purposes. Impedimetric responses were determined based on the graphene-based electrodes. Th electrocatalytic activities of the prototypes were studied where the oxidation of the target samples led to the enhancement of the significant peak current and the lowering of the oxidation overpotential. Further optimization on other experimental parameters, such as supporting electrolyte, the volume of the electrolyte suspending the electrodes, accumulation potential and time, assisted in decreasing the LOD and increasing the peak current. These devices had a high sensitivity for the detection of SY and TT, obtained a peak potential of around 260 mV with the pulse voltammetry technique. The LODs obtained for the measurements of SY and TT were 30 and 0.5 µg/L, respectively. This hybrid of graphene and PTA was also mixed with glassy carbon electrodesfor detecting other colorants, such as Sudan I and Orange II. The detection range was from 3.3 to 660 nM, and from 2.85 to 28.54 nM for Sudan I and Orange II, respectively. Similar to nanoparticles, certain compounds like tin dioxide could also be used with graphene to form nanocomposites for the detection of deadly contents like mercury in very low concentrations in food. The prototypes showed high electro-catalytic activities, performing the reduction of mercury ions during the experimental process.
Other forms of graphene-like quantum dots were mixed with certain nanoparticles like gold in glassy carbon electrodes to detect malachite green, having an LOD of 1 × 10 −7 mol/L. The magnetic composites were also used by other researchersfor the extraction of certain analyte and triazine herbicides, such as atrazine, prometon, propazine, and prometryn. Some of the other works using graphene-based devices for the detection of pesticides are. Apart from these biomolecules, other nanomaterials, such as metallic nanoparticlesand carbon nanomaterials, have also been used for the detection of harmful chemicals present in the food products. Another work depicting the use of graphene for inspecting the quality of food was shown by Migliorini et al., where reduced graphene oxide was used for the detection of malathion pesticide. The devices were formed using polymeric nanofibers, having a combination of polyamide 6 and polypyrrole, that were electrospun together along with graphene. Impedimetric experiments were carried out using CV and electrochemical impedance spectroscopy (EIS) to determine the changes in the responses of the developed prototypes for the tested concentrations of malathion pesticide. For CV, the graphene-based electrode served as the working electrode, whereas platinum foil and Ag/AgCl/KCl were used as the counter and reference electrodes, respectively.
EIS measurements were performed with KCl containing ferrocyanide ions as redox probes with a potential open circuit. The sensing surface of the prototypes was modified using chemically and electrochemically reduced graphene oxides to increase the sensitivity and improve the LOD. The LOD and signal-to-noise ratio were 0.8 ng/mL and 3, respectively. The three-electrode system consisted of 0.1 M sodium sulfate solution with a voltage and scan rate of −1.5-1 V and 50 mV/s, respectively. For IS measurements, the frequency sweep was done between 0.1 and 100 kHz, having 0.1 mol/L KCl as a redox probe. The buffer solution had a pH of 7.4, having input parameters of voltage, pulse width and a pulse period of 50 mV, 0.4 s and 0.5 s, respectively. The fabricated prototypes were able to detect malathion in real samples in the presence of other pesticides such as pestanal and cadusafos by depicting a change in current. The reproducibility and recoveries of malathion from the real-time samples were 1.4-2% and 99-105%, respectively.shows some of the other uses of graphene-based sensors for the detection of harmful biomolecules in food for assurance and safety. It is seen that the ligands vary from sensor to sensor, wherein nanoparticles, compounds and alloys have been used alongside graphene to form active sites for the electro-catalytic activity in the detection of graphene. Similar to, some of the research workshave used labeled sensors to perform the experiments. The labeling of these graphene sensors induces high complexity since some of the active sites generated through the developed immuno-complex does not respond to the intended analyte, which eventually decreases the overall sensitivity. Also, some of the disadvantages associated with the conjugates are the formation of aggregation and resistance to the used drugs. In comparison to that, label-free sensors can achieve lower detection limits and wider linear ranges, as shown in. The label-free sensors also increase the synergic effects of the sensors, which are blocked in the case of labeled sensors due to the secondary validation templates provided in them.
## Mentioned in
One of the works, as done by Nag et al., was related to the design, fabrication, and implementation of laser-induced graphene-based sensors for taste sensing purposes. The sensors were used to test the five fundamental tastes of sweet, sour, bitter, salty, and umami. The prototypes were formed by the laser-ablation of commercial polymer films.shows the schematic diagram of the fabrication process of the sensor patches. Commercial polymer films were attached to glass substrates and taken for the laser ablation process. Certain laser parameters, such as power, speed, and Z-axis, were optimized prior to the laser induction. Laser-induced graphene was produced at the end of the ablation process. This graphene was generated due to the conversion of the sp 3 hybridized carbon atoms in the polyimide films to the sp 2 hybridized carbon atoms in graphene. The change in hybridization took place as a result of the breakage of bonds of carbon with other elements due to the local heat generated by the laser. After the generation of graphene, this conductive material was manually transferred to sticky tapes to use them as electrodes. Sticky Kapton tapes were placed carefully over the formed graphene, followed by manual transfer, to transfer them from the polyimide films to the tapes. The pressure was applied carefully to avoid damaging the design of the formed electrodes.One of the works, as done by Nag et al., was related to the design, fabrication, and implementation of laser-induced graphene-based sensors for taste sensing purposes. The sensors were used to test the five fundamental tastes of sweet, sour, bitter, salty, and umami. The prototypes were formed by the laser-ablation of commercial polymer films.shows the schematic diagram of the fabrication process of the sensor patches. Commercial polymer films were attached to glass substrates and taken for the laser ablation process. Certain laser parameters, such as power, speed, and Z-axis, were optimized prior to the laser induction. Laser-induced graphene was produced at the end of the ablation process. This graphene was generated due to the conversion of the sp 3 hybridized carbon atoms in the polyimide films to the sp 2 hybridized carbon atoms in graphene. The change in hybridization took place as a result of the breakage of bonds of carbon with other elements due to the local heat generated by the laser. After the generation of graphene, this conductive material was manually transferred to sticky tapes to use them as electrodes. Sticky Kapton tapes were placed carefully over the formed graphene, followed by manual transfer, to transfer them from the polyimide films to the tapes. The pressure was applied carefully to avoid damaging the design of the formed electrodes..
Interdigitated electrodes were formed that responded towards the tested samples through the change in their impedance values. The five different tasted samples were tested with concentrations ranging between 1 and 1000 ppm in order to determine the change in the resistance and reactance values of the sensors. The EIS technique was used to determine the responses of the sensors. An impedance analyzer was connected to the sensors via Kelvin probes to determine the changes in resistive and reactive parts of the impedance values. The impedance analyzer, in turn, was connected to the laptop using a USB to store the data using an automated data acquisition algorithm. The frequency sweep was done between 1 and 10 kHz. The sensors were capable of distinguishing each of the concentrations for the entire frequency sweep. These experiments were conducted for five chemicals, with each of them corresponding to a specific taste. The response time was two minutes, whereas the recovery time was ten minutes. A particular frequency showed that the reactance values were different for the five chemicals for all the tested concentrations. Other works on the use of laser- Interdigitated electrodes were formed that responded towards the tested samples through the change in their impedance values. The five different tasted samples were tested with concentrations ranging between 1 and 1000 ppm in order to determine the change in the resistance and reactance values of the sensors. The EIS technique was used to determine the responses of the sensors. An impedance analyzer was connected to the sensors via Kelvin probes to determine the changes in resistive and reactive parts of the impedance values. The impedance analyzer, in turn, was connected to the laptop using a USB to store the data using an automated data acquisition algorithm. The frequency sweep was done between 1 and 10 kHz. The sensors were capable of distinguishing each of the concentrations for the entire frequency sweep. These experiments were conducted for five chemicals, with each of them corresponding to a specific taste. The response time was two minutes, whereas the recovery time was ten minutes. A particular frequency showed that the reactance values were different for the five chemicals for all the tested concentrations. Other works on the use of laser-induced graphene for the detection of biomolecules in food are. A summary of the above-mentioned research works on graphene-based sensors has been provided in.
## Other nanoparticle-based impedimetric sensors
Apart from graphene and other carbon allotropes, there are other nanomaterials as well, which have been used to form biosensors for the inspection of food quality. These materials have been largely used for the chemical and microbiological analysis due to their quick response, high porosity, inexpensive for roll-to-roll production and great reliability. These sensors have been able to operate on variant categories for food inspection, such as the detection of pathogens, toxins, adulterants, vitamins, pesticides, taste and smell.
Zappa et al. showed thatcertain types of nanomaterials, such as tungsten, copper and tin, were being used to form oxide-based nanowires for the detection of food preservatives. These nanowires were subsequently used to form sensor arrays for low-power detection processes. These nanowires were deposited silicon micro hot plates using the MEMS technique. The sensor arrays were used for the discrimination of four types of compounds, namely ethanol, acetone, nitrogen dioxide and ozone. These compounds are said to be added to the food products to increase their shelf life. Each of the sensors in the developed arrays consumed power less than 50 mW and operate on the principle component analysis (PCA) technique.shows the schematic diagram of the mounting of the nanowires on the hotplates to form the chemical sensors. The hotplates were initially diced and deposited with a metallic layer using the magnetron sputtering technique. The choice of the deposited metal depended on the nanowires as the former acted as a catalyst for the latter one. Then, the oxidation process was carried out to synthesize the nanowires directly on the hotplates. Finally, pads were connected to the pins using gold wires via the electro-soldering technique.-dshows the optical images of the nanowires. It is seen that the thermal oxidation process assisted in the patterning of the nanowires, with the help of simple shadow masking.. It is seen from the table that the sensors were able to detect the four compounds at very low concentrations. Another area in which many researchers have worked on food inspection was glucose sensing. Solanki et al.showed the development of impedimetric biosensors for the detection of esterified cholesterol. These sensors have been developed by forming a nanocomposite with MWCNTs and sol-gel-derived silica and chitosan. This nanocomposite was deposited into an indium-tin-oxide glass to form biosensors. Different concentrations of cholesterol were tested to obtain a linearity range and response time of 10-500 mg/dL and 10 s, respectively. The characterization was done using a scanning electron microscope and Fourier transforms infrared. The sensitivity and shelf-life of these sensors were 3.8 μA/mM and 10 weeks, respectively.. It is seen from the table that the sensors were able to detect the four compounds at very low concentrations.. Detection limits of the three types of nanowires for the four tested compounds.
## Nanowires
Nitrogen Dioxide (ng/mL) Ethanol (ng/mL) Acetone (ng/mL) Ozone (ng/mL) Another area in which many researchers have worked on food inspection was glucose sensing. Solanki et al.showed the development of impedimetric biosensors for the detection of esterified cholesterol. These sensors have been developed by forming a nanocomposite with MWCNTs and sol-gel-derived silica and chitosan. This nanocomposite was deposited into an indium-tin-oxide glass to form biosensors. Different concentrations of cholesterol were tested to obtain a linearity range and response time of 10-500 mg/dL and 10 s, respectively. The characterization was done using a scanning electron microscope and Fourier transform infrared spectroscopy. The sensitivity and shelf-life of these sensors were 3.8 µA/mM and 10 weeks, respectively.
One of the works focused on monitoring the quality of chocolate products was shown by L'ubomir Švorca. Impedimetric sensors were fabricated with boron-doped diamond electrodes to determine different concentrations of theobromine (TB). The experiments achieved sub-micromolar LODs using optimized Square-wave voltammetry (SWV) and delivery point validation techniques. The attributes of these sensors are simplicity, fastness, and reliable quantification of the particular analyte. The measurements were done using common voltammetry techniques, such as CV, DPV and SWV in order to evaluate the analytical performances of the sensors under optimized experimental conditions. Three-electrode system was employed, where miniaturized thick-film boron-doped diamond and ceramic substrate were used as working electrode and electrolyte, respectively. Ag/AgCl/3 M KCl was used as reference electrodes, beside screen-printed carbon auxiliary electrodes and silver pseudo electrodes. The pH values of the solutions were tested using a combined system consisted of a pH 1100 L meter and a reference glass electrode. The tested concentrations of TB were ranging between 0.99 and 54.5 µM. The sensitivity of the sensors was 0.07 µA/µM, whereas LODs and relative standard deviation were from 0.42 to 0.51 µM, and from 2.7% to 5%, respectively.
Parra et al.displayed the fabrication and implementation of sensors from bisphthalocyaninebased carbon paste electrodes for the discrimination of red wines. Three types of rare compounds, including letutium, gadolinium and praseodymium bisphthalocyaninates were used to chemically modify the sensing prototypes. The sensor arrays were used to differentiate six types of Spanish red wines that were made from a variety of grapes, but three different geographical regions and aging stages. The experiments were conducted inside a thermostatised cell where the temperature was kept constant at 20 - C using a liquid system. CV and SWV were used for the measurement purpose, where the carbon-based electrodes, platinum wire and Ag/AgCl/KCl were used as working, counter and reference electrodes, respectively. The measurements were taken at an average of 3-5 times in an aqueous solution of 0.1 M KCl, in order to obtain a stable voltammetric response. The input voltage, frequency of the SWV applied to the electrochemical cell were 100 mV and 15 Hz, respectively. PCA was further applied to the obtained signals, where a division of each wine sample was done to obtain seven of its replicates.
Another work with nanomaterial-based prototypes was done by Cinti et al.to showcase the use of paper-based nano-modified sensors. The sensors were fabricated using a screen-printing technique on common office paper. The processed materials included carbon black and Prussian blue nanoparticles for the quantification of ethanol in beer samples.shows the schematic diagram of the experimental setup for the fabrication process. A total of 100 µL was required to test the amount of hydrogen peroxide in the chosen analyte. CV and EIS techniques were used to test the samples, followed by comparing their results with standardized techniques. Low cost and simple disposal by incineration were two of the major advantages of these sensors. These paper substrates depicted high suitability for sensor development, while depicting similar properties when compared with polyester. Optimization was done on certain analytical parameters, including pH, enzyme, concentration and working potential of the fabricated biosensors. Four categories of beers, namely pilsner, weiss, lager and alcohol-free, were used for the experiments. The sensitivity and LOD of these sensors were 9.13 µA/mM cm 2 and 0.52 mM, respectively, for the quantified amount up to 10 mM. Similar paper-based work was done by Cinti et al.to show the development of three-electrode-based printed sensors. Screen printing and drop-casting techniques were used to fabricate the sensors. The attributes contributed to by the substrates of the sensors are affordability, lightness, portability and biodegradability. An impedimetric sensing process was involved with these sensors, where quantitative measurements were done via studying the change in the range of currents in microamperes levels. The prototypes involved two separate platforms for analysis purposes. The first one was the phosphate part where the voltammetric analysis was carried out. The second one was the nerve agent biosensor, where the amperometric analysis was carried out. Graphite was used to form the working and counter electrodes, whereas silver/silver chloride was used to form the reference electrodes. Similar to the previous work, the performance and specificity of these sensors were improved by carbon black and Prussian blue. Three types of papers were used to form substrates of the sensors..
Similar paper-based work was done by Cinti et al.to show the development of threeelectrode-based printed sensors. Screen printing and drop-casting techniques were used to fabricate the sensors. The attributes contributed to by the substrates of the sensors are affordability, lightness, portability and biodegradability. An impedimetric sensing process was involved with these sensors, where quantitative measurements were done via studying the change in the range of currents in microamperes levels. The prototypes involved two separate platforms for analysis purposes. The first one was the phosphate part where the voltammetric analysis was carried out. The second one was the nerve agent biosensor, where the amperometric analysis was carried out. Graphite was used to form the working and counter electrodes, whereas silver/silver chloride was used to form the reference electrodes. Similar to the previous work, the performance and specificity of these sensors were improved by carbon black and Prussian blue. Three types of papers were used to form substrates of the sensors.
Mishra et al.showed the design, fabrication and implementation of stretchable, wearable sensors for the on-site detection of organo-phosphorous chemical threats. These impedimetric sensors have been used as a wearable point-of-care screening tool by using lab-on-a-glove for enzyme-based sensing purposes. The sensing system operated on a real-time basis with wireless data transmission of the responses to a smartphone device. The prototypes were fabricated using stressenduring inks to form printed electrodes and long serpentine connections. The gloves showed very high resilience and compliance against tested mechanical deformations. CV was used for determining the changes in the responses of the sensors during the characterization and experimental process. The electrolyte was formed using 2 M of potassium ferricyanide in 1 M of KCl. The range of input voltage was fixed between −0.6 and +0.6 V, along with a scan rate of 0.1 V/s. The sensed data were recorded using SWV, having a frequency, amplitude and potential increment of 10 Hz. 25 mV and 4 mV, respectively. The organophosphorus hydrolase-based part of the index finger was used to detect the response of the organophosphate nerve-agent compounds, which was then collected using the thumb finger.
One of the interesting smart sensing systems that have been used to monitor the food quality is based on radio frequency identification (RFID) sensors, as shown by Huang et al., where wireless sensors have been developed to determine the pH level in the food. The sensors were fabricated with iridium oxide and silver chloride as the conductive material. The electrodes were developed on polyimide substrates to achieve convenient, long-term, and on-demand wireless insitu detection. Other attributes of these pH sensors were high sensitivity, stability and reversibility. Ubiquitous monitoring of food quality was done in large quantities through these low-cost sensors. The pH sensing mechanism was based on electrochemical reactions taking place with the assistance Mishra et al.showed the design, fabrication and implementation of stretchable, wearable sensors for the on-site detection of organo-phosphorous chemical threats. These impedimetric sensors have been used as a wearable point-of-care screening tool by using lab-on-a-glove for enzyme-based sensing purposes. The sensing system operated on a real-time basis with wireless data transmission of the responses to a smartphone device. The prototypes were fabricated using stress-enduring inks to form printed electrodes and long serpentine connections. The gloves showed very high resilience and compliance against tested mechanical deformations. CV was used for determining the changes in the responses of the sensors during the characterization and experimental process. The electrolyte was formed using 2 M of potassium ferricyanide in 1 M of KCl. The range of input voltage was fixed between −0.6 and +0.6 V, along with a scan rate of 0.1 V/s. The sensed data were recorded using SWV, having a frequency, amplitude and potential increment of 10 Hz. 25 mV and 4 mV, respectively. The organophosphorus hydrolase-based part of the index finger was used to detect the response of the organophosphate nerve-agent compounds, which was then collected using the thumb finger.
One of the interesting smart sensing systems that have been used to monitor the food quality is based on radio frequency identification (RFID) sensors, as shown by Huang et al., where wireless sensors have been developed to determine the pH level in the food. The sensors were fabricated with iridium oxide and silver chloride as the conductive material. The electrodes were developed on polyimide substrates to achieve convenient, long-term, and on-demand wireless in-situ detection. Other attributes of these pH sensors were high sensitivity, stability and reversibility. Ubiquitous monitoring of food quality was done in large quantities through these low-cost sensors. The pH sensing mechanism was based on electrochemical reactions taking place with the assistance of a three-electrode system. Iridium oxide and silver chloride were used as working and reference electrodes, respectively. The electrochemical potential was obtained via achieving a redox equilibrium between two oxidation states of iridium oxide. The sensing device was integrated with a battery-less transducer that had an operating principle similar to that of RFID. The wireless system was conducted via inductive coupling between the reader and tag coil antennas contained with tuning capacitors at a resonant frequency. The sensitivity of the sensors was −49.7 mV/pH, operating on inductive coupling between the reader and transponder circuits. This system was useful for products like meat and fish, where their storage over the course of time can change the pH.shows the summary of the vital parameters of nanomaterial-based sensors for the detection of food quality. Other works on the use of nanomaterials for the inspection of food quality are shown in [119]. It can be seen from the table that the utilization of nanomaterials has been done to a great extent for a stretch of analyte found in food products. . Uses of nanomaterials to form biosensors for sensing the analyte found in food products.
## Electrode
## Nanomaterials analyte limit of detection reference
Micro-comb 2-aminoethane thiol, gold nanoparticles Aflatoxin B 1 0.10 ng/mLIndium-tin-oxide Chitosan, titanium dioxide nanoparticles Ochratoxin A 10 ng/mLIndium-tin-oxide Chitosan, gold nanoparticles Cholesterol -Indium-tin-oxide Chitosan, cerium oxide nanoparticles Mycotoxin 0.25 ng/dLGlassy carbon
## Impedimetric sensors based on electronic nose and other smart sensing circuits
Irrespective of the nanomaterial-based sensors deployed for food quality measurement, other types of electrochemical sensors are also popular due to their fast response and high sensitivity towards a specific biomolecule. These sensing systems have been used on an industrial scale for the last two decades. The sections below describe a few of the important works done on electronic noses (e-noses) and other forms of electronic circuits used for food quality inspection. The other forms include LC circuits along with intelligent systems embedded with RFID for wireless data transmission.
## Electronic noses
One of the interesting phenomena in the sector of food testing has been done using e-noses. The development of these devices for food quality sensing has seen a rapid rise in the last decade. They provide a rapid and early assessment process by mimicking the operating principle and fundamental building blocks of the mammalian olfactory system. They offer advantages such as high sensitivity, non-invasive measurement techniques, rapid and cost-effective detection techniques, and correlation with the human sensor panel. Due to these attributes, many food-related applications have employed e-noses for quality control. These devices operate in two categories, namely the pattern recognition of the target analyte, and the determination of quality in terms of classification, flavor detection and spoilage. The sensors used within the e-noses have a strong commonality with the data processing algorithms that analyze the sensed data. They normally contain an array of sensors, where the prototypes are made up of polymers and metal oxides. Sensor arrays mounted inside the electronic noses operate on the impedimetric sensing. The sensing arrays were positioned on replaceable modules so that they can be easily replaced or modified without changing the platform, in case of any malfunction. The sensors presented cross-sensitivity to some of the target chemical compounds. High measurement frequencies were achieved through the reduction in the dead volumes of the sensor chamber. These devices have been widely used to detect the spoilage of a large variety of food products, such as grains, meat, fish and dairy products.
Some of the integrated sensing arrays used inside the e-noses include MOSFET, MOX and Taguchi sensors. In this area, Wojnowski et al.explained its use by developing an electronic nose to assess food quality. The nose was used in conjugation with the support vector machine (SVM) method to classify different samples of poultry meat. The prototypes were able to detect the shelf-life of the poultry and adulterations in extra virgin oil with an accuracy of 100% and 82%, respectively. Some of the attributes of the developed prototypes were portability and the ability to obtain solutions to analyze the tested food products. The sensing system consisted of a thermostated sample block, a two-way valve, four replaceable sensor modules that are connected in sequence, two gas filters, and a pump.shows the schematic diagram of the pneumatic assembly of the e-nose. The sensors were connected using a signal-conditioning circuit consisting of a microprocessor, 12 power supply and an analog-to-digital converter. The e-nose consists of sensors that determine temperature, humidity, pressure, and concentration of gases. The sensors that were used in the e-nose were fabricated on FR4 plates using a screen printing technique. Some of the gases measured were carbon monoxide, ethanol, hydrogen sulfide, nitrogen dioxide, sulfur dioxide, ammonia, and other volatile organic compounds. The use of the SVM technique was done to classify the samples of rapeseed oil on the basis of the amount of thermal degradation.
The specific inputs given to the SVM were selected through the interpretations done on the PCA loadings. The training of the SVM algorithm was done using 66% of the randomly chosen data, where 10-fold cross-validation was carried out to obtain the overall classification accuracy. The accuracy was of the SVM algorithm was as high as 98.7%, which could be even 100% via increasing repeatability and shelf-life. The SVM algorithm was also used in conjugation with Radial Basis Function (RBF) kernel to classify the samples of extra virgin olive oil as a function of the volume of the admixed sunflower oil. Statistical analysis was done to determine the features from the ANOVA signals obtained from the six sensors. Around 66% of the data from these signals were chosen at random and used to train the SVM, in order to achieve an accuracy of 82.4% via the cross-validation technique. The limitation of cross-sensitivity and high power consumption, as faced by other sensors, was overcome by the developed e-nose. The experimental process was conducted with a sample and purge time of 20 s and 100 s, respectively.
The aromatic profile of food products such as coffee has also been studied by Servini et al.. The nose was used to study, as an instrumental tool, the sensorial analysis of the coffee filter holders. The experimental process consisted of two holders, each containing one and two cups of coffee, respectively. Espresso coffee samples were used for this process, whose extraction was done in such a way that not many differences were observed in their overall aromatic profiles. The study helped to determine the differences obtained for the pH, titratable acidity, total solids, and caffeine, depending on the types of used filter holders. To monitor the freshness in fish samples, O'Connell et al.demonstrated the use of a portable e-nose. The fish samples taken from the Argentinean lake were introduced inside a chamber consisting of gas sensors. The commercial sensing prototypes were coated with tin dioxide to increase their specificity towards the samples kept inside the chamber. The change in the response of these sensors was recorded to monitor a change in their pattern with time. The analysis of the signals was done using principle component analysis (PCA) technique..
The specific inputs given to the SVM were selected through the interpretations done on the PCA loadings. The training of the SVM algorithm was done using 66% of the randomly chosen data, where 10-fold cross-validation was carried out to obtain the overall classification accuracy. The accuracy was of the SVM algorithm was as high as 98.7%, which could be even 100% via increasing repeatability and shelf-life. The SVM algorithm was also used in conjugation with RBF kernel to classify the samples of extra virgin olive oil as a function of the volume of the admixed sunflower oil. Statistical analysis was done to determine the features from the ANOVA signals obtained from the six sensors. Around 66% of the data from these signals were chosen at random and used to train the SVM, in order to achieve an accuracy of 82.4% via the cross-validation technique. The limitation of cross-sensitivity and high power consumption, as faced by other sensors, was overcome by the developed e-nose. The experimental process was conducted with a sample and purge time of 20 s and 100 s, respectively.
The aromatic profile of food products such as coffee has also been studied by Servini et al.. The nose was used to study, as an instrumental tool, the sensorial analysis of the coffee filter holders. The experimental process consisted of two holders, each containing one and two cups of coffee, respectively. Espresso coffee samples were used for this process, whose extraction was done in such a way that not many differences were observed in their overall aromatic profiles. The study helped to determine the differences obtained for the pH, titratable acidity, total solids, and caffeine, depending on the types of used filter holders. To monitor the freshness in fish samples, O'Connell et al.demonstrated the use of a portable e-nose. The fish samples taken from the Argentinean lake were introduced inside a chamber consisting of gas sensors. The commercial sensing prototypes were coated with tin dioxide to increase their specificity towards the samples kept inside the chamber. The change in the response of these sensors was recorded to monitor a change in their pattern with time. The analysis of the signals was done using principle component analysis (PCA) technique.
The intensity of the signals increased with the increase in the mass of fish as well as the storage time. Some of the samples got rotten with an increase in the number of days of storage, irrespective of the storage conditions. The pattern of the signals changed corresponding to those rotten samples, while the same remained indifferent for the non-rotten samples. Similar to this work, the meat quality was also analyzed by Wojnowski et al.using e-noses. Here, the sensors were formed using The intensity of the signals increased with the increase in the mass of fish as well as the storage time. Some of the samples got rotten with an increase in the number of days of storage, irrespective of the storage conditions. The pattern of the signals changed corresponding to those rotten samples, while the same remained indifferent for the non-rotten samples. Similar to this work, the meat quality was also analyzed by Wojnowski et al.using e-noses. Here, the sensors were formed using volatile compounds that indicated certain parameters of the meat products, such as spoilage and shelf life. These sensors operated on the unique fingerprinting technique based on pattern recognition algorithms.shows the schematic diagram of the main components of the e-nose.
Another work on the categorization of hard cheeses was done by Gursoy et al., where MGD-1 e-noses were used for carrying out the assessment procedure. The ion mobility spectrometric technique was employed to differentiate the hardness of cheese samples based on headspace analysis. Emmental cheese was also studied to determine the changes occurring with age, or due to geographical origin and varieties. The Emmental cheese showed varied responses for samples aged nine months in comparison to the samples aged three and six months. Another food product assessment, as shown by Dymerski et al.with e-noses, was for Polish honey. A few types of honey, such as that sourcing from acacia flower, linden flower, rape, buckwheat and honeydew, were classified using Figaro semiconductor sensors that were embedded inside the e-noses. The process was initiated by providing gradient temperature characteristics using a set of thermostatic modules. The classification process was then carried out using three techniques, namely PCA, linear discriminant analysis (LDA) and cluster analysis (CA). The measurement processes were carried out in optimized conditions by setting up certain parameters, such as volumetric flow rate to 15 lit/h, 35 - C of barbotage temperature, and a sensor signal acquisition time of 60 s.shows the design of the used e-nose. Four different modules were used to control the temperature of the heating jacket, sensors and the sensing chamber of the e-nose.
volatile compounds that indicated certain parameters of the meat products, such as spoilage and shelf life. These sensors operated on the unique fingerprinting technique based on pattern recognition algorithms.shows the schematic diagram of the main components of the e-nose..
Another work on the categorization of hard cheeses was done by Gursoy et al., where MGD-1 e-noses were used for carrying out the assessment procedure. The ion mobility spectrometric technique was employed to differentiate the hardness of cheese samples based on headspace analysis. Emmental cheese was also studied to determine the changes occurring with age, or due to geographical origin and varieties. The Emmental cheese showed varied responses for samples aged nine months in comparison to the samples aged three and six months. Another food product assessment, as shown by Dymerski et al.with e-noses, was for Polish honey. A few types of honey, such as that sourcing from acacia flower, linden flower, rape, buckwheat and honeydew, were classified using FIGARO semiconductor sensors that were embedded inside the e-noses. The process was initiated by providing gradient temperature characteristics using a set of thermostatic modules. The classification process was then carried out using three techniques, namely PCA, linear discriminant analysis (LDA) and cluster analysis (CA). The measurement processes were carried out in optimized conditions by setting up certain parameters, such as volumetric flow rate to 15 lit/hr, 35 °C of barbotage temperature, and a sensor signal acquisition time of 60 s.shows the design of the used e-nose. Four different modules were used to control the temperature of the heating jacket, sensors and the sensing chamber of the e-nose..
Adjustments were made to match the relative humidity of the prepared gas and the sample's temperature. The inert gas flow rate was fixed between 86% and 91%. The entire system was connected to Teflon tubes having a diameter of 4 mm. The sensed data were converted digitally and sent to the computer for analysis purposes using the data acquisition system. The reproducibility and range obtained by the three techniques were 96% and 4.9 to 8.6%, respectively. PCA and CA techniques were able to distinguish three types of honey, whereas LD was able to distinguish all the five types.
Similar to the previous work, an evaluation of agricultural distillates was also done using enoses, as shown by Dymerski et al.. Six semiconductor FIGARO sensors were used for the Adjustments were made to match the relative humidity of the prepared gas and the sample's temperature. The inert gas flow rate was fixed between 86% and 91%. The entire system was connected to Teflon tubes having a diameter of 4 mm. The sensed data were converted digitally and sent to the computer for analysis purposes using the data acquisition system. The reproducibility and range obtained by the three techniques were 96% and 4.9 to 8.6%, respectively. PCA and CA techniques were able to distinguish three types of honey, whereas LD was able to distinguish all the five types.
Similar to the previous work, an evaluation of agricultural distillates was also done using e-noses, as shown by Dymerski et al.. Six semiconductor FIGARO sensors were used for the experimental process, where the nose was embedded with an electronic circuit for analog to digital conversion of the sensed signal. Optimized measurement conditions were followed with a gas flow rate, thermostat temperature, and sensor signal acquisition time set as 15 lit/h, 15 - C, and 60 s respectively. Three techniques, namely PCA, single-linkage cluster analysis and cluster analysis with spheres, were employed for the interpretation of results. The entire system was connected to Teflon tubes, and a bottle reducer with a metal membrane. This membrane allowed the flow of non-corrosive gases to provide a chemically inert environment for the carrier gases. The recording of the signals was done at specific time instants of 20, 60, 90, 120 and 180 s. The reproducibility and coefficient of variation obtained from the results were 93%, and between 5.8% and 9.2%, respectively. The classification done using PCA showed a response to the change in temperature, where a decrease in temperature caused a corresponding separation of the points associated with high-class agricultural distillates, from that of middle and low-class distillates. Similar changes were observed for the corresponding changes in the volumetric flow rate, where the high-class distillates were away from those of the middle and low class.
In terms of determining beverages, e-noses have also been used to detect wine where the complexity in each sample, minute disparities present between the wines, and the presence of water and ethanol content had been detected. Certain parameters, such as the assessment of the quality of grapes and its crushing, the fermentation process, the aging of wines in oak barrels, determination of the organoleptic characteristics of the final product, adulteration, and spoilage, are being taken care of by the e-noses. MOSFET and MOX, both of which had been developed with conductive polymers, were subsequently used to analyze wines by monitoring the change in the resistance values. Some of the e-noses included surface acoustic wave sensors for precise detection purposes.
Another work related to the testing of wine was shown by Macías et al.. FIGARO sensors were used to detect aroma and subsequently classify wine samples having an alcohol volume of 12% and 14%. Around sixty prototypes with three different classes of every prototype were collected for the entire operation. The three classes were named alc10, alc12 and alc14. The sensors were embedded in a system alongside an LCD display, two small pumps, and two electro-valves. The cost of the entire system was around USD 200. The system also consisted of temperature and humidity sensors having accuracies of 0.4 - C and ±3.0% RH, respectively. PCA was used for the feature extraction purposes, whereas neural networks (NN) and SVM-based algorithms were used for classification. The accuracies obtained with NN algorithms for both alc10 and alc14 were 100%, and that for alc12 was 99.5%. Other major areas where the usage of the e-noses had been done were microbiological quality controland pharmaceutical industries. shows a summary of some of the critical aspects of the work done with respect to e-nose for the detection of food quality. . Summary of the above-mentioned works on e-nose on the basis of some vital parameters.
## Types of sensors target material reproducibility reference
Screen-printed FR4 sensors Rapeseed oil High (100%)α-FOX sensors Roasted coffee beans HighMGD-1 sensor Emmental cheese -FIGARO One of the earlier works related to the development of a sensing system for in situ monitoring of the quality of packaged food was done by Tan et al., where LC sensors were developed with interdigitated electrodes and a spiral inductor. The conducting elements were printed on large substrates, following which their resonant frequency was determined. The application of these prototypes was based on determining the amount of moisture inside packaged cereals. The sensors consisted of three layers: two conducting layers on top and the bottom, and one insulating layer separating them. The printing process was carried out using a toner transfer paper on papers backed with aluminum tapes having a thickness of 10 microns. The printing was completed in three steps. Initially, ferric chloride was used as an enchanting solution, and the printer toner was removed. Finally, the cellulose acetate layer having a thickness of 50 microns was applied to isolate the printed inductor. Thin strips of copper tapes, with a thickness of 20 microns, were finally placed over the insulating layer, in order to connect the ends of the capacitor and the inductor. A resonant frequency between 23-25 MHz was set to perform the experiments. This frequency was varied by varying the dimensions of the spiral inductor and interdigital electrodes.
The responses were monitored remotely, where the changes in the impedance values were measured for the two detection coils. The resonant frequencies of the sensors were calculated by determining the impedance of the detection coil with an impedance analyzer. The removal of the inductance value of the coil was done when the sensor was absent, in order to determine the values of the response of the sensor. The experiments were conducted by varying the humidity, ranging between 20% RH to 60% RH. The samples were placed inside a testing chamber with increasing humidity to determine the intensity of the spoilage of food content. The resonant frequency changed from 24.35 to 23.8 MHz with respect to the variation of relative humidity from 2% to 44%, respectively. The corresponding responses of the sensors were reversible in nature, having a response time of one and three hours for dry-to-wet and wet-to-dry cycles, respectively.
A similar work on the use of LC-based sensors for food quality detection was shown by Ong et al., where remote query resonant-circuit sensors were used to determine the growth of bacteria to access the quality control. In-situ detection of three different strains of bacteria, namely Bacillus subtilis, Escherichia coli JM109, and Pseudomonas putida, was done inside a biological medium. The sensed data were transmitted using a loop antenna to detect the complex permittivity of the medium. These sensors have high potential to be commercialized due to their low price and capability of remote query detection. The quality of milk, meat and beer were tested using these sensors. The sensors were placed within the food package in order to obtain the responses of the LC sensors through a loop antenna. The loop antenna had six turns with a total diameter of 9 cm. The solid and liquid testing mediums were treated differently during the experimental process. For the liquid medium, the prototypes were fully immersed, while they were placed on top of a solid medium such that the interdigital electrodes were facing the object. The impedance spectra were studied across the terminals of the loop antenna. The effects of certain factors, such as sensor location, temperature, coating thickness and the absorption of water on the sensing surface, as well as their impact on the responses of the sensors, were also studied.
Some of the impedimetric sensors that have been developed in this domain include aptasensors,, which have been developed with several types of conductive materials, depending on the type of food that has been tested.
## Current challenges and future opportunities
Although researchers have been working on the detection of food quality as can be seen above, there still exists some loopholes that need to be worked on in the current scenario. Firstly, most of the sensors described above, especially the ones with graphene and other nanoparticles, have not been embedded with a signal-conditioning circuit. This makes it difficult to deploy these sensors for real-time applications. These sensors can only be used in the laboratory environment, with customized experimental conditions. The impedimetric sensing in actual scenarios would be more complicated due to the intrusion of unwanted particles on the response of the sensors. Moreover, the structural dimensions of the above-mentioned sensors have not been standardized. The researchers have tried to develop prototypes with different processed materials for the sensors, fabrication techniques, and selective material to increase their specificity. This might be encouraging to opt for techniques to develop novel sensors, but the commercialization of these prototypes is not possible.
Secondly, not much stress on multifunctional sensors has been given yet. The use of multifunctional sensors is very important as it decides the total cost of the system. It also decides the cost of using a sensing system over a prolonged period, since the replacement of a defective sensor every time would neither be cost efficient nor time efficient. Moreover, multifunctional sensors can be employed for different real-time applications in a single scenario. These sensors can also be used as an array of electrodes with selective material for determining the constituents of different kinds of foods and beverages. It would be easier to deploy the multifunctional sensing systems with prototypes being embedded together in a single board in the food-processing industries, rather than using sensors at different places. If these aforementioned points are addressed efficiently, the sensors can be used to a better extent to monitor food quality.
There are certain issues with e-noses that need to be resolved as well. These instruments lose their sensitivity when they are operated in conditions of humidity. Their sensitivity also decreases when the analyte is present in a high amount. Other issues that the e-noses face are the sensor drift and lack of quantitative data for food products that differ only in smell. These devices need proper calibration before they can be deployed for each application for quality testing. The maintenance of sensitivity can be done by increasing the specificity of the sensors that are embedded within these devices. The array of electrodes used to fabricate the e-noses can be used to increase the versatility of these devices so that each of them can reach out to two or more applications.
Another area where further work is needed is the use of label-free sensors for biosensing applications. Although label-based detection has been largely popularized due to certain characteristics, such as the confirmation they provide during bio-molecular sensing, high sensitivity and specificity, researchers are nowadays inclined towards label-free detection due to certain reasons. A few of the major advantages of label-free sensors are their simple assessment technique for bio-sensing applications by reducing the liabilities created by the use of labels, the ability to screen the interaction of the chosen cells with specific drugs, quick detection processes due to the avoidance of an additional validation step, the capacity to enable the use of native cells for better biological reference, evaluation of difficult target classes, and the ability to determine the affinity of small molecule inhibitors to particular proteins. Due to the need for long-term operation and stability in the responses, robust techniques of label-free detection have become a necessity. These sensors also provide highly sensitive measurement techniques, due to which the necessity of any kind of tags or specialized reagents is eliminated. Since the researchers are currently working to develop sensors that can provide information about the early-stage diagnosis of a particular anomaly, label-free detection would help to determine the real-time kinetic analysis along with quality assessment. The label-free detection would also help to determine a wider diversity of molecules that plays a crucial role in acute and chronic diseases such as cancer, diabetes, obesity, inflammation, neuromuscular disorders, and pathological abnormalities.
Along with the label-free nature of sensors, the other attributes that should be considered while developing sensors for bio-sensing are the low cost of fabrication, and miniaturization in size. The cost of a single prototype should be as low as possible for roll-to-roll production on an industrial scale, which would be necessary for commercial purposes. Moreover, cheaper sensors would be easier to replace when they are being employed for real-time biomedical and industrial applications. The limited size of the overall dimension of a sensor would help in achieving high sensitivity, low power consumption, additional biomedical uses such as implantable applications, and a long lifetime. The decrease in the size, while maintaining the integrity of the sensors, also helps to reduce the roughness of the surface, which is critical for certain strain-related applications. The reduced cost and size of the sensors would carry out the detection process in a non-invasive manner, while the label-free nature would induce an additional level of sensitivity for biological interactions.
One of the primary goals in the future would be the commercialization of some of the above-mentioned sensing prototypes. This would assist the industries in employing the sensors for testing in real-time applications. There are some commercial sensorsthat have found success to a certain extent, where the freshness assessment of the different food products has been determined. These prototypes have been categorized as biological sensors having high potential for sensing in smart packing industries. The quality food products have been judged on the basis of their packaging time, before and during the time that is being sold in the stores. One of the cases can be exemplifiedin the use of sensors containing potassium permanganate (KMnO 4 ), where they have been employed as commercial scavengers to reduce the rapid aging of fruits and vegetables. Due to the emission of ethylene from fresh fruits and vegetables, rapid ripening and aging occur in them. The use of these KMnO 4 -based sensors was done for the chemisorption of the overproduced ethylene via the oxidation of ethylene to initially produce ethylene glycol, and subsequently carbon dioxide and water. Another work was shown by Jiang et al., where platinum catalyst on mesoporous silica was used as sensors for removal of ethylene from food products. The sensors were operated at low temperatures, in order to test the samples at a concentration of 50 ppm. Another major example is enzyme-linked immunosorbent assay, which has been used for the detection of biomolecules in food products.
The market for the use of sensors for the detection of food quality has been growing heavily over the last few years. These sensors are mainly sub-classified under biosensors, which are mainly used for the detection of biochemical substances that affect human beings. Some of them are wearable sensors for measuring the physiological parameters of a person, while some are sensors for smart homes. The use of these biosensors has been forecasted to increase from USD 19.2 billion in 2019 to USD 31.5 billion in the next five years, having a compound annual growth rate of 8.3%. This increase is mainly due to the ubiquitous monitoring done on human health to detect acute and chronic diseases such as heart diseases, cancer, obesity, and others. These problems can be avoided to a great extent, with the intake of high-quality food.
The other innovative forms of sensors that have been popularized are the paper-based sensors, which can be used as degradable and disposable prototypes. These sensors are said to be connected to the internet of things, which allows for the monitoring of the quality of packaged food. The association of these disposable sensors with certain communication protocols, such as RFID and Wi-Fi, makes it easier to determine the response, even from a distance, using a smart device for data collection. The biodegradability of these sensors, due to the use of certain polymers, such as poly (lactic-co-glycolic) acid and polyvinyl alcohol for their fabrication, makes them quite an efficient choice of sensors to be used for healthcare and environmental applications. The change in the dimension of the sensing prototypes from MEMS to smart sensors would make it easier for the detection of the quality of food and beverages to a great extent.
# Conclusions
This paper highlights the research work done on the inspection of food quality using impedimetric sensors. Different types of impedimetric sensors have been displayed, which have been differentiated on the basis of the processed materials used to fabricate them. One of the categories included electrodes formed with nanoparticles such as graphene, platinum and palladium, and other metallic nanowires. These sensors had enhanced the electrical and mechanical characteristics, high aspect ratio, and lightness in weight. The other category included e-noses and other smart sensing systems that have been standardized in the food quality industries due to their fast response and recovery time, high sensitivity and good precision. With the frequency of adulteration in food, which affects the health and quality of life, the need for multifaceted sensors with optimized physio-chemical characteristics has been dire. This makes it a compulsion for the researchers to deploy their sensors for both laboratory and field testing. The specificity and selectivity of the prototypes always seem to be challenged while operating in real-time conditions due to sudden intrusive factors. The monitoring of human health largely depends on equal participation on an industrial as well as on a consumer level, which would assist in the ubiquitous utilization of sensors for food quality tests. |
Genome organization and DNA accessibility control antigenic variation in trypanosomes
## For bayesian analysis, information on the choice of priors and markov chain monte carlo settings
For hierarchical and complex designs, identification of the appropriate level for tests and full reporting of outcomes Estimates of effect sizes (e.g. Cohen's d, Pearson's r), indicating how they were calculated
## Clearly defined error bars
State explicitly what error bars represent Our web collection on statistics for biologists may be useful.
## Software and code
Policy information about availability of computer code
## Custom code
The described analysis workflows and required custom made Unix Shell, Python and R scripts were deposited and are accessible at Zenodo (DOI 10.5281/zenodo.823671). All data is available via NCBI GEO (accession GSM2586510) and EBI ENA (accession PRJEB18945).
For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors/reviewers upon request. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Research guidelines for submitting code & software for further information.
## Data
Policy information about availability of data All manuscripts must include a data availability statement. This statement should provide the following information, where applicable:
-Accession codes, unique identifiers, or web links for publicly available datasets -A list of figures that have associated raw data -A description of any restrictions on data availability
The RNA-seq, scRNA-seq, ChIP-seq, ATAC-seq and Hi-C sequencing data used in this publication have been deposited in NCBI's Gene Expression Omnibus69
## Life sciences study design
All studies must disclose on these points even when the disclosure is negative.
## Sample size
Sample sizes were not statistically predetermined.
Data exclusions In the scRNA-seq analysis, only data from cells with more than 500 genes with more than 10 reads per gene were included. Therefore, data from 34 single cells were excluded because they were not matching the criteria.
## Replication
All attempts at replication were successful. Hi-C experiments were perfomed in triplicates for each cell line. RNA-seq experiments were performed in triplicates for each isolate. MNase-ChIP-seq experiments were performed in duplicates with an input control for each experiment. Scc1-ChIP experiments were performed in triplicates with an input control for each experiment. ATAC-seq experiments were performed in duplicates with different cell numbers, respectively. Two gDNA samples were included as internal control for accessibility. Single-cell RNA-seq analysis is based on 40 and 408 cells per cell line, respectively. Flow cytometry experiments were performed in triplicates for each cell line. Quantification of telomere clusters (FISH microscopy) was performed in duplicates (total number of cells: 1128).
Randomization Not relevant for this study as allocation of samples/organisms was not needed or intended.
## Blinding
Image acquisition and analysis was done in a blinded fashion. a) For IF images, pictures were taken randomly from the slide without "searching" for a suitable area. b) FISH images were taken and analyzed by an unbiased, external investigator. This person also chose representative images for the publication.
For other analyses investigators were not blinded. Validation Primary antibodies were validated regarding specificity and checked for cross-reactivity as described in the according publications: BB2 antibody Bastin, P., Bagherzadeh, A., Matthews, K. R. & Gull, K. A novel epitope tag system to study protein targeting and organelle biogenesis in Trypanosoma brucei. .
Mouse
## Chip-seq data deposition
Confirm that both raw and final processed data have been deposited in a public database such as GEO.
Confirm that you have deposited or provided access to graph files (e.g. BED files) for the called peaks. The axis labels state the marker and fluorochrome used (e.g. CD4-FITC).
## Data access links
The axis scales are clearly visible. Include numbers along axes only for bottom left plot of group (a 'group' is an analysis of identical markers).
All plots are contour plots with outliers or pseudocolor plots.
A numerical value for number of cells or percentage (with statistics) is provided.
# Methodology
Sample preparation 1x10*6 cells were centrifuged in a chilled microtube at 1,500 g for 4 min at 4 °C. Cells were resuspended in 100 ?l of ice cold HMI-11 and a VSG-specific antibody (Anti-VSG-2 or Anti-VSG Tick this box to confirm that a figure exemplifying the gating strategy is provided in the Supplementary Information.
[fig] - 13 ;: Figueiredo et al. (2008) was added. After 60 min of incubation at 4 °C with gentle shaking, cells were washed three times in 500 ?l of ice cold HMI-11, resuspended in 100 ?l of cold HMI-11 and incubated with an Alexa Fluor 488-conjugated secondary antibody for 20 min. The cells were washed twice with 500 ?abundance Post-sort fractions were not used in this studyGating strategyTrypanosomes are very homogenous. Only one population was gated in the SSC/FSC window. None-stained WT cells define the negative control. Stained WT cells define the positive population. [/fig]
|
Development and external validation of prognostic models for COVID-19 to support risk stratification in secondary care
Objectives Existing UK prognostic models for patients admitted to the hospital with COVID-19 are limited by reliance on comorbidities, which are under-recorded in secondary care, and lack of imaging data among the candidate predictors. Our aims were to develop and externally validate novel prognostic models for adverse outcomes (death and intensive therapy unit (ITU) admission) in UK secondary care and externally validate the existing 4C score. Design Candidate predictors included demographic variables, symptoms, physiological measures, imaging and laboratory tests. Final models used logistic regression with stepwise selection. Setting Model development was performed in data from University Hospitals Birmingham (UHB). External validation was performed in the CovidCollab dataset. Participants Patients with COVID-19 admitted to UHB January-August 2020 were included. Main outcome measures Death and ITU admission within 28 days of admission. Results 1040 patients with COVID-19 were included in the derivation cohort; 288 (28%) died and 183 (18%) were admitted to ITU within 28 days of admission. Area under the receiver operating characteristic curve (AUROC) for mortality was 0.791 (95% CI 0.761 to 0.822) in UHB and 0.767 (95% CI 0.754 to 0.780) in CovidCollab; AUROC for ITU admission was 0.906 (95% CI 0.883 to 0.929) in UHB and 0.811 (95% CI 0.795 to 0.828) in CovidCollab. Models showed good calibration. Addition of comorbidities to candidate predictors did not improve model performance. AUROC for the International Severe Acute Respiratory and Emerging Infection Consortium 4C score in the UHB dataset was 0.753 (95% CI 0.720 to 0.785).Conclusions The novel prognostic models showed good discrimination and calibration in derivation and external validation datasets, and performed at least as well as the existing 4C score using only routinely collected patient information. The models can be integrated into electronic medical records systems to calculate each individual patient's probability of death or ITU admission at the time of hospital admission. Implementation of the models and clinical utility should be evaluated.
# Background
The COVID-19 pandemic has placed exceptional strain on healthcare systems globally. Health systems, and especially critical care services, can be overwhelmed, given the number of patients and the duration and severity of their illness. A proportion of patients with COVID-19 can deteriorate rapidly. Clinicians need to differentiate between those with COVID-19 who are at Strengths and limitations of this study ► The University Hospitals Birmingham (UHB) development dataset represents one of the largest and most ethnically diverse patient cohorts within the UK. ► As part of the UHB COVID-19 response, all admitted patients underwent a wide range of investigations to support international research efforts examining prognostic markers allowing assessment of a wide range of possible predictors (demographic variables, symptoms, physiological measures, imaging and laboratory test results) with low levels of missing data. ► A limitation of the study was that the overall sample size was relatively small compared with that of the International Severe Acute Respiratory and Emerging Infection Consortium study and was limited to one UK geographical location. ► In the external validation cohort, we were unable to examine all of the predictors included in the original full UHB model due to only a reduced set of candidate predictors being available in CovidCollab. ► It was not possible to carry out stratified analysis by ethnicity as the UHB dataset contained too few patients in many of the strata, and no ethnicity data were available in the CovidCollab dataset.
Open access high risk of the most severe symptoms (requiring intensive care treatment/ventilation) or death, and those who can be considered at low risk and potentially managed in the community. Early identification of patients at highest risk of severe outcomes may provide opportunity to prioritise, intervene and improve outcomes. Objective prognostic tools for patients with COVID-19, based on patients' initial characteristics, symptoms, biomarkers and imaging at the time of hospital admission, which can be used at or just after admission, and which can accurately discriminate between patients who will progress to more severe symptoms or death and those who will not, can be used by clinicians to triage and manage patients. This could potentially reduce time to appropriate interventions and improve patient outcomes.
A rapid systematic review has identified a number of prediction models developed for COVID-19, including prognostic models. [bib_ref] Prediction models for diagnosis and prognosis of covid-19: systematic review and critical..., Wynants [/bib_ref] However, while these existing studies provided useful information on candidate predictors for further exploration, the review found substantial limitations: many models were developed exclusively in a Chinese population; many were at high risk of bias, particularly in terms of inclusion of non-representative control participants, inappropriate exclusion criteria and small sample sizes, leading to high risk of overfitting; and external validation was limited. [bib_ref] Prediction models for diagnosis and prognosis of covid-19: systematic review and critical..., Wynants [/bib_ref] Other studies have evaluated existing early warning scores such as the National Early Warning Score, but with conflicting findings regarding their utility in predicting COVID-19 outcomes. More recent models have since been developed, some of which overcome a number of these limitations, including the International Severe Acute Respiratory and Emerging Infection Consortium (ISARIC) model and corresponding (simplified) 4C score, which was developed in a UK secondary care population representing 260 hospitals in England, Scotland and Wales (the ISARIC dataset). [bib_ref] Risk stratification of patients admitted to hospital with covid-19 using the ISARIC..., Knight [/bib_ref] While the 4C score showed reasonable discrimination for mortality, there are some limitations, including a reliance on clinicians counting specific comorbidities, which may not be recorded at admission and which are known to be under-recorded in secondary care, [bib_ref] Reliability of comorbidity scores derived from administrative data in the tertiary hospital..., Li [/bib_ref] and an absence of imaging data among the candidate predictors.
## Aims and rationale
To date, there have been few prognostic models for patients admitted to the hospital with COVID-19 developed in a UK dataset. Furthermore, evaluation of the extent to which the inclusion of comorbidities, imaging and additional biomarkers improves model performance is required. It also remains to be determined whether updating the clinical parameters with evolving biomarkers improves prediction of the clinical course of patients as the disease evolves.
The overarching aim of this study was to develop prognostic models for patients admitted to the hospital with COVID-19 using routinely collected data at the point of admission, which can be used in a secondary care setting to support clinical decision-making. Specific objectives were (1) to develop novel prognostic models for calculating predicted probability of adverse outcomes (death and intensive therapy unit (ITU) admission) at an individual patient level in a UK secondary care setting; [bib_ref] Evaluation and improvement of the National early warning score (NEWS2) for COVID-19:..., Carr [/bib_ref] to externally validate these models in an international dataset (including data from UK hospitals); (3) to externally validate the existing UK ISARIC 4C score [bib_ref] Risk stratification of patients admitted to hospital with covid-19 using the ISARIC..., Knight [/bib_ref] ; and [bib_ref] Development and validation of a clinical risk score to predict the occurrence..., Liang [/bib_ref] to compare performance of the newly developed models with the UK ISARIC 4C score. In addition, we developed daily models using time series data from the first 8 days from admission to explore changes in predictors over time.
# Methods
## Data source
Data from University Hospitals Birmingham (UHB) NHS Foundation Trust were sourced via the PIONEER Health Data Research Hub for acute care and were used for model development and for external validation of the ISARIC 4C score. Data from patients with COVID-19 admitted to Queen Elizabeth Hospital, Birmingham (part of UHB), between 1 January 2020 and 16 August 2020 were included. Data included symptoms recorded at admission, comorbidities (from International Classification of Diseases, 10th revision (ICD-10) discharge codes), vital signs (eg, blood pressure and oxygen saturation), laboratory results (biochemistry, haematology, microbiology and pathology), imaging and outcomes (ITU admission and death).
External validation of the newly developed models was performed in the CovidCollab dataset. CovidCollab is an international project using routinely collected healthcare data to develop a better understanding of how best to treat and care for adults with COVID-19. The dataset includes symptoms, comorbidities, vital signs, laboratory results, imaging findings and outcomes.
## Study population
Patients of all ages diagnosed with COVID-19 and hospitalised were included. Diagnosis was defined as a positive test result for SARS-CoV-2 from one or more reverse transcription PCR or transcription-mediated amplification tests. In the CovidCollab dataset, COVID-19 diagnosis was by either PCR or antibody test. Anonymised data for all patients with COVID-19 admitted to UHB during the study period were included. For CovidCollab, data collection was dependent on the specific processes within individual participating hospitals and the capacity of the data collector. 7
## Study design
The study utilised retrospective cohort analyses; the index date (start of follow-up) was the hospital admission date. The study period was from 1 January 2020 to 12 September 2020 (the last admission date was 16 August to ensure a minimum of 28 days of follow-up).
## Open access
## Outcomes
The primary outcome was death within 28 days of admission (in-hospital or post-discharge). The secondary outcome was ITU admission within 28 days of admission.
Study follow-up Participants were followed up from index (admission) date until the earliest of outcome date or study end (latest available data, 12 September 2020). Participants were censored 28 days after the index date. Participants admitted after 16 August 2020 (less than 28 days prior to the study end date) were excluded.
## Candidate predictor variables
Candidate predictors were selected a priori following a review of existing literature, discussion with clinical experts (specialists in acute care, critical care and geriatric medicine), and based on availability of variables routinely collected in secondary care/UHB. These included demographic variables, symptoms, comorbidities, physiological measures, imaging findings and laboratory test results. Comorbidities are not reliably and completely collected at admission, with the most complete hospital record of comorbidities usually being the discharge ICD-10 codes; therefore, the development and performance of models with and without comorbidity predictors were compared in order to explore the potential for developing models which would require no additional data collection (other than routinely collected data) at the point of admission.
## Model development
Models were trained using UHB data (patients admitted up to and including 16 August 2020). We used a multistage model building process that assessed the impact of a range of feature representation and modelling choices to select important candidate predictors. All analyses were performed in R.
Three sets of models were fitted which incorporated continuous variables in three different ways, to explore the impact of treating these variables as continuous or categorical, and also to explore the impact of different methods of handling missing data: ► As continuous numeric values, with missing values imputed ('continuous'). ► As categorical values derived from the imputed continuous values ('categorical-imputed'). ► In secondary analysis, as categorical values, using clinically meaningful categories and reference ranges, with missing indicators as a separate category ('categorical'). For the three ways of handling numerical features and missing variables mentioned previously, we fitted outcomes of death within 28 days and ITU admission (within 28 days) to candidate predictors using a range of models, which allowed both linear relationships and complex interactions between variables: ► Logistic regression with (1) all baseline parameters (demographic variables, symptoms, vital signs/ physiological measures and laboratory test results);
(2) demographic variables only; and (3) all baseline parameters with the addition of recorded comorbidities (recorded up to the point of discharge). ► Logistic regression with stepwise Akaike information criterion (AIC) minimisation, both forward and backward. 9 ► Least absolute shrinkage and selection operator (LASSO, l1 penalised) logistic regression using all baseline parameters. ► Gradient boosted model (GBM) using all baseline parameters with default hyperparameter values of 150 trees, maximum interaction depth of 3, minimum of 10 observations in nodes and shrinkage of 0.1. 10 Further information on handling of continuous variables is presented in online supplemental appendix 1.
For each of these four variable selection models, in order to reduce overfitting and selection bias, we internally validated using fivefold cross-validation (80/20 train/test split) to derive the candidate variable list. To avoid sensitivity to imputation, this cross-validation was repeated for each of the five multiple imputations.
Due to the relatively small number of outcome events (<300), we did not attempt to systematically look for interactions between multiple variables.
## Model performance
Model performance (discrimination) was assessed by calculating the area under the receiver operating characteristic curve (AUROC or C-statistic). [bib_ref] Receiver operating characteristic curve in diagnostic test assessment, Mandrekar [/bib_ref] Calibration was assessed by plotting the observed probability of the outcome against predicted probability and by calculating the calibration slope and intercept. We also calculated sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) for the final models. For each feature set and each model, the final results for cross-validated (optimism-adjusted) AUROC and all other metrics (including calibration plots) were combined from all the multiple imputations of the dataset using Rubin's rules for the mean and CI (derived from the SD).Missing data Information on candidate predictors was collected at the point of admission; however, where information on physiological or laboratory measures was not available on the day of admission, measures recorded up to 72 hours after admission were used. Candidate predictors for which >40% of patients had missing data were excluded from the analysis. Further missing continuous variables (vital signs and laboratory tests) and symptoms were imputed using multiple imputation using chained equations (using the R 'mice' multiple imputation package). We performed five imputations and a maximum of 50 iterations. [bib_ref] Multiple imputation using chained equations: issues and guidance for practice, White [/bib_ref] Continuous variables were imputed with predictive mean matching, and categorical variables with logistic regression (logreg) or polytomous regression (polyreg). Input variables for the multiple imputation included all Open access available candidate predictor variables in the dataset; outcomes were not included in the imputation variables.
We also explored use of a missing category for missing test results. Absence of a record of a comorbidity was taken to indicate absence of the condition.
## External validation
To investigate the transferability of models, we performed external validation of logistic regression models derived from the UHB dataset in the CovidCollab dataset for predicting outcomes of 28-day mortality and ITU admission.
Not all candidate predictors were common to both datasets; therefore, new logistic regression models for death within 28 days and for ITU admission were refitted on the UHB data using only those variables also present in the CovidCollab data. We then performed an external validation of these UHB models in the CovidCollab dataset and ascertained the AUROC in both the UHB and Covid-Collab datasets. Based on model performance observed in the initial model derivation and in the interest of clinical utility, we used only categorical rather than continuous numerical variables, with imputed missing values (imputed prior to categorisation). To verify that predictors behaved similarly, we compared logistic coefficients from UHB to the same models fitted on the CovidCollab dataset. To account for sensitivity to missing values, we performed training and testing five times on fivefold multiple imputed datasets for both UHB and CovidCollab.
External validation of ISARIC 4C score A logistic regression using the 4C score was performed in the UHB dataset (following the same modelling methods used in the original ISARIC study). Model performance was assessed by calculating the AUROC and plotting calibration curves.
## Sensitivity analyses
Most patient records had some missing variables; we therefore performed a complete case analysis where we refitted the best forward stepwise selection model derived using the full set of UHB variables to complete case data, then data with ≤1, 2, 5 and 10 missing values, imputing missing values in the same way as previously mentioned, and examined AUROCs and logistic coefficients for stability.
In addition, we performed sensitivity analyses (1) within male and female strata by assessing performance (AUROC) of the final models in male and female patients separately; and (2) within age strata by assessing model performance in patients aged ≤60 and>60 years separately.
# Time series analysis
The UHB regression models used baseline measurement data collected on admission; where not available at admission, we accepted values up to 72 hours after admission. To investigate fine-grained temporal effects of data acquisition, we produced a series of separate logistic regression models using data collected at different time windows from within 24 hours of admission up to within 7 days of admission, in 1-day increments, for the mortality outcome. Each dataset included only those patients eligible at the end of the window (not dead or discharged). This created eight different sets of predictors, including baseline variables of age, gender, symptoms and the time-sensitive variables of the latest physiological and laboratory measurements available.
For missing data, data were carried forward from the first observation (last observation carried forward (LOCF)) and fivefold multiple imputation was performed for missing data after LOCF was done, within each separate time-window dataset. Each model was trained and tested in fivefold cross validation, within each imputation, and AUROCs averaged using Rubin's rule. We compared the AUROCs for each of the eight models for predicting 28-day mortality from the time of admission and compared the logistic coefficients for the models. For additional insight into possible effects of changing measurements, we produced an additional logistic model for 28-day mortality to time-sensitive data collected within 4 days of admission, augmented with predictors indicating an increase or decrease in the category of each time-sensitive predictor relative to the reference category from 0 to 4 days, for example, whether temperature had crossed from below to above 37.8°C in that period.
## Patient and public involvement
We engaged with members of the PIONEER patient and public involvement group during development of the study protocol. We will further engage with this group, as well as other local and national patient and public involvement groups, in order to discuss dissemination of the findings and the best way to communicate these to patients and the public. We also consulted with several secondary care clinicians before and during the study to ensure that the tools developed meet the needs of clinicians. We have engaged with local NHS trusts to ensure that the algorithms developed are implemented/tested in a hospital setting.
# Results
## Derivation cohort characteristics
A total of 1040 participants with COVID-19 admitted to UHB were included in the derivation cohort. A total of 288 (28%) died within 28 days of admission and 183 (18%) were admitted to ITU. Baseline characteristics are presented in table 1 (stratified by mortality outcome) and online supplemental table 1 (stratified by ITU admission). The mean (SD) age of participants was 68.2 (17.7) years; 57% (589) were male; and almost 90% had at least one comorbidity.
## Candidate predictors
After exclusion of seven candidate predictors with >40% missing data (D-dimer, ferritin, high-sensitivity troponin, fibrinogen, lactate dehydrogenase, vitamin D and Physiological measures and vital signs: body mass index (BMI, kg/m 2 ), systolic blood pressure (mm Hg), diastolic blood pressure (mm Hg), temperature (degrees Celsius), heart rate (beats/min), respiratory rate (breaths/min), oxygen saturation (%), partial pressure of CO 2 (kPa) and portable oxygen concentrator fraction of inspired oxygen (FiO 2 , %).
Imaging: chest X-ray finding (categorised as clear/ unchanged, local consolidation, ground-glass opacity/ bilateral infiltrates, other/no firm diagnosis, none performed/missing).
Scores: frailty score (Rockwood Clinical Frailty Scale) [bib_ref] A global clinical measure of fitness and frailty in elderly people, Rockwood [/bib_ref] ; Glasgow Coma Scale score ; laboratory test results: estimated glomerular filtration rate (eGFR, ml/min), pH (%), base excess (mmol/L), anion gap (mmol/L), white blood cell (WBC) count (10 9 /L), platelets (10 9 /L), lymphocytes (10 9 /L), neutrophil:lymphocyte ratio, mean corpuscular volume (fL), red cell distribution width (%), monocytes (10 9 /L), eosinophils (10 9 /L), haemoglobin (g/L), glucose (mmol/L), bicarbonate (mmol/L), C reactive protein (mg/L), albumin (g/L), bilirubin (μmol/L), alanine aminotransferase (U/l), alkaline phosphatase (U/l), urea (mmol/l), potassium (mmol/l), sodium (mmol/l), corrected calcium (mmol/l), lactate (U/l) and haematocrit (l/l); Comorbidities (binary, presence or absence of record in discharge ICD-10 codes): dementia, cancer, asthma, chronic obstructive pulmonary disease, sleep apnoea, cardiovascular disease, hypertension, diabetes without complications, diabetes with complications, peptic ulcer, liver disease, rheumatic/inflammatory disease, thyroid disorder.
Mortality outcome (28 days): UHB model and predictive performance Area under the ROC curve values for each of the logistic, LASSO and GBM models, treating continuous variables in one of three ways (as continuous variables with imputed missing values; as clinically meaningful categorical variables with imputed missing values; and as categorical variables with missing categories), are presented in online supplemental table 2.
The final model selected was a logistic regression using stepwise selection of variables with categorisation of continuous variables (with imputed missing values). The final 18 categorical predictors included in the model were: age, breathlessness, sputum, systolic blood pressure, temperature, respiratory rate, oxygen saturation, FiO 2 , alkaline phosphatase, C-reactive protein, corrected calcium, eosinophils, glucose, pH, urea, WBC count, platelets and frailty score.
AUROC for the UHB cross-validated model was 0.779 (95 % CI 0.744 to 0.813) [fig_ref] Table 2: AUROCs, calibration slopes and calibration intercepts for models developed in UHB data [/fig_ref]. At a 20% predicted probability of mortality, sensitivity was 83% (95% CI 81% to 85%); specificity was 58% (95% CI 55% to 61%); positive predictive value was 43% (95% CI 41% to 46%); and negative predictive was 90% (95% CI 88% to 91%) Open access [fig_ref] Table 3A: Sensitivity, specificity, PPV and NPV for mortality at 28 days after admission [/fig_ref]. Calibration was very good at low to medium predicted probabilities but was poorer at very high predicted probabilities; a calibration plot is shown in [fig_ref] Figure 1: Calibration plots [/fig_ref] ; the calibration slope was 0.79 (95% CI 0.64 to 0.94) (
## Open access
The final model selected was a logistic regression using stepwise selection of variables with categorisation of continuous variables (with imputed missing values). The final 16 categorical predictors included in the model were: age, gender, fever, new onset diarrhoea or vomiting, heart rate, respiratory rate, FiO 2 , temperature, albumin, C-reactive protein, eGFR, pH, monocytes, WBC, frailty score, and Glasgow Coma Scale score.
AUROC was 0.893 (95% CI 0.864 to 0.922) [fig_ref] Table 2: AUROCs, calibration slopes and calibration intercepts for models developed in UHB data [/fig_ref]. At a 20% predicted probability of ITU admission, sensitivity was 79% (95% CI 74 to 84), specificity was 83% (95% CI 81 to 84), positive predictive value was 49% (95% CI 46 to 52), and negative predictive was 95% (95% CI 94 to 96) (table 3B). Calibration was good; a calibration plot is shown in [fig_ref] Figure 1: Calibration plots [/fig_ref] , and the calibration slope was 0.91 (95% CI 0.80 to 1.01) (table 2). Model coefficients are presented in online supplemental table 5.
Addition of comorbidities to the predictors included in the model did not improve performance.
## Reduced uhb model and external validation in the covidcollab dataset
A total of 6099 patients admitted with COVID-19 were included in the CovidCollab external validation dataset; 1668 (27%) died and 722 (12%) were admitted to ITU [fig_ref] Table 1: Baseline characteristics of participants admitted with COVID-19 in the derivation [/fig_ref]. Not all variables included in the UHB model derived previously were available in the CovidCollab dataset. Therefore, revised and reduced models were developed in UHB data using the subset of candidate predictors common to both the UHB and CovidCollab datasets (reduced UHB dataset, UHB-R), using logistic regression with stepwise selection, and these were then externally validated in the Covid-Collab dataset.
The reduced set of 27 candidate predictors included demographic characteristics: age and gender; symptoms: cough, fever and delirium; physiological measures and vital signs: BMI, systolic blood pressure, diastolic blood pressure, heart rate, temperature, respiratory rate, oxygen saturation, FiO 2 and chest X-ray; frailty score; Glasgow Coma Scale score; laboratory test results: eGFR, pH, base excess, lymphocytes, neutrophil:lymphocyte ratio, haemoglobin, bicarbonate, C reactive protein, alanine aminotransferase, urea and lactate.
## Mortality (28 days)
For the 28-day mortality outcome, following stepwise selection, the final 10 categorical predictors (common to both datasets) included in the reduced logistic regression model were age, oxygen saturation, FiO 2 , respiratory rate, temperature, systolic blood pressure, C reactive protein, pH, urea and frailty score.
The selected predictors were a subset of those in the original UHB model derivation, but gave similar model performance. AUROC in the UHB-R dataset was 0.791 (95% CI 0.761 to 0.822), and AUROC in the CovidCollab external validation dataset was 0.767 (95% CI 0.754 to 0.780) [fig_ref] Table 2: AUROCs, calibration slopes and calibration intercepts for models developed in UHB data [/fig_ref]. At a 20% predicted probability of Open access mortality, in the UHB-R dataset, sensitivity was 86% (95% CI 85% to 88%); specificity was 54% (95% CI 51% to 57%); PPV was 42% (95% CI 40% to 43%); and NPV was 91% (95% CI 90% to 92%); in the CovidCollab dataset, sensitivity was 88% (95% CI 87% to 89%); specificity was 46% (95% CI 45% to 47%); PPV was 38% (95% CI 0.37% to 0.38%); and NPV was 91% (95% CI 91% to 92%) [fig_ref] Table 4A: Sensitivity, specificity, PPV and NPV for mortality at 28 days after admission... [/fig_ref]. Calibration was good for both derivation and external validation datasets; calibration plots are shown in [fig_ref] Figure 1: Calibration plots [/fig_ref] ,D, and calibration slopes were 0.89 (95% CI 0.81 to 0.97) and 0.85 (95% CI 0.75 to 0.94) for the UHB-R and CovidCollab datasets, respectively [fig_ref] Table 2: AUROCs, calibration slopes and calibration intercepts for models developed in UHB data [/fig_ref]. Model coefficients are presented in online supplemental table 6.
## Itu admission
For the ITU admission outcome, the final 11 categorical predictors (common to both datasets) included in the reduced model were age, gender, fever, respiratory rate, FiO 2 , C reactive protein, eGFR, pH, neutrophil:lymphocyte ratio, frailty score and Glasgow Coma Scale score.
AUROC in the UHB-R dataset was 0.906 (95% CI 0.883 to 0.929), and in the CovidCollab dataset was 0.811 (95% CI 0.795 to 0.828) [fig_ref] Table 2: AUROCs, calibration slopes and calibration intercepts for models developed in UHB data [/fig_ref]. At a 20% predicted probability of ITU admission, in the UHB-R dataset, sensitivity was 83% (95% CI 81% to 85%); specificity was 83% (95% CI 82% to 84%); PPV was 51% (95% CI 49% to 52%); and NPV was 96% (95% CI 95% to 96%); in the CovidCollab dataset, sensitivity was 64% (95% CI 62% to 67%); specificity was 80% (95% CI 79% to 82%); PPV was 30% (95% CI 29% to 32%); and NPV was 94% (95% CI 94% to 95%) [fig_ref] Table 4B: Sensitivity, specificity, PPV and NPV for intensive therapy unit admission within 28... [/fig_ref].
Calibration was good for both derivation and external validation datasets; calibration plots are shown in figure 1E,F; calibration slopes were 0.94 (95% CI 0.84 to 1.04) and 0.95 (95% CI 0.82 to 1.08) for the UHB-R and CovidCollab datasets, respectively [fig_ref] Table 2: AUROCs, calibration slopes and calibration intercepts for models developed in UHB data [/fig_ref]. Model coefficients are presented in online supplemental table 7.
External validation of the ISARIC 4C score in the UHB dataset The AUROC for the recently published ISARIC 4C score in the UHB dataset was 0.753 (95% CI 0.720 to 0.785). The calibration slope was 0.99 (95% CI 0.85 to 1.12) (table 2 and online supplemental [fig_ref] Figure 1: Calibration plots [/fig_ref].
It was not possible to externally validate the ISARIC 4C score in the CovidCollab dataset, as information on many of the comorbidities required to calculate the ISARIC comorbidity score was not available in the dataset.
## Sensitivity analyses
Analyses exploring different ways of handling missing data are reported in online supplemental appendix 2 and online supplemental figures 2 and 3.
# Complete case analysis
Few patients in the dataset had complete data (n=224/1040, 22%); model performance in this patient subset was slightly poorer for the mortality outcome: AUROC 0.696 (95% CI 0.597 to 0.795) for mortality and 0.892 (95% CI 0.844 to 0.940) for ITU admission. Including patients with missing variables, with missing values imputed, improved model performance for predicting mortality; allowing even a single missing/ imputed variable improved AUROC for mortality to 0.760 (95% CI 0.708 to 0.812) (online supplemental table 8).
Stratification by gender and age When patients were stratified by gender, the reduced models predicting mortality and ITU still performed well: AUROCs for mortality were 0.775 (95% CI 0.726 to 0.823) for males and 0.755 (95% CI 0.706 to 0.804) for females, and those for ITU 0.897 (95% CI 0.856 to 0.937) for males and 0.873 (95% CI 0.833 to 0.913) for females (online supplemental table 9). When patients were stratified by age, the models performed slightly better in patients aged >60 years (AUROC 0.778 (95% CI 0.722 to 0.834) and 0.897 (95% CI 0.864 to 0.930) for mortality and ITU admission, respectively) compared with those aged ≤60 years (AUROC 0.730 (95% CI 0.638 to 0.823) and 0.845 (95% CI 0.761 to 0.930) for mortality and ITU admission, respectively).
# Time series analysis
Online supplemental shows variation in logistic regression coefficients for the candidate predictors from day of admission and up to 7 days later. The majority of coefficients remained relatively constant over time. However, several (not necessarily statistically significant) trends in the modification of effects over the week of admission on mortality were visible, such as a decrease over the week of the effect of obesity on mortality, elevated effect of eosinophils, and an increase over the week of the effect of elevated haemoglobin, elevated potassium and elevated oxygen saturation. Some of these might be depletion effects related to relatively high patient mortality in the first few days, for example, the apparent protective effect of obesity and high eosinophils.
# Discussion
Using routinely collected data for more than a thousand patients admitted with COVID-19 at a large UK hospital trust, we have developed and externally validated prognostic models for mortality and ITU admission. The models showed good discrimination and calibration. The candidate predictors explored included a clinically informed, wider range of demographics, clinical observations, symptoms, comorbidities, biomarkers and radiological investigations than those included in the derivation of existing prognostic scores or models.
If integrated into hospital electronic medical records systems, the model algorithms will provide a predicted probability of mortality or ITU admission within 28 days of hospital admission for each patient based on their individual data at, or close to, the time of admission, which will support clinicians' decision making with regard to
## Open access
appropriate patient care pathways and triage. This information might also assist clinicians in explaining complex prognostic assessments and decisions to patients and their relatives, particularly at times when relatives are unable to see the patient and understand how unwell they are.
# Summary of results
The models developed using all 63 available candidate predictors from UHB performed well with an optimismadjusted AUROC of 0.779 (95 % CI 0.744 to 0.813) for mortality within 28 days of admission and 0.893 (95% CI 0.864 to 0.922) for ITU admission. Not all variables included in the UHB dataset are routinely collected at admission in other hospitals; therefore, reduced models using only variables common to both UHB and the CovidCollab external validation dataset were explored. Discrimination remained similar, with an AUROC of 0.791 (95% CI 0.761 to 0.822) for mortality and 0.906 (95% CI 0.883 to 0.929) for ITU admission in the UHB derivation dataset. These reduced models also performed well in the CovidCollab external validation dataset, with AUROCs of 0.767 (95% CI 0.754 to 0.780) and 0.811 (95% CI 0.795 to 0.828) for mortality and ITU admission, respectively. The models also performed well in gender-stratified and age-stratified patient subgroups.
Calibration of all models showed good agreement between observed and predicted probabilities, particularly at lower predicted probabilities in the range where the models would be of most clinical utility.
We found that addition of comorbidities to the model predictors did not improve overall model performance. This may be due to a correlation between presence of comorbidities and related physiological measurements and/or biomarkers which are already captured by the model.
## Comparison with existing literature
Two systematic reviews summarised the existing secondary care COVID-19 prognostic models or scores published until 31 May 2020. The majority of the reported models, along with several more recent ones, [bib_ref] Development and validation of a clinical risk score to predict the occurrence..., Liang [/bib_ref] were derived in Chinese cohorts. Many of the models included in the reviews demonstrated high discriminatory performance; however, all pre-existing models when assessed using the PROBAST score were at high risk of bias. Furthermore, few models were externally validated in suitable cohorts. By deriving our model from routinely collected data, we were able to reduce the risk of bias in patient selection as well as predictor and outcome measurements. Additionally, in this study, we were able to externally validate models in a large global heterogeneous cohort.
More recently, the most notable secondary care prediction model advised for uptake in UK hospitals was derived from the ISARIC-WHO collaborating cohort and has been externally validated. Both the full and reduced UHB-derived models for mortality had slightly better discrimination than the ISARIC 4C score in the UHB data (AUROC 0.753, 95% CI 0.720 to 0.785 for 4C). This compares with an AUROC of 0.767 (95% CI 0.760 to 0.773) for the 4C score reported in the original ISARIC validation cohort. [bib_ref] Risk stratification of patients admitted to hospital with covid-19 using the ISARIC..., Knight [/bib_ref] However, better performance may be expected for models evaluated in their development dataset compared with external datasets. The newly developed UHB model offers an advantage over the ISARIC 4C model in that it uses only routinely collected patient data recorded at admission and does not require additional assessment and recording of specific comorbidities (which are often not routinely fully recorded at the point of admission).
In our time series analysis, we did not find strong evidence for trends in predictor coefficients over the first 8 days of admission, particularly for variables included in the final models, suggesting that time-dependent effects due to effect modification or selection bias in the first week are small. Another recent model derived from patients with COVID-19 in a Hong Kong hospital adopted the use of time-dependent routinely collected predictors; the model in the Hong Kong study demonstrated high discrimination, with an AUROC of 0.91 when predicting severe COVID-19 outcomes.However, this model is yet to be peer-reviewed and externally validated.
# Strengths and limitations
The UHB dataset represents one of the largest and most ethnically diverse patient cohorts within the UK. Additionally, as part of the early UHB response to the COVID-19 pandemic, the hospital trust ensured that, on admission, all patients underwent a wide range of investigations to support international research efforts examining prognostic markers. This allowed us to examine a wide range of possible predictors (63 candidate predictors after exclusions). Lastly, a strength of this study was the good performance, in terms of both discrimination and calibration, of the simplified, reduced model in an externally validated cohort (CovidCollab), indicating its suitability for wider use, including potentially in LMICs.
Despite the strengths, the findings must be considered in light of the study's limitations. Although we were able to use a derivation dataset from UHB with low levels of missing data, the overall sample size was relatively small compared with that of the ISARIC study and was limited to one UK geographical location. However, we were able to externally validate the model in a larger external cohort. A second limitation was that in the external validation cohort, we were unable to examine all of the predictors included in the original full UHB model due to only a reduced set of candidate predictors being available in CovidCollab. Nevertheless, the model performed well and the results suggest it may be applicable in a wide range of datasets where only a reduced set of predictor variables is available. It was not possible to carry out stratified analysis by ethnicity as, in the UHB dataset, too few patients were included in most of the strata; ethnicity data were not available in the CovidCollab dataset. Our definition of 28-day COVID-19 mortality aligns with the current technical guidance from Public Health England and Open access the definition used by the UK government in reporting COVID-19 mortality statistics ; however, we acknowledge that this may not capture all COVID-19-related deaths, and some other studies have used a longer period of follow-up. [bib_ref] A fuller picture of COVID-19 prognosis: the added value of vulnerability measures..., Aliberti [/bib_ref]
# Conclusion
In this paper, we have described the development and external validation of novel prognostic models which predict mortality and ITU admission within 28 days of admission for patients admitted to hospital with COVID-19. The simple, reduced models used only routinely collected data gathered at admission, showed good discrimination and calibration, performed at least as well as the existing ISARIC 4C score and performed well in a validation cohort. The models can be integrated into existing electronic medical records systems to calculate each individual patient's probability of death or ITU admission at the time of hospital admission. The models should be further validated to determine their applicability in other populations. In addition, implementation of the models and clinical utility should be evaluated.
[fig] Figure 1: Calibration plots (observed probability (y-axis) against predicted probability (x-axis)): (A) UHB derivation dataset for mortality outcome, (B) UHB derivation dataset for ITU admission outcome, (C) UHB-R derivation/train dataset reduced model for mortality outcome, (D) UHB-R derivation/ train dataset reduced model for ITU admission outcome, (E) CovidCollab external validation dataset reduced model for mortality outcome, (F) CovidCollab external validation dataset reduced model for ITU admission outcome. ITU, intensive therapy unit; UHB, University Hospitals Birmingham; UHB-R, University Hospitals Birmingham reduced. [/fig]
[table] Table 1: Baseline characteristics of participants admitted with COVID-19 in the derivation (UHB) and validation (CovidCollab) datasets breathlessness, chest pain, cough, fever, headache, malaise, new-onset diarrhoea or vomiting, sputum and delirium. [/table]
[table] Table 2: AUROCs, calibration slopes and calibration intercepts for models developed in UHB data (full (UHB) and reduced (UHB-R) datasets) and externally validated in CovidCollab data, and for external validation of the ISARIC 4C score [/table]
[table] Table 3A: Sensitivity, specificity, PPV and NPV for mortality at 28 days after admission (University Hospitals Birmingham derivation dataset) [/table]
[table] Table 4A: Sensitivity, specificity, PPV and NPV for mortality at 28 days after admission for the reduced model (UHB derivation dataset and CovidCollab external validation [/table]
[table] Table 4B: Sensitivity, specificity, PPV and NPV for intensive therapy unit admission within 28 days after admission in the reduced model (University Hospitals Birmingham derivation dataset and CovidCollab external validation dataset, using predictors common to both datasets) FN, false negatives; FP, false positives; NPV, negative predictive value; PPV, positive predictive value; TN, true negatives; TP, true positives; UHB-R, University Hospitals Birmingham reduced model. [/table]
|
The Inactivation by Curcumin-Mediated Photosensitization of Botrytis cinerea Spores Isolated from Strawberry Fruits
# Introduction
Strawberries are one of the most popular fruits consumed worldwide, with global production in excess of 9M tons per annum and an estimated annual production value of USD 18B worldwide . These fruits are rich sources of antioxidants and vitamin C, but their commercial value is restricted by limited shelf-life due largely to susceptibility to microbial infections, particularly Botrytis grey mould. Botrytis cinerea is the causal agent of this disease, which effects not only strawberries, and is considered one of the most important phytopathogens of horticultural crops. This fungus is found in more than 200 ornamental and agriculturally important plants in the world, and ranks second in the list of 'top 10' fungal plant pathogens.
Antimicrobial photosensitization is an effective and promising approach to elicit cell death and kill microorganisms such as Gram-negative and Gram-positive bacteria and yeasts, as well as parasites and viruses. The basic principle of photosensitization is a photochemical process in which a photosensitizer dye such as crystal violet, methylene blue and safranin O are energized to an unstable singlet excited-state after absorbing light photons, and then loses its excess energy to produce free radicals including superoxide and hydroxyl radicals (Type I pathway) and singlet oxygen (Type II pathway). These reactive oxygen species (ROS) cause oxidative damage to biomolecules including amino acids, lipids, nucleic acids, with ensuing cell death. The application of natural plant additives such as curcumin as photosensitizer is considered a clean and green technology with minimal reported adverse impacts. By comparison fungicides, and other methods, such as irradiation, biocontrol agents, etc., that are currently used to extend food shelf life, show unfavourable side effects that can impact on consumer health. Photosensitization is thus being investigated as a novel technology to accomplish microbiological decontamination and extend the shelf life of food in an environmentally friendly way.
B. cinerea has many unique characteristics in its phytopathological behaviour, including the capacity to kill host cells through both the production of toxins and also oxidative burst (reactive oxygen species) generated by both the host and the pathogen. This fungus is both a pathogenic and saprophytic organism, initially infecting the weak or dead parts of the plant, and then extending to the rest of healthy plant tissue. B. cinerea produces characteristic fungal metabolites based on the botryane skeleton, principally botrydial (1) and its metabolite dihydrobotrydial (2), which have been isolated from fungal culture medium. Botrydial is a particularly powerful phytotoxin, produced during plant infection, which can induce chlorosis and cell collapse. It has been demonstrated to have phytotoxic and cytotoxic activity at ID 50 as equal to or lower than 5 µg/mL and antibiotic activity at 100 ppm. A further B. cinerea phytotoxin botcinic acid has also been studied and shown to have a redundant role in virulence with botrydial. Even though photosensitization has been demonstrated to have significant antimicrobial efficiency against an array of microorganisms, there is limited literature available as to whether photosensitization will inhibit the growth of B. cinerea, and moreover the impacts of photosensitization on the production of the phytotoxic metabolites botrydial and dihydrobotrydial. Previous studies have investigated the antifungal effect of 405 nm light at 60 mW cm −2 for 24 h on this phytopathogen. These authors suggested that the excitation of endogenous porphyrins and subsequent accumulation of singlet oxygen contribute to the light-mediated photoinactivation of grey mould, and suggested such treatment as a means of controlling plant disease. Such lengthy irradiation would, however, not be appropriate with a food commodity such as strawberries, and moreover these authors did not examine the impacts of this treatment on the phytotoxic metabolites.
In this study, B. cinerea was isolated from Queensland (Australia) grown strawberries and used to evaluate the effects of photosensitization on the growth of this fungus under a range of experimental conditions including different concentrations of photosensitizer (curcumin), incubation time, and light dose with irradiation periods of 10-30 min. At the same time, a new high-resolution accurate mass (HRAM) UPLC-MS/MS method was developed and validated to study the relative formation of the metabolites botrydial and dihydrobotrydial under these photosensitization conditions.
# Results
## Extraction of botrydial and dihydrobotrydial from b. cinerea cultures
Solvent extraction of B. cinerea cultures obtained from Queensland strawberries provided a mixed sample of the fungal metabolites botrydial and dihydrobotrydial after repeated silica chromatography. Unfortunately, attempts to further purify these compounds resulted in further degradation of these metabolites. Botrydial and dihydrobotrydial are known compounds for which both 1 H andC NMR have been reported, and literature data were used to both confirm the identity of the two components in our extract and also enabled quantification of the amount of each component within the mixed sample. Integration of 1 H NMR resonances with a measured aliquot of dioxane as internal standard were used to determine the amount of botrydial/dihydobotrydial present in the sample. The isolated mixed standard was determined to contain botrydial and dihydrobotrydial in a 1:9 ratio, and used without further purification as a mixed external standard in the development of and validation of a new HRAM UPLC-MS/MS analysis method for both metabolites.
## Hram uplc-ms/ms analysis of botrydial and dihydrobotrydial
HRAM UPLC-MS/MS analysis of the mixed botrydial/dihydobotrydial standard provided two peaks with molecular ions consistent with the molecular formula of the two components botrydial (1) The effect of increasing curcumin dye (D) concentrations with a photosensitization light dose (L) of 120 J/cm 2 on B. cinera growth on agar plates incubated at 26 - C for 8 days was investigated. The B. cinerea control D − L − treatment (D − = no curcumin; L − = no light) and D − L + treatment (no curcumin, light dose 120 J/cm 2 ) produced characteristic prolific greyish-brown spores, and visible differences in fungal growth was readily apparent with increasing curcumin concentrations after photosensitization treatment and incubation for 8 days . With increasing curcumin concentration, the growth of B. cinerea decreased. When treated with a curcumin concentration above 800 µM and a light does of 120 J/cm 2 , the fungus was not able to survive . As shown in, a significant percentage (p < 0.001) reduction in the spores was also observed when the spores were treated with curcumin-mediated photosensitization. Treatment with light or curcumin alone (D − L + or D + L − ) was not effective. The curcumin-mediated photosensitization (D + L + ) reduction in spore germination was directly proportionate to curcumin concentrations, with increasing concentrations of curcumin photosensitizer acting to drastically inhibit B. cinerea conidia germination. With curcumin concentrations above 600 µM and light dose of 120 J/cm 2 , spore germination was completely (100%) inhibited (p < 0.001). . Effect of curcumin-mediated photosensitization with different curcumin concentrations (0-1000 µM) and light dose (120 J/cm 2 ) on the concentration of botrydial and dihydrobotrydial in spores after incubation at 26 - C for 11 days (n = 6). The significances of difference in concentration of (a) botrydial and (b) dihydrobotrydial and ratio of (c) dihydrobotrdial/botrydial between different curcumin concentrations and light-only treatment (D − L + , curcumin = 0 µM) p < 0.05 (*), and compared to the control (D − L − ) p < 0.05 ( # ), tested with t-test.
## Effect of curcumin-mediated photosensitization on the b. cinerea spore using different light doses
As shown in, when treated with a light dose of 120 J/cm 2 and curcumin concentration of 800 µM, no spore germination was observed, which indicates that the spores were completely inactivated by this photosensitization mediated curcumin treatment. By comparison, the spores grew well after treatment with 800 µM curcumin without light (D + L − ) and under control conditions (D − L − ). The percentage reduction in total spores reached 100% with the curcumin-mediated light doses of 120, 240, and 360 J/cm 2. Similarly, neither botrydial nor dihydrobotrydial could be detected when light and dye (D + L + ) are used at the same time with light doses of 120, 240 and 360 J/cm 2. The botrydial concentration was observed to increase slightly from 6.2 ± 2.2 to 8.5 ± 0.9 µg/g in the curcumin-only treatment (D + L − ), comparing to the control (D − L − ) (p < 0.05). However, the dihydrobotrydial concentration was not significantly different between the curcumin-only treatment (D + L − ) and the control (D − L − ).
## The growth of b. cinerea over different incubation time after photosensitization
In studies of effect of different treatments on B. cinerea with increasing incubation time, it was observed that in the control treatment (D − L − ) mycelial growth was fast with white hyphae apparent within two days incubation. With prolonging incubation time, the three treatments (D − L − ; D − L + ; D + L − ) display the expected regular fungi morphology. In contrast, with the curcumin treatment (D + L + ), the spores did not germinate during the 16 days incubation. As shown in, in general the light-only and dye-only treatments (D − L + ) and (D + L − ) demonstrated a general increase in production of both secondary metabolites reaching 8.31 ± 0.54 and 11.18 ± 1.53 µg/g of botrydial, and 179.69 ± 30.04 and 205.29 ± 53.71 µg/g of dihydrobotrydial, respectively, at 15 days. The control treatment (D − L − ) produced a more variable botrydial concentrationwith a maximum of 7.41 ± 3.56 µg/g. At all incubation times, all samples from treatment (D + L + ) were below limits of detection for both botrydial and dihydrobotrydial, which is most likely due to the destruction of all B. cinerea spore by this treatment. In order to eliminate interference due to variability between plates, the ratio of dihydrobotrydial/botrydial was also calculated. The average metabolite ratio in the three treatments (D − L − ; D − L + ; D + L − ) is 19.42 ± 3.62, 21.83 ± 3.52 and 15.34 ± 3.20, respectively.
# Discussion
## Lc-ms/ms analysis of metabolites botrydial and dihydrobotrydial
High-resolution accurate mass (HRAM) spectrometers, such as the time-of-flight (TOF) and the Orbitrap spectrometer, has become more accessible and affordable to many laboratories in recent years. One of the biggest strengths of a HRAM instrument is its ability to acquire accurate mass data, which allows the user to relate the observed accurate mass to an equivalent elemental composition/s and molecular formula/s. Molecular formulas alone, however, are not sufficient to identify specific compounds. Without applicable standards, erroneous identifications can result due to the many possible compounds with the same molecular formula, and for this reason we have chosen to isolate authentic botrydial/dihydrobotrydial to use as standards in this study. These metabolites are not commercially available but can be obtained from B. cinerea cultures as described here.
Literature reports of UPLC-MS/MS analysis of similar authentic botrydial standards are lacking, but there are two recent literature reports on the detection of botrydial through the untargeted HRAM approach. Niedzwiecki et al.Notably, no loss of AcOH was observed, which in itself is inconsistent with botrydial structure (as is the loss of 2 × CO without loss of H 2 O). The MS 2 fragmentations of the metabolite detected by El Fellah et al.is seemingly different to that of our authenticated botrydial, and casts doubt on the water analysis results published in their report. Similarly the [M+Na] + and [M+NH 4 ] + molecular ions reported by Niedzwiecki et al.are not unique to botrydial and may or may not be actually botrydial.
Coincidentally, both the study by El Fellah et al.and the current investigation were conducted on the Orbitrap mass spectrometer platform with similar elution solvent systems. We, however, find the HRAM approach not ideal for the detection of botrydial (without standard) as accurate mass fragmentation data in the literature is lacking to support the direct identification of botrydial by accurate mass alone. During our own investigations we frequently noted a second background contaminant in our samples and/or in our LCMS system which unexpectedly gives an accurate mass which is in agreement with the [M+H] + molecular ion of botrydial under 5 ppm mass errors. This background compound did not co-elute with our authentic botrydial standard, nor did it exhibit the same MS fragmentation as our authenticated botrydial despite apparently having the same molecular formula accordingly to HRAM data. Consequently, without a botrydial standard we find it unreliable to identify botrydial based purely on the untargeted accurate mass data. Interestingly, dihydrobotrydial, which is a detoxification by-product of botrydial, was not reported from the two previous HRAM studies. With the obvious gap in the untargeted HRAM detection of botrydial, we have isolated the two diagnostic metabolites, botrydial and dihydrobotrydial, from B. cinerea and have provided here detailed HRAM and MS 2 accurate mass data for future comparison. The new analysis method described here also demonstrated good reproducibility and reliability in recoveries of both botrydial and dihydrobotrydial from spiked agar plates , enabling the use of this method to study effects of photosensitization on B. cinerea metabolite production.
## Effect of curcumin-mediated photosensitization on b. cinerea
This study demonstrated that the use of curcumin followed by irradiation with blue light (430 nm) promoted a significant reduction in the germination of B. cinerea. This result is in agreement with previous studies that have reported similar effects of curcumin and other photosensitizers with different food-borne pathogens. For example, Penha et al.reported that curcumin (75 µM) and blue illumination (470 nm) had synergistic effect on inactivation of food-borne pathogens with relative order of efficacy Aeromonas hydrophila > Salmonella aureus > Escherichia coli > Salmonella typhimurium. Josewin et al.similarly utilized blue light-emitting diodes (405-460 nm) with sodium chlorophyllin as photosensitizer to inactivate Listeria monocytogenes and Salmonella spp., minimizing the contamination risk of food-borne pathogens by consumption of cantaloupe. These authors achieved similar reductions with and without added photosensitizer on cantaloupe rind and speculated about the presence of an endogenous photosensitizer. Our research group found that curcumin combined with visible light (420 nm) was an effective treatment to inactivate spores of Aspergillus flavus both in vitro and in vivo in maize kernels. In subsequent studies, the percentage reduction in Aspergillus niger, A. flavus, Penicillium chrysogenum and Zygosacharomyces bailii remained above 50% with curcumin concentrations (100 to 400 µM) followed by illumination with a light dose of 96 J/cm 2. In the present study, at low curcumin concentration (50 to 400 µM) and under a similar light dosage, the percentage reduction in B. cinerea varied from 21.5 to 99.7%, which may reflect the differing susceptibility of the fungal species studied. Penicillium griseofulvum by comparison demonstrated less susceptibility in previous studies, than that recorded here for B. cinerea. In all cases, the phototoxic effect seems to depend on the curcumin concentration, which is in agreement with the previous reports. Fungi are usually enveloped by a thick rigid cell wall, which is made of chitin, mannoproteins and α-β glucans, and photoinactivation is therefore dependent on photosensitizer uptake through this cell wall and distribution to subcellular targets. The phototoxicity of curcumin is then directed through reactive oxygen species formed by the reaction between the light-activated dye molecules and the excited states of oxygen molecules. Sanita et al.reported that photodynamic therapy mediated by curcumin associated with LED light significantly reduced the metabolism of the biofilm organized cells of Candida dubliniensis. They also reported that exposure to curcumin alone in the highest concentrations (30 and 40 µM) significantly reduced the biofilm viability. Further, while curcumin was rapidly taken up by Candida dubliniensis cells, curcumin was reported to require a longer time interval to penetrate into biofilm cells. Araujo et al.reported that curcumin has a toxic effect on cariogenic pathogens only at appreciable concentrations (5.0 g/L) upon photoactivation. Providentially, curcumin is a food additive (E100) and a naturally polyphenol extracted from Curcuma longa with no significant reported toxicity, well-known for its beneficial biological activities, including antimicrobial anti-proliferative, anti-inflammatory and antioxidant activities. From an economic viewpoint, curcumin can be produced in high quantities at a reasonable cost, so it is an ideal photosensitizer. However, curcumin is a very hydrophobic molecule and requires some kind of formulation vehicle to enable higher curcumin load and improve interaction between dissolved curcumin and cell membranes. For consumer acceptance any change in flavour and colour of the treated commodity as a result of curcumin addition would also need to be considered.
The present study also evaluated the effect of different exposure time or light dose on curcumin-mediated photosensitization of B. cinerea and demonstrated that only a short 10 min period of illumination was required. The D + L + treatment achieved complete inactivation of B. cinerea when utilizing 800 µM curcumin and illumination time of 10 min, equivalent to a light dose of 120 J/cm 2. This result is analogous to previous studies with other microbial species. For example, Penh et al.reported that with increasing the illumination time (10 to 30 min) or light does (139 to 417 J/cm 2 ), curcuminmediated photosensitization induced a significant reduction in the counts of Gram-positive and Gram-negative bacteria, such as S. aureus, Pseudomonas aeruginosa and Aeromonas hydrophila. Similarly, Araujo et al.reported that a curcumin illumination time of 2 or 5 min was equally effective in reducing Streptococcus mutans and Lactobacillus acidophilus. It is, however, worth noting that some researchers have observed photobleaching with consequent decreases in effectiveness with increased photosensitization illumination times. For example, in evaluation of photodynamic therapy using a curcumin solution on root canals contaminated with Enterococcus faecalis, curcumin as sensitizer was effective with 5 min irradiation but not with 10 min irradiation, which the authors related to curcumin photobleaching.
This current study aimed at the identification of the most effective curcumin-mediated photosensitization treatment against B. cinerea. Therefore, the B. cinerea morphology was observed for a period of 16 days after treatment with a light does of 120 J/cm 2 and curcumin concentration of 800 µM. In comparison to curcumin-only or light-only treatment (D + L − and D − L + ), this study demonstrated that curcumin-mediated photosensitization (D + L + ) not only influences the mycelial growth of B. cinerea, but also causes a delay or inhibition in the germination of the fungal spore population. Curcumin has been reported to exhibit antifungal activity against Cryptococcus neoformans, Phytophthora infestans, Rhizoctonia solani, Candida albicans, and Erysiphe graminis. In addition, phototherapy has been suggested as a potential therapeutic alternative to antifungal treatment for the treatment of C. albicans biofilm infections. Many studies have demonstrated that blue light (a wavelength of 400-500 nm) alone exhibits significantly antimicrobial effects against methicillin-sensitive S. aureus, Acinetobacterand C. albicans. However, in the present study, neither curcumin nor light alone was effective in eliminating B. cinerea, which is a similar conclusion to the previous reports with A. flavus.
The exact mechanism of the photosensitizer inducer photodynamic inactivation has not been elucidated. However, many researchers believe that the photosensitization method is based on combined action of photosensitizer, visible light, and oxygen, producing the observed cytotoxic effect, in which light-activated photosensitizers transfer energy to oxygen to form reactive oxygen species and other free radicals which inactivate microorganisms by damaging proteins, cell membranes, and organelles.
In the present study, the effects of photosensitization on B. cinerea phytotoxin production were also considered. Phytotoxins, including alkaloids, terpenes, polyketides, non-ribosomal peptides, or metabolites of mixed biosynthetic origin, are secreted by necrotrophic pathogens to induce cell necrosis and leakage of nutrients. B. cinerea is one of most broad-host range fungal necrotrophs, and infects a large number of vegetables and fruit crops with subsequent economic loss both pre-and post-harvest. B. cinerea produces phytotoxins, ROS, as well as cell wall-degrading enzymes to induce necrosis of plant tissues. B. cinerea is also reported to trigger hypersensitive responses and ROS production in the host, which cause a series of programmed cell death to promote the infection process. Botrydial is the primary phytotoxic sesquiterpene metabolite secreted by B. cinerea, and triggers ROS production, hypersensitive response, and expression of defence genes. Therefore, it is considered that botrydial may act not only as a phytotoxin, but also as an elicitor of plant defence responses. For this reason, botrydial, as well as other B. cinerea-originated sesquiterpene metabolites, is regarded as an economically important fungal toxin for agriculturally and ornamental crops. Although dihydrobotydial has only moderate phytotoxic activity at high concentrations, it is one of the major B. cinerea metabolites and co-occurs with botrydial during isolation. The inhibition of the growth of B. cinerea has also been reported to be directly proportional to the botrydial concentration, with B. cinerea transforming botrydial to the less active phytotoxins such as dihydrobotrydial, botryenedial, and secobotrytrienediol. Therefore, both botrydial and dihydrobotrydial were chosen in the present study as detection index to investigate the function of our photodynamic treatment.
Measured metabolite concentrations in the present study did not exactly mirror the effects of photodynamic degradation of B. cinerea spores. With increasing concentration of curcumin between 0 and 200 µM, the botrydial concentration remained relatively unchanged despite the inhibition of fungal growth. However, when the concentration of curcumin was raised above 200 µM, the botrydial concentration decreased and the fungi growth was also inhibited. Interestingly the production of dihydrobotrydial was more impacted at the lower curcumin concentrations. One explanation is that fungal growth may cease or be inhibited as a result of relatively higher botrydial with lesser colonies and with the increase in the concentration of curcumin. Duran-Patron et al.similarly commented on the complexity of botrydial degradation pathways, with botrydial production during initial fungal growth stages regulating growth at higher concentrations, and metabolism to less inhibitory compounds, such as dihydrobotrydial, allowing growth to resume. It is noted that the botrydial concentrations measured in our agar platesare less than one tenth of the maximal botrydial levels reported in the liquid culture studies of Duran-Patron et al., and have not reached the levels reported by Duran-Patron et al. to regulate fungal growth. On the other hand, light-only (D − L + ) or curcumin-only (D + L − ) treatments may also effect the B. cinerea metabolite production, as seen in the varied ratio of dihydrobotrydial and botrydial produced. Lineiro et al.found that there is a connection between gene expression of virulence factors and B. cinerea culture and environmental conditions, whereby the botrydial biosynthesis can be inhibited when the culture uses a sole carbon, such as cellulose and tomato cell walls. Similarly, when the spore suspension was treated with either light or curcumin in this study, this change in B. cinerea culture conditions could lead to a changed relationship between the metabolites botrydial and dihydrobotrydial (and consequently the ratio of these metabolites).
In the food industry, the most widely employed postharvest fruit protection methods to reduce microbial spoilage include chemicals and physical methods such as controlled atmosphere, low temperature or modified atmosphere packaging. Photosensitization as an environmentally friendly method, represents a potential novel technology with application in food and beverage industries. De Oliveira et al.reported that combination of UV-A light and curcumin can significantly reduce bacterial cross-contamination of fresh produce. Tao et al.reported curcumin-based photosensitization inactivate E. coli on the surface of apple slices and preserve the quality of apple slices. Our in vitro study has demonstrated the effectiveness of curcumin-mediated photosensitization against B. cinerea, a common fruit and vegetable pathogen. Further studies are required to investigate the effectiveness of this treatment against B. cinerea in fresh fruit such as strawberries. Promisingly, recent studies in Australian-grown strawberries demonstrated that photosensitization extended the shelf-life and did not affect the physicochemical quality of the strawberry and retained key quality attributes.
# Conclusions
The observed reduction in B. cinerea spore germination was directly proportionate to curcumin concentrations, and both botrydial and dihydrobotrydial decreased with increasing curcumin concentration. In the treatment of D + L + with an illumination time of 10 min at curcumin concentration of 800 µM, the spores were completely inhibited and these botryane secondary metabolites could not be detected even after 16 days incubation. Under the other three treatments (D − L − , D − L + and D + L − ), the spores grew normally, with both botrydial and dihydrobotrydial being produced throughout the 16 days incubation period. Curcumin-mediated photosensitization therefore represents a potentially effective method to control B. cinerea infection, and warrants further investigation in vivo in susceptible fruit and vegetables.
# Materials and methods
# Fungal materials
Five filamentous fungi (Botrytis cinerea, Pestalotiopsis theae, Cladosporium sp., Penicillium raistrickii and Mucor rudolphii) were isolated and identified from four commercial varieties (Fortuna, Festival, Ruby Gem and Red Rhapsody) of strawberries grown on a Queensland farm, Australia. A randomly selected sample of ten fruits was washed thrice with potable water, and drained on clean paper towels. Then, each fruit was cut into four equal pieces using a sterilized scalper, and placed on potato dextrose agar (PDA) (Thermo Fisher Scientific, Victoria, Australia) plates. The plates were incubated at 25 - C. The samples were observed daily in a laminar flow cabinet for any fungal growth, and fungi were isolated as they appear on the cut pieces using sterile needles. B. cinerea isolated from these strawberries was identified by colony morphology. Pure cultures were observed under a light microscope (Leica, Germany) after staining with cotton blue lactophenol. Identification was confirmed by performing 18S rDNA analysis, as previously described. Specifically, the fungus was cultured on Czapek Dox agar (Thermo Fisher Scientific, Victoria, Australia) at 26 - C for 15 days, and the spores were harvested by flooding the plate with 10-15 mL 0.1% Tween 80 solution. The spore solution was then filtered through sterilized double folded cheesecloth to remove the hyphal fragments. The spore solution was diluted with sterile distilled water, and concentration determined by plating 0.1 mL aliquots of ten-fold dilutions on Dichloran Rose-Bengal Chloramphenicol Agar (DRBC) (Thermo Fisher Scientific, Victoria, Australia). The spores were stored in 15% glycerine at −20 - C. For the photosensitization experiments, spore suspensions of 10 4 CFU per mL were used, and cultured on Czapek Dox agar.
## Isolation of botrydial and dihydrobotrydial standards
Isolation of fungal metabolites was based on an adaption of the method of Lineiro et al.. B. cinerea was grown on PDA culture for 3 days and then transferred to individual 500 mL reagent bottles (20 bottles, 3 × 1 cm plugs of agar, ca. 0.75 g per bottle), each containing about 250 mL modified Czapek-Dox medium, which were continuously shaken at 250 rpm under constant room temperature (22 - C) with 12 h light/dark cycles for 13 days. All the liquid broths were combined (ca. 4.8 L) and then extracted with ethyl acetate (3 × 1 L) (Merck, New South Wales, Australia). The combined organic extract was dried over anhydrous Na 2 SO 4 (Merck, New South Wales, Australia) and concentrated under reduced pressure to obtain ca. 0.3 g oily residue. This residue was then subjected to silica flash column chromatography, eluting with an increasing polarity of hexane (Merck, New South Wales, Australia) and ethyl acetate mixtures from 100% hexane to 100% EtOAc. Flash column fractions (37 × 10 mL each) were collected and monitored by silica TLC under UV (254 nm) with vanillin and 2,4-dinitrophenyl-hydrazine (DNP) staining reagent applied to the air-dried TLC plates. Several fractions which were eluted with~70% hexane were subsequently combined and evaporated to provide a solid residue (10 mg), which was confirmed by HRAM UPLC-MS/MS and 1 H and 13 C NMR (as described below) to contain a mixture of botrydial (1) and dihydrobotrydial (2).
## Nmr analysis of botrydial and dihydrobotrydial standards
The botrydial/dihydrobotrydial mixture (10 mg) was analysed by 1 H and 13 C NMR in deuterated chloroform (Cambridge Isotope Laboratories, Inc.) on a 500 MHz Bruker Avance instrument (5 mm Selective Excitation Inverse (SEI) probe) with addition of a measured aliquot of dioxane as internal standard for NMR quantification. Integration of aldehyde and dioxane 1 H NMR resonances was used to determine the amount of botrydial present in the sample. The sample containing botrydial (0.8 mg) and dihydrobotrydial (7.2 mg) was recovered from the NMR solvent (CDCl 3 ), and used without further purification as a mixed external standard in HRAM UPLC-MS/MS analysis as described below.
## Photosensitizer and light source
A stock solution (2000 µM) of natural curcumin was prepared by dissolving 73.8 mg curcumin (Sigma Aldrich, St. Louis, MO, USA) in 50 mL propylene glycol (99.5%, Sigma Aldrich, St Louis, MO, USA), which was then diluted 1:1 in sterile water (50 mL). Our preliminary study showed that this solution had no inhibition effect on the spores (data not shown). The curcumin stock solution was stored in a dark cool place, and diluted as appropriate with sterile water to provide required curcumin concentrations for photosensitizer experiments. The illumination source was a 500-Watt xenon arc lamp (Polilight, PL 500, Rofin Australia Pty Ltd., Victoria, Australia) equipped with an optical fibre light with a range of 370-680 nm filters. This source was configured according to the previously described methodand a wavelength of 430 nm was selected. The light dose (J/cm 2 ) was calculated as irradiation time (s) multiplied by the light power (w) divided by the area of irradiation (cm 2 ).
## Determination of spore survival under differing photosensitization conditions
A wide concentration range of curcumin (50, 100, 200, 400, 600, 800, and 1000 µM) together with different levels of light doses (0, 120, 240, and 360 J/cm 2 , corresponding to an illumination time of 0, 10, 20 and 30 min, respectively) was utilized to determine an efficient photosensitization mediated curcumin treatment regime. Aliquots (1 mL) of both the fungal suspension and curcumin solution (1:1 v/v) were mixed in a Petri dish (30 mm diameter), and the light source was placed approximately 10 cm above the surface of the Petri dish. A magnetic stirrer bar was placed in the mixture (2 mL) to ensure constant stirring during illumination to expose all spores uniformly to the light energy. After illumination, aliquots (100 µL) of the treated mixture were transferred onto Czapek Dox agar (3 replicates per treatment) and incubated with 12 h light/dark cycles at 26 - C for 8 days to determine the colony forming units (CFU), with plates incubated under the same condition for a further 3 days for botrydial and dihydrobotrydial analysis. The percentage reduction in fungal spores after treatment was calculated by the number of the colony (CFU) as given below, with the results of 3 replicates averaged for each treatment.
[formula] %Reduction = (CFU control − CFU test )/CFU control × 100 [/formula]
The selected optimized curcumin concentration and light dose were then utilized to investigate the effect of curcumin-mediated photosensitization on B. cinerea spores using treatment regimes of D + L + : curcumin (800 µM) and light (120 J/cm 2 ); D − L − (control): no curcumin and no light; D − L + : no curcumin and light (120 J/cm 2 ); D + L − : curcumin (800 µM) and no light followed by incubation at 26 - C for 2-16 days.
## Determination of toxin production under differing photosensitization conditions
B. cinerea-inoculated agar plates after photosensitization treatment and incubation (as described above) were also analysed by LC-MS/MS to determine botrydial and dihydrobotrydial production. A 1 cm diameter plug of agar, with an average weight of 0.25 g, was selectively removed from the respective B. cinerea-inoculated agar plates and extracted with acetonitrile (2.5 mL) by sonication for 10 min. The sonicated sample was then filtered (0.2 µm polypropylene membrane syringe filter) and analysed by HRAM UPLC-MS/MS with results reported as µg botrydial or dihydrobotrydial per g of agar extracted. From each agar plate, three plugs were analysed as replicates.
Spiked recovery samples for metabolite analysis were prepared by addition of aliquots of the standard botrydial/dihydrobotrydial solution to 1 cm diameter plugs of agar from uninoculated agar plates at three different concentrations (n = 4), which were then extracted as above.
## Hram uplc-ms/ms conditions
UPLC separations were conducted out on an UltiMate 3000 RS UPLC system (Thermo Fisher Scientific, Bremen, Germany) equipped with an RS pump, a temperature control column compartment and an autosampler. An Agilent ZORBAX Eclipse Plus C 18 (2.1 mm × 100 mm; 1.8 µm) column was used for the LC separations. The UPLC parameters consisted of column oven at 20 - C, an injection volume of 2.0 µL, a flowrate of 0.2 mL/min and gradient elution with mobile phase (MP) A: 0.1% formic acid in H 2 O and MP B: 0.1% formic acid in MeOH (MP B). The mobile phase gradient was 5% MP B (t = 0) increasing to 40% MP B at 0.8 min, 80% MP B at 1.5 min, 95% MP B at 2.2 min before reaching 100% MP B at 4 min and held there until 7 min. MP B decreased to 5% at 7.3 min and was maintained at 5% until 10 min.
The HRAM MS/MS was acquired on a Q-Exactive mass spectrometer (ThermoFisher Scientific, Bremen, Germany) with a heated electrospray ionization (HESI) probe. The MS data were collected in the parallel reaction monitoring (PRM) positive mode at 35,000 FWHM mass resolution, AGC target at 2.0e5, maximum IT at 100 ms, mass isolation window
## Statistical analyses
All experiments were performed in triplicate. The value were expressed as mean ± SD. The analysis of statistical difference was performed using the TTEST function in Microsoft Excel 2010. Two-tails and 'type' for the test were choosen by the correspondence and variance of the data with FTEST. p < 0.05 (significant), or p < 0.001 (highly significant) were applied to determine any significant differences between specific means.
## Conflicts of interest:
The authors declare that there are no conflicts of interest. |
Occurrence of and factors influencing elderly homebound in Chinese urban community
Studies on the occurrence of homebound and the factors influencing it are available. However, the study of community homebound in China is still in its preliminary stage. No previous studies about this issue are available. This study aims to assess the occurrence of and factors influencing homebound elderly in Chinese communities and to provide a basis for effective intervention and prevention of homebound elderly people.One sample community from three provinces was randomly selected. Investigations were performed on the selected communities and 2180 elderly people were chosen as the research subjects. Unified survey scales were used. Home visit and face-to-face interviews were performed to ensure that no single qualified survey respondent was missed.The rate of morbidity in homebound elderly Chinese community was found to be 15.49% and it gradually increased with age, and also with a lower education or poorer Activities of Daily Living (ADL). Single factor analysis showed that general situation, living habits, physical condition, mental condition, society, social support, and other factors affected the occurrence of community homebound elderly. Women were more likely to be homebound than men (P < .05). Having a spouse or high income reduced the rate of morbidity in the homebound elderly (P < .05). Multifactor regression analysis revealed that poor ADL, depression, hearing impairment, being old, no exercise, and low social support are the main influencing factors.Appropriate measures should be taken based on the specific influencing factor to prevent the occurrence of homebound.Abbreviations: ADL =Activities of Daily Living, GDS-15 = Geriatric Depression Scale Short Form, SSRS = Social Support Rating Scale.
# Introduction
According to The National Economic and Social Development Statistics Announcement of the People's Republic of China, 2014, by the end of 2014, elderly people aged >60 years accounted for 15.5% of the total population of China. These elderly people and their health conditions have become the focus of the entire society. [bib_ref] Present situation of pension service demand in our country and long-term care..., Xie [/bib_ref] China's problem of aging population has become serious. As the elderly people grow even older, their physiological functions and immunity gradually decline, which leads to increasingly serious health problems. At the same time, homebound results in the reduction of elderly people's physical and cognitive functions, weakens their self-maintenance ability in daily living, and intensifies their psychological stress. The lack of communication with the outside world eventually result in lying on bed or dementia, seriously impairing the elderly people's physical and mental health and life quality, and also placing huge pressure on their families, society, and social medical treatment. [bib_ref] A review of social isolation: an important but underassessed condition in older..., Nicholson [/bib_ref] [bib_ref] Systematic review of outcomes from home-based primary care programs for homebound older..., Stall [/bib_ref] Therefore, the concern on elderly homebound is essentially important in maintaining and improving the quality of life of elderly people, and also alleviating the burden on family and society, as well as preventing lying on bed and dementia.
Researchers have performed a number of studies on the occurrence of homebound and the factors influencing it and developed a variety of measures. However, the study of community homebound in China is still in its preliminary stage; so far the available investigation is confined only to an elderly homebound survey of a particular urban community, with limited scope. No studies on the occurrence of and factors influencing elderly homebound in Chinese urban community are available. Hence in this study, Chinese urban elderly people were chosen for investigating the occurrence of and the factors influencing elderly homebound in Chinese urban community, and provide a basis for community medical personnel to give individual intervention.
# Subjects and methods
## Subjects
The inclusion criteria were as follows: elderly people (aged ≥60 years) living in urban communities (having resided in the community for >1 year) in China were enrolled in this study.
The exclusion criteria were as follows: people who had serious speech impediment, hearing impairment, or visual impairment and those who were laid up.
Informed consent was obtained from all enrolled subjects. The study was approved by the Ethics Committee of the Local Disease Control and Prevention Center (2013150101).
The sample size was calculated as follows: Based on the related foreign studies, the rate of morbidity in community elderly homebound was 5.9% to 18.6%. [bib_ref] Mental health services for homebound elders from home health nursing agencies and..., Zeltzer [/bib_ref] [bib_ref] Homebound older adults: prevalence, characteristics, health care utilization and quality of care, Musich [/bib_ref] In a Chinese study, the reported rate of morbidity in elderly homebound of a particular urban community was 17.59% to 18.8%. [bib_ref] Homebound status of older workers and influencing factors of the investigation and..., Xing [/bib_ref] In the present study, the rate of morbidity of 18% was used, and the sample size was calculated using the formula N = Z 2 PQ/(0.1P) 2 which is simplified as N = 400Q/P(Q = 1 À P), where Q = 1 À P, Z (=1.96 ≈ 2.00) represents the Chi-squared value.Accordingly, the sample size calculated was N = 1822. Considering the possible loss of data and nonresponse, the sample size was increased by 10%. Hence, the sample size of this study should no more than 2004 subjects.
# Sampling method
Random cluster sampling method was used in this study. At Stage 1, Anhui Province (South China) and Hebei Province (North China) were selected based on the geographical location. At Stage 2, the 11 cities of the Hebei Province were numbered and 2 cities, that is, Baoding City and Tangshan City, were randomly drawn; similarly, the 16 cities of Anhui Province were numbered and 1 city, that is, Bengbu City, was randomly drawn. At Stage 3, 1 sample community (Dajijia community) from the 159 communities of Baoding City, 1 sample (Diaoyutai community) from the 363 communities of Tangshan City, and another sample (Yan'an community) from the 204 communities of Bengbu City were randomly selected. From these communities, a total of 2180 qualified elderly people were enrolled. A total of 2133 questionnaires were handed out to the subjects and were retrieved. Among all the retrieved questionnaires, 28 were invalid. The remaining 2105 copies were regarded as valid. The percentage of valid questionnaires was 98.69%.
## Survey
From July 2013 to December 2013, a survey was conducted among 2180 qualified elderly people from the selected communities. Unified survey scales were used including homemade general situation questionnaire, Activities of Daily Living (ADL) ability scale, Social Support Rating Scale (SSRS), Geriatric Depression Scale Short Form (GDS-15), and homebound condition assessment scale in Chinese version. Home visit and face-to-face interviews were performed among these research subjects to ensure that no single qualified survey respondent was missed. Questionnaires were retrieved and examined on the spot to ensure that there were no unfilled/missed items in the questionnaire.
## Assessment
Twenty senior professors were selected from the Department of Health Statistics, Epidemiology, and Nursing, eventually an expert consultation group was constituted. All members of this group had the title of deputy senior or above and were engaged in professional teaching and research study for >10 years. The Delphi expert enquiry method was adopted to revise the content of the questionnaire. 2.4.3. Physical condition. Subject's hearing ability, eyesight, chronic diseases, medications, and ADL ability. Among these indexes, the ADL was assessed using the ADL ability scale described by Lawton and Brody in 1969. [bib_ref] Assessment of older people: self-maintaining and instrumental activities of daily living, Lawton [/bib_ref] The ADL scale has 14 items and 4 grades. The studied subjects were assigned a score based on their ability or assistance needed to perform these activities, that is, 1, 2, 3, and 4 points, respectively, for "can do it by yourself," "a bit difficult," "need help," and "cannot do." Hence, the final score ranged from 14 to 56. A final score of 14 points means totally normal; a final score >14 points implies a certain level of dysfunction; and a score of >22 points indicates distinct dysfunction. The Cronbach a coefficient of the scale was set as 0.85 while the repeatability was 0.93. This scale was used to investigate the daily living of the studied subjects for the recent week.
## Social conditions.
Social support, participation in community or social activities, and networking with friends, relatives, and neighbors. Among these indexes, social support was assessed using the SSRS designed by Xiao.This scale was designed and utilized first in 1986, and some minor changes were made in 1990 and further promoted. The scale contained 3 dimensions and 10 items, that is, objective support (3 items), subjective support (4 items), and utilization of support (3 items). The Cronbach a coefficient of the scale ranged from 0.89 to 0.94, while the repeatability was 0.92. The sum score of the 10 items was the final score of social support, which ranged from 8 to 66 points. A final score of 22 points was treated as low level; a final score of 23 to 44 points was treated as medium level; and a final score of 45 to 66 points was treated as high level. A higher score implies more social support can be obtained.
2.4.5. Psychological condition. Self-assessment of health condition, depression, and loneliness. GDS-15 [bib_ref] Ethnicity, social support, and depression among elderly Chilean people, Gallardo-Peralta [/bib_ref] was adopted to grade the level of depression in the elderly people. Out of 15 points, those who scored 0 to 5 points were considered as normal elderly people, while those who scored 6 points or higher were considered as depressed. A higher score indicates more serious depression.
2.4.6. Homebound status. People who go out of house less than once a week are considered as homebound. [bib_ref] Homebound status increases death risk within two years in the elderly: results..., Herr [/bib_ref] (The going out record of the month before the survey was used as the reference data. If the going-out counts for each week of the month differ, the total counts for that 1 month averaged by the number of weeks was considered.)
# Statistical analysis
All obtained data were statistically analyzed using SPSS 13.0 software. Measured data were expressed as x ± s and analyzed using t test, while count data were analyzed using Chi-square test. The statistically significant results of single factor analysis were further processed through stepwise logistic regression analysis. The inclusion criteria for a was 0.05, while the exclusion criteria for a was 0.10. P < .05 was considered statistically significant.
# Results
## General factors affect homebound
From the selected communities, 2180 qualified elderly people were initially enrolled. Among them 2133 people were investigated (response rate of 97.84%). Among all the retrieved questionnaires, 28 were invalid and hence the remaining 2105 were regarded as valid. The percentage of valid questionnaires was 98.69% [fig_ref] Figure 1: A flow diagram summarizes the selection of studies [/fig_ref]. The ages of respondents ranged from 60 to 101 years including 882 (41.9%) men with a mean age of 71.74 ± 8.18 years and 1223 (58.1%) women with a mean age of 72.00 ± 8.61 years. Among these elderly people, 326 people (209 women and 117 men) were regarded as homebound, providing a rate of morbidity of homebound of 15.49%. It was found that older and less educated people were probably more homebound (P < .05) [fig_ref] Table 1: General factors affect homebound [/fig_ref]. It was also found that women with low income and without a spouse showed a higher prevalence of homebound than men, high income group, and those with a spouse (P < .05).
## Living habits affect homebound
Those elderly people who have hobbies, exercised regularly, and do not smoke or drink showed a lower prevalence of homebound than the other groups.
Compare with the nonhomebound, those homebound elderlies have high rate of no hobby (87.4% vs 81.3%) and low rate of hobby (12.6% vs 18.7%), The difference is statistically significant (P < .05). The homebound were doing less exercise (no exercise 48.8%, used to do, but not now 11.3%, sometimes 9.2%, often 30.7%), then the nonhomebound were doing more exercise (no exercise 14.3%, used to do, but not now 2.2%, sometimes 12.8%, often 70.7%), The difference is statistically significant (P < .05). The rates of smoking and drinking of the 2 groups (homebound vs nonhomebound) are 13.5% versus 18.7% and 7.1% versus 15.5%, respectively. The P values of the 2 indicators (smoking and drinking) between the 2 groups are both less than .05, which means the differences between the 2 groups are statistically significant at a confidence level of 95%.
3.3. Physical, social connection, and psychological condition affect homebound 3.3.1. Physical condition. With visual impairment, hearing impairment, chronic diseases, and medication, the prevalence of homebound elderly people was higher than the other groups. With ADL ability, the prevalence of homebound was higher, and the difference was statistically significant (P < .05) [fig_ref] Table 2: Physical, social, and psychological condition affect homebound [/fig_ref].
Those elderly people with visual impairment, hearing impairment, chronic diseases, and medication showed a higher prevalence of homebound than the other groups (P < .05). A poorer ADL ability resulted in a higher prevalence of homebound (P < .05), and the differences were statistically significant [fig_ref] Table 2: Physical, social, and psychological condition affect homebound [/fig_ref].
## Social connection.
Less social support resulted in higher prevalence of elderly homebound, and the difference was statistically significant (P < .05). Those who have frequent contacts with friends, neighbors, and relatives, and frequently participated in social activities showed a lower prevalence of homebound than the other groups (P < .05), and the difference was statistically significant [fig_ref] Table 2: Physical, social, and psychological condition affect homebound [/fig_ref].
## Psychological condition.
Those homebound elderly people who graded themselves as relatively healthy in selfassessment of health level showed the lowest rate of morbidity (P < .05). Those homebound elderly people who usually felt lonely showed a high rate of morbidity (P < .05). Those homebound elderly people who experienced depression showed a higher rate of morbidity than those have no symptoms of depression (P < .05) [fig_ref] Table 2: Physical, social, and psychological condition affect homebound [/fig_ref].
## Logistic regression analysis of multiple factors influencing elderly homebound
The occurrence of homebound was taken as a dependent variable and those meaningful factors in the single factor analysis as the independent variables. Using the stepwise regression method, the multiple logistic regression analysis was performed, because the AIC value for this model is the smallest, which means the model is best fitted. Results proved that poor ADL, depressive symptoms, being old and hearing impairment are the risk factors for affecting the occurrence of community elderly homebound since the OR values for these variables are more than 1 while the BMI, low social support and lacking exercise are the protective factors since the OR values for these variables are less than 1 [fig_ref] Table 3: Unconditional logistic regression analysis of factors influencing elderly homebound [/fig_ref].
# Discussion
The results of the study showed that the prevalence of Chinese community elderly homebound was 15.49%. In the United States, the results of a survey conducted in 2012 [bib_ref] Homebound older persons: prevalence, characteristics, and longitudinal predictors, Cohen-Mansfield [/bib_ref] showed the prevalence of homebound elderly people to be 19.0%, while another survey conducted in 2015 showed that the prevalence rate was 5.9%. [bib_ref] Homebound older adults: prevalence, characteristics, health care utilization and quality of care, Musich [/bib_ref] The prevalence of urban homebound elderly people in Brazil was found to be 22.4%. [bib_ref] Prevalence of homebound elderly people in the urban region of Belo Horizonte..., Ursine [/bib_ref] The prevalence in big cities of Japan is 14.4%. [bib_ref] The prevalence of homebound individuals in the elderly population: a survey in..., Umegaki [/bib_ref] The results of the survey conducted in China and other countries, based on the same definition of homebound, on the prevalence of homebound urban community elderly people were different. The reasons could be first, the ages of selected study subjects were different; second, different sampling methods were used; third, regional differences, which meant the social environment, lifestyle, and economic levels of the Chinese elderly people were different from other countries, were found.
The results showed that the age of the elderly people affected the occurrence of homebound. With the increase in age, the prevalence of homebound increases gradually. Studieshave found that nearly 30% of the homebound people are aged ≥80 years. Physiological function and social activity ability weakened with the increase in age, [bib_ref] An operational definition of the homebound, Gilbert [/bib_ref] [bib_ref] The effect of homebound status on older persons, Cohen-Mansfield [/bib_ref] leading to a gradual increase in the occurrence of elderly homebound. Attention should be paid to the elderly, especially the health condition of the very old people. The results of the study showed that the prevalence of homebound women was higher than men. Many overseas studies have found that the prevalence of homebound women is significantly higher than men. The reasons could be complex, [bib_ref] Older homebound women's perceived risk of being unable to reach help quickly:..., Porter [/bib_ref] which may be due to diet habits, sexual hormone levels, work stress, and social pressure, and also due to the profound influence on their behaviors by the traditional role of a man and a woman in the family, that is, men are encouraged to work outside, while women are encouraged to work at home. Hence men preferred social activities outside the family, while women focused more on family and children and stayed at home.
The results of multifactor regression analysis showed that physical exercise is the main factor influencing homebound elderly people. Those elderly who exercised regularly had a low rate of morbidity among the homebound. Those elderly who exercise regularly tend to have a relatively correct health concept, pay attention to their physical condition, maintain a positive attitude and mindset every day, and have sufficient confidence in their life. [bib_ref] Homebound status of older workers and influencing factors of the investigation and..., Xing [/bib_ref] The results also showed that these elderly people had more friends and richer entertainment activities, hence these elderly would not stay at home every day.
The results showed that hearing impairment is also a main factor influencing the occurrence of homebound in elderly people. A number of elderly people choose to reduce their frequency of going out due to the impaired hearing, which seriously affects their communication with the outside world.The results of multifactor regression analysis showed that the ADL is an independent risk factor that affected the occurrence of homebound, a higher score of ADL ability resulted in a lower prevalence of homebound, and 1 study also showed that ADL was an independent factor for the occurrence of homebound.Studies [bib_ref] Physical and mental health of homebound older adults: an overlooked population, Qiu [/bib_ref] have showed that none of the elderly people who were able to take bus were homebound; 20.3% of the elderly people who can only walk 5 m were homebound; 75% of the elderly people who lost the ability to walk were homebound. A poor ADL ability of the elderly people makes their physical movement difficult, making it difficult to go out and hence more likely to result in homebound.
Results showed that social support affected the occurrence of homebound elderly people in Chinese community. A lower social support resulted in a higher prevalence of homebound. A number of overseas studies [bib_ref] Behavior change following a selfmanagement intervention for homebound older adults with arthritis:..., Nour [/bib_ref] have proven that elderly people with poor social support have a higher prevalence of homebound than those with strong social support. Social support can help the elderly people to cope with stress faced in daily living, alleviate the impact of negative events occurring in their lives, and protect their physical and mental health. Hence social support is an important factor that helps the elderly people in receiving treatment, overcoming disease, and maintaining a good psychological status, eventually improving their quality of life. [bib_ref] Factors associated with depression experience of immigrant population: a study of Korean..., Kim [/bib_ref] Social support to elderly people should be given attention, gradually complete the social support items, and reduce the prevalence of homebound.
The present study showed that those who often feel lonely scored their health condition to be poor and those who have depressive symptoms have a high rate of prevalence of homebound. The result of multifactor regression analysis showed that depression is a risk factor affecting the occurrence of homebound urban community elderly people in China. A number of overseas studies have also confirmed that depression leads to high prevalence of homebound elderly people. The incidence of depression in the homebound elderly people was higher than the nonhomebound elderly people. [bib_ref] Assessment of a brief CES-D measure for depression in homebound medically ill..., Gellis [/bib_ref] With the presence of a depressive disorder, the elderly people are usually depressed, frustrated, low in emotion, self-abased, reluctant to communicate with people, with limited outdoor activities. [bib_ref] Comparison of depressive symptoms between homebound older adults and ambulatory older adults, Choi [/bib_ref] Studies have shown that [bib_ref] Loneliness, health and social network among elderly people-a follow-up study, Holmen [/bib_ref] age is an important factor intensifying the feeling of loneliness in the elderly people. Older age leads to a stronger feeling of loneliness. A high self-rating on health can slow down the aging of physiology, while a poor subjective sense of health can worsen the self-maintenance ability. Hence the high incidence of homebound occurred among those who had a low self-rating of health.
## Summary
The study used the multistage layered random cluster sampling method and investigated the current situation and factors influencing homebound elderly people in Chinese urban community. The results are as follows: The prevalence of homebound urban community elderly people in China is 15.49%. Poor ADL, depressive symptoms, hearing impairment, being old, no exercise, and low social support are the independent factors influencing the occurrence of homebound urban community Chinese elderly people.
The incidence of homebound urban community elderly people is relatively high in China and the factors influencing are complex. Hence attention should be paid to the problem of homebound elderly, conduct homebound-related health education events in the community, introduce elderly people, families, and the community medical staff to homebound-related knowledge and its harmful effects, and guide the elderly to maintain good living habits. The community should regularly organize activities that cater to elderly people, encourage the elderly people to walk out of their houses to join activities and do exercise, and also to interact with other people. Taking appropriate measures based on the specific influencing factor can effectively prevent the occurrence of homebound or becoming severe in elderly people, therefore effectively improving their quality of life and alleviate the endowment stress on their family and the society.
[fig] Figure 1: A flow diagram summarizes the selection of studies. [/fig]
[table] Table 1: General factors affect homebound (n = 2105).BMI = body mass index. *Trend value x 2 = 262.362, P < .05. † Trend value x 2 = 3.162, P < .05. ‡ Trend value x 2 = 10.836, P < .05. [/table]
[table] Table 2: Physical, social, and psychological condition affect homebound. [/table]
[table] Table 3: Unconditional logistic regression analysis of factors influencing elderly homebound. ADL = Activities of Daily Living, CI = confidence interval, OR = odds ratio. [/table]
|
Lemierre syndrome: remember the forgotten disease.
# Introduction
Lemierre syndrome, described as the 'forgotten disease'' although very uncommon2 is an infection-related syndrome which is important to diagnose accurately since its mortality may be high.
CASEREPORT A 14-yearoldgirl,withahistory of chronic mastoiditis, who gave a three-week history of left otalgia and headache presented to hospital with worsening pain, dizziness, vomiting and rigors following seven days oforal amoxicillin for presumed acute otitis media. She was febrile and had cervical lymphadenopathy with neck pain and tenderness. A diagnosis of acute suppurative otitis media was made for which she received intravenous cefotaxime 2g 8-hourly and metronidazole 500mg 8-hourly. Over the following five days she had persisting fever and worsening left-sided neck pain. Furthermore, she became dyspnoeic, developing pleuritic chest pain and haemoptysis; measured oxygen saturation on room air was 88%. Computerised tomography (CT) revealed thrombus, bubbles of gas in the left internal jugular vein and several ill defined opacities in the lung fields consistent with septic pulmonary emboli . Further imaging identified bilateral sclerosis of the mastoids, greater on the left, consistent with chronic mastoiditis; small gas collections and soft tissue changes consistent with acute inflammation were also present [fig_ref] Fig 3: CT scan of head showing mastoid sclerosis, especially on left [/fig_ref]. Blood cultures were positive for Bacteroides fragilis and an unidentified anaerobic Streptococcus sp. The patient underwent a left sided modified radical mastoidectomy and removal of granulations. She received 14 days of intravenous piperacillinl/ tazobactam 4.5g 8-hourly and metronidazole 500mg 6-hourly followed by a further 14 days of oral co-amoxiclav and metronidazole. Initial anticoagulation was with subcutaneous enoxaparin 100mg daily and, subsequently, warfarin for a total of two months. She remains well six months later.
# Discussion
Although the correlation between oropharyngeal infection and sepsis was first described by Courmant and Cade in 1900, [bib_ref] Sur une septico-pyohemie de l'homme simulant la peste et causee par un..., Courmont [/bib_ref] intervention and broad spectrum antibiotic therapy with potent anaerobic activity is the cornerstone of management; examples may include either clindamycin or a combination of a beta-lactam and metronidazole. The preferred duration of antibiotic therapy is 2-6 weeks, with conversion from intravenous to oral administration when marked improvement with defervescence, resolution of leucocytosis and falling inflammatory indices are observed. Mortality today is less than 17%,6 much lower than in the preantibiotic era. Furthermore, the role of anticoagulation remains unclear since no randomised controlled trials exist to test the hypothesis that this is beneficial. Interestingly, case series suggest that those who are anticoagulated have similar outcome to those who are not; indeed a small number suffer adverse events attributable to warfarin therapy.7 Some have recommended that, except when cavernous sinus thrombosis is present, anticoagulation is withheld. [bib_ref] Human necrobacillosis, with emphasis on Lemierre's syndrome, Hagelskjaer [/bib_ref] Lemierre remarked, in his original description, that upon observation of the constellation of findings which characterise the syndrome "mistake is almost impossible".4 However, unfamiliarity with this disease may confound the diagnosis and lead to missed therapeutic opportunity. The clinical lesson is, therefore, straightforward: when presented with a young, previously well patient who has a primary head and neck infection but a clinical illness out of proportion with this, often with symptoms at a discrete site, remember the forgotten disease.
[fig] Fig 1 Fig 2: CT scan of neck highlighting the thrombus (manifest by the lack of contrast agent seen) in the left internal jugular vein. CT scan of chest showing several opacities with surrounding infiltrate attributed to septic emboli. [/fig]
[fig] Fig 3: CT scan of head showing mastoid sclerosis, especially on left. [/fig]
|
Synthesis and Electrochemical Proprieties of Novel Unsymmetrical Bis-Tetrathiafulvalenes and Electrical Conductivity of Their Charge Transfer Complexes with Tetracyanoquinodimethane (TCNQ)
The synthesis and properties of a series of bis-tetrathiafulvalenes (bis-TTFs) containing nitrophenyl, aminophenyl or dimethylaminophenyl is reported. The synthesis was carried out by using routes involving Wittig-type, cross-coupling, reduction and alkylation reactions. The electron donor ability of these new compounds has been measured by cyclic voltammetry (CV). Charge transfer complexes with tetracyanoquinodimethane (TCNQ) were prepared by chemical redox reactions. The complexes have been proven to give conducting materials.
# Introduction
Tetrathiafulvalene (TTF) and its derivatives have attracted attention for many years because of their electron donor ability and the electrical conductivity of their charge transfer salts, which was started in 1973 with the synthesis of the tetrathiafulvalene-tetracyanoquinodimethane complex (TTF-TCNQ) by Cowan and coworkers [bib_ref] Electron transfer in a new highly conducting donor-acceptor complex, Ferraris [/bib_ref]. Since then, the progress made in the synthesis of such molecules has been closely related to the discovery of new materials [bib_ref] New concepts in tetrathiafulvalene chemistry, Segura [/bib_ref] exhibiting conducting [bib_ref] Dimensionality and electrical properties in organic synthetic metals-Current results through selected recent..., Fabre [/bib_ref] [bib_ref] Hybrid molecular materials based upon organic π-electron donors and inorganic metal complexes...., Clemente-León [/bib_ref] [bib_ref] Single crystalline commensurate metallic assemblages of π-slabs and CdI 2 -type layers:..., Devic [/bib_ref] [bib_ref] Highly conducting crystals based on single-component gold complexes with extended-TTF dithiolate ligands, Suzuki [/bib_ref] , superconducting [bib_ref] A new organic superconductor, (DODHT) 2 BF 4 ·H 2 O, Nishikawa [/bib_ref] , magnetic [bib_ref] π-d Interaction-based molecular magnets, Miyazaki [/bib_ref] [bib_ref] Unconventional TTF-based molecular magnets, Enoki [/bib_ref] , or optical properties [bib_ref] Synthesis and properties of push-pull chromophores for second-order nonlinear optics derived from..., Otero [/bib_ref] [bib_ref] Synthesis of novel phthalocyanine-tetrathiafulvalene hybrids; intramolecular fluorescence quenching related to molecular geometry, Farren [/bib_ref].
The TTF core and its derivatives-due to their characteristics, in particular their stability and reversible redox character-have found a significant number of applications in materials chemistry [bib_ref] Functionalised tetrathiafulvalenes: New applications as versatile π-electron systems in materials chemistry, Bryce [/bib_ref] such as molecular switches rotaxanes and catenanes [bib_ref] A chemically and electrochemically switchable [2] catenane incorporating a tetrathiafulvalene unit, Asakawa [/bib_ref] , conductive materials [bib_ref] Multistability in a BEDT-TTF based molecular conductor, Laukhina [/bib_ref] and superconductors [bib_ref] A new organic superconductor, (DODHT) 2 BF 4 ·H 2 O, Nishikawa [/bib_ref] , complex with the C 60 [bib_ref] Stabilisation of charge-separated states via gain of aromaticity and planarity of the..., Martin [/bib_ref] , conductive polymers [bib_ref] Linearly extended π-donors: When tetrathiafulvalene meets conjugated oligomers and polymers, Roncali [/bib_ref] , materials for nonlinear optics [bib_ref] Synthesis of novel phthalocyanine-tetrathiafulvalene hybrids; intramolecular fluorescence quenching related to molecular geometry, Farren [/bib_ref] , sponges cations [bib_ref] Crown ether derivatives of tetrathiafulvalene, Hansen [/bib_ref] , ferromagnetic organic magnets, liquid crystals [bib_ref] Induction and variation of discotic columnar phases through doping with electron acceptors, Bengs [/bib_ref] , and dendrimers [bib_ref] Macromolecular tetrathiafulvalene chemistry, Bryce [/bib_ref]. Our research is focused on the conducting and superconducting materials in the hope to improve results in the field and to aim to find an organic superconductor at room temperature.
Dimeric tetrathiafulvalenes are currently of interest to chemists for their applications in organic molecular materials. They can be divides into two types: one is linked through conjugated π-systems and the other through non-conjugated σ-chains [bib_ref] New concepts in tetrathiafulvalene chemistry, Segura [/bib_ref].
In a continuation of previous work of our group [bib_ref] New TTF and bis-TTF containing thiophene units: Electrical properties of the resulting..., Abbaz [/bib_ref] [bib_ref] Characterization of the anion-ordering transition in (TMTTF) 2 ReO 4 by x-ray..., Subias [/bib_ref] [bib_ref] Quadratic nonlinear optical response in partially charged donor-substituted tetrathiafulvalene: From a computational..., Lamère [/bib_ref] , in this paper we now describe the synthesis and properties of some new unsymmetrically π-donors of bis-TTFs which linked directly by σ-bond with alkyl chains of different lengths, containing nitrophenyl, aminophenyl or dimethylaminophenyl groups, synthesized via cross-coupling, reduction and alkylation methods. The redox behavior of such precursors has been studied by cyclic voltammetry and finally the electrical conductivity of charge transfer complexes was measured.
# Results and discussion
Several steps are needed to convert the 4-(p-nitrophenyl)-1,3-dithiole-2-thione 1a in the corresponding 4-(p-nitrophenyl)-1,3-dithiolium tetrafluoroborate 4 (Scheme 1). The first step consists in methylating the thione 1a [bib_ref] The 1,2,3-thiadiazole route to new vinylogue tetrathiafulvalenes, Clausen [/bib_ref] with methyl triflate in anhydrous CH 2 Cl 2 at 0 °C. The desired salt, 2-methylthio-4-(p-nitrophenyl)-1,3-dithiolium trifluoromethane sulfonate 2, isolated as a precipitate by simple addition of cold ether in 95% yield. During the second step, the salt 2 placed in suspension in ethanol is reduced by sodium borohydride at 0 °C. 2-methylthio-4-(p-nitrophenyl)-1,3-dithiole 3 is isolated as oil in 75% yield, which is treated directly in the next step. Finally, dethiomethylation of the reduced product 3 with tetrafluoroboric acid in acetic anhydride at 0 °C leads to 1,3-dithiolium 4 expected with a yield of 45%.
The synthesis of electron donors 9, 10a and 10b based on a multi-step procedure were carried out as shown in Scheme 2. We have synthesized 2,3-bis(2-cyanoethylthio)-6-p-nitrophenyl tetrathiafulvalene 9 by two different methodologies. The first involved a Wittig-type reaction using a weak base such as triethylamine which does not remove cyanoethyl protecting groups. The action of triethylamine in acetonitrile at room temperature on the phosphonium salt 4-(p-nitrophenyl)-1,3-dithiole-2-yl-triphenylphosphonium tetrafluoroborate 5 [bib_ref] Reparation and chemistry of new unsymmetrically substituted tetrachalcogenofulvalenes bearing CN(CH 2 )..., Binet [/bib_ref] , generates the ylid, which reacts on the 1,3-dithiolium salt 4 to give the adduct intermediate. This is converted, by elimination of triphenylphosphine to 2,3-bis(2-cyanoethylthio)-6-p-nitrophenyl tetrathiafulvalene 9 in 30% yield.
The second route involved the reaction of chalcogenones 4-(p-nitrophenyl)-1,3-dithiole-2-thione 1a or 4-(p-nitrophenyl)-1,3-dithiole-2-one 1b [bib_ref] Dithiolium derivatives. II. Some new 1,3-dithiolium perchlorates, Campaigne [/bib_ref] with 5-bis-(2-cyanoethylthio)-1,3-dithiole-2-one 6 [bib_ref] Reparation and chemistry of new unsymmetrically substituted tetrachalcogenofulvalenes bearing CN(CH 2 )..., Binet [/bib_ref] , via a cross coupling method [bib_ref] Ethylenedioxy substituted 2,5-bis(1′,3′-dithiol-2′-ylidene)-1,3,4,6-tetrathiapentalenes and their conducting salts, Misaki [/bib_ref] in toluene at reflux in the presence of triethyl phosphite. As the reactivity of precursor 2-one was much higher than that of precursor 2-thione, mono-TTF 9 bearing two cyanoethylthio groups was obtained in 35% and 57% yield, respectively. The deprotection of bis(2-cyanoethyl) groups of 9 with the aid of CsOH and sequentially reaction with 10 equivalents of 1,2-dibromoethane or 1,3-dibromopropane afforded the unsymmetrically functionalized mono-TTF derivatives 2,3-bis(2-bromoethylthio)-6-p-nitrophenyl tetrathiafulvalene 10a or 2,3-bis(3-bromopropylthio)-6-p-nitrophenyl tetrathiafulvalene 10b in 45% and 44% yield, respectively after purification by column chromatography.
The syntheses of bis-TTFs 11-14, also based on a multi-step procedure, were carried out as shown in Scheme 3. The electrodonors 11a and 11b containing two TTF units were obtained in a mixture of cis/trans isomers in 40% and 38% yield, respectively, by using the reaction of 10a or 10b with 2,3-bis(2-cyanoethylthio)-6,7-di(methyl)tetrathiafulvalene 8 [bib_ref] Synthesis of new functionalized π-electron donors: Primary hydroxy and primary amino multisulfur..., Binet [/bib_ref] using cesium hydroxide in DMF at room temperature. In this reaction, the risk of formation of a polymer is high, and to limit the polymerization it is necessary to work at high dilution (use a syringe pump). Thus, compounds 12a and 12b were obtained in 88% and 94% yield, respectively, by treatment of compounds 11a or 11b by cesium hydroxide in DMF at room temperature.
After, the nitro group of bis-TTFs 12a or 12b was reduced into an amino group in ethanol at reflux in the presence of tin and hydrochloric acid. The amino substituted bis-TTFs derivatives 13a or 13b were obtained after purification by column chromatography in 52% and 48% yield, respectively.
Finally the alkylation of amino bis-TTFs 13a or 13b was effected by treatment with K 2 CO 3 (2 equiv.) and with 4 equivalents of iodomethane at room temperature, the dimethylamino bis-TTFs 14a or 14b were isolated by filtration and then extracted with CH 2 Cl 2 and washed with water. 14a and 14b were obtained in 76% and 75% yields, respectively.
In the 1 H NMR spectra of compound 9, protons of CH=C exhibit a singlet at δ 6.82 ppm, in addition, protons of CH 2 CN showed a triplet at 2.78 ppm and 2.83 ppm, with coupling constants of 7.07 Hz and 6.00 Hz, respectively. Thus, the protons SCH 2 showed a triplet at 3.14 ppm and 3.21 ppm, with coupling constants of 7.07 Hz and 6.00 Hz, respectively. The 1 H NMR spectra of the 10a,b revealed the absence of the CH 2 CH 2 CN group protons and the presence of (CH 2 ) n Br protons. 10a showed two triplets at 3.30 ppm and at 3.85 ppm with coupling constants of 6.72 Hz and 6.33 Hz, respectively. 10b showed a multiplet at 2.15 ppm and two triplets at 2.93 ppm and at 3.55 ppm, with coupling constants of 6.78 Hz and 6.36 Hz, respectively. Further confirmation for the structure of 12a,b was obtained from their mass spectral data, where they showed ion peaks at [M + H] + 739 and [M + H] + 767, respectively, and by their 1 H RMN spectra, the aromatic protons for 12a as two doublets at 7.55 ppm and at 8.25 ppm, with the same coupling constants of 9.37 Hz, as well as another characteristic triplet at 3.59 ppm for the two (CH 2 ) 2 groups protons. On the other hand, the spectrum of 12b exhibited a multiplet at 2.47 ppm for the two CH 2 CH 2 CH 2 groups. Moreover, the spectra of 13a,b showed amino group protons as multiplet around 3.50-3.85 ppm. The final products 14a,b showed the absence of the amino group proton signals and the presence of dimethylamino group proton signals as singlets at 3.35 ppm and at 3.20 ppm, respectively.
## Electrochemical studies
The redox behavior of these new functional mono-and bis-TTF was studied in solution by cyclic voltammetry (CV) and by square wave voltammetry (SQW). Measurements were performed under nitrogen at room temperature using a glassy carbon working electrode, a Pt counter electrode and a standard calomel electrode (SCE) as reference, with tetrabutylammonium perchlorate (n-Bu 4 NClO 4 , 0.1 M) in dry acetonitrile, as supporting electrolyte. A scan rate of 100 m·Vs −1 was used. The CV measurements showed reversible redox waves for all the compounds studied and the corresponding oxidation potentials Eox were determined by the SQW technique. The results are summarized in [fig_ref] Table 1: Oxidation potential of mono-and bis-tetrathiafulvalenes [/fig_ref].
The type I (a) SQW curve shows two sharp oxidation waves each with one electron process for the mono-TTF [fig_ref] Figure 1: Voltammogram of mono-TTF 9 [/fig_ref]. This kind of voltammogramm is observed for compounds 9, 10a and 10b. The type II (b) SQW curve is observed [fig_ref] Scheme 2: Synthetic route of mono-tetrathiafulvalenes [/fig_ref]. We can clearly see three oxidation peaks with respectively a 1, 1 and 2 electron process. The real distinction of the two first oxidation waves is clearly due to the difference of the two TTF units of the bis-TTFs. The oxidation potentials are almost identical for each pair of bis-donors (11a, 11b), (12a, 12b), (13a, 13b) and (14a, 14b). The first oxidation potentials in each pair are shifted cathodically by 20 mV. This may be due to the alkyl linked group. The oxidation potentials of compounds 14a,b are slightly higher than that of compounds 13a,b, on the other hand, the compounds 12a,b are slightly higher than that of compounds 14a,b. This should be attributable to the electronic properties of the nitrophenyl, aminophenyl and dimethylaminophenyl groups. All the oxidation potential values measured for mono-TTFs were found higher than the oxidation potentials of bis-TTFs. These results showed the good donor ability of this new series of mono-TTFs derivatives, which consequently should lead to conducting materials.
## Preparation and electrical conductivity of charge transfer complexes
The first report on the electrical conductivity in an organic solid appeared in 1954 [bib_ref] Organic semiconductors with high conductivity. I. Complexes between polycyclic aromatic hydrocarbons and..., Akamatu [/bib_ref] , namely, a perylene-bromine complex, which has a room-temperature conductivity of 0.1 S cm −1 . In 1960, the organic acceptor TCNQ [bib_ref] 7,7,8,8-tetracyanoqunodimethane and its electrically conducting anion-radical derivatives, Acker [/bib_ref] was synthesized as well as a great number of its conducting charge-transfer complexes and radical ion salts.
In the 1970s, the organic donor TTF [bib_ref] 1,3-Bis-dithiole chloride, Wudl [/bib_ref] led to the first organic metal TTF-TCNQ [bib_ref] Electron transfer in a new highly conducting donor-acceptor complex, Ferraris [/bib_ref]. Its room-temperature conductivity (500 S cm −1 ) increases with a decrease of the temperature to the value of 6000 S cm −1 at 60 K where a metal-insulator transition occurs. Since then, great interest has been devoted to this type of material, and a great number of new organic donors and acceptors have been synthesized as well as their charge-transfer salts. Therefore, the complexation of the donors 9-14 with 7,7,8,8-tetracyanoquinodimethane (TCNQ) in hot acetonitrile solution gave the corresponding charge transfer complexes (CTC). Most of the solids were isolated in powder forms. Electrical conductivity was only measured on compressed pellets at room temperature using a two probe technique. The results are reported in [fig_ref] Table 2: Electrical conductivity and melting points of charge transfer complexes [/fig_ref].
For this family of materials, only CTC 9-TCNQ, 10a-TCNQ and 10b-TCNQ resulting from mono-TTFs, can be classified in the area of conductors. In fact, they have a conductivity measured on powder compressed pellets of 8.60 10 −1 to 1.30 10 −2 S cm −1 , that allows a conductivity ten times greater on single crystal.
Other, CTC resulting from bis-TTFs can be classified in the category of semi-conductors materials with conductivities from 10 −3 to 10 −6 S cm −1 . This can be due to a structural disorder and/or a full charge transfer of an electron for each molecule.
## Experimental section
## General
NMR spectra were recorded on a Brucker AC 250 instrument. Microanalyses were performed in the Microanalysis Laboratory of ENSCM (Montpellier). FAB mass spectra were recorded on a JOEL JMS-DX 300 spectrometer. Uncorrected melting points were measured on a 510 Buchi apparatus. Cyclic voltammetry measurements were carried out on a PAR-273 potentiostat/galvanostat. All solvents were dried by standard methods and all commercial reagents used without purification. All reactions were performed under an inert atmosphere of nitrogen.
## 2-methylthio-4-(p-nitrophenyl)-1,3-dithiolium trifluoromethane sulfonate 2
A suspension of 4-(p-nitrophenyl)-1,3-dithiole-2-thione 1a (4.28 g, 16.81 mmol) in dry methylene chloride (25 mL) was treated with methyl triflate (2.4 mL, 19.
## 2-methylthio-4-(p-nitrophenyl)-1,3-dithiole 3
A suspension of 2 (7.73 g, 18.45 mmol) in dry ethanol (10 mL) was treated with NaBH 4 (3.41 g, 92.25 mmol) at 0 °C under nitrogen. After the reaction mixture was stirred for 2 h, the solvent was evaporated under vacuum. The residue was dissolved in dichloromethane (100 mL), washed three times with water and dried over magnesium sulphate; the product was obtained after evaporation. Yield = 75%; TLC: Rf = 0.85 (CH 2 Cl 2 ); dark red oil; [bib_ref] Electron transfer in a new highly conducting donor-acceptor complex, Ferraris [/bib_ref]
## 4-(p-nitrophenyl)-1,3-dithiolium tetrafluoroborate 4
Tetrafluoroboric acid (0.92 g, 10.60 mmol) was added dropwise under nitrogen to a solution of 3
## 2,3-bis(2-cyanoethylthio)-6-p-nitrophenyl tetrathiafulvalene 9
3.5.1. Method 1 4-(p-Nitrophenyl)-1,3-dithiole-2-thione 1a [bib_ref] The 1,2,3-thiadiazole route to new vinylogue tetrathiafulvalenes, Clausen [/bib_ref] or 4-(p-nitrophenyl)-1,3-dithiole-2-one 1b [bib_ref] Dithiolium derivatives. II. Some new 1,3-dithiolium perchlorates, Campaigne [/bib_ref] and 5-bis-(2-cyanoethylthio)-1,3-dithiole-2-one 6 [bib_ref] Reparation and chemistry of new unsymmetrically substituted tetrachalcogenofulvalenes bearing CN(CH 2 )..., Binet [/bib_ref] were synthesized as described in the literature. Under a nitrogen atmosphere, 25 mL of freshly distilled triethyl phosphite was added to the mixture of 1a (0.5 g, 1.96 mmol) or 1b (0.5 g, 2.09 mmol) and 6 (1 equiv.). The resulting mixture was heated over an oil bath up to 110 °C and stirred for a further 4 h. The solvent was then removed under reduced pressure. Compound 9 was obtained by column chromatography of the residue (silica gel, eluting with dichloromethane and petroleum ether (2:1)) in 35% and 57% yield, respectively.
# Method 2
A solution of the dithiolium salt 4 (0.5 g, 1.60 mmol) and (4,5-bis((2-cyanoethyl)thio)-1,3-dithiol-2yl)triphenylphosphonium tetrafluoroborate 5 (0.99 g, 1.60 mmol) in acetonitrile (30 mL) was treated with triethylamine (0.27 mL, 1.92 mmol) at room temperature under nitrogen and the reaction mixture was stirred for 4 h, the solvent was evaporated under vacuum. The residue was purified by column chromatography on silica gel with dichloromethane and petroleum ether (2:1) as the eluent to afford 9 in 30% yield.
TLC: Rf = 0.53 (CH 2 Cl 2 /petroleum ether) (2:1); dark-violet powder, mp = 175 °C; 1 H NMR (CDCl 3 ) δ ppm: 2.78 (t, 2H, CH 2 S, J = 7.07 Hz); 2.83 (t, 2H, CH 2 S, J = 6.00 Hz); 3.14 (t, 2H, CH [bib_ref] New concepts in tetrathiafulvalene chemistry, Segura [/bib_ref]
## Synthesis of mono-ttfs 10a and 10b
Cesium hydroxide monohydrate (0.62 g, 3.72 mmol) in dry methanol (10 mL) was added to tetrathiafulvalene dicyano derivative 9 (0.5 g, 1.69 mmol) dissolved in dry and degassed DMF (30 mL). The reaction mixture was stirred for 10 min, the colour becoming dark violet. Then, an excess of 1,2-dibromoethane or 1,3-dibromopropane (10 equiv.) was added in one portion. The colour of the reaction mixture became clear, and the reaction mixture was stirred at room temperature for 1 h. The solvent was removed in vacuo, the residue was dissolved in dichloromethane (100 mL), washed three times with water and dried over magnesium sulphate. The mixture was concentrated in vacuo and the residue was purified by chromatography on a silica gel column (eluent: dichloromethane).
## Synthesis of bis-ttfs 11a and 11b
Compounds 11a and 11b were synthesized by employing the same experimental process as 10 from 1 equiv. of 10a or 10b, 1 equiv. of 2,3-bis(2-cyanoethylthio)-6,7-di(methyl)tetrathiafulvalene 8 and 1 equiv. of cesium hydroxide.
## Synthesis of bis-ttfs 12a and 12b
Compounds 12a and 12b were synthesized by employing the same experimental process as 10 from 1 equiv. of 11a or 11b and 1 equiv. cesium hydroxide.
## Synthesis of bis-ttfs 13a and 13b
A stirred mixture of 4-p-nitrophenyl-bis-TTFs derivatives 12a or 12b (4 mmol), tin (0.94 g, 8 mmol), and aqueous solution of HCl (35%) (1.8 mL, 20 mmol) in ethanol (30 mL) was refluxed for 4 h under nitrogen. During this time the initial black solution turned light yellow. The solution was then concentrated in vacuo and treated with an aqueous solution (100 mL) of sodium hydroxide (0.1 M) and extracted with ether. The organic phase was washed with water, dried (MgSO 4 ), and concentrated in vacuo. The product was subjected to column chromatography on silica gel (CH 2 Cl 2 ), affording the expected compounds 13a and 13b as powder.
p-Aminophenyl bis-tetrathiafulvalene 13a: Yield = 52%; TLC: Rf = 0.62 (CH 2 Cl 2 ); dark-orange powder, mp = 138 °C; [bib_ref] Electron transfer in a new highly conducting donor-acceptor complex, Ferraris [/bib_ref]
## Synthesis of bis-ttfs 14a and 14b
To a stirred solution of 4-aminophenyl-bis-TTF 13a or 13b (3 mmol) and of iodomethane (0.75 mL, 12 mmol) in acetone (15 mL) under nitrogen, K 2 CO 3 (0.83 g, 6 mmol) was added. After 4 days of stirring at room temperature, the precipitate obtained was filtered, washed with acetone, and then extracted with CH 2 Cl 2 . The organic phase was dried (MgSO 4 ) and concentrated in vacuo, providing the expected compounds 14a and 14b as dark orange powder.
p-Dimethylaminophenyl bis-tetrathiafulvalene 14a: Yield = 76%; TLC: Rf = 0.78 (CH 2 Cl 2 ); orange powder, mp = 182 °C; [bib_ref] Electron transfer in a new highly conducting donor-acceptor complex, Ferraris [/bib_ref]
# Conclusions
In summary, we have successfully prepared via Wittig-type, cross-coupling, reduction and alkylation synthetic strategies some new mono-TTFs and bis-TTFs containing nitrophenyl, aminophenyl or dimethylaminophenyl groups. All donors synthesized during the course of this work have been characterized by cyclic voltammetry and their oxidation potentials were determined by cyclic voltammetry. Charge transfer complexes of the donors with TCNQ were prepared and the electrical conductivity of these materials was measured, some CTC are conductive.
[fig] Scheme 2: Synthetic route of mono-tetrathiafulvalenes (TTFs) 9 and 10. [/fig]
[fig] Figure 1: Voltammogram of mono-TTF 9 (type I) (a) and bis-TTF 14a (type II) (b). [/fig]
[fig] 2: mmol) at 0 °C. The mixture was stirred under nitrogen for 4 h and dry ether (150 mL) was added. The violet-orange salt 2 was obtained by filtration after 24 h, washed with more dry ether, and dried. Yield = 95%; orange powder, mp = 237 °C; 1 H NMR (CDCl 3 ) δ ppm: 3.10 (s, 3H, CH 3 S); 7.37 (s, 1H, HC-S); 7.53 (m, 2H arom ); 8.24 (m, 2H arom ); M.S: (NOBA, FAB > 0): 420 [M + H] + ; M = 419; Anal. Calcd for C 11 H 8 S 4 NO 5 F 3 : C, 31.49; H, 1.92; S, 30.57; found: C, 31.70; H, 2.14; S, 30.60. [/fig]
[fig] 64: mmol) in 30 mL of acetic anhydride at 0 °C. The reaction mixture was stirred for 15 min, then anhydrous ether was added. The 1,3-dithilium salt was precipitated, collected by filtration and washed with ether. Yield = 45%; white powder, mp = 253 °C; 1 H NMR (CDCl 3 ) δ ppm: 7.50 (m, 2H arom ); 7.85 (s, 1H, HC=C-S); 8.20 (m, 2H arom ); 9.35 (s, 1H); M.S: (NOBA, FAB > 0): 312 [M + H] + ; M = 311; Anal. Calcd for C 9 H 6 S 2 NO 2 BF 4 : C, 34.74; H, 1.94; S, 20.61; found: C, 34.98; H, 2.16; S, 20.96. [/fig]
[table] Table 1: Oxidation potential of mono-and bis-tetrathiafulvalenes (bis-TTF)s. [/table]
|
The knowledge and attitudes of general practitioners to the assessment and management of pain in people with dementia
Background: Pain in people with dementia is underdiagnosed and undertreated. General practitioners (GPs) play a pivotal role in dementia care but their perspectives on pain in people with dementia remains under-researched. The aim of this study was to explore GPs' knowledge and attitudes towards pain assessment and management in people with dementia. Methods: This was a descriptive cross-sectional study. A questionnaire was adapted from a previous study and piloted with 5 GPs. The questionnaire was posted to a census sample of all GPs in Cork city and county in the southern region of Ireland. The questionnaire collected demographic information, responses to a series of Likerttype statements assessing GPs' knowledge and attitudes, and provided an opportunity for the GP to give qualitative feedback on their experiences of managing pain in dementia. SPSS v25 was used for statistical analysis. Qualitative responses were thematically analysed. Results: Of the 320 questionnaires posted, 157 completed questionnaires were returned (response rate of 49%). The sample was representative of GPs nationally in terms of years in GP practice and practice location. Over twothirds (108/157) of respondents had a nursing home commitment. Only 10% of respondents (16/157) were aware of any dementia-specific pain assessment tools. The larger the nursing home commitment of the GP the more likely they were to be familiar with these tools (p = 0.048). The majority of respondents (113/157) believed people with dementia could not self-report pain. Respondents were uncertain about the safety of using opioid medications to treat pain in people with dementia with only 51.6% agreeing that they were safe. The qualitative comments highlighted the importance the GPs placed on surrogate reports of pain, GPs' uncertainty regarding the value of formal pain assessment tools and the challenges caused by under-resourcing in general practice. Conclusion: This study has highlighted aspects of pain assessment and management in dementia that GPs find challenging. Guidance on pain assessment and management in people with dementia do not appear to be translating into clinical practice. The findings will inform educational interventions being developed by our research team as part of the implementation of the Irish national dementia strategy. The knowledge and attitudes of general practitioners to the assessment and management of pain in people with dementia.
# Background
The global prevalence of dementia is increasing. In 2013 it was estimated that 44 million people worldwide were living with dementia and this figure is expected to reach 75 million in 2030 and 135 million by 2050. Although these estimates may be reduced by improvements in population health, such as reductions in smoking and hypertension, current evidence suggest that only 10% of the expected rise in incidence will be avoided by improvements in these disease control measures. A second major health issue facing the older population is that of chronic pain. The prevalence of pain is strongly correlated with increasing age [bib_ref] Guidance on the management of pain in older people, Abdulla [/bib_ref] [bib_ref] Prevalence of arthritis and joint pain in the oldest old: findings from..., Duncan [/bib_ref]. People with dementia appear to have a significantly increased risk of pain [bib_ref] Pain in community-dwelling older adults with dementia: results from the National Health..., Hunt [/bib_ref] , with up to half of people with dementia estimated to be living with chronic pain [bib_ref] Assessment and treatment of pain in people with dementia, Corbett [/bib_ref] [bib_ref] Pain in community-dwelling persons with dementia: frequency, intensity, and congruence between patient..., Shega [/bib_ref] [bib_ref] The prevalence of pain in nursing home residents with dementia measured using..., Zwakhalen [/bib_ref]. In one study of nursing home residents the prevalence of chronic pain in residents with dementia was almost double that of residents without dementia [bib_ref] Pain Assessment in Elderly with Behavioral and Psychological Symptoms of Dementia, Malara [/bib_ref]. Similarly, in the community setting, pain is more prevalent in people with dementia than in people without dementia [bib_ref] Pain in community-dwelling older adults with dementia: results from the National Health..., Hunt [/bib_ref] [bib_ref] Pain in community-dwelling persons with dementia: frequency, intensity, and congruence between patient..., Shega [/bib_ref] [bib_ref] Community pharmacists and people with dementia: a cross-sectional survey exploring experiences, attitudes,..., Barry [/bib_ref]. This increased prevalence is related, at least in part, to the significantly higher burden of co-morbid physical disease in people with dementia [bib_ref] Comorbidity and polypharmacy in people with dementia: insights from a large, population-based..., Clague [/bib_ref] [bib_ref] Health problems and correlates of pain in nursing home residents with advanced..., Black [/bib_ref] which contributes to increased musculoskeletal pain [bib_ref] Prevalence of pain in nursing home residents: the role of dementia stage..., Van Kooten [/bib_ref] [bib_ref] Pain management in patients with dementia, Achterberg [/bib_ref] , orofacial pain [bib_ref] Oral health and orofacial pain in people with dementia admitted to acute..., Van De Rijt [/bib_ref] and neuropathic pain [bib_ref] Prevalence, causes, and treatment of neuropathic pain in Dutch nursing home residents:..., Van Kollenburg [/bib_ref]. However, pain in people with dementia is often underdiagnosed, underestimated and undertreated [bib_ref] Pain Assessment in Elderly with Behavioral and Psychological Symptoms of Dementia, Malara [/bib_ref] [bib_ref] Management of noncancer pain in community-dwelling persons with dementia, Shega [/bib_ref] [bib_ref] Disparities in pain management between cognitively intact and cognitively impaired nursing home..., Reynolds [/bib_ref].
The negative impact of undiagnosed, untreated pain in dementia is substantial. In a person with dementia untreated pain can worsen cognitive function, lead to depressive symptoms, reduce quality of life and trigger or exacerbate behavioural and psychological symptoms of dementia (BPSD) [bib_ref] Assessment and treatment of pain in people with dementia, Corbett [/bib_ref] [bib_ref] Pain Assessment in Elderly with Behavioral and Psychological Symptoms of Dementia, Malara [/bib_ref] [bib_ref] The role of pain treatment in managing the behavioural and psychological symptoms..., Ballard [/bib_ref] [bib_ref] The interactive relationship between pain, psychosis, and agitation in people with dementia:..., Habiger [/bib_ref]. The aetiology of BPSD is often multifactorial [bib_ref] Assessment and management of behavioral and psychological symptoms of dementia, Kales [/bib_ref] , however, pain is one of the most important causal factors for BPSD [bib_ref] Assessment and treatment of pain in people with dementia, Corbett [/bib_ref]. Indeed interventions targeting pain have been shown to be effective at reducing BPSD [bib_ref] Interventions targeting pain or behaviour in dementia: a systematic review, Pieper [/bib_ref]. A person with advanced dementia may not either understand or be able to effectively communicate pain and so pain may be expressed through behaviours such as agitation or aggression [bib_ref] The relationship between pain and disruptive behaviors in nursing home resident with..., Ahn [/bib_ref]. In addition to the distress and discomfort untreated pain can cause, not correctly identifying pain as a trigger for these behaviours can lead to the inappropriate prescribing of potentially harmful psychoactive medications to people with dementia. In one study the presence of pain in nursing home residents was found to be significantly associated with the use of antipsychotic medication [bib_ref] Pain in care home residents with dementia: an exploration of frequency, prescribing..., Barry [/bib_ref]. In addition to the harmful side effects of psychotropic medication in people with dementia [bib_ref] Antipsychotics, other psychotropics, and the risk of death in patients with dementia:..., Maust [/bib_ref] [bib_ref] Efficacy and adverse effects of atypical antipsychotics for dementia: meta-analysis of randomized,..., Schneider [/bib_ref] these drugs often have a sedative effect which can mask behaviours that are indicative of pain [bib_ref] Behavioral manifestations of pain in the demented elderly, Cipher [/bib_ref] , further contributing to under-diagnosis.
The vast majority of people living with dementia either live at home in the community or in a residential care setting such as a nursing home. The general practitioner (GP) is the key healthcare professional for a person with dementia as they provide care in both the community and in nursing home settings. GPs play a pivotal role in assessing and managing pain in people with dementia, however, they are challenged by many aspects of dementia care including the management of BPSD [bib_ref] We're certainly not in our comfort zone": a qualitative study of GPs'..., Foley [/bib_ref] [bib_ref] General practitioners' knowledge, confidence and attitudes in the diagnosis and management of..., Turner [/bib_ref] [bib_ref] General practitioner's clinical practices, difficulties and educational needs to manage Alzheimer's disease..., Somme [/bib_ref] [bib_ref] General practitioners' knowledge, attitudes, and experiences of managing behavioural and psychological symptoms..., Jennings [/bib_ref] and providing end of life care [bib_ref] Barriers and facilitators for GPs in dementia advance care planning: a systematic..., Tilburgs [/bib_ref]. Both the management of BPSD and the provision of end of life care in dementia require GPs to assess for and manage any underlying pain. In a recent qualitative study conducted by the authors, GPs reported difficulty identifying pain as a potential trigger for BPSD [bib_ref] A qualitative exploration of the challenges general practitioners experience when managing behavioural..., Jennings [/bib_ref]. However, to the best of our knowledge no research to date has explored GPs' perspectives on pain management in dementia. The aim of this study was to explore the knowledge and attitudes of Irish general practitioners to the identification and management of pain in people with dementia.
# Methods
An anonymous postal questionnaire was sent to a census sample of all GPs in Cork city and county in the southern region of Ireland in May 2017. Ethical approval was granted by the Social Research and Ethics Committee in University College Cork (Log2016-050).
## Questionnaire
There was no previously validated questionnaire available to address this research question. However, we did identify an appropriate questionnaire that was used in a previous study that explored nurses' knowledge of and attitudes to pain management in dementia [bib_ref] Nurses' knowledge and attitudes towards pain assessment for people with dementia in..., Burns [/bib_ref]. With the original authors' permission we adapted this questionnaire for use with GPs. We (AJ, ML, TF), all practicing GPs with an academic interest in dementia care, reviewed the questionnaire to ensure its appropriateness and relevance to a general practice setting. Furthermore, the questionnaire was piloted with five GPs, all of whom have experience managing people with dementia in community and nursing home settings. Subsequently, minor amendments were made to enhance the clarity of the questionnaire. The final questionnaire developed for use in this study contained three sections. The aim of the first section was to capture demographic information from the GPs. The second section consisted of a series of five-point Likert-type statements exploring GPs' knowledge and attitudes to pain in people with dementia. The final section provided GPs with the opportunity to give free-text responses. (See Additional file 1 for questionnaire).
## Sample
The questionnaire was posted with an information sheet to all 320 GPs practicing in the Cork region. With a large urban and rural population, a census sample of Cork GPs is largely representative of Irish GPs nationally. The sample size was calculated based on this census population of 320 with desired precision estimates of +/− 5% around a prior estimate of 50%. Based on these calculations to adequately power the study the sample size required was 175 respondents. All GPs were identified from the Irish Medical Directory. A stamped addressed envelope was provided to return the questionnaire by post if the GP was willing to participate.
# Analysis
The data from the responses were coded and SPSS version 25 was used for statistical analysis. Chi-square tests and Mann-Whitney test were used to explore associations between demographic data and responses, with differences at a level of 95% probability and above regarded as statistically significant. The following demographic associations were explored; years in practice, presence of a nursing home commitment and the extent of the GPs nursing home commitment (as indicated by the number of residents the GP cared for). Free-text responses were entered into MS Word and were thematically analysed [bib_ref] Using thematic analysis in psychology, Braun [/bib_ref] by two of the authors (AJ,TF).
# Results
## Sample
Of 320 questionnaires sent, a total of 157 completed questionnaires were received, representing a response rate of 49% (157/320). The respondents had a broad mix of practice locations and years of experience. The demographic characteristics of respondents is displayed in [fig_ref] Table 1: Participant demographics [/fig_ref].
## Provision of care
Over two-thirds (108/157) of respondents had a nursing home commitment (mean number of nursing homes +/− SD = 1.35 +/− 1.19; range (0-5)). These GPs provided care to a total of 2393 people in nursing homes (mean number of patients = 15; range (1-112)). The respondents reported that just over half of these nursing home patients (1242/2393) had dementia. Of the GPs who provided care to nursing home residents, the majority (60.2%) did a regular round in the nursing home. Over half of the GPs with a nursing home commitment did 1 or more nursing home round per week (mean number of rounds per week = 1, range (0-5)). Rural GPs were significantly more likely to have a nursing home commitment (P-value = 0.042). There was no association found between the numbers of years a GP was in practice and having a nursing home commitment.
## Pain assessment in dementia
The overwhelming majority of GPs (98%) agreed that the presence of dementia can make pain difficult to assess [fig_ref] Table 2: Responses to Likert statements on assessment of pain in people with dementia [/fig_ref]. A smaller majority of respondents (68.7%) felt that pain was under-recognised in patients with dementia. The majority of GPs surveyed agreed that observing behavioural and physiological indicators of pain and obtaining surrogate reports are important when assessing pain in a person with dementia. However, most GPs were unfamiliar with dementia-specific pain assessment tools with only 10% reporting any knowledge of their existence. The larger the nursing home commitment of the GP, as indicated by the number of nursing home residents they cared for, the more likely they were to be familiar with pain assessment tools for people with dementia (P-value = 0.048). However, of the respondents with a nursing home commitment only 14% were aware of guidelines/policies on pain management in the nursing homes they attended. The numbers of years the GP was in practice was not associated with an increased familiarity with pain assessment tools. Despite the lack of awareness of pain assessment tools, the majority (73.2%) of respondents believed that a pain assessment tool would be useful to increase recognition of pain in patients with dementia in nursing home settings.
## Management of pain in dementia
The responding GPs appeared less certain about aspects of the management of pain in dementia [fig_ref] Table 3: Responses to Likert-type statements on management of pain in people with dementia... [/fig_ref]. The overwhelming majority (95%) agreed that the treatment of pain should follow a step-wise approach. However, 26% were either disagreed or neither agreed nor disagreed that optimal treatment of pain is achieved when analgesics are given on a regular basis. The respondents were particularly uncertain about the safety of using opioid medications in people with dementia. Just over half of GPs surveyed (51.6%) agreed with the statement that opioid analgesics are safe to use when treating pain in dementia, while 34.4% neither agreed nor disagreed with the statement and 14% believed opioids were unsafe in this patient group. There was no statistically significant association found between the GP's knowledge of the management of pain in dementia and their number of years in practice or the extent of their nursing home commitment.
## Free-text responses
Of the 157 respondents, 49 GPs (31% of respondents) provided additional qualitative feedback in the free text section of the questionnaire. There were a number of themes identified from the free-text responses. The three main themes identified were; (i) the role of pain assessment tools, (ii) the importance of input from carergivers and (iii) challenges of resource limitations. Verbatim quotations are presented here to When assessing pain in a resident with dementia, it is important to consider a family/care givers report 150 (95.5) 7 (5.5) 0 a Note: The original Likert scale options "strongly agree" and "agree" were combined to "agree", whereas the options "strongly disagree" and "disagree" were combined to "disagree" Note: The original Likert scale options "strongly agree" and "agree" were combined to "agree", whereas the options "strongly disagree" and "disagree" were combined to "disagree"
illustrate the themes.These particular quotations were selected as they were considered to be typical of the responses that underpinned the development of each theme.
## Theme 1: role of pain assessment tools
Several of the responding GPs expressed a desire for guidance in the area of pain assessment and management in dementia:
"Would love guidance on pain management in these patients" (Respondent_79, experienced GP, urban practice, no nursing home commitment)
Some respondents expressed apprehension about introducing pain assessment tools.
"I feel a lot of 'tools' can lead to unnecessary work if arbitrarily based on BP readings, heart rates etc e.g. MEWS score in hospital which is why I'd be wary of their use. Would need to be used judiciously" (Respondent_84, recently qualified GP, rural practice, with nursing home commitment)
Many GPs perceived pain assessment tools as yet another tool that would add to GPs' workload without necessarily improving care.
"A pain assessment tool wouldn't be helpful in nursing homes as it would add more workload to already onerous paperwork" (Respondent_128, recently qualified GP, urban practice, with nursing home commitment) Theme 2: The importance of input from caregivers A second theme identified was the value GPs placed on the input of relevant caregivers, who knew the person with dementia well, when assessing pain:
"Key component to good assessment depends on good collateral history -family/close friends, nursing staff/ carers." (Respondent_46, experienced GP, rural practice, with a nursing home commitment)
In particular many GPs highlighted the important role of nurses in pain assessment:
"Feedback from nurses helps to make appropriate prudent descriptions." (Respondent_75, experienced GP, mixed practice setting, with a nursing home commitment)
The GPs described how they relied on the nursing staff and trusted their opinion when assessing a person with dementia:
"An experienced nurse is the best person to rely on, they know as do the carerslisten to them and you'll get it right." (Respondent_140, experienced GP, urban practice, with a nursing home commitment)
## Theme 3: the barriers of under-resourcing
A third theme identified was the challenges GPs experienced providing care to people with dementia given the current underfunding of Irish general practice.
"Dementia [is] not currently a paid 'chronic disease' in GP, poor remuneration." (Respondent_114, mid-career GP, mixed practice setting, no nursing home commitment)
The GPs reported extensively about the impact of recent austerity cuts in Ireland which dramatically cut funding to GPs who provide care to nursing home residents:
# Discussion
This is the first study that has explored GPs' assessment and management of pain in people with dementia. Our findings suggest that GPs are confident in many aspects of assessing pain in people with dementia such as the value of observing behavioural and physiological indicators and the importance of surrogate reports but that they are challenged by many aspects of assessing and managing pain in dementia. The majority of GPs surveyed believed that a person with dementia cannot self-report pain and the vast majority of GPs were unfamiliar with dementia-specific pain assessment tools. In the absence of either a self-report or a standardized observational tool to assess pain, the responding GPs appeared to rely significantly on surrogate reports from family members and nursing home staff when assessing pain in dementia. Although the majority of responding GPs welcomed the idea of guidance in the area of pain assessment and management in dementia, in the free-text comments many questioned the value of a standardized, observational pain tool. Furthermore, when managing pain in people with dementia the GPs were particularly uncertain about the role of opioid medication and the consequences of the use of opioids in people with dementia. While the majority of GPs agreed that in order to achieve optimum pain relief analgesia should be prescribed regularly, over a quarter of GPs surveyed did not agree with this statement. The inference being that these GPs are favouring 'as required' analgesic medication, a sub-optimal method of pain control. Although we did find that GPs with a larger nursing home commitment were more likely to be familiar with dementia-specific pain assessment tools, in general, the experience level of the GP was not associated with increased levels of knowledge or a more positive attitude towards pain assessment and management in dementia.
## Comparison with existing literature
Our findings indicate that GPs' value good communication with family members and nursing home staff when managing pain in people with dementia. This finding is similar to previous studies with community pharmacists [bib_ref] Community pharmacists and people with dementia: a cross-sectional survey exploring experiences, attitudes,..., Barry [/bib_ref] and nurses [bib_ref] Nurses' knowledge and attitudes towards pain assessment for people with dementia in..., Burns [/bib_ref]. In previous research on GPs' educational needs in dementia [bib_ref] We're certainly not in our comfort zone": a qualitative study of GPs'..., Foley [/bib_ref] both GPs and family caregivers emphasized the importance of good channels of communication in dementia care. In our study, the GP respondents also emphasized the importance of a report from nursing staff or family carer when assessing pain in the qualitative free-text responses. In our study nursing staff were seen by the GPs to facilitate pain assessment. This is in contrast to findings from a previous study that examined nurses knowledge and attitudes to pain management in dementia where nurses identified a lack of GP support as a barrier to successful pain management [bib_ref] Nurses' knowledge and attitudes towards pain assessment for people with dementia in..., Burns [/bib_ref]. Previous research with GPs found that consistency of care was an important factor in improving relationships between GPs and nursing staff [bib_ref] A qualitative exploration of the challenges general practitioners experience when managing behavioural..., Jennings [/bib_ref]. In that study structured visits by the GP to the nursing home were seen to facilitate the provision of continuity of care and led to good communication channels between the GP and nurses [bib_ref] A qualitative exploration of the challenges general practitioners experience when managing behavioural..., Jennings [/bib_ref]. Effective communication between nursing staff and GPs is an essential component of any optimisation of pain assessment and management in nursing home settings.
The majority of respondents in our study believed that people with dementia could not accurately provide a self-report of pain. However, self-reporting of pain is considered the gold standard method of pain assessment [bib_ref] The Assessment of Pain in Older People: UK National Guidelines, Schofield [/bib_ref] and can be a reliable way of assessing pain in people with dementia [bib_ref] The Assessment of Pain in Older People: UK National Guidelines, Schofield [/bib_ref] [bib_ref] Pain assessment in elderly adults with dementia, Hadjistavropoulos [/bib_ref] [bib_ref] An interdisciplinary expert consensus statement on assessment of pain in older persons, Hadjistavropoulos [/bib_ref]. Best practice recommendations advise that where possible attempts should always be made to elicit self-reports of pain from the person with dementia [bib_ref] The Assessment of Pain in Older People: UK National Guidelines, Schofield [/bib_ref]. Although in the very advanced stages of dementia many individuals may be unable to self-report pain [bib_ref] Pain management in patients with dementia, Achterberg [/bib_ref] , people with mild-moderate [bib_ref] Pain in care home residents with dementia: an exploration of frequency, prescribing..., Barry [/bib_ref] [bib_ref] Observer-rated pain assessment instruments improve both the detection of pain and the..., Lukas [/bib_ref] and in some cases severe dementia [bib_ref] Pain in severe dementia: self-assessment or observational scales?, Pautex [/bib_ref] have been found to provide valid self-reports of pain. Our finding echoes previous research with nursing home managers which found that only 8.3% of respondents felt that people with dementia could self-report pain [bib_ref] An exploration of nursing home managers' knowledge of and attitudes towards the..., Barry [/bib_ref]. The large majority of GP respondents agreed with the statement that 'a person with dementia is not able to provide a self-report of pain'. This finding could mean that an attempt is not being made to elicit a self-report from a person with dementia. However, this needs to be explored further, ideally with qualitative research, to establish the impact this attitude has on how GPs assess pain in people with dementia.
Nearly all GP respondents agreed that patient observation was a critical part of pain assessment in dementia. Despite this belief the vast majority of respondents were not using any validated, standardised, observational approach. Observational methods are central to clinical assessment, especially when a person lacks the ability to self-report, however, there is a risk of observer bias if there is no standardized approach to the observation [bib_ref] Pain assessment in elderly adults with dementia, Hadjistavropoulos [/bib_ref]. A large array of pain assessment tools exist for use in people with dementia -a recent systematic review of pain assessment tools included 28 such tools [bib_ref] Pain assessment for people with dementia: a systematic review of systematic reviews..., Lichtner [/bib_ref]. However, the vast majority of GP respondents were unaware of these tools. Similar to previous findings from a study with community pharmacists [bib_ref] Community pharmacists and people with dementia: a cross-sectional survey exploring experiences, attitudes,..., Barry [/bib_ref] , the more experience the GP had with dementia the more likely he or she was to be aware of these tools. The lack of awareness of dementia-specific pain assessment tools is a particularly noteworthy finding since the majority of respondents thought that having such a pain assessment tool would be helpful. This highlights an incongruity between research in this area and real-life clinical practice. These tools appear to be rarely used in general practice.
The responding GPs lack of familiarity with pain assessment tools may be surprising to researchers in the area but may be unsurprising to front-line GPs. GPs do not readily embrace assessment tools [bib_ref] GPs' knowledge, use, and confidence in national physical activity and health guidelines..., Chatterjee [/bib_ref] [bib_ref] To score or not to score: a qualitative study on GPs views..., Pettersson [/bib_ref]. They are inductively trained to rely on their clinical skills. This is in contrast to nursing staff who are specifically trained to use and rely on assessment tools. In the free-text responses some GPs feared the additional workload such tools could bring and questioned what clinical value they would add. Similarly, recent qualitative research exploring GPs' , hospital physicians' and nurses' perspectives on the use of observational pain tools in people with dementia identified a number of barriers to using observational pain tools, one being the perceived lack of value in using them [bib_ref] A tool doesn't add anything". The importance of added value: Use of..., Witt [/bib_ref]. Furthermore, participants in that qualitative study described using the pain tools to comply with local recommendations but not actually using the results to inform treatment decisions [bib_ref] A tool doesn't add anything". The importance of added value: Use of..., Witt [/bib_ref]. This echoes some of the concerns raised by participants in our study that a pain tool would become another source of paper-work rather than a tool to aid clinical decision making. GPs are not usually the healthcare professional tasked with completing a pain assessment tool; in a nursing home setting that role would typically fall to the nursing staff. GPs do still need to be aware of these tools in order to interpret the findings and generate appropriate management plans in discussions with nursing staff. However, clinicians and nurses can find the results of pain assessment tools difficult to interpret [bib_ref] Pain in older adults with dementia : A survey across Europe on..., Zwakhalen [/bib_ref]. Implementing these observational pain assessment tools in isolation, without adequate guidance on how to interpret the results, will not lead to improved treatment of pain [bib_ref] Pain management in patients with dementia, Achterberg [/bib_ref]. The tools in themselves will not result in improved care unless they are combined with guidance that will help clinicians to translate a pain score into an appropriate treatment plan.
Respondents appeared unsure about the safety of opioid analgesics in people with dementia. This is similar to previous research conducted with nurses [bib_ref] Nurses' knowledge and attitudes towards pain assessment for people with dementia in..., Burns [/bib_ref] , nurse managers [bib_ref] An exploration of nursing home managers' knowledge of and attitudes towards the..., Barry [/bib_ref] and community pharmacists [bib_ref] Community pharmacists and people with dementia: a cross-sectional survey exploring experiences, attitudes,..., Barry [/bib_ref] all of whom had similar concerns regarding the safety of prescribing opioids to people with dementia. Guidelines on pain management in the older adult do recommend considering opioid analgesics for patients with moderate to severe pain, particularly if the pain is causing functional impairment or reducing quality of life [bib_ref] Guidance on the management of pain in older people, Abdulla [/bib_ref]. The uncertainty the GP respondents felt regarding the safety of opioid medication in dementia is perhaps understandable; prescribing opioids to an older adult is not without its complexities. Age is a significant predictor of opioid related harm [bib_ref] Treatment of chronic non-malignant pain in the elderly: safety considerations, Barber [/bib_ref] [bib_ref] Clinical pharmacology of analgesic medicines in older people: impact of frailty and..., Mclachlan [/bib_ref]. Adverse effects of opioids can include respiratory depression, sedation, constipation, nausea and dizziness [bib_ref] Outcomes associated with opioid use in the treatment of chronic noncancer pain..., Papaleontiou [/bib_ref]. Many of these adverse effects increase with age [bib_ref] Side effects of opioids during short-term administration: effect of age, gender, and..., Cepeda [/bib_ref] and frailty [bib_ref] Clinical pharmacology of analgesic medicines in older people: impact of frailty and..., Mclachlan [/bib_ref] both of which are associated with dementia. Furthermore, these adverse effects can be particularly problematic to identify in a person with advanced dementia because of their reduced ability to communicate. However, when managing pain in people with dementia the adverse effects of opioids needs to be weighed up against the harmful effects of undertreating pain. Although a narrow majority agreed with the statement that opioids were 'safe in people with dementia' , a larger majority of respondents agreed with the statement that there is 'a greater risk of side effects from opioid analgesics when used in people with dementia'. Many respondents appeared to feel that, despite the increased risks, the use of opioids in people with dementia was still "safe". How this belief influences the GP's prescribing is unclear. Previous research suggested that people with dementia may receive less opioids [bib_ref] Who suffers most? Dementia and pain in nursing home patients: a cross-sectional..., Husebo [/bib_ref]. A qualitative study which examined GPs' perspectives on prescribing opioids for chronic pain found that fear of causing harm was a barrier to prescribing opioids to the older adult [bib_ref] Primary care providers' perspective on prescribing opioids to older adults with chronic..., Spitz [/bib_ref]. Despite this a recent systematic review identified that, internationally, prescribing of opioids to nursing home residents has increased over time [bib_ref] Temporal trends in analgesic use in long-term care facilities: a systematic review..., Frenais [/bib_ref]. Similarly, a large Danish study found that buprenorphine and fentanyl patches were more commonly prescribed to people with dementia [bib_ref] Frequent use of opioids in patients with dementia and nursing home residents:..., Jensen-Dahm [/bib_ref]. It is possible that these opioid patches are perceived as having less side effects than oral opioids or as being more tolerated by people with dementia. However, these transdermal opioid patches typically contain a stronger opioid dose than oral opioids, and since the adverse effects of opioids are dose related these patches are likely to result in more, not less, side effects. Prescribers' perspectives of the role of different opioid medications in the management of pain in dementia is an important area that could be explored in future research.
Although optimal treatment of pain is achieved when analgesics are given on a regular basis, over a quarter of respondents in our study did not agree with this statement. A similar finding was reported in previous research with nurses [bib_ref] An exploration of nursing home managers' knowledge of and attitudes towards the..., Barry [/bib_ref] where 20% of respondents either agreed, or neither agreed nor disagreed, with a statement that 'optimal pain treatment of pain relief is achieved when analgesics are given in a PRN (or as required) way'. If the value of regular administration of analgesia is not recognized this could result in the prescribing of analgesics in the less-effective 'when required' way. Research conducted in nursing homes in Northern Ireland in 2015 found that in the majority of residents with dementia, analgesic medication was prescribed 'when required' and not regularly [bib_ref] Pain in care home residents with dementia: an exploration of frequency, prescribing..., Barry [/bib_ref]. Likewise, research from the U.S. found that cognitively impaired nursing home residents were less likely than their cognitively intact peers to be prescribed regular, scheduled analgesia and were more likely to be prescribed analgesic medication in an 'as required' way [bib_ref] Disparities in pain management between cognitively intact and cognitively impaired nursing home..., Reynolds [/bib_ref]. To receive this 'as required' medication a person would either need to ask the nurse for pain relief or the nurse would need to identify that the person was in pain and give them pain relief [bib_ref] Disparities in pain management between cognitively intact and cognitively impaired nursing home..., Reynolds [/bib_ref]. In the case of a person with advanced dementia neither of these situations are very likely to occur. In view of their cognitive and communication difficulties a person with advanced dementia is unlikely to self-request analgesia and we know from existing evidence that there are a number of barriers to nurses identifying and initiating pain relief in a person with dementia [bib_ref] Nurses' knowledge and attitudes towards pain assessment for people with dementia in..., Burns [/bib_ref] [bib_ref] An exploration of nursing home managers' knowledge of and attitudes towards the..., Barry [/bib_ref].
## Areas for future research & implications for policy & practice
This study highlights several areas of GPs assessment and management of pain in dementia that could be explored with future research. There were some findings that warrant further exploration in future research such as the use of 'as required' medication and the belief that people with dementia cannot self-report pain. Future qualitative research with GPs in this area would help gain a deeper understanding of the context and nuances that are involved in this complex area. It would be important to ascertain how GPs' knowledge of and attitudes towards pain in people with dementia impacts on their actual prescribing of analgesia to people with dementia. A quantitative study of current prescribing of analgesia to people with dementia both in nursing homes and in the community would be an important complementary study. These studies would lead to a more extensive understanding of the problem and would help to inform the development of effective interventions to improve the management of pain in dementia in primary care settings.
There is a role for educational interventions for GPs that focus on pain assessment and management in dementia. The results of this current study will inform the ongoing, national roll-out of educational interventions for GPs in dementia care that have being developed by our research team [bib_ref] The development and evaluation of peer-facilitated dementia workshops in general practice, Foley [/bib_ref] as part of the implementation of the Irish National Dementia strategy. Previous educational interventions for GPs in the area of pain management in dementia have used teleconferencing technology based on the Project ECHO© model and have been found to improve healthcare professionals' knowledge and self-efficacy of pain assessment and management in advanced dementia [bib_ref] Evaluation of the impact of telementoring using ECHO(c) technology on healthcare professionals'..., Witt [/bib_ref]. Such educational interventions would be particularly acceptable to general practitioners as they eliminate the need for travel and don't require any prolonged periods away from practice. Another important finding was in relation to educational initiatives was that GPs see nurses as facilitators to optimum pain management, whereas in previous studies [bib_ref] Nurses' knowledge and attitudes towards pain assessment for people with dementia in..., Burns [/bib_ref] [bib_ref] An exploration of nursing home managers' knowledge of and attitudes towards the..., Barry [/bib_ref] nurses identified GPs as barriers to optimum pain management. This highlights a communication gap between these two professional groups, both of whom play a pivotal and complementary role in the assessment and management of pain in dementia. This finding underlines the importance of inter-professional educational initiatives in this area. Palliative care is another professional group that have a significant role to play in the management of pain in dementia, particularly in complex cases, yet we know that many people with dementia are not routinely assessed to determine their palliative care needs [bib_ref] Better palliative care for people with a dementia: summary of interdisciplinary workshop..., Fox [/bib_ref]. An inter-professional educational approach including all the relevant professional groups could help improve communication and bridge professional divides [bib_ref] A BEME systematic review of the effects of interprofessional education: BEME guide..., Reeves [/bib_ref] , which would play an important role in improving the care provided to people with dementia who are living with chronic pain.
Although relevant and important we know that educational interventions alone have limited effect in changing GPs' behaviour in dementia care [bib_ref] Dementia diagnosis and management: a narrative review of changing practice, Koch [/bib_ref]. To effectively improve GPs' performance in dementia care, education needs to be combined with adequate reimbursement and organisational incentives [bib_ref] Effects of educational interventions on primary dementia care: a systematic review, Perry [/bib_ref]. A previous systematic review highlighted how time constraints and inadequate remuneration act as barriers to optimum diagnosis and management of dementia in primary care [bib_ref] Rapid appraisal of barriers to the diagnosis and management of patients with..., Koch [/bib_ref]. In Ireland general practice receives only 4.5% of the overall health budget, significantly lower than other European countries. Additionally, recent government-led austerity cuts in Ireland have dramatically cut funding to general practice, these cuts have particularly affected general practitioners providing care to nursing home residents. In our study the dissatisfaction of GPs with the current under-resourcing of dementia care was evident in the free-text responses. Many GPs discussed the challenges of providing optimum care to people with dementia in the context of current resource limitations, reporting that they can no longer provide care to nursing home residents. Like many other European countries [bib_ref] From practice employee to (co-)owner: young GPs predict their future careers: a..., Gisler [/bib_ref] [bib_ref] Addressing the crisis of GP recruitment and retention: a systematic review, Marchand [/bib_ref] [bib_ref] Career choices of the United Kingdom medical graduates of, Svirko [/bib_ref] , Irish general practice is currently facing a recruitment crisis [bib_ref] Future career intentions of recent GP graduates in Ireland: a trend analysis..., Pericin [/bib_ref]. This recruitment crisis is particularly affecting rural general practice, therefore, it was notable that rural based GPs were significantly more likely to provide care to nursing home residents than non-rural GPs. In the face of this recruitment crisis and in the context of current inadequate reimbursement the future of GP led nursing home care is uncertain. Future policy needs to focus on adequate resourcing of dementia care in both community and nursing home settings.
The study results have several immediate practical implications for GPs caring for people with dementia. The implications for GPs assessing pain in people with dementia include; being more aware of the increased risk of pain in people with dementia, considering pain as a potentially reversible trigger for BPSD, providing people with dementia an opportunity to self-report pain and familiarising themselves with the pain assessment tools that may already be in place in the nursing homes they attend in order to facilitate more effective communication with the nursing home staff. From a pain management perspective one practical implication for GPs is that if pain is identified, or suspected, then it is important that the person's pain is not under-treated. Inappropriate prescribing in people with dementia does not just mean over-prescribing. It also pertains to under-prescribing of appropriate medications that can improve comfort and overall quality of life.
# Strengths and limitations
Although a 49% response rate is modest, it is typical of postal surveys with this professional group [bib_ref] Short report: encouraging GPs to complete postal questionnaires--one big prize or many..., Thomson [/bib_ref]. Our response rate is similar to the response rate in a previous national study on GPs attitudes to diagnosis in dementia [bib_ref] The attitudes and practices of general practitioners regarding dementia diagnosis in Ireland, Cahill [/bib_ref] and significantly higher than a recent online survey of Irish GPs referral patterns in dementia care [bib_ref] Dementia Diagnosis and Referral in General Practice: A Representative Survey of Irish..., Dyer [/bib_ref]. To adequately power this study a sample size of 175 respondents was required. Therefore, with 157 respondents the study was marginally underpowered. Furthermore, this study was a census study of a specific geographical area in the southern region of the Republic of Ireland. This may affect the generalisability of the study, however, the respondents' demographic characteristics, in terms of years of experience and practice location, are representative of Irish GPs nationally. A large proportion of respondents had a nursing home commitment. There is currently no Irish data available on the number of GPs who attend nursing homes, therefore, it is difficult to ascertain whether this represents a respondent bias. It is possible that GPs with a nursing home commitment were more likely to respond to the survey as they may have been more interested in the research topic. Finally, this was a cross-sectional survey of self-reported knowledge and there is evidence to suggest that when self-reporting physicians may underestimate their knowledge in an area [bib_ref] Selfassessment of medical knowledge: do physicians overestimate or underestimate?, Jankowski [/bib_ref]. The nuances of clinically complex areas such as this are difficulty to fully address with a single study that rely primarily on self-reporting measures. However, since there is very little research exploring GPs experiences of managing pain in people with dementia our chosen methodological approach is a necessary and reasonable place to start.
# Conclusions
Despite the pivotal role GPs play in dementia care their experience of managing pain in dementia is greatly under-researched. Prior to this study very little was known about GPs' knowledge of and attitudes towards the assessment and management of pain in dementia. This study enriches existing literature in the area of pain management in dementia care and also raises some important issues that should be explored in future research. Pain management in a person with dementia touches on some of the most fundamental aspects of a GP's role as a patient advocate. It is, therefore, challenging to identify aspects of care that could, perhaps, be improved upon. However, without identifying these areas we cannot design effective interventions to appropriately address them. The results of this study will be used to inform the development of interventions to improve the management of dementia in general practice as part of the wider implementation of the Irish National Dementia Strategy.
## Additional file
Additional file 1: Questionnaire A blank copy of the questionnaire (DOCX 19 kb)
Abbreviations BPSD: Behavioural and psychological symptoms of dementia; GPs: General practitioners
[fig] ": Usual gripefee for attending nursing home residents in now 1/3 of what it was in 2008! Hhhmmmm….. (Respondent_70, mid-career GP, rural practice, with a nursing home commitment)Several GPs also stated that they no longer provided care to nursing home patients as a result of these reductions in remunerations:I have given up nursing home care. Funding poor. Bureaucracy a problem. (Respondent_26, experienced GP, urban practice, no nursing home commitment) [/fig]
[table] Table 1: Participant demographics [/table]
[table] Table 2: Responses to Likert statements on assessment of pain in people with dementia [/table]
[table] Table 3: Responses to Likert-type statements on management of pain in people with dementia Paracetamol is the best analgesic to use for residents with dementia who are experiencing chronic pain.Residents with dementia are less likely to become addicted to opioid analgesics than cognitively intact patients.There is a greater risk of side effects from opioid analgesics (e.g. respiratory depression, confusion) when used in residents with dementia.Non-drug based methods of pain control (e.g. TENs, Heat/Cold, massage, complimentary therapy) are useful in the management of pain in residents with dementia. [/table]
|
Access to basic drinking water services, safe water storage, and household water treatment practice in rural communities of northwest Ethiopia
Protecting water from cross contamination at source and point of use is an important strategy to improve water quality. However, water safety measures at the source and point of use may not be implemented in the rural communities. This community-based cross-sectional study was, therefore, conducted among 1190 randomly selected households in a rural setting of northwest Ethiopia to assess access to basic drinking water services, safe water storage, and household water treatment practices. Water service level was determined using JMP criteria and practices that prevent cross contamination of water at point of use were used to determine safe water storage. Results showed that 23.0% of the households had access to basic water services; 37.0% practiced safe water storage; and 15.4% practiced one or more household water treatment methods. Public taps (54.5%) and protected spring (25.1%) were the common water sources to rural communities in northwest Ethiopia. Boiling (43.2%), chlorination or water guard (26.8%), and plain sedimentation (23.0%) were among the household water treatment methods commonly practiced in the area. In conclusion, rural households in the studied region has low access to basic water services. Safe water storage practice was also low in the area and household water treatment is not commonly practiced.OPEN
Lack of safe water remains one of the world's most urgent health issues. People in developing countries have no access to safe and adequate drinking water despite access to safe drinking water is a global priority agenda. In 2015, it was estimated that 56% of the world's population had an unsafe water source [bib_ref] Global, regional, and national comparative risk assessment of 79 behavioural, environmental and..., Forouzanfar [/bib_ref]. Unsafe water sources are important sources of infectious diseases transmission [bib_ref] Global distribution of outbreaks of water-associated infectious diseases, Yang [/bib_ref] [bib_ref] Public health hazards due to unsafe drinking water, Pal [/bib_ref]. The global burden of disease study estimated that in 2015, an unsafe water source resulted in 1.2 million deaths and 71.7 million disability-adjusted life years (DALYs) [bib_ref] Global, regional, and national comparative risk assessment of 79 behavioural, environmental and..., Forouzanfar [/bib_ref]. Access to safe drinking water together with hygiene and sanitation is fundamental to global health. Almost one tenth of the global disease burden could be prevented by increasing access to safe drinking water and improving sanitation and hygiene. Annually, safer water could prevent 1.4 million child deaths from diarrhea; 500,000 deaths from malaria; 860,000 child deaths from malnutrition; and 280,000 deaths from drowning. In addition, 5 million people can be protected from being seriously incapacitated from lymphatic filariasis and another 5 million from trachoma.
Limited access to drinking water services continues to be a major public health problem in Ethiopia. In Ethiopia, provision of safe, accessible, and reliable water is very critical. In rural areas of Ethiopia during 2016, 4% have safely managed services, 30% have basic services, and 26% have limited services. This leaves 40% of the country's rural population with unimproved water services. According to Ministry of Water, Irrigation and Electricity (MoWIE) estimates, rural water supply coverage reached 63% by mid-2016 7 (or 57% according to the Ethiopian Demographic and Health Survey 2016), but these assessments were not made based on the JMP service delivery categories.
Source-based water safety measures and protecting water from cross contamination at point of use are important strategies to improve water quality in the water supply system and to minimize associated health www.nature.com/scientificreports/ consequences. Source-based interventions include designing and construction of improved source that have the potential to protect water from contamination and deliver safe water; community-driven sanitation to protect pollution of the catchment area from human, animal, and agricultural wastes; and source-based water treatment. Water safety measures at point of use include safe water storage and household water treatment [bib_ref] Safe water treatment and storage in the home: A practical new strategy..., Mintz [/bib_ref] [bib_ref] Household water treatment and safe storage to prevent diarrheal disease in developing..., Clasen [/bib_ref] [bib_ref] Review of low-cost point-of-use water treatment systems for developing communities, Pooi [/bib_ref]. However, these safety measures may not be implemented in the rural communities due to limited knowledge, misinformation, negative attitude, and lack of experience toward best practices of alternative water treatment technologies and safe storage [bib_ref] Knowledge, attitude, and practice of mothers/caregivers on household water treatment methods in..., Bitew [/bib_ref]. Accordingly, this community-based cross-sectional study was conducted to assess access to basic drinking water services, safe water storage, and household water treatment practice in rural communities of northwest Ethiopia.
# Methods
Study design and setting. A community-based cross-sectional study with structured observation was conducted among rural households in Central and North Gondar administrative zones of the Amhara national regional state, Ethiopia in May 2016 . Central Gondar zone covers thirteen districts and North Gondar zone covers seven districts. The total population residing in Central Gondar is estimated to be 2,896,928 and it is estimated to be 912,112 in North Gondar zone 14 .
Sample size calculation and sampling procedures. The sample size (i.e., 1210 rural households)
was calculated using single population proportion formula and the target households were included in the study using systematic random sampling technique. The sample size calculation and sampling procedures are described in more detail elsewhere [bib_ref] Human Ectoparasites are highly prevalent in the rural communities of Northwest Ethiopia:..., Gizaw [/bib_ref].
Data collection tools and procedures. www.nature.com/scientificreports/ language, and back-translated into English to check consistency. The questionnaire was organized in to three parts: (1) socio-demographic information; (2) access to WASH information; and (3) drinking water sources, handling practice, and household water treatment. Environmental health experts were participated in the data collection process after getting a one day training on the tool. The data collection process and completeness of data was closely supervised.
Measurement of study variables. Access to basic drinking water services, safe water storage, and homebased water treatment were the primary outcomes of this study. Access to basic drinking water services was defined as drinking water from all year round improved source that have the potential to deliver safe water by nature of their design and construction, and include: piped water, public taps, protected wells, protected springs, and protected rain catchments, provided collection time is not more than 30 min for a roundtrip including queuing. Safe water storage was defined as storing water in clean narrow-mouthed and properly covered containers plus withdrawing water from the storage containers by tilting or pouring [bib_ref] Access to improved water and household water treatment practice in rural communities..., Azage [/bib_ref]. Household water treatment is the application of different water treatment options including solar disinfection, chlorination (water guard), filtration, plain sedimentation, or boiling that improve water quality at the point of use [bib_ref] Household water treatment and safe: storage options in developing countries, Lantagne [/bib_ref].
Data processing and analysis. Ethics approval and consent to participate. Ethical clearance was obtained from the Institutional Review Board of the University of Gondar (reference number: V/P/RCS/05/1520/2016). There were no risks due to participation and the collected data were used only for this research purpose with complete confidentiality. Written informed consent was obtained from household heads. All the methods were carried out in accordance with relevant guidelines and regulations.
# Results
Sociodemographic characteristics. A total of 1190 households were participated in the current study, with a response rate of 98.3%. The family size in 513 (43.1%) of the households was more than five and 1013 (85.1%) of the households had children. Three-forth, 888 (75.3%) and 643 (59.3%) of the female and male heads, respectively did not receive formal education. About half, 565 (47.5%) of the households reported that they received WASH education and 967 (81.3%) of the households reported that they have been regularly supervised by health professionals. Furthermore, 812 (68.2%) of the households reported that they regularly discussed about health and sanitation issues with their family. Similarly, 524 (44%) of the households reported that they discussed about health and sanitation issues with village health groups [fig_ref] Table 1: Sociodemographic characteristics of households [/fig_ref].
## Water sources.
In the current study, 957 (80.4%) of the households had access to improved water sources and more than half, 649 (54.5%) of the households collected drinking water from public taps [fig_ref] Figure 2: Drinking water sources for households [/fig_ref]. The water sources for 156 (13.1%) households are not all year round and 837 (70.3%) of the households reported that the time for water collection is more than 30 min for a roundtrip including queuing time. The volume of water collected in 1154 (97.0%) of the households was below 20 L per capita per day. Accordingly, 274 (23.0%) (95% CI: 20.7, 25.3%) of the households had access to basic water services [fig_ref] Table 2: Access to basic water services among households [/fig_ref].
Water handling at point of use. Two-third, 795 (66.8%) of the households primarily stored drinking water using narrow-mouthed containers, such as Jerricane. The water storage containers were clean at the time of the survey in 859 (72.2%) of the households and 544 (45.7%) of the households reported that they daily washed or cleaned the water storage containers. Moreover, the water storage containers were properly covered at the time of the survey in 1046 (87.9%) of the households and 768 (64.5%) of the households withdraw water from the storage containers by pouring or tilting. Accordingly, 440 (37.0%) (95% CI: 34.2, 39.6%) of the households practiced safe water storage [fig_ref] Table 3: Water handling practices of households [/fig_ref].
Household water treatment. were among the household water treatment methods commonly practiced in the rural households. We also investigated the reasons why households did not treated water at household-level and found that 780 (77.5%) of the households reported that they did not practiced household water treatment methods since they believed that the water is safe and 232 (23.0%) of the households did not practiced household water treatment methods due to knowledge or awareness gap [fig_ref] Table 4: Household [/fig_ref].
Factors associated with safe water storage and household water treatment. Health education, health supervision, family discussion, maternal education, paternal education, and family size were entered in to the multivariable model to identify factors associated with safe water storage. In the adjusted model, safe water storage was significantly associated with health education, health supervision, and family size. The odds of safe [fig_ref] Table 5: Factors associated with safe water storage practice among households [/fig_ref]. Health education, health supervision, family discussion, maternal education, water sources, family size, and presence of children in the household were entered in to the multivariable model to identify factors associated with household water treatment. In the adjusted model, household water treatment was statistically associated with health professionals close supervision and family discussion on WASH issues. The odds of practicing household water treatment was 1.91 times higher among households who have been frequently supervised by health professionals compared with their counterparts . Similarly, the odds of practicing household water treatment was 2.15 times higher among households who regularly discussed about
# Discussion
This is a community-based cross-sectional study conducted to assess access to basic water services, safe water storage, and household water treatment practice among households in a rural setting of northwest Ethiopia. This study found that 23.0% (95% CI 20.7, 25.3%) of the households had access to basic water services; 37.0% (95% CI 34.2, 39.6%) of the households practiced safe water storage; and 15.4% (95% CI 13.3, 17.5%) of the households practiced one or more household water treatment methods. Boiling (43.2%), chlorination or water guard (26.8%), and plain sedimentation (23.0%) were among the household water treatment methods commonly practiced in the rural households. www.nature.com/scientificreports/ The proportion of households who had access to basic water services in the current study (i.e., 23%) is comparable with a report of JMP, i.e., 30% of households in rural areas of Ethiopia have basic services 6 . However, the proportion of households with basic services in the current study is lower than reports of the Ministry of Water, Irrigation and Electricity (MoWIE) and the 2016 Ethiopian Demographic and Health Survey. Ministry of Water, Irrigation and Electricity reported that rural water supply coverage reached 63% by mid-2016 7 and 57% according to the Ethiopian Demographic and Health Survey 2016). This differences might be due to the assessment methods used. In our study, we used the JMP definition to basic services as elaborated in the method part, whereas assessments in the aforementioned reports were not made based on the JMP service delivery categories, i.e., they only considered improved sources, which overestimates the coverage. Moreover, the lower access level to basic services in the current study might be due to the resilience of water sources in dry season since we collected data in the dry season. Most water sources in dry season are unreliable, which leads the community to use unimproved water sources in long distances.
The proportion of households who safely stored water in the current study (i.e., 37%) is comparable with findings of a study in rural households of Oshimili North Local Government Area of Delta State, Nigeria, 40% [bib_ref] Drinking water quality and handling practices among women in rural households of..., Oloruntoba [/bib_ref]. On the other hand, the proportion of households who safely stored water in the studied region is lower than findings of studies in three districts of Amhara region, 58.8% 17 and Bona District of Sidama zone, 78.1% 20 . This low-level safe storage practice in the study area might be due to the fact that water storage is affected by traditions such as use of wide-mouthed clay pots. Rural communities preferred to store water in wide-mouthed traditional clay pots because they believe that the use of a clay pot makes the water cool and thus "sweet to drink" [bib_ref] Improving households knowledge and attitude on water, sanitation, and hygiene practices through..., Wasonga [/bib_ref] [bib_ref] Sociocultural determinants to adoption of safe water, sanitation, and hygiene practices in..., Wasonga [/bib_ref]. However, the problem associated with this practice is the way they withdraw water, i.e., dipping of mugs, which largely cross contaminate the water. Moreover, rural communities may not have access to detergents to wash water storage containers due to poor socioeconomic status 23-25 that makes the storage containers unsafe.
The proportion of households who practiced household water treatment in the current study (i.e., 15.4%) is comparable with findings of studies in Degadamot Woreda, northwest Ethiopia, 14% [bib_ref] Assessment of knowledge and practice of house hold water treatment and associated..., Tsegaye [/bib_ref] and Assosa Woreda of Benishangul Gumuz Region, 13.2% [bib_ref] Assessment of drinking water accessibility, handling and treatment practice in Assosa Woreda,..., Merga [/bib_ref]. On the other hand, it is lower than findings of studies in Southern Ethiopia, 29.9% 28 ; Ameya district of Oromia region, 30.3% 29 ; Gibe District of Southern Ethiopia, 34.3% 23 ; India, 53% 30 ; Zambia, 50% 31 ; Nigeria, 54% 32 ; and Uganda, 76% [bib_ref] Perfomance of household water treatment methods for microbial removal under household conditions..., Saturday [/bib_ref]. The low-level practice of household water treatment in the studied region might be due to knowledge or awareness gap, perceived quality of drinking water, unavailability of treatment options, and cost. As documented in the current study, rural households did not practice household water treatment because of the following reasons: believing that the water is safe (77.5%), knowledge or awareness gap (23.0%), unavailability of treatment options (6.5%), and treatment options are expensive (5.4%). Over all, the low-level practice of household water treatment in the area can be due to psychological factors. Attitude . Factors associated with household water treatment practice among households (n = 1190) in a rural setting of northwest Ethiopia, May 2016. AOR adjusted odds ratio, CI confidence interval, COR crude odds ratio. *Statistically significant at p < 0.05, *** statistically significant at p < 0.001, Hosmer and Lemeshow test = 0.412. www.nature.com/scientificreports/ towards water-related technology or behavior is the most important psychological factor to make people treat the drinking water [bib_ref] Barriers and enabling factors associated with the implementation of household solar water..., Bitew [/bib_ref] [bib_ref] Endogeneity in water use behaviour across case studies of household water treatment..., Daniel [/bib_ref] [bib_ref] Socio-economic and psychological determinants for household water treatment practices in indigenous-rural, Daniel [/bib_ref]. This study revealed that safe water storage was significantly associated with health education, health supervision, and family size. The odds of safe water storage was higher among households who received health education, who have been regularly supervised by health professionals, and who had small family size. Similarly, household water treatment was associated with health supervision and family discussion. The odds of practicing household water treatment was higher among households who have been frequently supervised by health professionals and among households who regularly discussed about WASH issues with their families. The effect of health education can be justified as health education encourages changes in healthy behaviors and it is an effective strategy to create demand for water safety measures and thereby increase good practice [bib_ref] Endogeneity in water use behaviour across case studies of household water treatment..., Daniel [/bib_ref] [bib_ref] Advocating for behavior change with education, Arlinghaus [/bib_ref] [bib_ref] The influence of education on health: An empirical assessment of OECD countries..., Raghupathi [/bib_ref] [bib_ref] Does better education mitigate risky health behavior? A mendelian randomization study, Viinikainen [/bib_ref]. Moreover, health supervision is effective in improving or maintaining households' WASH practices. Health supervision is critical in area where there is no other sources of health information and low self-determination to improve WASH [bib_ref] Effect of supervision and education on community health workers performance in the..., Ziabari [/bib_ref]. The effect of large family size can be justified that large family number diverts attention of household heads to routine family supports than investing in water safety measures [bib_ref] Socioeconomic predictors of intestinal parasitic infections among under-five children in rural Dembiya,..., Gizaw [/bib_ref] [bib_ref] Family and social determinants of health-seeking behaviour of caregivers of febrile children..., Lovelyn [/bib_ref]. Moreover, large family sized households may have economic constrains and so that households may not have opportunities to invest on water safety measures [bib_ref] Socioeconomic predictors of intestinal parasitic infections among under-five children in rural Dembiya,..., Gizaw [/bib_ref] [bib_ref] Assessing income-wise household environmental conditions and disease profile in urban areas: Study..., Rahman [/bib_ref].
Lastly, to increase the degree to which inferences from the sample households can be generalized to a larger group of population (i.e., population validity), we recruited households at random or in a manner in which they are representative of the population that we wish to study and we granted that every household had an equal chance to be included in the study. In addition, we calculated adequately powered sample size using sample size determination procedures appropriate to objective with appropriate assumptions. Furthermore, our findings could be applicable to other situations and settings which have similar characteristics with the study populations of the current study, such as rural settings in developing countries (i.e., ecological validity). As limitations, the self-reported data may not be reliable since the study subjects may make the more socially acceptable answers rather than being truthful and they may not be able to assess themselves accurately, which might result reporting bias. Moreover, variables we included in the current study to identify factors associated with water handling/ management practices are not complete.
# Conclusion
Rural households in the studied region has low access to basic water services. This low access to basic water services implies that the community is collecting water from unimproved water sources that results contamination of water with disease causing pathogens and chemicals at the source. Moreover, the proportion of households who safely stored water is low in the area, which may intensify the level of water contamination in the water supply chain. Furthermore, household water treatment is not commonly practiced in the study area that indicates protection of water sources from contamination and source-based water treatment are effective approaches to improve drinking water safety in the area. All these imply that access to safe water in a rural setting of northwest Ethiopia is very critical and the spread of water-borne diseases in the community might be high. The local health department in collaboration with the community and other stakeholders need, therefore, strongly work to design and construct communal water sources that have the potential to deliver safe and adequate water all year rounds. Moreover, maintaining the constructed water sources is important since most of the water infrastructures were damaged. In addition, promotion of water safety measures at point of use, such as safe water storage and household water treatment through health education, health supervision, and village discussions is critical.
## Data availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.
# Funding
The research project was funded by the University of Gondar (grand number: R/T/T/C/Eng./250/08/2016).
[fig] Figure 2: Drinking water sources for households (n = 1190) in a rural setting of northwest Ethiopia, May 2016. [/fig]
[table] Table 1: Sociodemographic characteristics of households (n = 1190) in a rural setting of northwest Ethiopia, May 2016. [/table]
[table] Table 2: Access to basic water services among households (n = 1190) in a rural setting of northwest Ethiopia, May 2016. l/c/d liters per capita per day. [/table]
[table] Table 3: Water handling practices of households (n = 1190) in a rural setting of northwest Ethiopia, May 2016. [/table]
[table] Table 4: Household [/table]
[table] Table 5: Factors associated with safe water storage practice among households (n = 1190) in a rural setting of northwest Ethiopia, May 2016. AOR adjusted odds ratio, CI confidence interval, COR crude odds ratio. [/table]
|
Impact of a soy drink on climacteric symptoms: an open-label, crossover, randomized clinical trial
Objectives: The objective of this study is to evaluate the effects of a soy drink with a high concentration of isoflavones (ViveSoy Õ ) on climacteric symptoms.Methods: An open-label, controlled, crossover clinical trial was conducted in 147 peri-and postmenopausal women. Eligible women were recruited from 13 Spanish health centers and randomly assigned to one of the two sequence groups (control or ViveSoy Õ , 500 mL per day, 15 g of protein and 50 mg of isoflavones). Each intervention phase lasted for 12 weeks with a 6week washout period. Changes on the Menopause Rating Scale and quality of life questionnaires, as well as lipid profile, cardiovascular risk and carbohydrate and bone metabolism were assessed. Statistical analysis was performed using a mixed-effects model. Results: A sample of 147 female volunteers was recruited of which 90 were evaluable. In both sequence groups, adherence to the intervention was high. Regular consumption of ViveSoy Õ reduced climacteric symptoms by 20.4% (p ¼ 0.001) and symptoms in the urogenital domain by 21.3% (p50.05). It also improved health-related quality life by 18.1%, as per the MRS questionnaire (p50.05). Conclusion: Regular consumption of ViveSoy Õ improves both the somatic and urogenital domain symptoms of menopause, as well as health-related quality of life in peri-and postmenopausal women.
# Introduction
Over the last decades, consumption of plant-based beverages has become popular in Western countries [bib_ref] Soy milk and dairy consumption is independently associated with ultrasound attenuation of..., Matthews [/bib_ref] [bib_ref] The origins of western obesity: a role for animal protein?, Carty [/bib_ref]. The phytoestrogenic effects of soy isoflavones and isoflavone supplements have been proposed as alternatives to conventional hormone therapy [bib_ref] Soy foods, isoflavones, and the health of postmenopausal women, Messina [/bib_ref].
Several studies showed that the regular consumption of isoflavones may be beneficial for peri-and postmenopausal women when it comes to the reduction of climacteric symptoms [bib_ref] Nonhormonal therapies for menopausal hot flashes: systematic review and meta-analysis, Nelson [/bib_ref] , protection against bone decalcification [bib_ref] Soy milk and dairy consumption is independently associated with ultrasound attenuation of..., Matthews [/bib_ref] [bib_ref] Isoflavones: effects on bone health, Castelo-Branco [/bib_ref] [bib_ref] Phytoestrogens and bone health at different reproductive stages, Castelo-Branco [/bib_ref] , and reduction of markers of ischemic heart disease [bib_ref] Soy foods, isoflavones, and the health of postmenopausal women, Messina [/bib_ref].
A Cochrane systematic review identified six studies that found a significant reduction in the frequency and severity of hot flushes and night sweats in perimenopausal and postmenopausal women [bib_ref] Phytoestrogens for menopausal vasomotor symptoms, Lethaby [/bib_ref] [bib_ref] Short-term effects of phytoestrogen-rich diet on postmenopausal women, Brzezinski [/bib_ref] [bib_ref] The effect of dietary soy supplementation on hot flushes, Albertazzi [/bib_ref] [bib_ref] Isoflavone treatment for acute menopausal symptoms, Cheng [/bib_ref] [bib_ref] Assessment of soy phytoestrogens and exercise on lipid profiles and menopause symptoms..., Hanachi [/bib_ref] [bib_ref] Evaluation of isoflavone rich soy protein supplementation for postmenopausal therapy, Radhakrishnan [/bib_ref] [bib_ref] The effect of dietary soy supplementation compared to estrogen and placebo on..., Carmigiani [/bib_ref].
Isoflavones may lower LDL-C levels [bib_ref] Benefits of soy isoflavone therapeutic regimen on menopausal symptoms, Han [/bib_ref] [bib_ref] Effect of soy protein from differently processed products on cardiovascular disease risk..., Matthan [/bib_ref] [bib_ref] Effect of two types of soy milk and dairy milk on plasma..., Gardner [/bib_ref] , improve endothelial function, and slow the progression of atherosclerosis [bib_ref] Soy foods, isoflavones, and the health of postmenopausal women, Messina [/bib_ref] , although there is still controversy about their impact on hypercholesterolemia [bib_ref] Isoflavones for hypercholesterolaemia in adults, Qin [/bib_ref] and cardiovascular risk [bib_ref] Cardiovascular risks in relation to daidzein metabolizing phenotypes among Chinese postmenopausal women, Liu [/bib_ref].
In vitro studies also showed that isoflavones may inhibit the mechanisms involved in adipose tissue growth and regulate adipogenesis [bib_ref] The effects of soy isoflavones on obesity, Ørgaard [/bib_ref].
The benefits of soy drink can not only be attributed to isoflavones but also to its modification of diet through a reduction in the saturated fat intake [bib_ref] The origins of western obesity: a role for animal protein?, Carty [/bib_ref] [bib_ref] Weight loss without losing muscle mass in pre-obese and obese subjects induced..., Deibert [/bib_ref].
The objective of our study was to assess the effects of a commercially available soy drink (ViveSoy Õ ) on climacteric symptoms in peri-and postmenopausal women.
# Materials and methods
## Study design and setting
A randomized, open-label, controlled, crossover clinical trial was conducted to assess the effects of including a commercially available soy drink with a high concentration of isoflavones (ViveSoy Õ ) in the diet of peri-and postmenopausal women.
Participants were selected and followed up by their primary care physician at 13 primary care centers. Subject assignment to the intervention type was conducted using a block randomization list with four subjects in each block. They were assigned in a 1:1 ratio to two sequence groups. Every participant was assigned to the intervention type (group 1: control -ViveSoy Õ ; group 2: ViveSoy Õ -control). Each phase (control or ViveSoy Õ ) lasted for 12 weeks followed by a 6-week washout period.
Women in both groups were instructed by their physicians to follow a balanced diet without soy-based products or soy supplements. In the ViveSoy Õ phase, women added a daily consumption of 500 mL of soy drink (two 250 mL packs, which provided a minimum of 50 mg of isoflavones and 15 g of protein).
After the washout period, the second phase started. At the beginning and the end of each phase, daidzein and genistein concentrations were evaluated to assess the consumption of soy isoflavones.
## Participants
Perimenopausal or postmenopausal women according the STRAW criteria [bib_ref] Executive summary of the Stages of Reproductive Aging Workshop + 10: addressing..., Harlow [/bib_ref] aged 45 or more, with climacteric symptoms (hot flushes and/or sweating) and non-consumers of soy-based foods or soy supplements were selected. The exclusion criteria included hormone replacement therapy at any time within 6 months prior to the study, hysterectomy and/or ovariectomy, among others.
## Sample size calculation
It was assumed that regular consumption of soy protein or soy isoflavones may result in an approximately 30% reduction in climacteric symptoms [bib_ref] The effect of dietary soy supplementation on hot flushes, Albertazzi [/bib_ref]. Due to the lack of previous studies, a standard deviation of ¼ 1.2 was assumed, aiming at achieving a power of 90% at a significance level of 5% while allowing for an expected dropout rate of 20%. The estimated sample size to be recruited was 152 patients.
## Study variables
The primary variable, climacteric symptoms, was analyzed using the specific MRS quality of life questionnaire [bib_ref] Evaluation of climacteric symptoms (Menopause Rating Scale), Hauser [/bib_ref] [bib_ref] International versions of the Menopause Rating Scale (MRS), Heinemann [/bib_ref] [bib_ref] The Menopause Rating Scale (MRS) scale: a methodological review, Heinemann [/bib_ref]. The scores for the urogenital domain and the psychological domain were also evaluated as secondary variables as well as MRS total score. The Short Form Health Survey (SF-12) questionnaire was performed for generic quality of life assessment [bib_ref] A 12-Item Short-Form Health Survey: construction of scales and preliminary tests of..., Jr [/bib_ref].
The effects of regular consumption of ViveSoy Õ soy drink on anthropometric and clinical parameters, cardiovascular risk (SCORE) [bib_ref] Estimation of ten-year risk of fatal cardiovascular disease in Europe: the SCORE..., Conroy [/bib_ref] [bib_ref] Fourth Joint Task Force of the European Society of Cardiology and other..., Graham [/bib_ref] , total cholesterol, LDL and HDL-cholesterol, triglycerides, C-reactive protein, glucose and HbA1c, S-equol [bib_ref] Dietary phyto-oestrogens: molecular mechanisms, bioavailability and importance to menopausal health, Cassidy [/bib_ref] [bib_ref] Randomized controlled trial of whole soy and isoflavone daidzein on menopausal symptoms..., Liu [/bib_ref] , and bone resorption metabolites were evaluated.
At the end of each period (visits 3 and 6), a structured interview was conducted to assess the dietary habits.
Monitoring of adherence to the recommended diet and to regular consumption of ViveSoy Õ was performed using a leading question adapted from the Haynes-Sackett test and by comparing the quantity of drink supplied with the number of barcodes returned.
# Statistical analysis
All the analyses were conducted on the intention-to-treat (ITT) population, while the main analysis was also conducted in the perprotocol (PP) population.
The Chi-square test or Fisher's exact test was used to evaluate any potential association between categorical variables. Relationships between quantitative and categorical variables were assessed using T-tests, ANOVA, or the non-parametric Wilcoxon and Kruskal-Wallis tests.
Change analysis was performed using a mixed-effects model that accounted for the assigned sequence, period of consumption, variability between individuals and their baseline values.
All statistical analyses were performed using SAS Õ statistical package version 9.3 (SAS Inc., Cary, NC).
# Results
A total of 147 volunteers were recruited between January and June 2012, 57 of whom were excluded from the analysis primarily because they dropped out or failed to meet the selection criteria. A total of 90 women were randomized to a sequence group (45 to group 1: control -ViveSoy Õ and 45 to group 2: ViveSoy Õcontrol) and constituted the ITT evaluable population. Of these, 60 constituted the PP population.
Baseline demographic, clinical and laboratory characteristics as well as the MRS questionnaire scores were homogeneous in both sequence groups with no significant differences between them [fig_ref] Table 1: Demographic data and baseline clinical and analytical characteristics [/fig_ref].
No significant differences in S-equol levels were found in both groups (Kruskal-Wallis test p ¼ 0.1035) at the end of each phase.
Women in the ITT sample ranged from 45 to 62 years of age, with a mean of 51.6 (SD ¼ 3.3) years and a mean BMI of 26.12
[formula] (SD ¼ 3.80). [/formula]
Results from the follow-up surveys showed that 77.3% of women from group 1 (control-ViveSoy Õ ) followed the recommended diet during the control phase and 73.3% during the ViveSoy Õ phase (p ¼ 0.8067). In group 2 (ViveSoy Õ -control), 66.7% correctly followed the recommended diet during the ViveSoy Õ phase and 65.9% during the control phase (p ¼ 1.0000). When it came to consuming the soy drink on a regular basis during the ViveSoy Õ phase, the adherence rate was 66.7% regardless of which sequence group participants had been assigned to (p ¼ 1.000).
In the ViveSoy Õ phases, significant decreases were recorded in the climacteric symptom scores, while in both washing periods, scores remained stable.
In the ViveSoy Õ phase, a mean percentage reduction of climacteric symptoms of 4.4% (median 10%) and 25% was observed in group 1 (control -ViveSoy Õ ) and group 2 (ViveSoy Õcontrol), respectively.
Analysis of the mixed-effects model results revealed that the effect soy drink had on climacteric symptoms was statistically significant (p50.001), regardless of the temporal order of soy drink consumption in the sequence (p ¼ 0.479). Regular consumption of ViveSoy Õ soy drink during 12 weeks reduced climacteric symptoms by 20.4%. Similar results were observed in the PP population.
In participants with urogenital symptoms, the regular consumption of soy drink caused a statistically significant (21.3%) reduction in symptoms (p ¼ 0.019).
Analysis of the mixed-effects model results indicated that regular consumption of ViveSoy Õ soy drink improved the healthrelated quality of life of women, as per the total scores on the MRS questionnaire, regardless of the order of consumption. Thus, total scores on the questionnaire decreased by 18.1% when ViveSoy Õ was consumed (p ¼ 0.005) [fig_ref] Figure 2: Changes in total score on the MRS questionnaire [/fig_ref].
No significant changes were observed in the psychological subscale scores (p ¼ 0.205) as well as in the mental (MCS) and physical health (PCS) dimension scores on the SF-12 questionnaire (MCS, p ¼ 0.196; PCS, p ¼ 0.900) when ViveSoy Õ was consumed.
Regular consumption of ViveSoy Õ soy drink for 12 weeks did not cause any statistically significant changes in anthropometric measurements, lipid parameters, and the atherogenic index.
# Discussion
Although the effects of soy isoflavones on menopause are well known [bib_ref] Posicionamiento de la Asociación Española para el Estudio de la Menopausia sobre..., Juliá [/bib_ref] , few studies have evaluated the efficacy of commercially available soy-based products.
This study aimed to assess, for the first time, the effect of soy drink (ViveSoy Õ ) on peri-and postmenopausal women using a balanced diet-controlled design with the objective of minimizing potential interindividual variability.
The results showed the homogeneity of the populations enrolled and no significant differences were found in the different variables.
With respect to the primary variable, the soy drink under study significantly reduced climacteric symptoms in both evaluated sequences. Also consistent with these results, consumption of soy drink improved women's health-related quality of life and reduced scores in the urogenital domain.
These data are consistent with the results of previous studies in which placebo was compared with soy isoflavone capsules [bib_ref] Nonhormonal therapies for menopausal hot flashes: systematic review and meta-analysis, Nelson [/bib_ref] [bib_ref] Benefits of soy isoflavone therapeutic regimen on menopausal symptoms, Han [/bib_ref] [bib_ref] Evaluation of soy phytoestrogens for the treatment of hot flashes in breast..., Quella [/bib_ref] [bib_ref] Effects of genistein on hot flushes in early postmenopausal women: a randomized,..., Crisafulli [/bib_ref] [bib_ref] Improved cognitive function in postmenopausal women after 12 weeks of consumption of..., Duffy [/bib_ref] [bib_ref] Effects of a standardized soy extract on hot flushes: a multicenter, double-blind,..., Faure [/bib_ref] [bib_ref] A randomised double-blind controlled trial of oral soy supplements versus placebo for..., Gregor [/bib_ref] [bib_ref] A randomized placebo-controlled crossover trial with phytoestrogens in treatment of menopause in..., Nikander [/bib_ref] [bib_ref] Effect of soy-derived isoflavones on hot flushes, endometrial thickness, and the pulsatility..., Penotti [/bib_ref] [bib_ref] Clinical effects of a standardized soy extract in postmenopausal women: a pilot..., Scambia [/bib_ref] [bib_ref] Vasomotor symptom relief by soy isoflavone extract tablets in postmenopausal women: a..., Upmalis [/bib_ref] [bib_ref] Soy isoflavones and melatonin for the relief of climacteric symptoms: a multicenter,..., Secreto [/bib_ref] and are also in line with the results of placebo-controlled studies that evaluated consumption of soy protein by postmenopausal women at a dose of 40 g per day for 12 weeks [bib_ref] The effect of dietary soy supplementation on hot flushes, Albertazzi [/bib_ref].
In a previous study with a randomized parallel group design and other sources of soy protein, no statistically significant changes in climacteric symptoms were found [bib_ref] Randomized controlled trial of whole soy and isoflavone daidzein on menopausal symptoms..., Liu [/bib_ref]. In this regard, the choice of soy drink as the source of soy protein and isoflavones and the crossover design may explain the favorable results of our study on climacteric symptoms, especially with respect to hot flushes.
Hot flushes have a negative impact on quality of life in periand postmenopausal women [bib_ref] Psychosocial and socioeconomic burden of vasomotor symptoms in menopause: a comprehensive review, Utian [/bib_ref] and are a frequent reason for consultation in primary care and in gynecology clinics [bib_ref] Alternatives to hormone therapy for hot flashes: many choices but science is..., Richardson [/bib_ref] [bib_ref] Symptom experience after discontinuing use of estrogen plus progestin, Ockene [/bib_ref]. It is important for health professionals who care for peri-and postmenopausal women to consider a holistic approach in the management of climacteric symptoms that also includes dietary measures. The results of this study support recommending soy drinks to minimize climacteric symptoms during these periods.
Although the protective effect of soy isoflavones against cardiovascular disease has been described [bib_ref] Cardiovascular risks in relation to daidzein metabolizing phenotypes among Chinese postmenopausal women, Liu [/bib_ref] [bib_ref] The effects of soy isoflavones on obesity, Ørgaard [/bib_ref] [bib_ref] Weight loss without losing muscle mass in pre-obese and obese subjects induced..., Deibert [/bib_ref] , we did not find significant differences in total cholesterol or cardiovascular risk (SCORE) after consumption of soy drink. Nevertheless, it is possible that, based on the findings of previous studies [bib_ref] Benefits of soy isoflavone therapeutic regimen on menopausal symptoms, Han [/bib_ref] [bib_ref] Effect of soy protein from differently processed products on cardiovascular disease risk..., Matthan [/bib_ref] [bib_ref] Effect of two types of soy milk and dairy milk on plasma..., Gardner [/bib_ref] , statistically significant differences would be observed if consumption of soy drink was maintained over a longer period of time. Similarly, these differences would perhaps also appear in HbA1C levels, anthropometric measurements (weight and BMI) or bone metabolism markers (BAP) in certain subpopulations of women.
Despite the methodology used in the study being the same as in drug clinical trials, results should be interpreted taking into account the limitations of an open-label study; therefore, the placebo effect cannot be entirely discarded. Because of the limited evaluable sample size, further studies with a larger sample are needed in order to extrapolate our conclusions. Moreover, as the intervention was implemented only in Spanish primary care centers, it will be important to evaluate the effects of soy drink with other ethnically and nutritionally diverse populations.
In summary, the results of this study demonstrate the favorable effect of ViveSoy Õ soy drink on menopause-related symptoms and health-related quality of life in peri-and postmenopausal women, particularly on symptoms that are most bothersome to women, such as hot flushes. Further studies are needed to confirm our findings, including the effect of soy drink on lipid, glucose, and bone metabolism.
[fig] Figure 2: Changes in total score on the MRS questionnaire. [/fig]
[fig] Figure 1: Changes in climacteric symptoms (symptoms subscale on the MRS questionnaire, items 1, 2, 3, and 11). [/fig]
[table] Table 1: Demographic data and baseline clinical and analytical characteristics.MRS, Menopause Rating Scale; BMI, body mass index; SBP, systolic blood pressure; DBP, diastolic blood pressure; BAP, Bone alkaline phosphatase. [/table]
|
Focal vitreomacular traction: Resolution after ocular massage
A B S T R A C TPurpose: Vitreomacular traction (VMT) is a relatively common ocular disorder that may distort the foveal structure causing visual symptoms. The influence of ocular massage (OM) on this condition has not been considered yet. We aim to report clinical and OCT features of VMT release associated with OM. Observations: A 70-year-old woman complained about blurred vision and metamorphopsia in her right eye for one month. Her best-corrected visual acuity (BCVA) was 20/50. Macular OCT showed focal VMT in this eye. Moderate intensity, digital OM was performed by an ophthalmologist. However, the traction was still present. The patient was instructed to perform the same OM every 8 hours at home herself. Four days later she indicated disappearance of metamorphopsia, her BCVA increased to 20/25 and OCT showed VMT release with 39-μm foveal thinning. Conclusions and importance: OM may be useful for focal VMT release.
# Introduction
Vitreomacular traction (VMT) is caused by partial posterior vitreous detachment associated with persistent vitreous attachment to the macula. The traction may distort the foveal structure inducing decreased visual acuity and metamorphopsia. The diagnosis of this entity is confirmed by means of optical coherence tomography. The treatment options for VMT include observation, pars plana vitrectomy (PPV) and intravitreal injections (IVIs) of expansive gas (pneumatic vitreolysis) or ocriplasmin (enzymatic vitreolysis). [bib_ref] The charles Schepens lecture: management options for vitreomacular traction: use an individualized..., Jr [/bib_ref] However, as far as we know, ocular massage (OM) has not been considered so far as an adjunctive treatment to achieve VMT release (VMTR). In this case we describe the resolution of VMT in relation to OM.
## Case report
A 70-year-old woman complained about blurred vision and metamorphopsia in her right eye for one month. Her best-corrected visual acuity (BCVA) was 20/50. Macular OCT showed focal VMT with tiny intraretinal cysts [fig_ref] Figure 1: Progression of the vitreomacular traction [/fig_ref]. The horizontal diameter of VMT was 114 μm. The other ophthalmic examinations were normal, including careful funduscopy of the peripheral retina. Moderate-intensity, digital OM on the affected eye was applied by the ophthalmologist (JJGM) for 1 minute in order to try to release this small adhesion. OM was performed placing the two index fingertips on the nasal and temporal side of the eyeball, with the eyelid of the patient shut, and pressing alternatively with both fingers. Then OCT was repeated but traction was still present. The patient was instructed to perform the same OM (1 minute, moderate intensity massage) every 8 hours at home herself.
Four days later she indicated disappearance of metamorphopsia, her BCVA increased to 20/25, OCT showed VMTR with tiny intraretinal cysts decrease [fig_ref] Figure 1: Progression of the vitreomacular traction [/fig_ref] and 39-μm foveal thinning [fig_ref] Figure 1: Progression of the vitreomacular traction [/fig_ref]. The patient assured that she had accomplished OM as instructed.
# Discussion
In this case, OM was performed with the aim of solving the disorder. OM induces an intermittent vitreous movement and probably a mechanical tension/distension cycle over the VMT. Besides, a diminution in the volume of the vitreous with a reduction of intraocular pressure (IOP) due to vitreous water loss (dehydration) is observed after OM. [bib_ref] Ocular hypotension and massage of the eyeball, Francois [/bib_ref] Vitreous rehydration and subsequent IOP recovery occurs after some minutes. [bib_ref] Recovery of intraocular pressure and vitreous weight after ocular compression, Obstbaum [/bib_ref] It is very unlikely that VMTR had occurred spontaneously because this possibility has been described to happen many months (but not few days) after the diagnosis and in a moderately low percentage of eyes, according to the results of recent studies concerning the natural course of VMT. 4-7 John et al. found that only 30.23% of VMTs (13 out of 43 eyes) without intraretinal cysts or clefts and 30.35% (17 out of 56 eyes) with intraretinal cysts or clefts released spontaneously after a median follow-up period of 18 months. [bib_ref] Clinical course of vitreomacular adhesion managed by initial observation, John [/bib_ref] Additionally, Theodossiadis et al. [bib_ref] Spontaneous resolution of vitreomacular traction demonstrated by spectral-domain optical coherence tomography, Theodossiadis [/bib_ref] demonstrated that 28.5% of VMTs (12 out of 46 eyes) proceeded to spontaneous resolution during a mean follow-up of 8.75 ± 6.06 months. They also found that when the horizontal diameter of VMT is narrow (< 400 μm), a greater force is exerted upon the fovea, which facilitates the VMTR. Focal VMT has also been associated to a better final visual acuity after PPV. [bib_ref] Predictive factors of visual outcome for vitreomacular traction syndrome after vitrectomy, Yang [/bib_ref] In our case the diameter of VMT was very narrow (114 μm).
In another study Dimopoulos et al. [bib_ref] Rate and timing of spontaneous resolution in a vitreomacular traction group: should..., Dimopoulos [/bib_ref] noted that 43% of eyes affected with VMT (20 out of 46 eyes) presented a spontaneous VMTR during a median follow-up period of 594 days (longer than 19 months), most of them after 6-12 months of observation, being the median duration from baseline examination to the VMTR of 375 days. Plus, Errera 7 et al. recently observed that only 20% of 183 eyes with VMT resolved spontaneously in a 17.4-month follow-up period (occurring on average at 15 months with a range between 1 and 48 months).
Thus, in the present clinical case VMTR, that occurred 4 days later, seemed rather to be related to mechanical effects of external OM.
Otherwise, it is known that internal vitreous manipulation effected through IVIs can induce VMTR. IVIs produces hyperhydration of the vitreous and transient IOP elevation. [bib_ref] Incidence and management of elevated intraocular pressure with antivascular endothelial growth factor..., Abedi [/bib_ref] These changes seem to promote VMTR. In fact, Stalmans et al. [bib_ref] Enzymatic vitreolysis with ocriplasmin for vitreomacular traction and macular holes, Stalmans [/bib_ref] showed that IVIs of 0.1 ml of saline resulted in VMTR in 10.1% of eyes compared with 26.5% of VMTR in eyes treated with IVIs of ocriplasmin at day 28. Although a biological effect is attributed to ocriplasmin the authors admitted some treatment response to placebo injections. Recently, a study by showed macular structural changes and subretinal fluid in the eyes treated with ocriplasmin, not seen in eyes that had PPV. Visual improvement and VMTR rate (50% with ocriplasmin versus 100% with PPV) were better with vitrectomy.
Plus, previous IVIs of anti-VEGF has been associated with a higher incidence of VMTR. Almeida et al. [bib_ref] Factors associated with spontaneous release of vitreomacular traction, Almeida [/bib_ref] showed that 52% of eyes with VMTR received IVIs of anti-VEGF during the observation period (mean of 9.1 ± 8.9 injections during 13.7 ± 11.4 months) versus only 13% of eyes in the group of persistent VMT (mean of 2.8 ± 1.8 IVIs during 10.0 ± 6.6 months). More recently, Yu et al. [bib_ref] Efficacy and safety of treatment options for vitreomacular traction: a case series..., Yu [/bib_ref] showed in a metaanalysis that IVIs of expansile gas induced a higher rate of VMTR (87.5%) in comparison with IVIs of ocriplasmin (42.9%) at day 28. The authors attributed the VMTR associated with IVIs of expansile gas to the internal massaging action of the bubble over VMT. Similar results using expansile gas were found by Chan et al. 14 (86% of VMTR after a single IVI) at a median of 3 weeks. All these evidences lead to think that internal manipulation of the vitreous may help to release VMT.
It would have been interesting in this case to determine whether OM increases vector forces that might have contributed to the release of the vitreomacular adhesion. An early attempt to study VMT optically was done by Schepens, Trempe and Takahashi, using a noncontact lens (never commercialized), which predated the 90-diopter and similar lenses we have today. [bib_ref] Atlas of Vitreous Biomicroscopy, Schepens [/bib_ref] A possible approach to study these vector forces in our patient could have been using a hand-held 90-diopter lens and oblique slit-lamp illumination, or perhaps with dynamic B-scan ultrasound or OCT. The eye could have remained stationary during OCT, while digital pressure could have been intermittently applied. Several cautions must be considered before OM is initiated. It is reasonable to avoid this technique in pseudophakic eyes with posterior capsular disruption or eyes with peripheral retinal tears or degenerations, for example. All these were ruled out in our case. Although it seems to be rare, other potential harmful effects of OM (or eye rubbing) have been exceptionally reported such as hypotony maculopathy, [bib_ref] Hypotony maculopathy after eyelid massage for overcorrected blepharoptosis, Nguyen [/bib_ref] subretinal hemorrhage, 17 retinal tear/detachment 18 or lens dislocation. [bib_ref] Bilateral rupture of the posterior capsule and intraocular lens dislocation from excessive..., Bassily [/bib_ref] Complications might also be more likely when the ocular massage is performed by the patient, rather than in a clinical setting. Therefore, a careful explanation of the technique should be provided to the patient before starting ocular self-massaging.
In conclusion, although our experience is limited because this is the first and only time we have treated a focal VMT in this way, we consider that this safe and inexpensive technique may help to achieve VMTR. Further controlled studies are needed in this sense.
## Patient consent
The patient consented the publication of the case. This report does not contain any personal information that could lead to the identification of the patient.
[fig] Figure 1: Progression of the vitreomacular traction. (A) OCT showing focal vitreomacular traction and (B) release of the traction after 4 days of self-applied digital ocular massage. (C) 39-μm foveal thinning after vitreomacular traction release. [/fig]
|
Gynecomastia: A Rare Adverse Effect of Methylphenidate in an Adolescent Boy
Gynecomastia is a benign condition developing in association with localized fat deposition and glandular tissue proliferation in the breast in males, and characterized by breast growth. Drug is one of the most important factors in the etiology of gynecomastia. Methylphenidate is a commonly preferred and well-tolerated drug in the treatment of attention deficit hyperactivity disorder in children and adolescents. Gynecomastia is an uncommon side-effect of methylphenidate use. We report a case of bilateral gynecomastia developing in a dose-dependent manner during methylphenidate monotherapy and resolving with discontinuation of medication in a 15-year-old patient with a history of a similar side-effect during previous use of the drug. To the best of our knowledge this is one of the few case reports of gynecomastia developing in association with methylphenidate.
# Introduction
Gynecomastia is a benign condition characterized by breast growth in males. It develops in association with localized fat deposition and glandular tissue proliferation in the breast. Proliferation of male breast tissue may be seen at any age, and may be uni-or bilateral. Neyzi et al.reported a prevalence of gynecomastia of 7%. Several factors may be involved in the etiology of gynecomastia. Among these factors, the drug-related gynecomastia was one of the leading causes. Various medications are capable of causing gynecomastia, including diazepam, haloperidol, phenothiazine and tricyclic antidepressants. Methylphenidate (MPH), the first-line treatment for attention deficit hyperactivity disorder (ADHD), is effective and well-tolerated. Gynecomastia is a rare side-effect during MPH therapy. To the best of our knowledge, there have been few case reports of the condition developing as a side-effect of MPH. We report a case of an adoles-cent boy developing bilateral, dose-dependent gynecomastia during MPH monotherapy.
## Case
A 15-year-old boy presented with his mother due to the complaints of symptoms of ADHD including hyperactivity, impulsivity and inattentiveness. Anamnesis taken from the mother revealed that the patient had used the osmotic-release oral system (OROS) MPH (Concerta; Jansen-Cliang Manufacturing LLC., Gurabo, Porto Riko) formulation for ADHD, which he had benefitted from MPH at that time, but that they had discontinued the treatment of their own volition for the previous two years. Since the patient had previously benefited from MPH for ADHD, he was started on OROS MPH (Concerta) at 27 mg/day and invited to follow-up one month later. At follow-up we learned that he had failed to benefit from 27 mg/day OROS MPH (Concerta), and the dosage was raised to 36 mg OROS MPH (Concerta). At follow-up one month subsequently, the patient reported swelling and growth in the breast following dosage enhancement, and stated that this had a psychologically adverse impact on him. The pediatric endocrinology department was then consulted. At consultation, the patient's prolactin was 7.86 ng/ml (normal range, ＜ 15 ng/ml), and no tenderness or galactorrhea were present. The patient weighed 52 kg (10th−25th percentiles) and was 175 cm in height (10th −25th percentiles). His body mass index was 16.97 kg/m 2 . He denied the use of any drug, herbal product or substance. Bilateral gynecomastia was diagnosed by the endocrinology department, and the drug discontinuation was recommended. OROS MPH (Concerta) therapy was stopped, and the patient was started in atomoxetine (ATX) 50 mg/day. When we re-evaluated his previous history, the same side-effect had previously been observed with 27 mg OROS MPH (Concerta), but that the patient had been reluctant to disclose this at the initial interview due to embarrassment. At one-month follow-up we learned from the family that the breast growth had been reversed, but that the patient's attention problems persisted, that he was short-tempered and irritable, and that he had been involved in arguments with teachers at school. Risperidone 1 mg/day was then added to treatment, and ATX was increased to 60 mg/day. At the subsequent follow-up his irritability had decreased. The patient was monitored with ATX 60 mg/day and risperdal 1 mg/day for a period of one year. He benefited from treatment during that period, and no recurrence of gynecomastia was observed.
# Discussion
MPH is a commonly preferred and well-tolerated drug in the treatment of children and adolescents. However, some side-effects may occur during treatment. Commonly reported side-effects during MPH therapy include nausea, lack of appetite, weight loss, and sleep disturbances. In addition to these widespread side-effects, MPH has also been reported to cause rare side-effects such as skin eruptions, inappropriate sexual behavior, obsessive compulsive symptoms, hallucinations, painful muscle cramps, hyperhidrosisand excessive and frequent menstrual bleeding.
Only a few reports of MPH-induced gynecomastia have appeared in the literature. Coşkun et al.reported gynecomastia during OROS MPH (Concerta) use in a boy of 10. Ensat et al.reported unilateral gynecomastia associated with MPH use in a six-year-old boy. The case of Coşkun et al.used paroxetine and MPH therapy, and gynecomastia occurred six months after dosage increase. No regression in gynecomastia occurred after the drug discontinuation. Similarly, no regression was observed after discontinuation in the case of Ensat et al.In our case, however, MPH-related gynecomastia improved after drug stoppage. In a review study, Bowman et al.also reported that drug-related gynecomastia frequently improves after stoppage. From that perspective, our case is in line with previous reports. In our case, gynecomastia emerged immediately after the drug dosage was increased. In the two other cases, the condition emerged independently of dosage and six months after dosage increase. In our case, only the OROS MPH (Concerta) formulation was being used when gynecomastia developed. Our case was evaluated with the Naranjo Adverse Drug Reaction Probability Scale (NADRPS). On this scale, a score ≥ 9 is regarded as definite, a score between 5 and 8 is considered probable, a score between 1 and 4 is considered possible, and a score of 0 is regarded as doubtful. Our case's NADRPS score was 9. Gynecomastia is a physically and socially disturbing psychosocial problem for adolescents with adverse impacts on quality of life and self-esteem. Although gynecomastia had developed previously in our patient in association with the same medication, he was too embarrassed to mention this at interview. Arslan et al.revealed that gynecomastia impaired body perception, self-esteem, and psychosocial functionality in adolescents. Nuzzi et al.also reported adverse psychosocial impacts of gynecomastia in adolescents.
The pathophysiological mechanism by which MPH gives rise to gynecomastia is uncertain. Various mechanisms have been proposed in the development of the condition, including increased estrogen production (increased concentrations in serum and tissue), decreased androgen production or effects thereof, and hypersensitive breast tissue. Impaired hormone balance plays a key role in all these hypotheses. Low testosterone to estradiol and adrenal androgens to estradiol ratios have been determined in cases of prepubertal gynecomastia. In an animal study of the androgenic effects of MPH in rats, Adriani et al.reported a marked decrease in testosterone concentrations with MPH. Testosterone levels decreased significantly, by 40%, in rats treated with MPH compared to control animals. This study suggests that a decreased androgen effect associated with decreased testosterone levels may be responsible for MPH-induced gynecomastia. Treatment in our case was switched to ATX following the development of MPH-related gynecomastia. The patient was followed-up for a one-year period after switching to ATX, and no gynecomastia was observed. To the best of our knowledge, there have been no reports of ATX-related gynecomastia. We therefore elected to use ATX, and no problems were encountered in management of treatment.
Although MPH-related gynecomastia is a rare situation, it developed following an increase in MPH dosage in our case. Clinicians should therefore consider the possibility of gynecomastia with MPH therapy. It will also be beneficial for clinicians to consider ATX as an alternative in the management of MPH-related gynecomastia.
No potential conflict of interest relevant to this article was reported. |
Anticancer and Immunomodulatory Benefits of Taro (Colocasia esculenta) Corms, an Underexploited Tuber Crop
Citation: Ribeiro Pereira, P.; Bertozzi de Aquino Mattos, É.; Nitzsche Teixeira Fernandes Corrêa, A.C.; Afonso Vericimo, M.; Margaret Flosi Paschoalin, V. Anticancer and Immunomodulatory Benefits of Taro (Colocasia esculenta) Corms, an Underexploited Tuber Crop. Int. J. Mol. Sci. 2021, 22, 265. https://
# Introduction
## Underutilized and neglected crops worldwide
Crops that have been neglected over the years are currently being revalued based on modern technologies used to extract, identify, estimate, and assay a great number of compounds displaying claimed pharmacological effects. The study of the composition of such food matrices has stimulated the recognition and reevaluation of so-called "orphan" crops, reaffirming the knowledge that traditional communities have practiced since ancient times by considering the vital role of those crops not only in supporting diets but also in promoting the health and treating these populations. In most cases, neglected or underutilized species have been substituted by those cultures in huge demand, although sometimes, those crops are poorer not only in nutritional aspects but mainly in bioactive compounds.
## Figure 1.
Global distribution of taro production reproduced from FAOSTAT (http://www.fao.org/faostat/en/#data/QC). Quantitative taro production per country in 2018, repreScheme 230. tons followed by the USA, Canada, and Cyprus with production lower than 1600 tons. Uncolored countries represent production areas under 1000 ha. Quantitative taro production per country in 2018, repreScheme 230. tons followed by the USA, Canada, and Cyprus with production lower than 1600 tons. Uncolored countries represent production areas under 1000 ha.. na-Not applied ** [bib_ref] Sources and intake of resistant starch in the Chinese diet, Chen [/bib_ref].
Additionally, taro is a rich source of antioxidants, mainly phenolic compounds, both regarding diversity and quantity, distributed in the edible portion of taro. In addition to antioxidants, taro phytochemicals display immunomodulatory, antioxidant, antitumoral, antimetastatic, antimutagenic, anti-hyperglycemic, and anti-hypercholesterolemic bioactivities. Moreover, taro is a potential alternative staple source, with a lower glycemic index than potato, and its consumption may decrease the incidence and prevalence of several diseases, including certain types of cancers [bib_ref] Glycaemic index of selected staples commonly eaten in the Caribbean and the..., Ramdath [/bib_ref] [bib_ref] Food processing methods influence the glycaemic indices of some commonly eaten West..., Bahado-Singh [/bib_ref] [bib_ref] International table of glycemic index and glycemic load values: 2002. Am, Foster-Powell [/bib_ref].
Despite being considered an orphan crop, taro is a sacred food in some cultures, such as in Hawaii, Melanesia, and Micronesia, where it is known as a Gift of Ancient Gods. In these places, taro is consumed daily and included in several special occasions and rituals due to its symbolic importance. Taro is formulated according to the cultural traditions of each local population. For example, taro stems, petiole, corms, and leaves can be consumed as a common practice in Hawaii. However, taro corms are conventionally considered the edible portion of this plant, and they are consumed worldwide [bib_ref] Potential Immunomodulator and COX-Inhibitor Lectin Found in Taro (Colocasia esculenta), Pereira [/bib_ref]. Some cultivars can exhibit high calcium oxalate contents, which is considered an antinutritional factor that confers an acrid taste to the tubercles, causes skin irritation, and can decrease calcium absorption [bib_ref] Oxalate content of foods and its effect on humans, Bsc [/bib_ref]. For this reason, taro should be preferentially consumed after cooking in order to avoid these undesired effects.
In Hawaii, taro is cooked and smashed with a little water to prepare a starchy paste, which may be consumed immediately (fresh poi) or after 2-3 days of fermentation producing a sour taste paste (sour poi), which is a typical Hawaiian dish [bib_ref] Poi history, uses, and role in health, Brown [/bib_ref]. Achu, another ancient taro paste, preferentially prepared by women, is mostly consumed in Africa. Taro and bananas are boiled together, peeled, and pounded to form a smooth and homogeneous starchy paste. Then, typical sauces are mixed in, such as yellow sauce (achu soup), jaune sauce, black sauce (black soup), and pepper sauce [bib_ref] Traditional preparation of Achu, a cultural keystone dish in western Cameroon, Grimaldi [/bib_ref].
In other parts of the world, especially Brazil, taro can be served fried or steamed, prepared as a soup, or mashed. The corms are also marketed in a variety of commercial products such as flour, chips, fermented alcoholic beverage, ice bar, ice cream and canned taro, among others [bib_ref] Composition and food uses, Maga [/bib_ref] [bib_ref] Utilization of taro (Colocasia esculenta): A review, Kaushal [/bib_ref]. These taro derivatives are not globally available, as taro crops are concentrated in China, Taiwan, and Hawaii. Taro flour can be used as an ingredient for many other preparations including bread, cakes, cookies, noodles, and cereals, or even as a partial substitute for traditional whey flour [bib_ref] Utilization of taro (Colocasia esculenta): A review, Kaushal [/bib_ref] [bib_ref] Bread characteristics and descriptive analysis of taro (Colocasia esculenta, Cv Fouê): Wheat..., Hyacinthe [/bib_ref] [bib_ref] Using of taro flour as partial substitute of wheat flour in bread..., Ammar [/bib_ref].
The main carbohydrate present in taro is starch found in polygonal and small granules, averaging 1.3-2.2 µm in diameter, although granules measuring 5 µm can be observed [bib_ref] Chemical and physical properties of flour extracted from taro Colocasia esculenta (L.)..., Tattiyakul [/bib_ref]. As a starchy vegetable, taro presents part of the starch in resistant form, which can escape small intestine digestion and be directed to colon fermentation. This resistant-starch results in several health effects, including the augmented absorption of minerals, contribution in controlling blood glycemia, and reduction in plasma triglycerides and cholesterol [bib_ref] Production of resistant starch from taro (Colocasia esculenta L. Schott) corm and..., Simsek [/bib_ref].
Since taro is free of gluten and displays low protein and high calorie content, as well as low fat levels, taro consumption can benefit individuals with dietary restrictions such as those presenting allergies, especially in children and gluten-intolerant individuals, while contributing to reduce the risk of obesity and type II diabetes. In addition, the presence of soluble and non-soluble dietary fibers can improve intestinal transit and possibly aid in colorectal cancer prevention. As a result of its gluten-free nature, taro flour has arisen as a promising substitute for wheat flour, boosting the Brazilian market for gluten-free derivatives [bib_ref] Glycaemic index of selected staples commonly eaten in the Caribbean and the..., Ramdath [/bib_ref] [bib_ref] Food processing methods influence the glycaemic indices of some commonly eaten West..., Bahado-Singh [/bib_ref] [bib_ref] International tables of glycemic index and glycemic load values, Atkinson [/bib_ref] [bib_ref] Saura-Calixto, F. A starch hydrolysis procedure to estimate glycemic index, Goñi [/bib_ref] [bib_ref] Resistant starch-A review, Sajilata [/bib_ref].
To encourage and reinforce the importance of taro consumption, this study aims to discuss the benefits of the biofunctional compounds found in taro in promoting health, especially considering their potential against cancer, as well as in the control of other physiopathological conditions that compose the risk factors for cancer burden, including obesity and type II diabetes.
## Risk factors associated with cancer and the impact of healthy dietary habits
Cancer, obesity, and type II diabetes are the main causes of mortality and morbidity worldwide. Over the past decades, the prevalence of these diseases has increased considerably in most countries, as high body mass index and diabetes are responsible for 5-7% of all cancers, which is equivalent to 804,100 new cases in 2012 [bib_ref] Retraction and republication-Worldwide burden of cancer attributable to diabetes and high body-mass..., Pearson-Stuttard [/bib_ref].
The risk of cancer is directly associated with both intrinsic and non-intrinsic factors. Intrinsic factors are inherent to human cell biology and, therefore, non-modifiable, such as spontaneous modifications in DNA that can contribute to the cancer initiation process. Nonintrinsic elements are entirely modifiable and comprehend endogenous and exogenous aspects of cultural habits acquired by society, such as unhealthy lifestyles, exposure to carcinogenic/mutagenic chemicals, and the generation and accumulation of free radicals throughout life [bib_ref] Evaluating intrinsic and non-intrinsic cancer risk factors, Wu [/bib_ref].
Concerning personal individualities, endogenous agents may affect cell proliferation mechanisms and genome integrity triggered by biological aging, genetic susceptibility (heritable cancer genes), hormone and growth factor levels, metabolic dysfunctions, chronic inflammatory status, oxidative stress, and compromised immunological competences. Exogenous agents contribute to about 70% risk of cancer development compared to endogenous factors, by causing new mutations, the activation of oncogenes, or the inhibition of tumor suppressor genes [bib_ref] Substantial contribution of extrinsic risk factors to cancer development, Wu [/bib_ref]. Exogenous agents include infectious diseases that can be prevented by vaccination, protected sex, and no needle sharing. Among contagious diseases, Helicobacter pylori is associated with gastric cancer, human papillomavirus (HPV), with cervical, head and neck cancers, and hepatitis virus B and C, with hepatocellular carcinoma. Mutagens/carcinogenic agents comprise environmental physical or chemical agents such as UV radiation, air pollution substances (benzene, radon), and tobacco (containing 69 types of carcinogenic chemicals), which, if avoided, can prevent DNA damage. Many other substances are related to specific work environments (wood dust, nickel compounds), secondhand tobacco smoke, food (aflatoxins), and hormone therapy (estrogens), and they should be avoided or minimized when possible. Other exogenous agents can be grouped in the lifestyle category, including smoking, alcohol, and dietary habits [bib_ref] Evaluating intrinsic and non-intrinsic cancer risk factors, Wu [/bib_ref].
A healthier lifestyle concept includes physical activity and adequate nutritional habits while excluding smoking and alcohol consumption. Chronic inflammatory diseases linked to dietary habits such as type II diabetes and obesity are among the exogenous risk factors frequently associated with cancer cases [bib_ref] Dietary patterns in association to cancer incidence and survival: Concept, current evidence,..., Bamia [/bib_ref] [bib_ref] Healthy lifestyle and life expectancy free of cancer, cardiovascular disease, and type..., Li [/bib_ref]. On the other hand, a diet including vegetables, fruits, and greens is highly associated with a reduced risk of cancer development in humans [bib_ref] Evaluating intrinsic and non-intrinsic cancer risk factors, Wu [/bib_ref] [bib_ref] Dietary patterns and cancer risk, Steck [/bib_ref] [bib_ref] Diet and Nutrition in the Etiology and Prevention of Cancer, Clinton [/bib_ref] [bib_ref] Food systems approach to cancer prevention, Vanamala [/bib_ref]. These food sources are widely known to be rich in bioactive compounds and displaying pharmacological properties such as antioxidant, anti-inflammatory, anti-obesity, anti-diabetes, immunomodulatory, antimetastatic, and antitumoral activities which, together, can contribute to delay cancer burden, since they prevent oxidative stress, modulate the proinflammatory status, and control metabolic dysfunction [bib_ref] Dietary patterns in association to cancer incidence and survival: Concept, current evidence,..., Bamia [/bib_ref] [bib_ref] Diet and cancer prevention, Davis [/bib_ref]. Several of those pharmacological bioactivities can be found in taro (Colocasia esculenta [L.] Schott), which is a highly nutritious food that is capable of preventing starvation and malnutrition in countries below the poverty line [bib_ref] A critical review of the role of taro Colocasia esculenta L.(Schott) to..., Akwee [/bib_ref].
## Bioactive compounds and pharmacological properties of taro−popular medicinal knowledge with anticancer potential
The use of taro to treat multiple unhealthy conditions and diseases such as diabetes, hemorrhage, diarrhea, arterial hypertension, alopecia, among others, dates from ancient times [bib_ref] Colocasia esculenta: A potent indigenous plant, Prajapati [/bib_ref].
Studies on Colocasia esculenta have been conducted worldwide, but especially in geographic regions where taro cultivation is notably expressive, such as the USA, Nigeria, South Africa, Samoa, Korea, Japan, Philippines, Bangladesh, Indonesia, and New Zealand [fig_ref] Figure 1: Global distribution of taro production reproduced from FAOSTAT [/fig_ref] and [fig_ref] Table 2: [/fig_ref]. Taro's health-promoting potential has been confirmed by in vitro and in vivo preclinical assays, by assaying raw or cooked corms and taro derivatives in the form of flour or extracts ( Proliferation of total mice spleen cells in 5 days
[51] Proliferation of B220+ lymphocytes from mice spleen in 5 or 10 days Proliferation of total mice bone marrow cells in 10 days
## Immunomodulatory
Crude taro extract
In vitro Proliferation of total mice spleen cells Brazil
## Antioxidant
See [fig_ref] Table 3: Antioxidants in Colocasia esculenta from different geographic regions [/fig_ref] ( The bioactivities found in taro have been mainly attributed to its polyphenols, proteins, mucilage, polysaccharides, lipids, and non-polyphenol antioxidants. Many of these bioactive principles have already been identified and singly assayed, proving their participation in these claimed activities. Bioactive molecules identified in taro include tarin, taro-4-I polysaccharide, taro polysaccharides 1 and 2 (TPS-1/TPS-2), A-1/B-2 α-amylase inhibitors, monogalactosyldiacylglycerols (MGDGs), and digalactosyldiacylglycerols (DGDGs). Together, these data clearly indicate that the biological effects exerted by taro are possibly a synergic effect of multiple compounds displaying effectiveness not only against several cancer cell lines but also against some of the main external cancer risk factors, such as free radicals, mutagenic and carcinogenic agents, and physiopathological conditions such as obesity and type II diabetes [fig_ref] Table 2: [/fig_ref].
Interestingly, many of the bioactivities were demonstrated with cooked taro formulations, even following oral administration, suggesting that the recommendation to include taro in the daily human diets, alongside other healthy eating habits, could contribute to the efforts to reduce cancer risks.
## Taro antioxidants
Antioxidants, under appropriate concentrations, protect, prevent, or delay the oxidation of biomolecules such as nucleic acids (DNA and RNA), protein, and lipids. Foods enriched in antioxidants may contribute to health maintenance, especially concerning comorbidities caused by oxidative stress elicited by an excess of oxygen/nitrogen reactive species [bib_ref] Protective role of antioxidative food factors in oxidative stress caused by hyperglycemia, Osawa [/bib_ref] [bib_ref] The role of antioxidants in the chemistry of oxidative stress: A review, Pisoschi [/bib_ref].
Reactive species may be produced in physiological conditions, through aerobic metabolism and macrophage activation, or in pathological states following exposure to xenobiotics, such as toxins, pollutants, cigarettes pesticides, or radiation. To counterbalance oxidative stress, phytochemicals with antioxidant activity found in foods such as fruits, vegetables, cereals, and tubers, are noteworthy as a relevant diet intervention topic. Evaluating the total antioxidant capacity (TAC) of foods is useful to improve functional diet quality and assist in health maintenance [bib_ref] Food selection based on total antioxidant capacity can modify antioxidant intake, systemic..., Valtuena [/bib_ref].
Many phytochemicals found in taro display the potential to reduce oxidative stress. The H-ORAC (Hydroxyl Radical Antioxidant Capacity), ABTS (2,2 -Azino-bis[3ethylbenzothiazoline-6-sulfonic acid] diammonium salt), FRAP (ferric reducing antioxidant power), and DPPH (2,2 -diphenyl-1-picrylhydrazyl radical assay) methods have been applied to evaluate the TAC of taro extracts or taro-food derivatives. Flavonoids, tannins, saponins, alkaloids, carotenoids, phenols, vitamins, and fatty acids seem to contribute to overall taro antioxidant capacity [fig_ref] Table 3: Antioxidants in Colocasia esculenta from different geographic regions [/fig_ref] [bib_ref] Effect of cocoyam (Colocasia esculenta), unripe plantain (Musa paradisiaca) or their combination..., Eleazu [/bib_ref] [bib_ref] Ameliorative potentials of cocoyam (Colocasia esculenta L.) and unripe plantain (Musa paradisiaca..., Eleazu [/bib_ref] [bib_ref] Identification of phenolic compounds in edible aroids, Agbor-Egbe [/bib_ref] [bib_ref] Antioxidant activity and profiles of common fruits in Singapore, Isabelle [/bib_ref] [bib_ref] Impact of Taro Corms on Functional Low Fat Ice Cream Properties, El-Dardiry [/bib_ref] [bib_ref] Effect of steaming on the functional compounds and antioxidant activity of Fijian..., Kim [/bib_ref] [bib_ref] Colocasia based cropping systems affects the antioxidant properties and productivity of colocasia, Tuti [/bib_ref] [bib_ref] Cytotoxicity and antimicrobial activity of Colocasia esculenta, Chakraborty [/bib_ref] [bib_ref] In vitro micropropagation and antioxidant assay in Colocasia esculenta, Akshatha [/bib_ref] [bib_ref] Process optimization for extraction of bioactive compounds from taro (Colocasia esculenta), using..., Kumar [/bib_ref] [bib_ref] Hydrophilic antioxidant capacities of vegetables and fruits commonly consumed in Japan and..., Takebayashi [/bib_ref] [bib_ref] Antimicrobial, antioxidant, anticancer property and chemical composition of different parts (corm, stem..., Lee [/bib_ref] [bib_ref] Bioactive constituents and antioxidant activities of raw and processed cocoyam (Colocasia esculenta), Awa [/bib_ref] [bib_ref] Characterization of the natural products in cocoyam (Colocasia esculenta) using GC-MS, Eleazu [/bib_ref] [bib_ref] Dizon, I. The nutritional value and phytochemical components of taro [Colocasia esculenta..., Alcantara [/bib_ref] [bib_ref] Screening of South African food plants for antioxidant activity, Lindsey [/bib_ref] [bib_ref] Determination of Antioxidant Activity of Ethanol Extract of Gölevez, Akyuz [/bib_ref]. Several studies point out distinct antioxidants in variable amounts in taro and taro derivatives and these differences might be attributed to multiple factors, such as (i) genetic background between variants of the same species; (ii) the applied farming cultivation system; (iii) post-harvest processing, including the handling, cutting, peeling, drying, cooking, and storage stages; and (iv) non-standard extraction of bioactivities [bib_ref] Influence of taro (Colocasia esculenta L. Shott) growth conditions on the phenolic..., Goncalves [/bib_ref] [bib_ref] Further knowledge on the phenolic profile of Colocasia esculenta (L.) Shott, Ferreres [/bib_ref] [bib_ref] Effects of thermal treatment on the phenolic content and antioxidant activity of..., Chipurura [/bib_ref] [bib_ref] Effect of different blanching times on antioxidant properties in selected cruciferous vegetables, Amin [/bib_ref] [bib_ref] Influence of extraction solvents on antioxidant activity and the content of bioactive..., Bae [/bib_ref].
Few studies go beyond TAC determination and phytochemical group characterization to identify specific compounds within the aforementioned class, and at what concentrations, in taro [bib_ref] Identification of phenolic compounds in edible aroids, Agbor-Egbe [/bib_ref] [bib_ref] Antioxidant activity and profiles of common fruits in Singapore, Isabelle [/bib_ref] [bib_ref] Cytotoxicity and antimicrobial activity of Colocasia esculenta, Chakraborty [/bib_ref] [bib_ref] Process optimization for extraction of bioactive compounds from taro (Colocasia esculenta), using..., Kumar [/bib_ref] [bib_ref] Antimicrobial, antioxidant, anticancer property and chemical composition of different parts (corm, stem..., Lee [/bib_ref] [bib_ref] Characterization of the natural products in cocoyam (Colocasia esculenta) using GC-MS, Eleazu [/bib_ref]. The taro matrix presents a complex set of antioxidants and, although they have been identified and measured, it is hard to forecast which role a single compound will play on the human body. Taro antioxidants may function in combination to promote synergistic or antagonistic effects [bib_ref] Antioxidant intervention as a route to cancer prevention, Collins [/bib_ref].
Chlorogenic acid, catechin, epicatechin, epigallocatechin, pro-anthocyanidins, and gallic acid have been identified in taro by thin-layer chromatography (TLC) and highperformance liquid chromatography (HPLC) [bib_ref] Identification of phenolic compounds in edible aroids, Agbor-Egbe [/bib_ref]. Polyphenols represented by 1-O-feruloyl-D-glucoside; 3, 5-DiCQ acid; vitexin; isovitexin; cyanidin-3-glucoside; luteolin-7-O-rutinoside; vicenin-2; caffeic acid; cyanidin-3-rhamnoside; chlorogenic acid; quercetin and hyperoside were identified by LC-MS (liquid chromatography-mass spectrometry) [bib_ref] Process optimization for extraction of bioactive compounds from taro (Colocasia esculenta), using..., Kumar [/bib_ref].
Polyphenols and other antioxidants present in taro, besides acting as common freeradical scavengers, other molecular and enzymatic mechanisms triggered by polyphenols, are considered complementary for health promotion. Few mechanisms of action delineate how polyphenols participate in inflammatory cascades by increasing pro-inflammatory cytokines release. Subsequently, inflammation can, for example, activate the NF-κB (nuclear factor kappa B) transcription factor and stimulate the production of TNF-α, two critical factors that, when upregulated, may participate in cancer development. Thus, polyphenols could aid in controlling cancer progression due to their anti-inflammatory effects, including the activation of antioxidant enzymes, inhibition of pro-oxidant enzymes, and prevention of free radical attacks [bib_ref] Molecular mechanism underlying anti-inflammatory and anti-allergic activities of phytochemicals: An update, Bellik [/bib_ref] [fig_ref] Figure 2: Therapeutic potential of taro phytochemicals [/fig_ref] Several studies point out distinct antioxidants in variable amounts in taro and taro derivatives and these differences might be attributed to multiple factors, such as (i) genetic background between variants of the same species; (ii) the applied farming cultivation system; (iii) post-harvest processing, including the handling, cutting, peeling, drying, cooking, and storage stages; and (iv) non-standard extraction of bioactivities [bib_ref] Influence of taro (Colocasia esculenta L. Shott) growth conditions on the phenolic..., Goncalves [/bib_ref] [bib_ref] Further knowledge on the phenolic profile of Colocasia esculenta (L.) Shott, Ferreres [/bib_ref] [bib_ref] Effects of thermal treatment on the phenolic content and antioxidant activity of..., Chipurura [/bib_ref] [bib_ref] Effect of different blanching times on antioxidant properties in selected cruciferous vegetables, Amin [/bib_ref] [bib_ref] Influence of extraction solvents on antioxidant activity and the content of bioactive..., Bae [/bib_ref].
Few studies go beyond TAC determination and phytochemical group characterization to identify specific compounds within the aforementioned class, and at what concentrations, in taro [bib_ref] Identification of phenolic compounds in edible aroids, Agbor-Egbe [/bib_ref] [bib_ref] Antioxidant activity and profiles of common fruits in Singapore, Isabelle [/bib_ref] [bib_ref] Cytotoxicity and antimicrobial activity of Colocasia esculenta, Chakraborty [/bib_ref] [bib_ref] Process optimization for extraction of bioactive compounds from taro (Colocasia esculenta), using..., Kumar [/bib_ref] [bib_ref] Antimicrobial, antioxidant, anticancer property and chemical composition of different parts (corm, stem..., Lee [/bib_ref] [bib_ref] Characterization of the natural products in cocoyam (Colocasia esculenta) using GC-MS, Eleazu [/bib_ref]. The taro matrix presents a complex set of antioxidants and, although they have been identified and measured, it is hard to forecast which role a single compound will play on the human body. Taro antioxidants may function in combination to promote synergistic or antagonistic effects [bib_ref] Antioxidant intervention as a route to cancer prevention, Collins [/bib_ref].
Chlorogenic acid, catechin, epicatechin, epigallocatechin, pro-anthocyanidins, and gallic acid have been identified in taro by thin-layer chromatography (TLC) and highperformance liquid chromatography (HPLC) [bib_ref] Identification of phenolic compounds in edible aroids, Agbor-Egbe [/bib_ref]. Polyphenols represented by 1-O-feruloyl-D-glucoside; 3, 5-DiCQ acid; vitexin; isovitexin; cyanidin-3-glucoside; luteolin-7-Oruti-noside; vicenin-2; caffeic acid; cyanidin-3-rhamnoside; chlorogenic acid; quercetin and hyperoside were identified by LC-MS (liquid chromatography-mass spectrometry) [bib_ref] Process optimization for extraction of bioactive compounds from taro (Colocasia esculenta), using..., Kumar [/bib_ref].
Polyphenols and other antioxidants present in taro, besides acting as common freeradical scavengers, other molecular and enzymatic mechanisms triggered by polyphenols, are considered complementary for health promotion. Few mechanisms of action delineate how polyphenols participate in inflammatory cascades by increasing pro-inflammatory cytokines release. Subsequently, inflammation can, for example, activate the NF-κB (nuclear factor kappa B) transcription factor and stimulate the production of TNF-α, two critical factors that, when upregulated, may participate in cancer development. Thus, polyphenols could aid in controlling cancer progression due to their anti-inflammatory effects, including the activation of antioxidant enzymes, inhibition of pro-oxidant enzymes, and prevention of free radical attacks [bib_ref] Molecular mechanism underlying anti-inflammatory and anti-allergic activities of phytochemicals: An update, Bellik [/bib_ref] [fig_ref] Figure 2: Therapeutic potential of taro phytochemicals [/fig_ref]. Under normal conditions, NF-κB is held inactive in the cytosol by the inhibitor of NF-κB proteins, inhibitor kappa B (IκB) (α, β, ε). Upon binding to different immune receptors, the IκB kinase complexes phosphorylate IκB and thereby induces its proteasomal degradation and the release of NF-κB. The released transcription factor NF-κB can translocate to the nucleus and mediates the expression of target genes. Polyphenols may act either by suppressing kinases complex phosphorylation, which avoids NF-κB translocation to the nucleus or inhibits the interaction of this transcription factor with targeted DNA genes. Both mechanisms can down-regulate the inflammatory cascade, inducing apoptosis, controlling both cell proliferation and metastasis [bib_ref] Targeting NF-κB signaling pathway in cancer by dietary polyphenols, Khan [/bib_ref].
Food processing, such as cooking, frying, steaming, fermentation, and other conservation methods, including freezing, pasteurization, and drying, interfere in antioxidant concentrations, releasing more antioxidants, as observed with phenolic acids bound to complex structures such as lignin and cellulose [bib_ref] Processed sweet corn has higher antioxidant activity, Dewanto [/bib_ref] [bib_ref] Solid-state fermentation of apple pomace using Phanerocheate chrysosporium-Liberation and extraction of phenolic..., Ajila [/bib_ref]. Freeze-dried foods can preserve the antioxidant compounds [bib_ref] A review on the effect of drying on antioxidant potential of fruits..., Kamiloglu [/bib_ref].
Lipid-soluble antioxidants found in taro have been identified by HPLC. Vitamins and carotenoids, among them ascorbic acid, violaxanthin, lutein, β-carotene, δ-γ-α-tocopherol, and δ-γtocotrienol are responsible for protecting cellular structures against lipid peroxidation caused by free radicals [fig_ref] Table 3: Antioxidants in Colocasia esculenta from different geographic regions [/fig_ref] [bib_ref] The anti-cancer effects of poi (Colocasia esculenta) on colonic adenocarcinoma cells In..., Brown [/bib_ref] [bib_ref] Crude extract from taro (Colocasia esculenta) as a natural source of bioactive..., Pereira [/bib_ref] [bib_ref] Comparative bio-antimutagenicity of common vegetables and traditional vegetables in Kyoto, Nakamura [/bib_ref].
Several fatty acids present in taro are suggested as potential antioxidants, including 9,12,15-octadecatrienoic acid, 8,11-octadecadienoic acid, methyl ester; hexadecanoic acid, methyl ester; 9-octadecenoic acid, methyl ester (E); 3,5-Di-tert-butyl-4trimethylsiloxytoluene; cyclohexanol, 2-nethyl-5-(1-methylethenyl)-(1. alpha., 2. beta., 5. alpha.) [bib_ref] Cytotoxicity and antimicrobial activity of Colocasia esculenta, Chakraborty [/bib_ref] [bib_ref] Antimicrobial, antioxidant, anticancer property and chemical composition of different parts (corm, stem..., Lee [/bib_ref].
Despite positive results, recent studies have indicated that reactive oxygen species (ROS) scavenging in cancer cells, particularly during chemotherapy, interferes with the primary mechanism, which triggers apoptosis in these cells. In this way, antioxidant supplementation is a double-edged sword. To be used as a supportive therapy during cancer treatments, it might take into account the type of cancer, general patient status, and antioxidant dosage and type to reach the best performance. Another suggestion is to administer a single antioxidant prescribed by a unique kind of cancer, since antioxidants, as mentioned previously, can play synergic or antagonistic effects and even lead to uncontrollable effects in the human body when administrated as a non-controlled combination.
The data in [fig_ref] Table 3: Antioxidants in Colocasia esculenta from different geographic regions [/fig_ref] summarize the available studies concerning the antioxidant capacity of taro per geographic region, as well as the compounds supposed to be involved in the activity. Most investigations have been carried out in Asia, Oceania, and Africa, where taro plays a vital role as a staple food. However, the restricted distribution of taro-related studies reinforces the claims that taro is still a neglected crop worldwide, even though it displays a rich composition concerning health-influencing compounds [fig_ref] Figure 1: Global distribution of taro production reproduced from FAOSTAT [/fig_ref] [bib_ref] National Research Council. Underexploited Tropical Plants with Promising Economic Value: Report of..., Wang [/bib_ref]. In the beginning of the 1980s, complaints about the lack of research efforts focused on taro were noted. At that time, taro was considered the most underrated root crop in a number of reports and studies [bib_ref] Tropical root crops and their potential for food in the less developed..., Chandra [/bib_ref]. Nowadays, this number has increased, and more science institutes are carrying research on taro and its potential benefits for human health, although it is still concentrated in under-developed areas and not worldwide, as it should be. In the next section, each traditional claimed health benefit was harbored in the experimental evidence observed in cell cultures or animal models.
## Taro protection against mutagenic and carcinogenic agents
Antimutagens or anticarcinogens can decrease or even prevent the deleterious effects of physical, chemical, and biological agents that cause genome modifications or mutations, triggering the carcinogenesis process when a mutation occurs in somatic cells [fig_ref] Figure 3: Putative targets of taro-derived components in metabolic pathways [/fig_ref]. A variety of plant compounds exhibit this ability, including those found in taro such as antioxidant molecules belonging to polyphenol classes or not, dietary fiber, luteoline-derivatives, gallic acid, vitamins (ascorbic acid, b-carotene, and a-tocopherol), anthocyanins, catechins, flavonoids, and others [bib_ref] Antimutagenic compounds and their possible mechanisms of action, Słoczyńska [/bib_ref] [bib_ref] The correlation between antimutagenic activity and total phenolic content of extracts of..., Makhafola [/bib_ref]. The antimutagenic mechanism is diverse and can occur in the extra or intracellular compartments. Extracellular mechanisms include the inhibition of mutagen uptake, complexation, and/or deactivation, inhibition of endogenous mutagen formation, and favoring the absorption of protective agents. Intracellular mechanisms involve the blocking of reactive species, transmembrane transport modification, metabolism modulation, DNA metabolism and repair modulation, signaling pathway regulation, apoptosis enhancement, genomic stability maintenance, and trapping and detoxification stimulation in non-target cells. The National Cancer Institute (https://www.cancer.gov), the American Institute for Cancer Research (https://www.aicr.org/cancer-prevention/healthy-eating/), and the American Cancer Society (https://www.cancer.org) recommend the intake of dietary fibers, which are found in vegetables and fruits, since their protective effects against gastric and colorectal cancer risks have been demonstrated by several reports, including meta-analysis of case-control and cohort studies [bib_ref] Dietary fiber intake and risks of proximal and distal colon cancers: A..., Ma [/bib_ref] [bib_ref] The benefits of dietary fiber intake on reducing the risk of cancer:..., Mcrae [/bib_ref] [bib_ref] Dietary fiber intake reduces risk for gastric cancer: A meta-analysis, Zhang [/bib_ref] [bib_ref] Dietary fiber intake reduces risk for colorectal adenoma: A meta-analysis, Ben [/bib_ref]. The protective mechanism of dietary fibers depends on their composition, and it can be attributed to the reduced time of intestinal tissue exposure to carcinogens and short-chain fatty acids generated from fermentation by gut microbiota [bib_ref] Mechanisms linking dietary fiber, gut microbiota and colon cancer prevention, Zeng [/bib_ref] [bib_ref] From Dietary Fiber to Host Physiology: Short-Chain Fatty Acids as Key Bacterial..., Koh [/bib_ref] [bib_ref] The behavior of dietary fiber in the gastrointestinal tract determines its physiological..., Capuano [/bib_ref].
Dietary fibers found in crude taro are mainly composed of neutral monosaccharides, uronic acid, galacturonic acid, glucuronic acid, and neglectable concentrations of lignin and no starch, and they seem to be involved in protection against chemical and physical mutagenic agents [fig_ref] Figure 3: Putative targets of taro-derived components in metabolic pathways [/fig_ref]. Crude taro extract, rich in dietary fibers, can adsorb the hydrophobic compound 1,8 dinitropyrene (DNP), which is a mutagenic and carcinogenic pollutant found in the environment, making it ineffective in intestinal cells [fig_ref] Table 2: [/fig_ref] [bib_ref] Adsorption of a hydrophobic mutagen to dietary fiber from taro (Colocasia esculenta),..., Ferguson [/bib_ref] [bib_ref] Photochemical Relaxation Pathways in Dinitropyrene Isomer Pollutants, Brister [/bib_ref]. Certainly, absorbing mutagenic compounds can reduce the risk of gastrointestinal tract cancers. Nevertheless, cancer risk reduction by dietary fiber intake can be extended to several organs, i.e., breast, pancreas, and prostate [bib_ref] Tzoulaki, I. Dietary fiber and health outcomes: An umbrella review of systematic..., Veronese [/bib_ref].
Other evidence do not ascribe antimutagenic activity to taro dietary fibers, attributing it to different compounds in taro. A heptane extract from cooked taro prevented the deleterious mutations caused by the heterocyclic amine 2-amino-3-methylimidazo ]quinoline (IQ), which is a potent mutagenic and carcinogenic agent formed during meat and fish cooking [bib_ref] Antimutagens in food plants eaten by Polynesians: Micronutrients, phytochemicals and protection against..., Botting [/bib_ref]. Similarly, an aqueous extract from crude taro obtained from two different cultivars, a traditional one and the Ebi-taro from Kyoto (JPN), displayed antimutagenic effects against physical agents when assayed in E. coli cells exposed to UV radiation [bib_ref] Comparative bio-antimutagenicity of common vegetables and traditional vegetables in Kyoto, Nakamura [/bib_ref].
## Anticancer, anti-inflammatory, and immunomodulatory effectiveness of taro extracts or their components
Antitumoral and antimetastatic bioactivities associated with immunomodulatory effectiveness can be provided by the set of molecules found in taro, as previously demonstrated by in vivo and in vitro preclinical trials. A soluble extract from poi (cooked taro), a popular food traditionally consumed by Hawaiians, inhibits the proliferation of rat YYT colon cancer cells in a dose-dependent manner, reaching its maximum effect at 25 mg poi/mL. Rat YYT cells displayed morphological apoptosis characteristics, confirmed by TUNEL (TdT-Mediated dUTP Nick End Labeling) nucleus staining, which is indicative of DNA damage. In contrast, no toxicity to healthy mice spleen cells was observed, reinforcing suitability for antitumorigenic use. Moreover, splenocytes were stimulated to proliferate, displaying a TCD4 phenotype, followed by TCD8, B, and natural killer (NK) cells [bib_ref] The anti-cancer effects of poi (Colocasia esculenta) on colonic adenocarcinoma cells In..., Brown [/bib_ref]. The presence of these cells at the tumor microenvironment is a good prognosis, especially during the early stages of carcinogenesis. CD8+ T lymphocytes differentiate into cytotoxic cells by antigen-presenting cell (APC) activation, in order to promote the direct destruction of cancer cells. CD4+ T lymphocytes contribute to the fight against cancer by the release of pro-inflammatory cytokines IL-2, TNF-α, and INF-γ, which, in turn, activate T lymphocytes, NK, and macrophage cells, while enhancing antigen presentation [bib_ref] Roles of the immune system in cancer: From tumor initiation to metastatic..., Gonzalez [/bib_ref]. Poi antitumoral activity may be due to both a direct effect of its bioactive compounds on cancer cells associated with an indirect effect on immune system activation.
Crude taro extract inhibits the in vitro proliferation of several human breast cancer lines, i.e., MCF-7, MDA-MB-231, and MCF10A, as well as murine breast cancer lines 66.1, 410.4, and EpH4. The antiproliferative effects on murine cells were accompanied by morphological changes such as cell rounding and a reduction of foot projections. Crude taro extract displayed antimetastatic effects following intraperitoneal administration, before and after the establishment of cancer, exerting therapeutic and protective effects against heart and lung colonization by breast cancer lineages.
The anticancer effects promoted by taro extract have been attributed to a lectin named tarin (PDB id. 5T1X and 5T20). This protein is able to reproduce the anticancer and antimetastatic effects of the crude extract when assayed in vitro and in vivo by inhibiting human hepatoma cells (HepG2 lineage) proliferation as well as lung and heart colonization [bib_ref] Liposomal Taro Lectin Nanocapsules Control Human Glioblastoma and Mammary Adenocarcinoma Cell Proliferation, Corrêa [/bib_ref].
Tarin nano-encapsulated in liposomes improved anticancer activity compared to the free protein against human breast cancer (MDA-MB-231) and glioblastoma (U87 MG) lineages, resulting in 41 and 65% inhibition, respectively. Nano-encapsulated tarin was as effective as cisplatin and temozolomide in controlling glioblastoma cell proliferation. However, the advantage relies on the non-cytotoxicity of both forms of tarin, free or nano-encapsulated, in effective concentrations when added to healthy cells [bib_ref] Liposomal Taro Lectin Nanocapsules Control Human Glioblastoma and Mammary Adenocarcinoma Cell Proliferation, Corrêa [/bib_ref].
These findings indicate that tarin exhibits high potential as a supportive anticancer therapy [bib_ref] A Cytokine-Inducing Hemagglutinin from Small Taros, Yau Sang [/bib_ref] [bib_ref] Liposomal Taro Lectin Nanocapsules Control Human Glioblastoma and Mammary Adenocarcinoma Cell Proliferation, Corrêa [/bib_ref]. Moreover, tarin is considered a relatively stable protein, maintaining activity under a wide range of pH and temperatures, which are mandatory physicochemical features for candidate molecules for pharmaceutical purposes [bib_ref] Structural analysis and binding properties of isoforms of tarin, the GNA-related lectin..., Pereira [/bib_ref].
This lectin has become a notorious molecule after being extensively studied and fully characterized as an anti-tumoral and immunomodulatory effector [bib_ref] Potential Immunomodulator and COX-Inhibitor Lectin Found in Taro (Colocasia esculenta), Pereira [/bib_ref] [bib_ref] Purification and characterization of the lectin from taro (Colocasia esculenta) and its..., Pereira [/bib_ref] [bib_ref] Structural analysis and binding properties of isoforms of tarin, the GNA-related lectin..., Pereira [/bib_ref] [bib_ref] High-resolution crystal structures of Colocasia esculenta tarin lectin, Pereira [/bib_ref]. Not surprisingly, glycan microarrays indicate that this protein binds to specific glycan chains that are part of antigens found in many types of cancer cells, such as CA-125 in ovarian cancer, paucimannose in human cancerous cells, Lewis-y epitope, a typical antigen found in the colon, stomach, ovary, breast, pancreas, prostate and lung cancer, hematopoietic progenitor cells, peripheral blood granulocytes, and to the antigen H2, which is found in leukemic cells and hematopoietic progenitors [bib_ref] Structural analysis and binding properties of isoforms of tarin, the GNA-related lectin..., Pereira [/bib_ref]. Although binding recognition has not yet been tested in culture cells, these data profoundly reinforce that tarin may contribute to the antiproliferative, antimetastatic, and immunomodulatory responses observed upon taro extract intake as aforementioned and further discussed.
Taro extracts have also demonstrated anti-inflammatory activity against human and murine breast cancer cells by modulating the immune response through the reduction of prostaglandin E2 (PGE2) release accompanied by the abolishment of mRNA synthesis of cyclooxygenase-2 (COX-2), and decrease of COX-1. The tarin down-regulation of COX-2 gene depressed E-series prostaglandins by almost 70%, especially PGE2, which is an inflammatory mediator known to exert pro-tumorigenic effects, contributing to (i) the generation of additional mutations in cancer cells by stimulating the release of ROS; (ii) upregulation of a series of critical molecules such as Bcl-2, conferring resistance to apoptosis, vascular endothelial growth factor (VEGF) involved in angiogenesis, type 2 and 9 matrix metalloproteinases (MMPs), which confer invasive ability, epidermal growth factor receptor (EGFR), facilitating cell proliferation, extracellular signal-regulated kinase (ERK) and membrane proteases also involved in invasion; (iii) suppression of antitumoral response by downregulating the immune system; (iv) activation of wingless-related integration (WNT)/β-catenin pathway that favor metastasis and maintain stem cell skills;
(v) activation of phosphatidylinositol 3-kinase/protein kinase B (PI3K/AKT) pathway that promotes cell proliferation and survival. As a result, the angiogenesis process, cancer cell proliferation, differentiation, and migration would be prevented [bib_ref] Cyclooxygenase-2 promotes tumor growth and suppresses tumor immunity, Liu [/bib_ref] [bib_ref] Cyclooxygenase-1 (COX-1) and COX-1 inhibitors in cancer: A review of oncology and..., Pannunzio [/bib_ref] [bib_ref] Cyclooxygenase-2 in cancer: A review, Hashemi Goradel [/bib_ref].
Previous studies indicate that the inhibition of COX-2 resulted in an overall 70% reduction in cancer risk for breast, lung, prostate and, colon cancers, and also in cancers whose etiology is associated with several mutagenic conditions such as tobacco, alcohol, UV light, oxidative stress, and infections by viruses or bacteria [bib_ref] Cyclooxygenase-2 in cancer: A review, Hashemi Goradel [/bib_ref] [bib_ref] Reduction in cancer risk by selective and nonselective cyclooxygenase-2 (COX-2) inhibitors, Harris [/bib_ref].
The selective downregulation of COX-2 by taro lectin seems to be very promising in cancer treatment, since the reduction in PGE2 synthesis attenuates pro-tumorigenesis. Further studies are required in order to understand the molecular mechanisms behind COX-2 regulation by tarin, which can occur in different ways, considering that the COX-2 promotor region harbors multiple binding sites for enhancers, including NF-κB, β-catenin, interleukins, and Hu antigen-R, among others [bib_ref] Cyclooxygenase-2 in cancer: A review, Hashemi Goradel [/bib_ref].
Since tarin belongs to the Galanthus nivalis agglutinin (GNA)-related lectins, sharing specificity and structural characteristics with them, its mechanism of action could be extrapolated . Different mechanisms of action were described for these lectin members and attributed to their binding to mannose-containing antigens. These mechanisms can be linked to PGE2 overexpression, which is a typical condition in cancer. GNA-related lectins exert their cytotoxic activities on cancer cells mainly by deactivating rat sarcoma-rapidly accelerated fibrosarcoma (Ras-Raf) and PI3K-Akt pathways previously activated by PGE2 autocrine action through the prostaglandin E receptor 1-4 (EP1-4) receptor, inhibiting the proliferation, migration/invasion, and survival of cancer cells. Once the anti-apoptotic pathways are inhibited, cell death can be induced through apoptosis or autophagy triggered by mitochondria injury following the accumulation of ROS and cytochrome c release, culminating in the activation of ROS-p38-p53 and caspase-dependent pathways [bib_ref] Anti-tumor and anti-viral activities of Galanthus nivalis agglutinin (GNA)-related lectins, Wu [/bib_ref] [bib_ref] Plant-Derived Lectins as Potential Cancer Therapeutics and Diagnostic Tools, Mazalovska [/bib_ref] [bib_ref] Eicosanoids and cancer, Wang [/bib_ref]. Taro components are traditionally known to boost the immunological response, which could be an indirect way to contribute to reducing cancer risk or controlling tumorigenesis . Indeed, in mice, tarin proved to stimulate the in vitro and in vivo proliferation of bone marrow (BM) and spleen cells while protecting BM progenitor cells from death. B lymphocytes and granulocytic cells were identified among the proliferating cells when taro extract was intraperitoneally administered (1 mg/animal). Moreover, tarin-sensitized splenocytes were stimulated to release IL-2, IL-1b, TNF-α, and INF-γ, which are essential cytokines involved in the anticancer response, as mentioned previously [bib_ref] Effects of thermal treatment on the phenolic content and antioxidant activity of..., Chipurura [/bib_ref] [bib_ref] National Research Council. Underexploited Tropical Plants with Promising Economic Value: Report of..., Wang [/bib_ref].
Considering that tarin is found in the edible part of C. esculenta, the regular intake of this tubercle could be a useful dietary intervention to boost the immune system in healthy individuals or to accelerate the recovery from leukopenia in immunosuppressed patients, including those under chemotherapy. Tarin immunomodulatory potential has been evidenced by the administration of the purified protein (200 µg/animal) to cyclophosphamideimmunosuppressed mice, where BM cells were stimulated to proliferate and differentiate into granulocytic cell lineage, promoting a faster recovery from leukopenia. Tarin also protects the BM erythroid progenitors from the cyclophosphamide cytotoxicity [bib_ref] Tarin stimulates granulocyte growth in bone marrow cell cultures and minimizes immunosuppression..., Merida [/bib_ref].
In addition to tarin, taro contains a mixture of other anti-neoplastic molecules, such as polyphenols and polysaccharides, as mentioned previously [bib_ref] Secondary metabolite diversity in taro, Colocasia esculenta (L.) Schott, corms, Muñoz-Cuervo [/bib_ref] [bib_ref] Isolation and Characterisation of Water Soluble Polysaccharide from Colocasia esculenta Tubers, Pawar [/bib_ref]. An ethanolic extract from freeze-dried taro exhibited an antiproliferative effect against adult T-cells leukemia lineages and, in some of them, effectiveness was superior to the apoptotic effect of genistein, an isoflavone found in soy, which is a popular grain recognized to trigger apoptosis in estrogen-dependent cancers [bib_ref] Inhibition of proliferation by agricultural plant extracts in seven human adult T-cell..., Kai [/bib_ref] [bib_ref] Genistein induced apoptotic cell death in adult T-cell leukemia cells through estrogen..., Yamasaki [/bib_ref]. A water-soluble 200 kDa polysaccharide (taro-4-I) composed of neutral sugar and uronic acid isolated from taro crude extract was able to inhibit mice lung colonization by B16BL6 melanoma cells [bib_ref] Anti-metastatic effect of polysaccharide isolated from Colocasia esculenta is exerted through immunostimulation, Park [/bib_ref].
Similar to the tarin modulatory effect, polysaccharides extracted from taro also enhanced the immune response. Taro-4-I, as well as TPS-1 and TPS-2, was able to activate the complement component 3 (C3 protein) through classical and alternative pathways, which is a prerequisite to trigger cellular lysis. In addition, macrophages and NK cells were activated, while macrophages were stimulated to release IL-12, IL-6, and TNF-a cytokines and nitric oxide (NO) [bib_ref] Anti-metastatic effect of polysaccharide isolated from Colocasia esculenta is exerted through immunostimulation, Park [/bib_ref] [bib_ref] Structure characterization of two novel polysaccharides from Colocasia esculenta (taro) and a..., Li [/bib_ref].
## Taro compounds effect on type ii diabetes and obesity
Obesity or overweight is the primary risk factor for developing type II diabetes. Diabetes, obesity, hyperglycemia, and hypercholesterolemia, occurring together or independently, have been directly associated with cancers in the kidney, bladder, thyroid, ovary, breast, endometrium, stomach, liver, pancreas, colon, and rectum, as well as leukemia. The adoption of a healthier lifestyle is the recommended strategy to prevent overweight or obesity and avoid cancers related to these comorbidities [bib_ref] De la Vieja, A. From obesity to diabetes and cancer: Epidemiological links..., García-Jiménez [/bib_ref] [bib_ref] Hypercholesterolemia Increases Colorectal Cancer Incidence by Reducing Production of NKT and γδ..., Tie [/bib_ref]. Taro can be a powerful ally considering its medium glycemic index, since 33% of total taro starch is composed of SDS and RS after cooking, and high dietary fiber content, which are two essential properties to manage these metabolic dysfunctions [fig_ref] Table 1: Nutritional composition of taro analyzed raw, cooked and baked [/fig_ref]. Moreover, other taro bioactive molecules can synergistically reinforce the metabolic effects of taro, as demonstrated in preclinical trials [fig_ref] Figure 3: Putative targets of taro-derived components in metabolic pathways [/fig_ref] and [fig_ref] Table 2: [/fig_ref].
Taro flour offered to streptozotocin (STZ)-induced hyperglycemic rats restored glycemia after four weeks of intake. Proteinuria and glucosuria, kidney function, relative kidney weight, hepatic function, glycated hemoglobin, and body overweight, which are associated with type II diabetes, were also attenuated following taro consumption. Plasmatic levels of total cholesterol, VLDL-and LDL-cholesterols, triacylglycerol, serum pancreatic lipase, atherogenic, and coronary risk were all reverted, and the HDL-cholesterol levels were enhanced [bib_ref] Effect of cocoyam (Colocasia esculenta), unripe plantain (Musa paradisiaca) or their combination..., Eleazu [/bib_ref] [bib_ref] Ameliorative potentials of cocoyam (Colocasia esculenta L.) and unripe plantain (Musa paradisiaca..., Eleazu [/bib_ref] [bib_ref] Biochemical basis of the use of cocoyam (Colocassia esculenta L.) in the..., Eleazu [/bib_ref]. Similar effects were obtained by the administration of an ethanolic extract from crude taro flour, which decreased glucose tolerance and glycemia in a murine model [bib_ref] Antihyperglycemic activity of methanolic extracts of corms of Colocasia esculenta var esculenta, Islam [/bib_ref]. The anti-hyperglycemic and anti-hyperlipidemic effects observed herein point out that taro may have the potential to manage diabetes and obesity. These effects are mainly attributed to flavonoids, alkaloids, saponins, steroids, and tannins, but the participation of minerals (Mg, Ca, K, P, Fe, and Zn) and crude dietetic fiber have not been discarded [bib_ref] Ameliorative potentials of cocoyam (Colocasia esculenta L.) and unripe plantain (Musa paradisiaca..., Eleazu [/bib_ref] [bib_ref] Antihyperglycemic activity of methanolic extracts of corms of Colocasia esculenta var esculenta, Islam [/bib_ref].
Taro mucilage, mainly composed of neutral sugar, absorbs mutagenic/carcinogenic agents and also takes part in anti-hyperglycemic and anti-hyperlipidemic effects. A mucilage-rich extract from crude taro flour inhibits the starch-hydrolytic enzymes αglucosidase and α-amylase, and pancreatic lipase. [bib_ref] A comparative study on the physicochemical, anti-oxidative, anti-hyperglycemic and anti-lipidemic properties of..., Chukwuma [/bib_ref] [bib_ref] Non-covalent dietary fiber-Polyphenol interactions and their influence on polyphenol bioaccessibility, Jakobek [/bib_ref]. Arabinogalactan extracted from taro flour mucilage and incorporated into hypercaloric rat diets decreased lipid levels in serum and tissues, and it reduced hepatocyte synthesis/secretion of apoBcontaining lipoproteins, mainly VLDL [bib_ref] Hypolipidaemic effect of chemically different mucilages in rats: A comparative study, Boban [/bib_ref]. The viscous mucilage fibers are known to reduce the bolus motility through the gastrointestinal tract and, as a consequence, the increasing digestion and absorption of macronutrients, especially lipids and carbohydrates, promoting satiety [bib_ref] Dietary fiber role in type 2 diabetes prevention, Ismaiel [/bib_ref]. Glucose entrapment by plant mucilage is a widely described phenomenon [bib_ref] Hypoglycemic and Hypolipidemic Activities of Aloe vera Leaf Mucilage in Alloxan-Induced Diabetic..., Sefi [/bib_ref] [bib_ref] Hibiscus esculentus seed and mucilage beneficial effects in reducing complications of diabetes..., Hajian [/bib_ref].
The polysaccharide matrix can retain and deliver many bioactive compounds, such as phenolic compounds and peptides, which are both detected in mucilage-rich extracts from crude taro flour [bib_ref] A comparative study on the physicochemical, anti-oxidative, anti-hyperglycemic and anti-lipidemic properties of..., Chukwuma [/bib_ref] [bib_ref] Non-covalent dietary fiber-Polyphenol interactions and their influence on polyphenol bioaccessibility, Jakobek [/bib_ref]. The importance of phenolic compounds in the management of type 2 diabetes has also been widely reported [bib_ref] An Overview of Plant Phenolic Compounds and Their Importance in Human Nutrition..., Lin [/bib_ref] [bib_ref] Mechanisms of Action of Phenolic Phytochemicals in Diabetes Management, Hoda [/bib_ref] [bib_ref] Polyphenols and their effects on diabetes management: A review, Aryaeian [/bib_ref]. Flavonoids, phenolic acids, and tannins are generally associated with the reduction of starch digestion due to the inhibition of α-glucosidase and α-amylase activities, as mentioned previously. In the end, a reduction in glucose release and absorption can occur, controlling post-meal glycemic levels.
Lipid digestion may also be regulated by phenolic compounds through the inhibition of pancreatic lipase, followed by excretion in feces, consequently reducing body mass. Both effects have been demonstrated in clinical trials, being considered adequate for diabetes and obesity control, since they share similar mechanisms of action with many drugs currently in use [bib_ref] Cyclooxygenase-1 (COX-1) and COX-1 inhibitors in cancer: A review of oncology and..., Pannunzio [/bib_ref] [bib_ref] Anti-tumor and anti-viral activities of Galanthus nivalis agglutinin (GNA)-related lectins, Wu [/bib_ref] [bib_ref] Plant-Derived Lectins as Potential Cancer Therapeutics and Diagnostic Tools, Mazalovska [/bib_ref].
Additionally, proteins A-1 and B-2, with molecular masses of 17 and 19 kDa, respectively, obtained from defatted flour of crude taro, inhibit α-amylase from human saliva and porcine pancreas in vitro, thus being potentially able to control diabetes and obesity [bib_ref] Alpha-amylase inhibitor of amadumbe (Colocasia esculenta): Isolation, purification and selectivity toward-amylases from..., Mcewan [/bib_ref]. The inhibition of starch-hydrolytic enzymes by taro compounds is especially important to prevent or minimize the digestion of the rapidly-digestible starch (RDS) portion that accounts for 67% of the total starch and contributes to the sudden increase of glycemia.
The anti-hyperlipidemic activity exerted by taro could be explored as a source of natural inhibitors of human 3-hydroxy-3-methyl-glutaryl coenzyme A (HMG-CoA) reductase, which is the key point enzyme in cholesterol synthesis and the molecular target for drugs used to treat hypercholesterolemia. The continued use of such drugs by healthy individuals for hypercholesteremia prevention or long-term treatments may cause side effects. Based on this, the search for natural inhibitors was focused on enzymes downstream to HMG-CoA reductase, such as lanosterol synthase. From an extensive list of 130 plant extracts, one obtained from freeze-dried taro was able to cause a 55% inhibition of lanosterol synthase activity, which is an impressive result considering that less than 32 extracts showed a very slight effect, under 5% inhibition. Eight lipids purified from taro extract, three of them classified as monogalactosyldiacylglycerols (MGDGs), and five as digalactosyldiacylglycerols (DGDGs), inhibited in vitro lanosterol synthase activity by 28-67%. The therapeutic potential of MGDGs and DGDGs should be further explored in preclinical studies, since similar lipids isolated from microalgae have already been reported as inhibiting tumor-promoting agents [bib_ref] Inhibition of human lanosterol synthase by the constituents of Colocasia esculenta (taro), Sakano [/bib_ref].
The anti-hypercholesterolemic effect of tiwul, a traditional Indonesian dish prepared from cooked taro flour, has been demonstrated in rats. Hypercholesterolemic rats exhibited a 36% reduction in total cholesterol following two weeks of tiwul intake [bib_ref] Instant Tiwul Made of Colocasia esculenta (L.) Schott as A Current Functional..., Fidyasari [/bib_ref].
Taro should be included in the earliest ancient crops in the world, because the first evidence of taro consumption is dated between 28,700 and 20,100 years BP in Salomon Islands, Oceania, where taro starch remains were found in stone tools possibly used to cut raw corms, indicating that taro may have been consumed at Kilu Cave, and probably it was part of the prehistoric diet [bib_ref] Direct evidence for human use of plants 28,000 years ago: Starch residues..., Loy [/bib_ref]. Over time, these populations have identified several health benefits from consuming taro, many of which are empirical observations that were confirmed in animal models and/or in cell lines testing for its effectiveness and no toxicity to healthy tissues. However, to date, there is a scarcity of clinical studies leaving gaps in the translation of the positive health effects identified in preclinical studies to human beings. The better exploitation and understanding of taro bioactivities, following dietary interventions in humans could be helpful in developing new functional compounds.
Clinical trials on healthy non-diabetic young adults evidenced that taro shows a medium glycemic index, low glycemic load, and moderate glycemic response [bib_ref] Glycaemic index of three cocoyam varieties consumed in imo state, Amadi [/bib_ref]. Moreover, the addition of other food components, such as vegetables, oils, and rich protein food, during cooking, can reduce the taro glycemic index to a lower rate [bib_ref] The concept of low glycemic index and glycemic load foods as panacea..., Eleazu [/bib_ref].
Taro is reported to have anticancer potential through several preclinical analyses, as aforementioned. The Japanese population traditionally consumes starchy roots, such as taro, that are associated with a decrease in the risk of kidney cancer death [bib_ref] Risk factors for kidney cancer in a Japanese population: Findings from the..., Washio [/bib_ref] [bib_ref] Risk factors for renal cell cancer in a Japanese population, Washio [/bib_ref] [bib_ref] Risk factors for renal cell carcinoma in a Japanese population. Asian Pac, Washio [/bib_ref].
Taken together, the intake of taro or their derivatives can provide bioactive compounds capable of promoting health benefits, especially in the control of hypercholesterolemia, which is not only a complication of diabetes and overweight but also a risk for cerebrovascular and cardiovascular diseases, which are important causes of death worldwide, along with cancer [bib_ref] Comprehensive risk management for the prevention of cerebro-cardiovascular diseases in Japan, Teramoto [/bib_ref] [bib_ref] Hypercholesterolemia induced cerebral small vessel disease, Kraft [/bib_ref]. Moreover, the overall benefit against type II diabetes and obesity could certainly aid in reducing the risk factors for cancer.
# Conclusions
Although neglected, taro is a valuable source of several health-promoting compounds, such as taro lectin or tarin, bioactive-complex carbohydrates, and natural polyphenols and other antioxidants. In general, these molecules act through individual or synergic pathways and play a role in the modulation of cellular proliferation, differentiation, apoptosis, angiogenesis, and invasion of cancer cells. The antiproliferative and apoptosis-inducing activities in tumorigenic cells are still poorly understood, despite the evidence obtained from in vitro and in vivo assays. The bioactive compounds from taro may interact with plasmatic cell membranes and intracellular receptors, interfere with signaling cascades, regulate enzyme activities, interact with oncogenes and oncoproteins, and bind to gene promoter sequences. The result of such vast interactions can contribute to ameliorate systemic health status by managing oxidative stress imbalance, reducing systemic inflammation, modulating metabolic dysfunctions, and boosting the immune response. The molecular mechanisms of taro active compounds are not yet fully understood, which is probably due to the few studies on taro, which are restricted to specific regions around the globe, as shown along with this review. Further investigation is necessary to determine which signaling pathways are stimulated to achieve cellular immunity through the activation of T lymphocytes, promoting the apoptosis of tumor cells, how macrophages are activated and progenitors protected, as well as how taro bioactivities can modulate gene expression. Thus, many mechanisms remain to be elucidated to better exploit taro extracts, taro derivatives, or individual taro components. The non-toxicity of these molecules toward healthy cells turns taro components into potential candidates for supportive target therapies when associated with traditional drug treatments. In addition, since taro is a food matrix rich in bioactive compounds, spreading its benefits worldwide may enhance its consumption and consequently production while resulting in better population health maintenance.
# Methodology
This review article collected data regarding the anticancer potential of taro corms components in order to show the importance of including this crop in human diet and includes taro as a promising source of therapeutic agents, with especial attention to tarin. Scientific data were freely collected and followed four main steps: (i) guiding questions formulation; (ii) database searching; (iii); eligibility; and (iv) inclusion.
## Guiding questions
The data search was guided by the main questions:
## 1.
What are the biological activities associated with taro corms? 2.
What are the bioactive compounds responsible for taro corm health-promoting effects, and what is their mechanism of action? 3.
What is the impact of the cooking process on the presence of nutritional components and bioactive compounds of taro corms? 4.
How can taro corm bioactive compounds contribute to cancer fighting or prevention? 5.
What could be the benefits provided by taro corm consumption?
## Databases, descriptors, and/or keywords
A literature search was performed using certain keywords at Scopus (https://www. scopus.com/home.uri?zone=header&origin=searchbasic), PubMed (www.ncbi.nlm.nih. gov/pubmed), and Google Scholar (https://scholar.google.com.br), using the advanced option.
The main keywords used in the search were: "Colocasia esculenta" OR "taro" OR "poi" combined or not to "tuber" OR "corm" OR "extract" OR "tarin" OR "antitumoral activity" OR "immunomodulatory activity" OR "antidiabetic" OR "anti-hyperlipidemic" OR "antimutagenic" OR "anti-hyperglycemic" OR "anti-hypercholesterolemic" OR "antioxidant" Duplicated studies were excluded before eligibility criteria application and references of eligible studies were carefully analyzed in order to obtain additional information not covered by the primary search.
## Eligibility criteria for each of the articles consulted
Articles found according to the search methodology were primarily selected based on titles and abstracts. When these were not enough, the entire study was carefully read.
Articles were excluded when presenting:
1.
Biological activities not directly related to cancer fighting or prevention and cancer risk factors (antimicrobial, anti-insect, antiviral, anti-helminthic, and others) for this study; 2.
Studies performed with parts of the plant other than the corm, such as leaves, petioles or roots; 3.
Unclear or wrong data.
Criteria used to include articles in the present review should comprise:
1.
Experimental studies (in vitro, in vivo or clinical trial) that analyzed biological properties considered important for cancer fighting or prevention, such as antioxidant, antitumoral, antimetastatic, immunomodulatory, anti-hyperglycemic, antidiabetic, antimutagenic, and anti-hyperlipidemic activities; 2.
Studies performed with corms, edible part of taro; 3.
Articles published up to 2020 in the English language with no restriction regarding time period; 4.
Review articles describing Colocasia esculenta characteristics, production, nutritional importance, medicinal uses and other general information; 5.
Studies that purify or identify any taro component that has been proven to exhibit the claimed biological activities specified in item 1.
## Conflicts of interest:
The authors declare no conflict of interest.
[fig] Figure 1: Global distribution of taro production reproduced from FAOSTAT (http://www.fao.org/faostat/en/#data/QC). [/fig]
[fig] nearly 80 -: 90%)-Calculated percentage. NK-natural killer. Yac-1 cells-mouse lymphome cell line sensitive to NK cells. NO-nitric oxide. BM-bone marrow. CY-cyclophosphamide. MGDGmonogalactosyldiacylglycerols. DGDG-digalactosyldiacylglycerols. DNP-1,8 dinitropyrene. IQ-heterocyclic amine 2-amino-3-methylimidazo [4,5-f ]quinoline. PGE2-prostaglandin E2. COX 1/2cyclooxygenase 1 and 2. TNF-α-tumour necrosis factor alpha. IL-interleukin. INF-γ-interferon gamma. VLDL-very low-density lipoprotein. HDL-high-density lipoprotein. LDL-low-density lipoprotein. IC50-half maximal inhibitory concentration. GU50-glucose uptake. hOSC-human oxidosqualene-lanosterol cyclase. apoB-apolipoprotein B. ID50-inhibitory dose for 50% of relative mutagenic activity. [/fig]
[fig] [ 83 ]: TPC-total phenolic content; TFC-total flavonoid content; CTC-condensed tannin content; TA-total anthocyanin content; RPA-reducing power assay; TAA-total antioxidant activity; DPPH-2,2 -diphenyl-1-picrylhydrazyl radical) assay; HORAC-Hydroxyl Radical Antioxidant Capacity; FRAP-ferric reducing antioxidant power; ABTS-2,2 -Azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) diammonium salt; TLC-Thin-layer chromatography; HPLC-High performance liquid chromatography; GC-MS-Gas-Chromatography-Mass Spectrometry; LC-MS-Liquid chromatography-mass spectrometry. Numerals indicate the kind of taro formulation and letters indicate the type of antioxidant compound or the method used for antioxidant evaluation. [/fig]
[fig] Figure 2: Therapeutic potential of taro phytochemicals: Anti-carcinogenic effects of taro bioactive compounds on the non-canonical NF-κB pathway could be mediated by polyphenols. Polyphenols [/fig]
[fig] Figure 3: Putative targets of taro-derived components in metabolic pathways. Lipid and carbohydrate metabolisms can be modulated, mainly, by taro mucilage, arabinogalactan, monogalactosyldiacylglycerols (MGDG), digalactosyldiacylglycerols (DGDG) and A-1/B-2 proteins, which may act in conjunction to control glycemia, lipidemia, and downstream effects such as body weight, glucose tolerance, and glycosuria, hepatic function, atherogenesis, and coronary risk. Inhibition of human lanosterol synthase (hOSC) affects the cholesterol synthesis of pancreatic lipase (PL), reducing triacylglyceride hydrolysis to monoglycerides (MG) and free fatty acids (FA), down-regulation of salivary α-amylase, glucose release by α-glucosidase, glucose absorption, reduction of very low-density lipoprotein (VLDL) formation, and enhancement of muscle glucose uptake. Mucilage can entrap mutagenic/carcinogenic agents, 1,8 dinitropyrene (DNP) and heterocyclic amine 2-amino-3-methylimidazo[4,5-f] quinoline (IQ), avoiding their absorption and consequent effects. HMG-CoA-β-Hydroxy β-methylglutaryl-Coenzyme A; IDL-intermediate-density lipoprotein. [/fig]
[fig] 1, Figure 4: Hypothetical antitumoral tarin mechanism. (left-hand panel) Tarin binds to specific carbohydrate antigens, typically overexpressed in cancer cells, down-regulating COX-2 (cyclooxygenase 2) expression, culminating in decreased PGE2 (prostaglandin 2) synthesis and derepression of the antitumoral response. Tumor progression can be inhibited by inactivation of Ras-Raf or PIK3-Akt pathways and cell death by the activation of ROS-p38-p53 or caspase-dependent pathways, inducing apoptosis or autophagy, commonly described for GNA-related lectins. Red lines show already known tarin mechanisms. Dashed red lines indicate the main mechanisms reported for GNA-related lectins. PLA2-phospholipase A2; PG synthase-prostaglandin synthase; ROS-reactive oxygen species; Cyt. c-cytochrome c; EP1-4-prostaglandin E receptor 1-4; NK cells-natural killer cells. (right-hand panel) Anticancer effects may be boosted by the immunomodulatory abilities of tarin and other taro derivatives, whose effects are highlighted in green and include complement activation through alternative and classical pathways, natural killer (NK) cell activation, T cell activation to TCD4+ cells (Th1) and TCD8+ (CTLs-cytotoxic T lymphocytes) effectors, cytokines release by polysaccharide-activated macrophages and tarinactivated splenocytes, progenitor cell protection through the attenuation of cyclophosphamide (CY) cytotoxic effects and the proliferation of spleen and bone marrow cells. Red lines indicate tarin and polysaccharide effects, while the dotted red lines indicate the downstream effects that might be triggered by released cytokines. CLP-common lymphoid progenitor, CMP-common myeloid progenitor, BM-bone marrow. [/fig]
[fig] Author: Contributions: P.R.P.; É.B.d.A.M. and A.C.N.T.F.C. collected and interpreted the data, and drafted the manuscript. V.M.F.P. helped draft the manuscript, interpreted data and made the final revision. M.A.V. helped draft the manuscript. All authors have read and agreed to the published version of the manuscript.Funding: This research was funded by Fundação de Amparo à Pesquisa do Estado do Rio de Janeiro (FAPERJ), grant number E-26/203.039/2015, E-26/202.815/2018 and E-26/210.865/2019; Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) Universal grant number 406601/2018-6. A.C.N.T.F.C. was funded by Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) Finance Code 001, grant number 88882.461696/2019-01. P.R.P. and É.B.d.A.M. were funded by FAPERJ under the grant number E-26/202.459/2019. and E-26/200.567/2020, respectively. [/fig]
[table] Table 1: Nutritional composition of taro analyzed raw, cooked and baked. [/table]
[table] Table 2: Protective and therapeutic potential of taro corms. [/table]
[table] Table 3: Antioxidants in Colocasia esculenta from different geographic regions. [/table]
|
Genetic Engineering as a Strategy to Improve the Therapeutic Efficacy of Mesenchymal Stem/Stromal Cells in Regenerative Medicine
Mesenchymal stem/stromal cells (MSCs) have been widely studied in the field of regenerative medicine for applications in the treatment of several disease settings. The therapeutic potential of MSCs has been evaluated in studies in vitro and in vivo, especially based on their anti-inflammatory and pro-regenerative action, through the secretion of soluble mediators. In many cases, however, insufficient engraftment and limited beneficial effects of MSCs indicate the need of approaches to enhance their survival, migration and therapeutic potential. Genetic engineering emerges as a means to induce the expression of different proteins and soluble factors with a wide range of applications, such as growth factors, cytokines, chemokines, transcription factors, enzymes and microRNAs. Distinct strategies have been applied to induce genetic modifications with the goal to enhance the potential of MCSs. This review aims to contribute to the update of the different genetically engineered tools employed for MSCs modification, as well as the factors investigated in different fields in which genetically engineered MSCs have been tested.
# Introduction
In the field of stem cell therapy, mesenchymal stem/stromal cells (MSCs) have been widely used in a large number of in vitro and in vivo studies, as well as in approximately 1000 clinical trials. These are multipotent stem cells that must meet minimum criteria, such as plastic adherence, expression of specific surface markers and the ability to differentiate in adipocytes, chondrocytes and osteocytes. Compared to other stem cells, such as embryonic and induced pluripotent stem cells, the use of MSCs has the advantage of being considered safer, regarding the possibility of tumor formation. Additionally, these cells are easier to obtain from different autologous or allogeneic sources. Moreover, MSCs have the capacity to secrete a repertoire of factors with immunomodulatory and anti-apoptotic potential, as well as factors associated to angiogenesis and tissue regeneration.
While in numerous studies the therapeutic effects of MSCs have been shown, in others the transplantation of MSCs did not lead to the desired effect, as evaluated in different disease models. Some studies have shown that transplanted MSCs presented a poor survival rate and proliferation , possibly due to the hostile microenvironment of lesioned tissues, which could lead to nutrient deprivation and cell death. Moreover, partial beneficial effects may be enhanced by expression of specific factors capable of promoting desired effects in the target disease setting. Therefore, approaches aiming to enhance the efficacy of MSC transplantation are needed in order to achieve the suitable results.
Genetic engineering of MSCs has been studied in the past years with the purpose of enhancing the therapeutic potential of these cells and improving the outcomes after transplantation. This has been achieved using non-viral or viral vectors to induce the expression of different factors, depending on the desired results, such as increasing their survival and proliferation rate and improving their pro-regenerative capacity. In this review, we discuss the current methods employed in the generation of genetically modified MSCs and the results obtained with the expression of different factors and the main disease settings in which this modality of cell and gene therapy has been investigated.summarizes the modification agents, cell source, genetic engineering method and applications of diverse studies described in this review.
## Msc as a cell target for genetic modification and gene therapy
Several genetic engineering methods to modify the MSCs gene expression profile have been described, as seen in. These techniques can be classified as those using viral vectors or nonviral methods. Replication-deficient viruses are the most used gene transfer tools, mainly due to their high efficiency in DNA transfer when compared to non-viral methods. However, the use of viral vectors in the clinical practice has been limited by the high cost of cell line production and the possibility of adverse immunological reactions, or even the occurrence of insertional mutagenesis, which may lead to the activation of oncogenes .
Non-viral methods, on the other hand, can be manufactured on a large scale and have low immunogenicity. Currently, the genetic modification of MSCs using non-viral vectors can be performed by physical or chemical methods. The physical methods used in MSCs include electroporation, nucleofection and sonoporation. Chemical methods use lipidic agents, polymers and inorganic nanoparticles. Although the use of non-viral vectors has some advantages for the process of genetic modification of MSCs compared to viral vectors, the impairment of cell viability, low efficiency and transient expression of transgenes make these methods less used in the clinical practice. Viral vectors are the most used tool in the MSC genetic modification protocols, and it has been demonstrated that the high efficiency of viral transduction in these cells (approximately 90%) does not affect their immunophenotypic characteristics, as well as their potential for cell differentiation and secretion of bioactive molecules, which are preserved after genetic modification. In addition, viral transduction ensures stable and longterm transcription of the gene of interest and, consequently, a greater efficiency compared to other methods that do not use viral vectors for genetic modification of MSCs. Currently, there is extensive clinical experience with several types of vectors that include mainly vaccinia, measles, vesicular stomatitis virus (VSV), polio, reovirus, adenovirus, lentivirus, retrovirus, adeno-associated virus (AAV), and herpes virus simplex (HSV). Among those, the most predominant vectors used for cell transduction and transplantation are the lenti-and retroviral vectors. Beyond those applications, AAV vectors have been used as favored vehicle for direct gene delivery to specific tissues.
Adenoviral vectors do not integrate into the host genome and can transduce both dividing and quiescent cells with high efficiency. Obtaining high titers of the recombinant vectors is relatively easy with little cytotoxic effect on the packaging cells. However, the high immunogenicity and transient expression of the transgene limits its application in clinical practice. Adeno-associated viruses, on the other hand, have no viral gene in its recombinant form used for gene therapy, a characteristic that contributes to a low immunogenicity and pathogenicity. Also, they are dependent on co-infection with other viruses, mainly adenoviruses, in order to replicate. Recombinant adeno-associated viruses episomal DNA is unable to integrate in the host genome, therefore reducing the consistency of transgene expression along the time in proliferating cells. However, a limiting factor for the use of AAVs is the action of neutralizing antibodies present in a large part of the population, which drastically reduces their effectiveness in vivo.
Retroviruses are RNA viruses made up of three essential genes: gag, pol and env, which encode the structural protein, reverse transcriptase / integrase and glycoprotein of the viral envelope, respectively. These genes are arranged in separate plasmids and separately for the packaging cells in order to avoid recombination or generation of retroviruses competent for viral replication.
Obtaining high viral titers is relatively easy using these vectors; however, the biggest limitation of retroviral vectors is their inability to transduce quiescent cells. Lentiviruses, however, have the property of transducing dividing and quiescent cells. Currently, lentiviral vectors are the most used in gene therapy, whether in preclinical or clinical studies, as they guarantee high process efficiency, as shown by the increased number of phase I clinical studies evaluating the safety of gene therapies using lentiviruses conducted in recent years. Recently, new genetic modification tools arose in order to promote insertion, deletion or correction of genes at specific sites in the genome, and MSCs have been used as a target for these new modifying tools for applications in different diseases. The sitespecific integration of genes can be achieved using tools such as Zinc Finger Nuclease (ZFN), Transcription Activator-Like Effector Nuclease (TALENS) or Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR/Cas9), which are nucleases capable of recognizing and direct the integration of genes in a site-specific manner. These molecular-editing tools can be further explored to improve safety and efficiency of gene insertion/expression .
## Applications in cancer treatment
Cancer is the second leading cause of global death (WHO, 2020), and conventional chemotherapies have shown poor efficacy for the treatment of cancer in advanced stages. Cell therapy emerged in the last years as a promising tool for cancer treatment. Furthermore, a growing number of cell therapies are being tested in combination with other therapeutic agents, such as checkpoint inhibitors.
Along the past years, many studies have been focused on MSCs potential to act as "Trojan horses, " promoting the delivery of anticancer immunostimulatory agents, such as chemokines and cytokines, to cancer site due to their tumoral tropism. The chemokine receptor 4 (CXCR4) plays a critical role in MSCs homing and survival to tumor sites. Overexpression of CXCR4 in adipose tissue (Ad-MSCs) and bone marrow (BMSCs) MSCs promoted anti-tumor activity in different studies. In a mouse model of colitis-associated tumorigenesis, BMSCs transduced with lentiviral vector carrying either CXCR4 had an enhanced homing (2-fold) to inflamed intestinal tissues when compared to control MSCs . This higher migratory and engraftment capacity was also observed in an in vivo breast cancer mouse model, in which MSCs administered systemically via the tail vein reached the tumors, even though part of the cells remained in the lungs. Moreover, the overexpression of CXCR4 in MSCs increased the accumulation in the tumor compared to non-CXCR4-overexpressing MSCs. A study using BMSCs genetically modified to express the CXC receptor 7 (CXCR7), a newly discovered chemokine ligand 12 (CXCL12) receptor that promotes migration to tumor, showed an increase in cell migration and proliferation, which may be attributed to the CXCL12 secreted by MSCs, thus promoting a positive feedback for CXCL12/CXCR7 axis .
Another explored cancer therapy strategy was the genetic modification of Ad-MSCs to overexpress the cytokine interferonbeta (IFN-β) in an in vitro model of canine melanoma. The cytokine family in which IFN-β belongs is known for their antiviral, immunomodulatory and antiproliferative effects. This approach was tested as an adjunctive therapy in order to begin and maintain skin lesion remission caused by pemphigus foliaceus. Ad-MSCs overexpressing IFN-β was associated with pro-apoptotic and growth-inhibitory effects on canine melanoma cells when compared with non-modified MSCs .
The tumor homing of MSCs has been used to deliver TNF-related apoptosis-inducing ligand (TRAIL), known by its anticancer properties. MSCs overexpressing TRAIL (MSC-TRAIL), when co-cultured with tumor cells, induced a 6-fold increase in apoptotic cell death when compared to non-modified MSCs. In addition, these cells were tested in a metastatic lung mouse model, promoting a tumor free rate of 37.5%, whereas no animals were tumor-free in the control MSCs group. Confirmation of the efficacy of MSC-TRAIL in vivo were shown by, which tested Ad-MSC-TRAIL in pancreatic cancer, finding a 37% reduction in tumor size in Ad-MSC-TRAIL group compared to non-modified Ad-MSCs, a result which was similar to the that found in the group treated with recombinant TRAIL. Moreover, an increased cytotoxicity of TRAIL-expressing MSCs compared to control MSCs was also seen in a lung cancer in vivo model, possibly by enhancing the apoptosis of CLDN7-negative non-small cell lung cancer cells.
The effects of MSCs-derived exosomes, which are secreted vesicles with an average size of 100 nm originated from the endosomal compartment, have also been explored in the context of cancer therapy . Exosomes isolated from MSCs overexpressing microRNAs (MIRs) were also evaluated in different cancer models. Treatment with BMSCs-derived exosomes overexpressing miR-16-5p in a mouse model of colorectal cancer (CRC) inhibited invasion, migration and proliferation of malignant cells, and induced a 4-fold higher apoptosis rate in vitro compared to control cells . Exosomes from BMSCs overexpressing miR-34 were 2-fold more potent than control MSCs in inducing the inhibition of invasion, migration, proliferation and tumorigenesis of glioblastoma cells, in vitro and in vivo . Exosomes from miR-101-3p and miR-199a-overexpressing BMSCs also inhibited invasion, migration and proliferation of oral cancer cells and glioma cells . Furthermore, inhibition of the tumor progression was also observed in vivo in a glioma model .
In a prostate cancer model, administration of BMSCs overexpressing sirtuin 1 led to a decrease in tumor growth and an increase in immune inflammatory response associated with the IFN-γ-induced recruitment and activation of tumoricidal macrophages when compared to non-modified MSCs . The effects of conditioned medium of umbilical cord-derived MSCs (UC-MSCs) overexpressing hepatocyte nuclear factor 4α (HNF4α), a transcription factor which acts as a master regulator of hepatic differentiation and liver metabolism, has also been evaluated in a mouse model of hepatocellular carcinogenesis (HCC). UC-MSC-HNF4α suppressed metastasis and proliferation of cancer cells when compared to untreated and control MSCs-treated groups. Furthermore, in an in vitro assay, MSC-HNF4α reduced approximately 5-fold the migration and invasion potential of cancer cells compared to controls .
## Applications in cardiovascular diseases
Novel treatments for improving the heart function are of immense clinical importance, and cell-based therapies show a great promise. A variety of cell types, including MSCs, have been used in strategies for inducing cardiac regeneration. Initially, it was proposed the differentiation of transplanted MSC into cardiomyocytes and vessels as the main mechanism underlying their therapeutic action in cardiovascular diseases. However, the number of MSCs-derived cells has been shown to be too low to contribute to the functional improvements, and evidence supports the hypothesis that MSC-mediated paracrine mechanisms play an essential role in tissue repair. Among the mechanisms promoted by MSCs are neovascularization, cytoprotection, endogenous cardiac regeneration, modulation of inflammatory and fibrogenic processes, cardiac contractility and cardiac metabolism.
Mesenchymal stem/stromal cells have been genetically modified to express factors that resulted in significant improvements in cardiac recovery. Several authors selected factors to be introduced in a gene therapy approach, based on the putative beneficial effects of MSCs described in infarct myocardial models, such as growth factors. MSCs overexpressing vascular endothelial growth factor (VEGF), when delivered by intramyocardial injection, reduced the infarcted area in 31% and improved left ventricular function to almost its baseline when compared with non-transfected MSCs. In a mouse model of myocardial infarction, transplantation of umbilical cord-MSCs overexpressing hepatocyte growth factor (HGF), a pro-regenerative factor, reduced by approximately half the infarcted area when compared with non-transduced cells, in addition to significantly increase the survival rate post-transplant. In addition, the pro-survival cytokine insulin-like growth factor (IGF)-1, when overexpressed by BMSCs, reduced by approximately 50% the number of apoptotic myocardial cells post-transplant compared to animals treated with control MSCs, and increased MSC retention in the tissue in an experimental model of myocardial infarction. Similar results were observed after C1q/tumor necrosis factor-related protein-3 (CTRP3) and CCR1-modified MSCs transplant. MSCs overexpressing CTRP3, which is associated with protective effects after myocardial infarct, had an enhanced survival and retention 7 days post-transplant, approximately three times higher when compared to unmodified control cells . Additionally, overexpression of CCR1 increased by ∼90% the chemokinesis of MSC in vitro and in vivo, also increasing by more than 50% the amount of viable MSCs in the infarcted myocardium area.
The therapeutic effects of MSCs overexpression growth factors were also studied in a mouse model of cardiomyopathy induced by chronic infection with Trypanosoma cruzi, the causative agent of Chagas disease. Administration of genetically modified mesenchymal cells overexpressing granulocyte colonystimulating factor (G-CSF) were about 2-fold more potent in reducing the number of inflammatory cells and the percentage of fibrosis than control MSCs, acting by increasing the number of myeloid-derived suppressor cells and T regulatory cells. In contrast, no increase in anti-inflammatory or antifibrotic activity was seen after transplantation of IGFoverexpressing MSCs in T. cruzi-infected animals. However, an increased pro-regenerative activity in skeletal muscle was observed after MSCs-IGF-1 transplantation, compared to control MSCs, increasing the number of myofibers similar to that of uninfected mice.
The overexpression of the regulators of cardiogenesis Csx/Nkx 2.5 and GATA-4 was also investigated in a myocardial infarct model. Genetically modified MSCs overexpressing these factors increased the cardiac function by inducing pro-angiogenic effects, significantly increasing microvessel density in the peri-infarct regions and reducing cell loss by apoptosis by approximately 10% compared to animals treated with control MSCs.
In addition to their pro-angiogenic effects, the cellular benefits of MSCs may also be mediated by activating the survival kinase pathways, including Akt activation, in cardiomyocytes in response to MSC-secreted cytokines, promoting a reduction programmed cell death. Interestingly, MSCs overexpressing Akt promoted the maintenance of metabolism, glucose uptake and cytosolic pH and prevention of cardiac metabolism remodeling in a myocardial infarct model. Remarkably, for Akt-MSC hearts, systolic performance was only 12% (72 h) to 17% (2 week) lower compared with shamoperated hearts while diastolic performance remained normal. In contrast, for unmodified MSC-treated and untreated infarcted hearts, markers of both systolic and diastolic performance were significantly lower than for sham-operated hearts. Akt-modified amniotic fluid derived MSCs were also capable of alleviating myocardial injury and contribute to cardiac regeneration in an ischemia-reperfusion injury in a rabbit model, promoting angiogenesis by capillary density enhancement and VEGF expression 2-fold more than control MSCs. Moreover, Akt-expressing MSCs promoted a significant increase in cTNT, GATA-4 and connexin 43 compared to controls . In addition, purified exosomes from TIMP2-and Aktmodified umbilical cord MSCs promoted improvements in cardiac function, by activating TIMP/Akt pathwayand increasing platelet-derived growth factor D expression , in rats submitted to myocardial infarction.
The expression of CXCR4, the stromal cell-derived factor (SDF)-1 receptor, which is largely involved in progenitor homing and survival, has been tested to improve the insufficient cell homing and tissue persistence observed in several preclinical studies. CXCR4-MSCs had a 3-fold increase in cell migration capacity in vitro and in vivo in a myocardial infarct model when compared to control MSCs, correlating with an increased expression of metalloproteinases. Additionally, a 4-fold increase in capillary density was found in the hearts of animals treated with CXCR4-MSCs when compared to control MSCs groups . A significant improvement in cardiac function in CXCR4-MSCs treated animals compared to those treated with control MSCs, which may be explained by the reduction of heart fibrosis and increased angiogenesis post-injury.
Furthermore, anti-inflammatory interleukins and regulators of oxidative factors were explored in the production of genetically engineered MSCs. IL-10 overexpression improved the therapeutic effects of MSCs by reducing by 2.5-fold the production of proinflammatory cytokines (TNF and IL-1β) and promoting a 3-fold reduction of the heart infarct size when compared to control MSCs. In addition, injection of MSCs modified to overexpress IL-33, a cytokine known to induce Th2 and Treg responsesresulted in a significant improvement of heart function and reduced inflammation compared to vector control MSCs in rats with myocardial infarction . MSCs modified to overexpress endothelial nitric oxide synthase (eNOS) were more potent than control MSCs and eNOS adenoviral vector in reducing the myocardial infarct size (4 and 2-fold increase, respectively), corrected hemodynamic parameters and increased the capillary density by increasing nitric oxide production . Moreover, heme oxygenase-1 (HO-1), an enzyme acting on heme degradation, as well as in the generation of cytoprotective agents. Transplantation of HO-1-overexpressing MSCs promoted not only an improvement in angiogenesis in scar areas, but also an increase in connexin 43-positive gap junctions and a higher tyrosine hydroxylase-positive cardiac sympathetic nerves sprouting in a myocardial infarction model.
Other molecules with diverse mechanisms of action were also explored as modification targets were Follistatin-like 1 and Islet-1. MSCs overexpressing Follistatin-like 1, a prosurvival cardiokine for cardiomyocytes, had increased survival, proliferation and engraftment, thereby improving in about 2fold their therapeutic efficacy when compared to mcherry transgenic control MSCs in a myocardial infarction model . Similarly, overexpression of Islet-1, a transcription factor involved in cardiogenesis regulation and a marker of cardiovascular progenitor cells, promoted MSCs survival post-transplant and enhanced their paracrine function.
Another strategy studied to improve cell therapies for cardiac diseases is the overexpression of microRNAs in MSCs. Overexpression of mIR-126 in MSCs induced the production of Notch ligand Delta-like-4, which activates the Notch pathway and promotes the production of pro-angiogenic factors, ameliorating the cell therapy efficacy in infarcted hearts by 2-fold compared to control MSCs. In a similar way, intramyocardial transplantation of microRNA-1-transfected MSCs was more effective to promote repair of the infarct injury and improvement in heart function by enhancing transplanted cells survival and cardiomyogenic differentiation when compared to mock-transfected MSCs. Furthermore, in a model of myocardial ischemic-reperfusion injury, mIR-181a overexpression enhanced the immunosuppressive capacity of MSCs-derived exosomes, improving their therapeutic effect by 2 to 3-fold in reducing the infarct area and increasing the ejection fraction when compared to exosomes produced by control MSCs . Genetic modification of MSCs to overexpress miR-21 not only promoted an enhanced migration and proliferation rates, but also in angiogenesis and cardiac function. This may be explained by an increase in Bcl-2, Cx43 and VEGF levels and a decrease in Bax, BNP and troponin T, seen in a model of adriamycin-induced myocardial damage . Transplantation of MSCs overexpressing miR-133 in a rat model of myocardial infarction reduced fibrosis and promoted an improvement in cardiac function more potently than control vector-MSCs, possibly by reducing Snail 1 expression . Finally, in a model of myocardial infarction, intravenous injection of MSCs genetically engineered to overexpress miR-211, known to influence cell migration, promoted a significant increase in migration of MSCs to the injured area and reduction of the infarct size, while PBS or control MSCs did not reduce the infarct size nor promoted functional recovery.
## Applications in lung diseases
Pulmonary tract pathologies are commonly due to the frequent exposure of the respiratory system to different factors, such as tobacco, toxic smoke or polluted air, which may contribute to the development of acute and chronic disorders, such as infections or autoimmune and inflammatory diseases. Thus, MSCs have been used and genetically modified to express different factors aiming to promote the patients' recovery, since many diseases still do not have an efficient treatment.
Pulmonary arterial hypertension (PAH) has been a target of therapies with modified mesenchymal cells. Guo and collaborators generated rat bone marrow derived MSCs expressing human HGF (MSC-HGF) by transduction with recombinant adenoviral vector (ad-HGF). The treatment with MSC-HGF or MSC-HGF+ recombinant G-CSF promoted a significant reduction in mean pulmonary arterial pressure and hypertrophy in the right ventricle when compared with the control MSCs and untreated groups. In addition, pulmonary perfusion in the MSC-HGF+G-CSF group was improved by increasing the number of blood vessels (∼2-fold more vessels compared with the MSC and untreated group). Treatment with MSCs overexpressing the secreted Klotho protein (SKL), a β-glucuronidase that regulates oxidative stress and inflammation, restored pulmonary endothelial dysfunction in a model of monocrotaline-induced PAH, acting by reducing 25% blood vessel thickness and doubling the lumen when compared to the MSCs, MSCs-GFP and untreated groups. MSCs-SKL also slightly attenuated systolic pressure and right ventricular hypertrophy and had anti-inflammatory effect, reducing by 50% the macrophage infiltration in the lung tissue.
In a severe acute lung injury (ALI) mouse model, treatment with bone marrow-derived MSCs modified to express human angiopoietin 1 (Ang1) promoted a significant reduction in airspace inflammation, with 96 % less neutrophils than groups treated with control MSCs or saline. Moreover, Ang 1-MSCs reduced the production of pro-inflammatory cytokines IFN-γ, TNF-α, IL-6 and IL-1β, equal to baseline values observed in naive mice. In a similar study, MSCs transduced by lentiviral vector to overexpress angiopoietin 1 (MSC-Ang1) also promoted benefits for ALI, reducing the levels of the proinflammatory cytokines TGF-β1 and IL-1β and increasing the expression of the anti-inflammatory cytokine IL-10 in the serum and bronchoalveolar lavage fluid, in addition to improving the lung function by increasing the expression of surfactant protein C.
In vitro studies with alveolar epithelial cells subjected to cigarette smoke extract (CSE) revealed that BMSCs overexpressing Notch1 receptor (MSC-N1ICD) maintained alveolar cells proliferation levels close to those observed in the control not exposed to CSE by activation of PI3K/Akt pathway. Additionally, MSC-N1ICD doubled the expression of CXCR4, promoted a 2-fold greater cell migration when compared to the control MSC groups and prevented apoptosis, as shown by the 50% reduction in caspase-3 expression in alveolar cells when compared to MSC control cells. In another study,reported that MSCs genetically modified to express Wnt/β-catenin contributed in vivo with the alveolar epithelium protection, promoting the retention of a greater number of MSCs in the lungs for up to 14 days when compared to the control cell line, as well as an increase (approximately 3-fold) of differentiation of MSCs in type II alveolar epithelial cells and improvements in alveolar epithelial permeability, resulting in a reduction in lung damage.
Mesenchymal stem/stromal cells from different sources have been genetically modified to express different cytokines and chemokines in order to explore the immunomodulatory potential of these cells in lung injury models. The studies demonstrated that MSCs overexpressing IL-10, CXCR4 or CXCR7 caused a decrease in the number of alveolar neutrophils and levels pro-inflammatory cytokines, such as tumor necrosis factor α (TNFα), transforming growth factor (TGF-β1) and IL-6. Moreover, an increase in the survival rate of the animals treated with these genetically modified cells was observed, likely due to the reduction of inflammatory severity and damage to the lungs . Differentiation of MSCs into type II alveolar epithelial cells and an increase of alveolar macrophages were also observed.
Overexpression of enzymes and receptors by MSCs was another approach used in the management of ALI. In a model of LPS-induced lung injury, mouse BMSCs modified by a lentiviral vector to express the angiotensin-converting enzyme 2 (ACE2) induced an improvement in the lung tissue, with decreased inflammatory infiltrate, edema and interstitial hemorrhage and reduction of lung injury score from 13 to 5.7 in WT mice, and to 8.6 in ACE-knockout mice when compared to the MSC and untreated groups. In addition, a decrease in IL-6 and IL-1β levels, as well as an increase in IL-10 levels, was observed in the MSC-ACE2 group compared to the control MSCs . Similarly,observed that human MSCs derived from adipose tissue overexpressing the soluble IL-1 receptor type 1 (sST2) caused a decrease in histopathological changes associated with lower levels of pro-inflammatory cytokines TNF-α, IL-6, and MIP-2 (1.5 to 3-fold reduction), and 6-fold higher levels of IL-10 in models of ALI and asthma. In a model of ALI in rats, BMSCs overexpressing the enzyme heme-oxygenase-1 (HO-1) promoted the survival and anti-apoptotic activity greater than wild-type BMSCs . The transplantation of MSCs overexpressing the type 2 angiotensin II receptor (AT2R) increased cell migration in vitro by 3fold, inhibited inflammation by decreasing pro-inflammatory cytokines (IL-1β and IL-6) and recovered the injured lung in a model of ALI in mice, while overexpression of the orphan receptor tyrosine kinase 2 (ROR2) by MSCs not only improved their ability to relieve inflammation and histopathological changes, but also increased their retention in lungs.
Exosomes derived from MSCs are important sources of microRNAs that provide protection against various diseases. Therefore, increasing the expression of microRNAs in MSC is an interesting alternative for some pulmonary pathologies. In a study that used exosomes derived from MSCs with overexpression of miR-30b-3p in co-culture with alveolar epithelial cells (AECs) challenged with type II LPS, high endocytosis of exosomes rich in miR-30b-3p was found, and this phenomenon was accompanied by reduced expression of SAA3 (protein highly expressed in ALI), increased cell proliferation and inhibition of apoptosis. BMSCs overexpressing let-7d, an antifibrotic microRNA, were tested in an ALI model in mice, promoting an increase in the production of the anti-inflammatory cytokine IL-1RN (when compared to wild type MSC), a reduction in the number of CD45 + cells (when compared to the untreated group) and decreased collagen transcription levels, even though no significant differences were observed in the Ashcroft score and OH-proline. UC-MSCs overexpressing p130 or E2F4 also proved to be more potent in decreasing tissue damage in the lungs, promoting retention of MSC-p130 and MSC-E2F4 7 days after intratracheal transplantation (50% more retained cells than control groups) and inhibiting fibrosis in a model of severe acute respiratory distress .
Transcription factors are important regulators of gene expression. The effects of human amniotic MSCs (hAMSCs) modified to express the nuclear factor erythroid-derived 2-like control MSCs in reducing apoptosis, fibrosis, edema and pro-inflammatory cytokine levels when injected intravenously .
## Applications in gastrointestinal and liver diseases
Genetic modification of MSCs is a promising approach widely explored in diverse models of gastrointestinal and liver diseases. The improvement in cell homing and target ability to inflammatory sites, together with the increase in immunosuppression and tissue repair capacity after transplant, are some of the improvements seen in gastrointestinal experimental models of oral mucositis and inflammatory intestinal diseases.
Growth factors have been used as targets for gene therapy in the field of liver disease, as shown in a study in which HGFoverexpressing BMSCs (BMSC-HGF) promoted an improved recovery from liver damage in a model of CCl 4 -induced cirrhosis in rats, increasing the expression of hepatic proteins HNF-4α, CK18 and ALB (2 to 3.5-fold greater than untreated control) and reducing the presence of liver injury markers, aspartate aminotransferase (AST), alanine aminotransferase (ALT), and total bilirubin, compared to the other groups tested . The therapeutic effects of HGF overexpressing MSCs were also studied in an acute model of radiation-induced liver injury. MSC-HGF showed greater migration and permanence in the injured tissue, increased the proliferation of hepatocytes (about 4-fold greater than untreated control), completely blocked the increase in ALT and AST and prevented apoptosis and liver fibrosis when compared with the other treatments . In a similar model of chronic injury, MSCs overexpressing fibroblast growth factor 4 (FGF4) also contributed with liver regeneration, promoting greater migration of MSC-FGF4 to cirrhotic livers and increasing the expression of the liver progenitor marker EpCAM by about 4-fold when compared to control groups .
Hepatic macrophages (Kupffer cells) play a central role in liver fibrosis, being critical in both its promotion and resolution. In a model of chronic thioacetamide poisoning (TAA), MSCs transduced with a lentiviral vector for overexpression of IGF-1 (Ad-MSC-IGF-I) were able to reverse a pro-fibrotic phenotype in Kupffer cells, reducing by half the levels of collagen deposition in the area of the lesion when compared to the vehicle-treated group and downregulating the expression of the pro-fibrogenic genes TGF-β1, α-SMA and collagen 1A2 (COL1A2). Neuregulin 4 (Nrg4) acts in the liver, where it modulates the lipogenesis in hepatocytes by activating the ErbB3/ErbB4 signaling. Thus, the therapeutic potential of Nrg4-overexpressing Ad-MSCs was tested in a model of hepatic steatosis induced by a high-fat diet. Transplantation of MSC-Nrg4 reduced weight gain, decreased serum glucose and insulin levels, in addition to improving glucose intolerance more significantly than treatment with control MSC .
Cytokines and chemokines have also been evaluated in liver and gastrointestinal diseases. Uncontrolled hepatic immune activation is the primary pathological mechanism of fulminant hepatic failure (FHF). The interleukin-1 receptor antagonist (IL-1Ra) plays an anti-inflammatory and anti-apoptotic role in acute and chronic inflammation. The transplantation of MSCs that overexpress this molecule increased survival (63.6% survival in the MSC-IL-1Ra group versus 30.0% and 22.2% in the MSC and vehicle groups). MSC-IL-1Ra increased the fraction of proliferating hepatocytes (23.5%) compared to , and improved liver function by negatively regulating inflammatory responses activated by IL-1 in vivo. In a TAA-induced model of chronic liver fibrosis in mice, engineered MSCs that secrete high levels of IL-10 had a more potent antifibrotic activity and caused the improvement in liver function than control MSCs, which were also able to significantly promote histopathological and functional improvements.
The role of CXCL2 in oral mucositis was investigated from transplantation of human BMSCs with increased expression of CXCR2 by lentiviral transduction. MSC-CXCR2 exhibited improved ability to migrate in response to CXCL2 in vitro (almost 2 times greater than MSC control), indicating greater targeting of these cells to the inflamed mucosa in animal models. MSC-CXCR2 also had a longer residence time in the oral cavity than control MSC and promoted accelerated ulcer healing by reducing pro-inflammatory cytokines, such as TNF-α, IL-1β, and IL-6 and levels of reactive oxygen species (ROS) in epithelial cells were 2-fold lower, while ROS levels were 4-fold lower in tongue fibroblasts . In a mouse model of inflammatory bowel disease (IBD) induced by sodium dextran sulfate, increased expression of the Intercellular Adhesion Molecule (ICAM)-1 by MSCs have been shown to alleviate inflammatory damage in IBD mice, reducing colon shortening to levels closer to the control without IBD, in addition to halving the number of inflammatory cells when compared to the untreated IBD group. Moreover, ICAM-1 also stimulated the migration of modified MSC to the affected colon .
In addition, the transfer of genes coding for important transcription factors regulating hepatogenesis, such as HNF4α and forkhead box protein (Foxa2), was investigated in the context of liver diseases in association with stem cell therapy. MSCs overexpressing Foxa2, cultured in a 3D system of poly-lactic-co-glycolic acid (PLGA) scaffold, promoted hepatic differentiation, in vitro and in vivo when implanted into nude mice. Importantly, when implanted into dorsal subcutaneous tissues, Foxa2-overexpressing Ad-MSCs/scaffolds reduced the alterations in liver function markers in the blood and in liver tissue, in an acute liver injury model induced by TAA. Moreover, Foxa2-MSCs promoted the recovery expressions of liver enzymes close to normal levels, and reduced the fibrotic areas more potently when compared to MSC alone or MSC/vector groups . Similarly, HNF-4α-modified MSCs showed increased therapeutic effects in a liver cirrhosis mouse model, where MSC-HNF-4α positively regulated iNOS expression by activating the NF-κB signaling pathway .
## Applications in kidney diseases
Nephrological disorders can lead to an increase in mortality and morbidity, especially in cases of acute kidney injury, and may affect the functionality of the organs. The use of genetically modified MSCs has also been investigated in the context of kidney diseases.
Ad-MSCs and BMSCs genetically engineered to overexpress glial cell line-derived neurotrophic factor (GDNF) promoted the switching of macrophages to a reparative phenotype, reduced renal fibrosis and contributed to a recovery in renal function in models of nephrotoxic serum nephritis and unilateral ureteral obstruction . In another study, injection of MSC-HGF, in a model of bladder outlet obstruction, not only prevented collagen deposition, with a reduction in collagen area of almost half when compared to the untreated animals, as well as increased cystometric parameters in transplanted animals . In a model of renal failure-induced anemia, MSC-IGF-1 co-implanted with MSCs genetically engineered to secrete erythropoietin (MSC-EPO) promoted a significant hematocrit elevation when compared to the group which received the MSC-EPO + MSC-null, with a difference of approximately 20% after 98 days. Additionally, an improvement in heart function was observed in the animals treated with MSC-EPO + MSC-IGF-1.
Overexpression of cytokines and chemokines in MSCs have also been evaluated in models of kidney diseases. In a model of renal ischemia/reperfusion injury, MSCs genetically modified to overexpress TGF-β1 increased renal function and reduced inflammation by a significant decrease in the proinflammatory cytokine expression, such as IL-2, IL-6, and TNF-α, in comparison to the PBS and MSC groups, and an enhancement in IL-10 expression of about 2.5-fold when compared to the sham group. Lipocalin 2 (Len2), secreted in high levels into the blood and urine after kidney injury may be an important cytoprotective agent against injuries caused by oxidative stress. BMSCs overexpressing Len2 prevented cytotoxicity and apoptosis when co-cultured with HK-2 and HEK293 cells in a model of cisplatin-induced kidney injury in vitro, with a cell viability of 60% after 70 days when cocultured with HK-2 cells and 40% with HEK293 cells, while the control groups with only kidney cells co-cultured with MSCs or Mock-MSCs had 20% and less than 10% viability for kidney cells, respectively. Additionally, the percentage of apoptosis in the groups treated with MSC-Len2 was significantly decreased, while HGF, IGF-1 and FGF-2 concentrations were higher in this group when compared to the other treatments. BMSCs-CXCR4, were also able to stimulate HGF production when compared to the MSCs alone, although not in the same intensity as the MSC-Len2. Furthermore, an increase in IL-10 and bone morphogenetic protein 7 (BMP-7) was also observed. Increasing in proliferating cells and a decrease in apoptotic cell markers when co-cultured with hypoxia/reoxygenationpretreated renal tubular epithelial cells (HR-RTECs), as well as renal functional improvement and a decrease in tubular cell death was seen in animals treated with BMSCs-CXCR4 in a model of acute kidney injury .
Transplantation of BMSCs modified to express the Klotho gene, which is known to be related with the aging process , promoted antifibrotic effects and led to an increase in proliferation and immuno-regulatory ability of MSCs in a model of acute kidney injury . Moreover, in a model of ischemia/reperfusion injury, administration of exosomes derived from BMSCs expressing miR-199a-3p had protective effects and decreased inflammatory mediators, as well as the number of apoptotic cells, although this reduction was similar to the group treated with the MSCs .
## Applications in neurological disorders
Due to the complexity of the nervous system, many neurological disorders do not have an effective treatment in order to reduce or prevent the damages caused in the nervous tissue. However, MSC cell therapy has been investigated as an option for the treatment of several neurological disorders, and many articles have studied their effects on the nervous system and possible strategies to enhance their therapeutic actions, including genetic engineering.
Mesenchymal stem/stromal cells secrete diverse growth factors related to neurogenesis and neuroprotection, such as brain derived neurotrophic factor (BDNF), Glial cell-derived neurotrophic factor (GDNF), fibroblast growth factor-2 (FGF-2), HGF, IGF-1, nerve growth factor (NGF) and platelet-derived growth factor (PDGF)and, therefore, many studies have been evaluating the effects of MSCs overexpressing these factors for the treatment of neurological disorders in different in vitro and in vivo models. In a model of diabetic cistopathy, rats that received intrathecal injections of human MSCs derived from umbilical cord modified by lentivirus to overexpress NGF (MSCs-NGF), had a significant improvement in voiding dysfunction, with a 12% and 45% voiding interval higher when compared to the MSC control and untreated diabetic groups, respectively. This effect was related to the differentiation of MSCs-NGF in neurons and glial cells, increasing the control of neuronal functions along the urinary pathway. In the in vitro model of Parkinson's disease, the conditioned medium MSCs overexpressing HGF showed an increase in the viability of neuron culture 48 h after treatment (about 14% greater than control MSC and 42% greater than untreated cells). In addition, the presence of intracellular free calcium was 2.4-fold lower in the MSC-HGF group and about 2-fold lower in the MSC group when compared to the untreated control, indicating a greater ability of MSC-HGF to maintain the cellular homeostasis . Supernatant of MSCs overexpressing BDNF from three different donors showed a higher neuroprotective effect in vitro, with a neuronal survival rate around 40% when compared to the control-vector infection group, while in another study, MSCs also overexpressing BDNF started to express neuronal phenotype markers 1 day after transfection and in the following days, presented a neuronlike morphology when compared to the MSC control, which maintained their fibroblast-like morphology.
Interestingly, these findings demonstrate that the overexpression of BDNF by MSCs seems to be able to either increase their neuroprotective effects or induce changes in their phenotype to one closer to a neuronal one, suggesting more than one possible effect in the cells.
In another study, MSCs with GDNF overexpression increased the differentiation of neural stem cells (NSCs) when these cells were co-cultured with a 3D microfluidic system. After 5 days in culture, a significant increase in the expression of neuronal markers, such as Tuj1 and MAP2, and in the formation of neurites in NSCs co-cultured with MSC-GDNF when compared to MSC control group was observed. In addition, the authors performed an animal model of hypoxic-ischemic stroke to assess neurobehavioral function. However, even demonstrating promising effects in vitro, transplantation of MSC-GDNF did not demonstrate a significant functional increase when compared to control MSC, being superior only to the untreated group . On the other hand, transplantation of MSCs with increased expression of FGF21 (an important regulator of metabolic pathways) in the brain of mice after traumatic brain injury (TBI) led to a decrease in memory deficits and increased the level of FGF21 in the ipsilateral hippocampus, naturally decreased after TBI. In addition, the MSC-FGF21 group had enhanced neurogenesis and restored the morphology of immature newborn neurons in the hippocampus when compared to the MSC and vehicle groups, which exhibited impaired maturation and deficiency in the neurogenesis process.
In a model of spinal cord injury (SCI), bone marrow-derived MSCs overexpressing or not ciliary neurotrophic factor (CNTF) promoted functional recovery, as shown by an improvement in the Basso, Beattie, and Bresnahan (BBB) motor score during the weeks following transplantation, and the H&E staining of the injury site demonstrated that both groups showed a greater cell density in the cavity when compared to untreated group. Nevertheless, even with both cell groups seeming to be able to preserve the nervous tissue, the BMSC-CNTF group displayed a profile more similar to the sham group in comparison to the others. Another study evaluating the effects of MSCs on SCI showed that transplantation of MSCs overexpressing HGF reduced the activation of astrocytes by inhibiting the expression of TGFβ1 and β2, promoting an almost 3-fold decrease in GFAP+ cells in the lesion area, when compared to the control or wild type MSCs, in addition to inducing the reduction of the glial scar, one of the main factors that hinder the functional recovery after SCI. BMSCs genetically modified to overexpress the growth factor IGF-1 showed promising effects on neuroprotective and antioxidant activity in mice after SCI. When compared to control groups, IGF-1 secretion promoted a 3-fold increase in graft survival in the lesion area, which facilitated the recruitment of endogenous neural progenitor cells, as well as the positive regulation of antioxidant genes, resulting better preservation of neural tissue (myelinated area about 50% greater than that observed in control groups) and significant motor improvement using Basso Mouse Scale (BMS; 6 at the end of the experiment for IGF-1 group vs 4 in control groups). In another SCI model, BMSCs overexpressing sonic hedgehog (Shh), a growth factor with multiple functions in the nervous system, were able to reduce tissue damage, with 2 times greater expression of neurofilament 200 (NF200), in addition to promoting functional recovery significant in animals when compared to BMSCs or vehicle groups. Finally, mice treated with BMSCs genetically modified with brain dopamine factor (CDNF) showed significant functional gains in the BBB score, from week 3 to week 6, when compared to control groups, which did not show significant differences between themselves. In addition, histopathological analyzes indicated greater tissue preservation, with a 6 to 7fold reduction in the area of cavitary lesions in the MSC-CDNF group when compared to MSC control and untreated groups. Neural fiber recovery was also observed, with improvements in remyelination levels and reduction of neuroinflammation after SCI .
A study evaluating the effects of neurotrophin-3 (NT3) overexpression by BMSCs in a model of Parkinson's disease revealed that genetically modified cells protected neuronal tissue and induced differentiation of BMSCs into cells similar to dopaminergic neurons (neuron type affected by Parkinson's disease), as seen by a 5-fold increase in the levels of nurr-1 and wnt-1 in transfected cells, when compared to treatment with wild-type MSCs.
VEGF-overexpressing MSCs have been used in different disease models, including neurological diseases. In the treatment of peripheral nerve damage, MSC-VEGF has shown promising results both in vitro and in vivo, promoting the extension of neurites and maintaining high expression of VEGF in the nerves 2 weeks after grafting. In a model of Alzheimer's disease, also evaluating the effects of MSC-VEGF, genetically modified cells promoted neovascularization in the hippocampus and reduced the presence of beta-amyloid plaques in the dentate gyrus when compared to the vehicle-treated group, although no significant differences were found compared to the MSC group, which also appeared to induce neovascularization .
The immunomodulatory effects of genetically modified MSCs to express interleukins have been studied to treat autoimmune diseases, such as multiple sclerosis, or to reduce the damage caused by trauma to the brain or spinal cord in cases of TBI, SCI or stroke. Genetically modified MSCs to express IL-4 (responsible for modulating the autoimmune inflammatory response), for example, exhibited protective effects in a model of multiple sclerosis when transplanted in the early stage of the disease. MSC-IL-4 reduced the expression of pro-inflammatory cytokines such as IFNγ and IL-6, leading to a reduction in disease severity when compared to control groups. In an acute ischemic stroke model, BMSCs overexpressing IL-10 ameliorate the motor function, posture score and forelimb grip strength. In addition, 72 h after treatment, a decrease in the number of Iba-1 and TNF-α positive cells, of approximately 2-fold more, was seen in BMSC-IL-10 group when compared to the vehicle, while this reduction was of almost half in comparison to the MSC, which also presented a significant difference in relation to the vehicle group. The effects of BMSC-IL-10 were also evaluated in a TBI model, in which these cells and MSCs alone decreased significantly the expression of cell death markers, such as Bax, cytochrome-C, caspase-3 and p53 when compared to the vehicle group and increase in those related to cell survival (Bcl2) and synaptic transmission (PSD95 and synaptophysin) was seen in BMSC-IL-10, but not in MSCs alone and vehicle groups, which presented differences only in synaptophysin levels. Thus, these results showed the promising effects of BMSC-IL-10 in promoting the protection of neuronal cells following TBI. C-C Motif Chemokine Receptor 2 (CCR2) is positively regulated in the first 24 h after ischemic stroke. Therefore, the effects of MSCs genetically modified to express CCR2 were evaluated in a stroke model. In this study, overexpression of CCR2 increased 5-fold approximately the migration of MSCs modified to ischemic region, and it was able to not only decrease ischemic injuries and promote neurological recovery in vivo, but demonstrated an important role in protecting the blood-brain barrier and in promoting endothelial regeneration, when compared to the control groups of MSC .
Overexpression of enzymes and receptors by MSCs, in order to improve the protective effects in tissue regeneration or increase cell survival, has also been studied for the treatment of different neurological disorders. MSCs derived from different sources have been genetically modified to express important enzymes for the CNS. MSCs overexpressing arginine decarboxylase (an enzyme that regulates the synthesis of agmatine, known to confer neuroprotection after brain damage) were tested in an SCI model in rats. Tissue analyzes showed an increase in BDNF expression at the lesion site, and this effect was accompanied by a reduction in the area of fibrotic scar and functional improvement in the group treated with the modified cells. MSCs overexpressing heme oxygenase-1 (an enzyme that modulates the response to oxidative stress after spinal cord trauma) were tested in canine SCI. Overexpression of this molecule led to a functional recovery after transplantation, related to the anti-oxidative effect and to a decrease in the presence of infiltrated cells, fibroblastlike cells and apoptotic cells in the area of the lesion, as well as an increase in the expression of NeuN and β3-tubulin markers when compared with the control groups. In addition, MSCs with increased tropomyosin kinase A (TrkA) receptor, a receptor highly expressed in sensory fibers, were used to repair peripheral nerves in rats. 8 weeks after transplantation, the MSC-TrkA-treated group showed greater axonal growth, with significantly higher expression of the basic myelin protein and superior results of the density of myelinated fibers, in addition to superior functional performance than observed in the control groups .
In another study, MSCs derived from Wharton's jelly genetically modified to express PARKIN (an ubiquitin ligase capable of protecting dopaminergic neurons against stress) were tested in the Parkinson's disease model produced by 6-hydroxydopamine-induced toxicity. The overexpression of PARKIN in MSCs was able to significantly reduce the expression of markers of cell death and oxidative stress, in addition to significantly reducing the production of reactive oxygen species (∼ 50% reduction) when compared to wild type and control groups. Studies in stroke models showed that MSCs overexpressing extracellular regulating kinase 1/2 or integrin A4 not only were able to show induction of neuronal differentiation, proliferation of neural stem cells and a significant functional recovery in mice after the treatment , but also demonstrated a significant decrease in cell aggregation and improvement of cerebral embolism in rats.
MicroRNAs have been receiving attention during the past years due to their involvement in the regulation of different and important cellular processes, as shown in a mouse model of ischemic brain injury, where BMSCs overexpressing miR-705 contributed to ameliorate neurological deficits and to suppress neuronal death, associated with a 2fold increase in BDNF and VEGF expression when compared to sham and vehicle groups. Several studies evaluated the effects of exosomes derived from genetically modified MSCs for the treatment of neurological diseases. Exosomes derived from MSCs overexpressing pigment epithelium-derived factor (PEDF) or miR-25 showed neuroprotective effects in models of cerebral ischemia reperfusion and ischemic spinal cord by mechanisms involved in axonal preservation, such as regulation of autophagy and apoptosis or reduction of inflammation and oxidative stress. MSCs overexpressing miR-21, miR-124, or miR-133b in models of intracerebral hemorrhage, SCI and stroke, respectively, increased the expression of neuronal markers, such as β-III tubulin and NF200, induced neurite remodeling and outgrowth, reduced neurological damages and promoted functional recovery after treatment.
Transcription factors have also been expressed by MSCs for the treatment of neurological diseases. BMSCs genetically engineered to express hypoxia-inducible factor 1α (HIF-1α) led to motor functional improvement, decreased cerebral infarction and increased by 4-fold the VEGF expression, when compared to the control groups in a model of cerebral artery occlusion . In another study, Mash1 overexpressing-MSCs enhanced neuronal markers expression, such as NeuN and GAD67, when compared to groups treated with MSCs control cells. In addition, these cells differentiated into cells similar to neurons that showed action potential, as seen by electrophysiological characterization, demonstrating that they were functional in vitro.
Genetically modified MSCs have also been combined with other methodologies in order to enhance their effects, as observed in a SCI model, in which the transplantation of BMSCs overexpressing BDNF combined with platelet-rich plasma (PRP) scaffolding promoted an increase in the expression of axonal markers such as NF200, 4 and 8 weeks after transplantation, when compared to groups that received control BMSC cells. Regarding GFAP levels, the BMSC groups, when combined with PRP, showed an increase in the expression of this molecule compared to groups without PRP. In addition, a better functional recovery was observed in the BMSC-BDNF group combined with PRP, with a significant gain in the BBB index, when compared to the control groups. Two studies evaluated the effects of transplanted MSCs genetically engineered to express tyrosine kinase receptor type 3 (TrkC) in combination with electroacupuncture (EA) for the treatment of SCI. An increase in the expression of NT3, an important neurotrophic factor, was seen in the spinal cord, as well as an improvement in the functional recovery in the animals which received this combined therapy. Differentiation of the MSC-TrkC into neuron-like and oligodendrocyte-like cells was also seen, with an expression of NF150 and MOSP almost twice as higher when compared to the control groups. Additionally, a reduction of degenerated myelin and an increase in the number of remyelinating and normal myelinated axons either in MSC combined with EA was observed.
## Clinical trials using genetically modified mscs
To date, genetically engineered MSCs are being tested only in few clinical studies for the treatment of different diseases. Either by viral or non-viral modification, the clinical use of genetically modified MSCs raises flags about the safety of cell products and quality control that should be carefully executed to guarantee the development of safe and effective alternative treatments.
Application of modified MSCs for genetic diseases treatments is being explored in the clinical research field in cases of severe combined immunodeficiency (SCID). The standard therapy for defective adenosine deaminase (ADA) derived SCID is bone marrow transplant. However, this procedure involves high risks and compatible donors are scarce. To overcome these problems, the collection of hematopoietic stem cells and/or MSCs, derived from the patient's bone marrow, followed by lentiviral transduction to induce correct defective ADA expression and reimplantation into the patient can be considered as a promising strategy for SCID treatment. Therefore, safety and efficiency of an improved self-inactivating lentiviral vector system for therapeutic gene delivery to patients with SCID, due to a defective ADA gene, is being studied in an interventional clinical trial (NCT03645460 ClinicalTrials.gov).
Another ongoing example of safety and efficacy evaluation of a gene transfer phase I/II clinical trial is for treating Fanconi anemia, a rare, inherited disease that is caused by a gene defect that primarily affects an individual's bone marrow, resulting in decreased production of blood cells. In this trial, autologous bone marrow derived MSCs and hematopoietic stem cells are being transduced with a self-inactivating lentiviral vector to functionally correct the defective gene FANCA. Furthermore, the effects of an infusion of these modified cells in order to promote immune reconstitution and long-term correction of Fanconi anemia associated disease symptoms is being evaluated (NCT03351868 ClinicalTrials.gov).
In the oncology field, a first clinical trial in solid tumors was conducted for treatment of advanced gastrointestinal cancer. Subjects were treated with a combination of genetically modified autologous bone marrow derived MSCs expressing the herpes simplex virus thymidine kinase (HSV-TK) (MSC_apceth_101) and ganciclovir, in order to generate a toxic metabolite for the tumor cells. In this study, from the 10 patients treated with this genetically modified MSC, five of them reached a stable disease and a post-study observation demonstrated a median overall survival of 15.6 months. In relation to the immunological markers, a slight increase in baseline levels of IL-6 was seen in 4 of the patients, while 3 demonstrated a slight enhancement in the levels of this cytokine at the end of the study, when compared to the baseline. Additionally, a moderate improvement in baseline levels of IL-8 was seen in 5 patients and 3 of them presented an increase during the time.
Also an enhancement in TNFα/IL-10 ratio, which suggests a higher inflammatory effectiveness and an anti-tumor capacity, was observed in 5 patients as well.
In addition, new anticancer immunotherapies are being developed based on the use of recombinant type I IFNs, type I IFN-encoding vectors and type I IFN-expressing cells. Based on that, a phase I study evaluating the effects of MSCs secreting IFN-β for ovarian cancer therapy is ongoing. The study aims to determine the highest tolerable dose of human MSCs-IFN-β that can be given to patients with ovarian cancer therapy (NCT02530047 ClinicalTrials.gov). The safety and antitumor activity of a modified human MSCs are also being studied in a phase I/II study investigating the outcomes of TRAIL-overexpressing MSCs in addition to chemotherapy, in metastatic non-small cell lung cancer (NCT03298763 ClinicalTrials.gov). In another study, the maximum tolerable dose, safety and efficacy of an intratumoral injection of the IL-12-expressing Human Mesenchymal Stem Cell Vaccine GX-051 are being investigated in subjects with advanced head and neck cancer. IL-12 is a cytokine that induces the production of IFN-γ, an important anti-tumoral factor. This study will evaluate the anti-tumor response, as well as possible changes of IFN-γ and IL-12 levels in blood comparing to the baseline and changes of immune cell distribution in tumor tissue after GX-051 intratumoral injection (NCT02079324 ClinicalTrials.gov).
# Conclusion
The development of genetic engineering technologies and the continuous improvement in the knowledge of MSCs gene expression enabled the use of genetically modified MSCs with enhanced therapeutic properties compared to wild-type MSCs in a variety of disease models, as discussed in this review. Moreover, many studies described here showed that genetic modification of MSCs improved their paracrine effects, protecting viable cells in a lesion area from further damage and promoting tissue regeneration and significant functional and behavioral recovery in many animal models. However, due to issues concerning the safety and efficacy regarding the use of genetically modified MSCs for treatments, there is a gap between the experimental models and clinical trials, and the few clinical studies were conducted or are being initiated, mainly in the hematology and oncology fields. Additional preclinical studies are needed to ensure the safety and demonstrate the therapeutic potential of engineered MSCs in more relevant animal models, especially those using medium or large animal models, giving support to the translation of this therapeutic modality into a broad variety of clinical settings.
# Author contributions
PD, TS, GS, IO, DS, JA, GGo, GGr, and MP participated in the literature search, wrote the manuscript parts, and prepared the figures and tables. RS, MD, and MS conceived the manuscript concept, wrote, and final edited the manuscript. All authors read and approved the final manuscript. |
Current methods for the assessment of skin microcirculation: Part 2
A b s t r a c tMicrocirculation accounts for about 99% of blood vessels in adults and mediates between the arterial and venous parts of the cardiovascular system, both structurally and functionally. Skin microcirculation consists of two vascular plexuses: superficial and deep. Microcirculation includes vessels with a diameter of less than 150 μm, i.e. arteries, small veins, lymphatic vessels and arteriovenous anastomoses, which build the microcirculation unit. Skin microcirculation may be affected both in systemic pathologies and specific skin disorders. Several non-invasive techniques are available to assess the skin microcirculation. Methods used in clinical practice were presented previously in Advances in Dermatology and Allergology. The current article describes methods that may be used in clinical research.
## Photoplethysmography
Photoplethysmography (PPG) measures volumetric changes in blood in peripheral microcirculation. PPG is simple, painless and inexpensive and it does not require much experience of the person who conducts the test. PPG allows only for a point assessment of microcirculation and the result of the test is not given in absolute units. PPG involves a probe equipped with a source of infrared light and a set of optic sensors. The infrared light is absorbed by the examined tissue. Depending on the volume of blood at the moment of the test in the examined skin fragment, more or less light is absorbed. In consequence, the amount of the light reflected relates to the point changes in the blood volume. Thus, the alterations in blood flow can be registered as alterations in light intensity. The measurement may be performed on fingers and toes on both hands and feet. The test is conducted in a silent room at > 20ºC temperature, allowing time for the end of the vascular response, connected to the locomotive muscle tension and heat adaptation.
The PPG shows alterations in blood flow in the form of a chart presenting point alterations in the volume of the flowing blood. Its waveform consists of two phases: the rapid anacrotic phase and the slow catacrotic phase . The anacrotic phase is primarily related to systole, while the catacrotic phase is related to diastole and reflexes from the periphery. Typically in the subjects with healthy artery compliance we see a dicrotic notch in the catacrotic phase. Reduction in microcirculation flow detected in the course of the photoplethysmography exam can show growth retardation, rounded top and a very slow decline without the dicrotic notch.
Photoplethysmography is currently the most commonly used method of diagnosis of chronic venous insufficiency. It allows for the assessment of the functioning of valves in the superficial and deep vein systems by the measurement of the venous refilling time (VRT). The photoplethysmographic test can also be applied in diagnostics of peripheral arterial occlusive disease. Allen et al. have demonstrated the clinical value of photoplethysmography pulse measurements on the toes of both feet in the diagnostics of the disease in the lower limbs. PPG may be also used in the diagnostics of alopecia areata (AA) in order to confirm the vascular aetiology of this disease. Sudnik has demonstrated alterations in the photoplethysmographic curve denoting mild flow distor-tions in 63.5% and severe flow distortions in 36.5% of AA patients.
## Orthogonal polarization spectroscopy
Orthogonal polarization spectroscopy (OPS) is a relatively new technique of human microcirculation imaging without using fluorescent dyes. OPS imaging shows a wide range of clinical applications with a strong emphasis on establishing the precise diagnosis, prognosis and treatment implementation. Polarized light with wavelength of 548 nm, which is reflected from the examined tissue, is captured by a camera and used to obtain the optimal microcirculation imaging in OPS. The light scattered by the superficial layer of tissue is blocked in the process of second polarization. As a consequence, only the light returning from the deeper layers of tissue passes second polarization and improves visibility of erythrocytes. Computer analysis of erythrocyte movement enables assessment of perfusion, diameter of vessels and capillary density.
The orthogonal polarization spectroscopy technique can be applied not only for real-time studying of microcirculation in skin vessels, but also for submucosal microcirculation. The test is usually conducted in the sublingual area and skin of newborns. Moreover, OPS technology allows for examination of tissues covered by a thin epithelial layer and internal organs during surgical operation. The obtained measurements have rather a quantitative character, however, OPS results are susceptible to artefacts due to organ movement and changes in blood pressure. De Backer et al. using the OPS technique have proved significant changes in skin microcirculation in patients with sepsis and septic shock. OPS proved to give a more objective visualization of microcirculation in critically ill patients compared to videocapillaroscopy and laser Doppler techniques. Orthogonal polarization spectroscopy has been also applied in monitoring of hypovolemic patients and in term and preterm infants as an effective non-invasive method of tissue perfusion assessment. Further development of the OPS is bound to bring a broader clinical application of this method, especially in the field of surgery, mainly neurosurgery and plastic surgery, as well as neonatology, anaesthesiology and dermatology.
## Near-infrared spectroscopy
Near-infrared spectroscopy (NIRS) is a specific form of tissue reflectance spectrophotometry. This method uses infrared light of 700-1000 nm which easily penetrates soft tissues and is subsequently partly diffused and partly absorbed by chromophores such as haemoglobin, myoglobin and cytochrome aa3. On this basis the tissue haemoglobin oxygen saturation (tHbO 2 ) assessment is performed. This technique may be used for assessment of various organs, mostly for the measurement of blood flow through the brain and muscle tissue. Advantages of this method are the simplicity of use and short time of examination. The devices for clinical measurement which are currently available are easy to operate. Unfortunately, NIRS does not allow for measurements of absolute values of tHbO 2 . In addition, the resolution of the signal is poor and the exact depth of light tissue penetration is not known. Originally this method was used for continuous monitoring of tissue haemoglobin oxygen saturation in the brain during carotid endarterectomy. NIRS technique is also currently recognized as a good tool for assessing superficial tissues, including skin microcirculation. NIRS is used primarily in critically ill patients for monitoring oxygen supply to the skeletal muscles in ICUs. Lima and Bakker proved that monitoring of circuit perfusion may show early indications of tissue hypoperfusion. The NIRS method has been reported to assess skin microcirculation in patients with sepsis and in septic shock, however its prognostic value has not been confirmed yet. Hartwig et al. in a pilot research successfully used this technique for the assessment of microcirculatory dysfunction in systemic sclerosis both in static and dynamic conditions. They also demonstrated clinical use of NIRS in assessment of skin microcirculation in connection with vascular occlusion testing. NIRS, similarly to laser Doppler flowmetry and transcutaneous oxygen tension, may be applied in the assessment of microcirculation in patients with diabetic foot. The assessment of microvasculature may help to make a therapeutic decision.
## Tissue reflectance spectrophotometry
Tissue reflectance spectrophotometry (TRS) is based on the detection of backscattered light in specific wavelength spectra for oxygenated haemoglobin at two peaks of 542 nm and 577 nm and deoxygenated haemoglobin at a peak of 556 nm. This technique allows for the assessment of haemoglobin oxygen saturation and haemoglobin concentration in capillaries so that it reflects skin microcirculation functioning in real time. Measurements may be taken from the surface of any organ, although for obvious reasons it is most commonly performed on the surface of the skin and gastric mucosa. A short time of examination, expression of the results in absolute units and the possibility to perform repetitive measurements are the main advantages of TRS. Limitations include the fact that the result depends on presence of tissue chromophores other than haemoglobin i.e. melanin or cytochromes. Erroneous results in measuring blood content and oxygenation might occur in persons of significant skin pigmentation as melanin in the epidermis reduces the amount of backscattered light and has a characteristic absorption spectrum, different in every person. Thus, TRS is presently applied predominantly for assessment of tissue healing.
## Optical coherence tomography
Optical coherence tomography (OCT) is a developing technique of imaging, mainly used in medical diagnostics. This method allows for non-invasive penetration into the examined tissue, which is why it is also called optical biopsy. Optical coherence tomography uses the phenomenon of interferometry of light with the tissues. In the OCT method, sources of light of special coherence and spectrum of considerable spectral width. Light used in this method has an intensity lesser than a couple of milliwatts. Thus, the OCT examination is entirely non-invasive and it may be safely applied repeatedly in many places. This technique is analogous to the ultrasonographic technique, with the use of light instead of sound. The most important advantage of this method is high resolution of 10-20 µm, allowing for obtaining images comparable to histopathology. Other advantages include the simplicity of use of the device, imaging in real time and capability of continuous registration. The OCT does not require prior preparation of a patient. Unfortunately, OCT is most effective for optically transparent tissue assessment. It covers a relatively small surface and is limited to imaging only 1-2 mm into the surface of the tissue.
The OCT is a constantly developing method and is applicable in many fields of medicine. In ophthalmology it is regarded as the most modern technique of imaging in diseases of retina and anterior segment of the eye. In pa- Works best for optically transparent tissues, it covers a relatively little surface and is limited to imaging only 1-2 mm into the surface of the tissue tients with diabetic retinopathy, OCT is used as an objective monitoring technique of the macular thickening before and after therapy as well as for vitreous assessment. In interventional cardiology it is applied in the coronary artery disease diagnostics. In dermatology, OCT is used to assess the extent of tissue damage, to measure pathologic skin lesions such as angioma and for precise assessment of scars and wrinkles. Comparison of correlation mapping optical coherence tomography (cmOCT) performed on the forearm of healthy persons and psoriasis patients demonstrates characteristic microcirculatory alterations in the presence of psoriasis. |
Femoral anteversion: significance and measurement
## | b iomechanic al s ig nific an ce of fn a
A change in FNA affects the position of the trochanter and therefore the line of action of the muscles surrounding that region.
Regional torsional changes along the femur also result in a change of lever arms. A higher FNA results in a slightly shorter hip extension moment arm and an increase in hip flexion moment arm of the abductor muscles. Furthermore, high FNA results in a shorter abductor lever armand also considerably increases internal rotation moment length by an average of 96.5% for all hip muscles , apart from the iliopsoas, which was not evaluated, and the gluteus maximus anterior, which decreases internal rotation moment arm length by 86%. A higher FNA also affects muscle activation, as lower gluteus medius and vastus medialis activity has been recorded during isometric hip abductionprobably due to the change in moment arm length. The higher FNA, and therefore the shorter abductor lever arm, also changes the mechanics of the hip joint resulting in up to 24% higher hip contact forces during gait with an anteversion of 30° and 8% higher forces with FNA of 14°, when compared with an anteversion of −2°. On the other hand, reduced FNA results in higher shear forces on the femoral neck-head junction, quantifiable as a 42% increase with an FNA of 0° and 86%
with an FNA of −12.5° and 12.5°, respectively.
Distally, increased FNA is associated with a progressive increase in patellofemoral contact pressures.
## | con s equen ce s of altered fna for he alth
Altered movement associated with differences in FNA also appears to have consequences for musculoskeletal health. The shorter hip extension moment arm and longer moment arm for hip flexion shown with increased FNA are consistent with the gait pattern in individuals with cerebral palsy, and the shorter abductor and adductor lever arms are likely to produce pelvic instability during gait. The increased internal rotation moment arm length, combined with a decreased external rotation moment arm length, is likely to be part of the cause of in-toeing gait in children with cerebral palsyas a strategy to increase the abductor lever arm during movement. Self-adjustment of in-toeing gait is often accompanied by the compensatory external rotation of the tibia. Greater FNA is also associated with a number of orthopaedic pathologies, including increased risk of anterior cruciate ligament injury, which might be related to altered knee kinematics during landing, lower hip abductor and vastus medialis activity, impaired tracking of the patella and femoral trochlear dysplasia. Greater hip load and the altered relationship with the acetabulum--resulting from increased FNA--may play a role in the genesis of osteoarthritis. This suggestion is reinforced by a prevalence of unilateral osteoarthritis in limbs with higher FNA. F I G U R E 1 Axial schematic representation of the right femur and of the femoral neck anteversion (FNA). The grey area represents the femoral neck and the white area represents the distal condylar region. FromThe decreased congruity could also result in hip dysplasia, a condition that displays FNA averages of 6°-18° above normal, whereas hip congruityand loadingmight be a contributors to femoral acetabular impingement. On the other hand, the aforementioned increased shear forces occurring with reduced FNA could explain the association of slipped capital femoral epiphysis with populations which have low FNA. Not only is increased or decreased FNA a risk factor for clinical conditions, but asymmetries in FNA also appear to influence musculoskeletal health, as shown by. In this study, patients with unilateral osteoarthritis of the hip with higher anteversion reported lower back pain, whereas unilateral osteoarthritic subjects with symmetrical FNA did not. FNA has been shown to affect the accuracy of clinically relevant bone mineral density measures. However, little is known about whether FNA and altered biomechanics could also affect bone mineral density, other bone strength indicators or the risk of femoral fractures through these factors or by altered fall mechanics.
## | epidemi ology
Normative data for FNA in the healthy adult population is highly dependent on the landmarks identified and imaging technique used, with mean values in the range of 7°-24°. In addition, there is substantial variation within the population, with individual values ranging by more than 30°, independent of the method used. FNA is at least partly hereditary, with a polygenetic influence on this and other features of proximal femur shape.
However, another key factor is the influence of mechanical loading during everyday movements and exercise. Both the greater trochanteric and the epiphyseal growth plate are accountable for shaping the proximal femur. Bone growth has been shown to be directed perpendicularly to the direction of the growth plate, which is orientated in line with the forces acting on it. The growth rate of growth plate cartilage is influenced by mechanical loading, such that increased compressive and tensile loading increases growth rate up to a point, with additional loading leading to reduced growth rate and potential damage.
This effect of mechanical loading likely contributes to the dramatic changes in FNA observed throughout prenatal development and childhood. An increase of around 30° in FNA during fetal life has been observed , particularly during the second trimester. In the womb, the hip has a high angle of flexion and F I G U R E 2 Effects of altered femoral neck anteversion (FNA) on reference moment arm length (MAL), calculated as the difference between subject-specific model and a general model. The subject-specific model is taken as the reference and therefore negative values indicate a higher value in subject-specific models and vice versa. The subject-specific model is an average of subjects with a high FNA (ranging from 25 to 51). The moment arm length is averaged over the whole range of motion (10° extension 90° flexion, 50° abduction, 20° adduction, 40° external and 40° internal rotation). Only the main function is recorded for every muscle and the internal/external rotation. Conventionally positive values are used for flexion abduction and internal rotation. Figure created using data from. GMaA, gluteus maximus anterior; GMaM; gluteus maximus medialis; GMaP Gluteus maximus posterior; GMeA, gluteus medius anterior; GMeM, gluteus medius medialis; GMeP, Gluteus medius posterior; GMiA, gluteus minimus anterior; GMiM, Gluteus minimus posterior; AdL, adductor longus; AdB, adductor brevis; AdMS, adductor magnus superior; AdMM, adductor magnus middle; AdMP, adductor magnus inferior; TFL, tensor fascia latae; Gra, gracilis; SemiM, semimembranosus; SemiT, semitendinosus; BiFem, biceps femoris long head; Sar, sartorius; RF, rectus femoris. Asterisks denote statistical significance:: *p < .05 and **p < .01 fetal life could also result in increased anteversion, and the opposite is true for external rotation. This was confirmed in animal studies with forced internal rotationand by the 10° higher FNA found in children born with breech presentation; these children often have an internally rotated position in the womb resulting in reduced kicking forces and lower femoral stress and strain during fetal movements. During childhood, a steady ~1.5° a year decrease in anteversion until completion of growth has been recorded . This decrease during growth might depend on the action of hip muscles during gait, which may shape the FNAand keep the resultant forces during the maximal weight-bearing period perpendicular to the growth plate.
There is some evidence from large cohort cross-sectional studies to suggest that FNA also decreases, at a lower rate, during adulthood. This raises the question of whether 50-70 years ago children were more active, and whether we are observing secular rather than within-individual changes. Furthermore, a recent longitudinal study in individuals with hip osteoarthritis suggests that FNA decreases with time over a period of 3 years. This might be due to localised addition of bone on the periosteal surface with increasing age, microfractures or the bony erosion due to the osteoarthritic condition. Most studies show a higher FNA in the female population with sex differences ranging from 2° to 8°. It is known that growth plate fusion occurs at an earlier age in women than men, therefore a shorter growth period could be a cause of the higher FNA in women.
Although FNA values reported in individuals from different ethnic F I G U R E 3 Epiphyseal growth plate: Radiography and computer tomography (CT) of cadaveric proximal femur of 13-year-old individual. Coronal view on top panels, axial view in bottom panelsF I G U R E 4 Femoral neck anteversion (FNA) means and standard deviation of fetuses at different stages of gestation. Bottom panel shows photos of typical fetal femur samples at different developmental stages (12 weeks to term). Figures adapted fromF I G U R E 5 Mean values and normal/pathological limits of femoral neck anteversion (described as antetorsion or AT angle) in children of different ages as measured by different investigators groups have differed substantially, this may relate to the use of different measurement methods. More recent CT studies have found no differences in FNA based on ethnicity.
The importance of mechanical loading for FNA during development is also evident from altered values in children with compromised motor development and movement. Notably, children with CP do not show a decrease in FNA during development . Typically, FNA is around 10°
higher in older children with CP than unaffected children, with similar differences evident between the affected and unaffected limbs in children with hemiparetic CP. This is thought to be caused by spasticity or decreased activation of certain muscle groups, which is frequent in the clinical spectrum of CP subjects. In particular, it was suggested that increased activity of adductor and extensor muscles predicts higher FNA, as does reduced activity of hip flexors. This was confirmed in animal studies resecting either internal or external rotator muscles.
Interestingly, ambulant children with CP have higher FNA than non-walking children with CP. This suggests that the alterations in muscle activity and subsequent joint loading during gait in children with CP contribute to development of FNA and therefore that healthy motor development is an important factor in development of the proximal part of the femur. This is supported by reports suggesting that differences in FNA between children with CP and normally developing children emerge at around 12 months, the typical onset of independent walking.
FNA has also been reported to differ from typical values in children with other conditions affecting neuromuscular development, such as Down syndrome, with an average of 33°and Charcot-Marie-Tooth disease, where the mean FNA is 28°. In addition, higher values are observed in children with a range of disorders affecting skeletal development.
For example, Blount's disease causing bowing of the tibia, Legg-Calvé-Perthes disease resulting in avascular necrosis of the femoral headand achondroplasia , which results in substantially reduced limb length. On the other hand, obesity in adolescence is associated with an FNA of only 0.4° ± 13°. This could be due to the increased muscular forces required to move a greater body mass during development..
## | tre atment
Idiopathic altered anteversion in early childhood usually corrects itself without intervention. In cases where increased FNA does not correct itself and results in in-toeing and tripping, the most effective method to change FNA is femoral de-rotational osteotomy. Clinical considerations such as indications, imaging, surgical techniques and associated results, and anatomical considerations have been reviewed in detail byand are discussed only briefly here. De-rotational osteotomy can be performed either at a distal supracondylar levelor at a proximal sub-trochanteric or intertrochanteric level. The contribution of these different regions to total anteversion differs within and between clinical groups. Therefore it has been suggested that the planning of osteotomies should take into account this segmental variation in order to ensure healthy postoperative hip biomechanics and prevent further clinical problems. De-rotational osteotomy has been shown to be a successful technique in the treatment of in-toeing children with cerebral palsyand patellar instability. However, it must be taken into account that complications might arise and the whole recovery process could be a traumatic experience. Therefore, surgery is suggested only in disabling or symptomatic instances, only after the age of 10 and, depending on the condition, in cases of a measured anteversion above 20°-50°and internal rotation higher than 80°. Non-operative methods to lower FNA such as shoe wedges, twister cables and night splints have been proposed but do not appear to be effective. The effects of movement and motor development on FNA described earlier suggest that physical therapies and targeted exercises may alter FNA during growth, but to our knowledge this remains unexplored.
## | femor al a xe s
There is evidence that femoral torsion occurs throughout the femoral shaft, below the lesser trochanter, and at the intertrochanteric level. As well as variation in clinical cases identified above, substantial variation in torsion within each of these regions has also been identified in non-clinical populations F I G U R E 6 Femoral neck anteversion (FNA) in children with cerebral palsy (CP) and typically developing controls during growth, presented as mean and standard deviation. Adapted from. FNA is considered the "total" femoral torsion.
The definition of FNA and the chosen femoral axes determine the measurement. The femoral axes are defined as follows: the neck shaft axis, the femoral shaft axis and the condylar axis. The shape of the proximal part of the femur is complex, as the lateral part of the femoral neck is elliptical and its major axis tilts anteriorly, and the femoral head is not usually centred on the femoral shaft. The femoral neck axis can be defined as the line connecting the femoral head centre to the femoral shaft axis, the line going through the centre of the femoral head to the narrowest part of the neck, the centre of the femoral neck, the centre of the greater trochanteror the edge of the greater trochanter, the latter being referred to as a functional axis, as it takes into account the contact point of the hip and the insertion of the adductor muscles. The femoral neck axis could also be defined as the line parallel to the femoral neck, without taking into account the trochanter or the femoral head . These reconstructions are available in single cross-sections of the femoral neck or on two different slices, further increasing differences even with similar definitions. For 3D models, the femoral neck axis may either be determined using a line of best fit of the centroids of the slices defining the neck as identified F I G U R E 7 Top: Examples of different methods of femoral neck anteversion (FNA) assessment and how they affect the assessed geometry: A, B, C, D, E are transverse slice methods, and F and G use oblique slices. The location of the slices in the coronal plan is shown in the lower panel. H shows that the posterior condylar line was taken as reference for all methods. Figure from. Below: left, the proximal and distal part of the femur is superimposed in this picture. The lines through the neck depict different neck axes and table top condylar axes looking along the shaft axis: 'Neck' refers to the Berryman method, a semiautomatic method taking into account the femoral head centre, the base of the femoral neck and the cluster of points of the neck. The Lee 2D method uses a straight line connecting the femoral head centre and the most cephalic junction of the greater trochanter on one axial slice. The Reikerasmethod uses a line connecting the centre of the femoral head on one slice and centre of the femoral neck on the slice that has the posterior and anterior edges of the neck running parallel. Murphyuses a line connecting the femoral head centre on one axial slice and the centre of the base of the neck on another axial slices. Figure from. Right: column 1 axial slice cranial, column 2 axial slice through the neck centre, column 3 axial slice through the base of the neck with little head left. Row A neck axis defined as centre of femoral head and centre of femoral neck. Row B neck axis defined as line connecting the two centres of the width of the neck; row C above I is the line connecting the femoral head and the greater trochanter lateral edge, and the line below is the anterior border of the femoral neck as in ultrasound methods
[formula] (a) (b) (c) (d) (e) (f) (g) (h) [/formula]
by handor using principal component analysis on a point cloud covering the femoral neck determined semi-automatically.
The anatomical femoral shaft axis is difficult to define accurately, due to the anterior curvature of the shaft. We can find its distal point at the anterolateral border of the posterior cruciate ligament, at the centre of the medial and lateral articular margins, at the midpoint of the centroids of each condyle, at the centre of the segment joining the two midpoints of the line connecting the anterior and posterior points of each condyle, at the centroid of an axial cross-section of the femoral condyleor at the most proximal aspect of the intercondylar fossa. The proximal point of the femoral axis has been defined using the centroid of the slice taken between the lesser trochanter and the greater trochanter, under the lesser trochanteror at the head centre if the functional weight-bearing axis is used as the rotational axis. The femoral shaft axis in most cases is used as the rotation axis of the femur, as a fixed point to determine the neck axisor as a reference for femoral rotation. In 3D models, it is possible to establish the anatomical femoral axis via the interpolation of the centroids of multiple slices taken along the femoral shaft.
The distal femoral axishas been defined in various ways: as the posterior edge of the lateral and the medial condyle (posterior condylar line), the midline between the anterior condylar line and the posterior condylar line, or as a line connecting the peak of the epicondyles on a transverse view (epicondylar line). The posterior condylar line has been shown to be the least location-dependent and the most repeatable method to define the distal femoral axis among those outlined. However, the most relevant axis from a biomechanical perspective remains to be determined. The functional distal axis has been proposed to lie along the epicondylar line, alternatively a variation of this axis using the sulcus of the medial epicondyles has been defined as the logical reference for rotation of the femoral component in knee arthroplasty.
## | me a surement of fna
## | computed tomography (ct)
Computed tomography can be used to acquire images of single cross-sections or volumes of bone. Because the X-ray attenuation is very different between bone and soft tissue, CT provides a sharp contrast between bone and soft tissue and is therefore good in depicting mature, well-ossified bones. CT scanning times per slice are lower than 2 s and for 3D-rendering of hip structures, which requires multiple slices, this can go up to 40 s.
CT is considered cheap compared with MRI due to shorter scanning time. CT results in ionising radiation exposure of 0.3-0.5 mSv for six slices in adults; this dose would be double in neonates.
Single axial slices using CT can be used to define the neck axis. However, the heightand
angle of slices used for the 3D reconstructionaffect the final result, as the femoral neck has an asymmetrical shape.
Alternatively, two axial slices can be taken to define the femoral neck axis, one at the femoral head and the other at different heights of the distal femoral neck. Where scans are not aligned with the femoral shaft axis, calculation may be needed to transform the torsion to a reference plane, as positioning-related measurement errors can be as large as 8.8°. After 3D rendering of the proximal femur it is possible to take an oblique slice along the femoral neck angle in the coronal planeor compute the femoral neck axis in 3D. The distal femoral axis can be evaluated with one single axial slice with reference to one of the axes described earlier.. Method C identifies the centroids of the medial and lateral condyles. Method D bisects the angle formed by the posterior and anterior condylar lines. Figure from
## | magnetic resonance imaging (mri)
Magnetic resonance imaging obtains similar features of the transverse femoral cross-section to CT. MRI uses strong magnetic fields and radio waves to exploit paramagnetic properties, mostly of freely movable protons, to generate images and is therefore free of ionising radiation. Different from the bone approach with CT, which measures the presence of bone apatite, MRI measures the absence of freely movable protons in bone. Moreover, the MRI magnetic field often limits the application to subjects who do not have metal implants, pacemakers or other contraindications. In addition, undergoing an MRI measure of the femur means remaining still within a narrow, confining tube for between 5and 20 min, depending on the sequence type, or up to 45 min where 3D rendering is required, which limits its application in young children without sedation. Furthermore, MRI is usually expensive and not available in all research and clinical facilities.
It has been suggested that MRI can be superior to CT in depicting the proximal and distal femoral contours in children with immature bones. The orientation of scan slices is virtually equally achievable with CT and MRIalthough alignment of CT scans requires post-scan reconstruction. In contrast, MRI scans can be directly aligned to anatomical features such as the femoral neck, thereby allowing the depiction of the whole region. This is particularly useful at high femoral neck inclination angles, where positioning bias is greatest. The distal femoral axis can be evaluated with one single axial slice with reference to one of the axes described earlier.
## | ultrasound imaging (us)
Ultrasound imaging with most clinical scanners only enables two- Some methods place the probe horizontally and measure the inclination on the image on the screen or later on the printed imagebut results are not consistent at high angles of anteversionwhere the distal part of the femoral neck becomes deeper and harder to image. To adjust for this issue, others use inclinometers mounted on the probeand take the measurement when the chosen features are showing horizontally on the screen, or use additional hardware to place the femur in internal rotation. Free-hand US couples the ultrasound with video or motion capture localisers.
Several features have been used to determine the proximal femur axis: the head-trochanter tangent, the femoral neck, and the intertrochanteric plane .
However, taking only the anterior border into account means that inter-individual differences in the angle between the anterior border and centre of the femoral neck are ignored. The features used to draw the inter-condyle axis are the posterior condyles, the epicondyles, the epicondyles, and the anterior condylesor the anterior condyles only. The posterior condylar line can also be inferred using the tibia as a perpendicular reference, bearing in mind that varus or valgus knee deformation could affect the end result.
## | radiography
The images yielded from radiography are projections of the bone structures in the space between the generator and the detector. The detection of the femoral neck axis can be done by single-plane radiographs giving a 2D image, biplanar radiography allowing two different projectionsor a 3D computer reconstruction. The actual image acquisition takes seconds, whereas positioning depends on the participant but is usually 1-2 min. Analysis of single-plane images involves drawing a line through the femoral neck axis and is therefore quick. In biplanar and 3D models, the postprocessing can take from 2 min for approaches where only gross features are identified, to 20 min for techniques requiring the identification of a large number of landmarks and evaluation of the whole lower limb. In the biplanar method, lines are drawn through the chosen landmarks and then converted using trigonometric conversion tables. In 3D reconstructions, the images are evaluated by the software. The recently introduced EOS imaging technique (a lowdose radiography system) semi-automatically draws the silhouette of the bone and 3D models of the lower limb.
Furthermore, radiography is cheap and available in most clinical facilities. The downside of the method is that it uses ionising radiation, reported to be in the range of 0.25 mSV for low-dose radiography to 7.5 mSV for regular radiography.
Biplanar methods involve an initial anteroposterior radiograph in a supine, proneor standing position. The second image is taken with the hip flexed at 90,° with various degrees of abduction for each method, with the detector parallel to the femoral neck inclinationor standing. In most methods, the proximal axis is taken as parallel to the femoral neckbut in others the femoral head-trochanter lineor head-neck centre is assessed. The distal axis can be defined using the posterior condylar line; alternatively, the tibia is considered perpendicular to the condylar line.
## | functional assessment
It is reported to be possible to measure the FNA without any imaging methods by assessing the angular range of motion (ROM) of the hip joint in the axial plane. FNA can be assessed using the ratio of internal rotation over external rotation, with greater internal rotation associated with greater FNA. In this case, the assumption is that the end of the internal and external rotation gives information about where the femoral head stops gliding in the acetabulum and the femoral neck prevents further rotation by touching the contour of the acetabulum. Another way of measuring FNA is by measuring the angle of rotation at the point where the greater trochanter feels most prominent, via palpation of the lateral hip. In this case, the assumption is that when the trochanter is most lateral during the rotation of the femur, the femoral neck is parallel to the floor and the angle of the tibia will indicate the FNA. This method is called the trochanteric prominence angle test (TPAT), or Craig's test. The downside, however, is the accuracy and precision of these functional methods. These are cheap and convenient, requiring only a goniometer or camera to take measurements. These methods are indirect indicators of femoral version and are dependent on both capsular and muscular restraints, as well as acetabular version.
The measurement of the rotation has been performed with both an extended and a flexed hip, and the degree of flexion can affect the measured FNA by 15°. A hip flexion of 45° was used to yield intermediate results between 0° and 90° of flexion. It has been suggested that performing functional tests in extended and in flexed positions will give additional F I G U R E 9 Left panel adapted from. Frontal view of different ultrasound approaches: the head-trochanter line approach features the peak of the red 'head' section BD and the peak of the red 'trochanter' section AC. The femoral neck approach assesses the region parallel to the blue CD section. The intertrochanteric plane approach assesses the bone parallel to the yellow GE line. information on capsular restraint and acetabular involvement. In all of these methods, the tibia is taken as the perpendicular reference of the posterior condyles.
## | differences between methods
FNA measurements are dependent on the imaging technique and the landmarks used (as listed in with differences in mean values between methods of up to 10°, which can increase to 20° when people with exaggerated anteversion are tested. In a study comparing repeatability of measurements using different landmarks on the same CT scans, mean intra-observer error was between 0.8° and 2.9° for the methods of Waidelich, Jarret, Yoshioka, Murphy and Hernandez, with a range up to 11.4° in Hernandez's method.
Interobserver repeatability is excellent for both oblique (interclass correlation ICC 0.95)and axial CT techniques (ICC 0.87, 0.93-0.96).
Inter-and intra-operator reliability of MRI in the measurement of FNA has been shown to be comparable to CT (ICC 0.90-0.97). A number of studies have compared CT and MRI measures with different methodsand, in some cases, different landmarksor even different samples. However, studies comparing the same methods, meaning the same slice height, orientation and landmark choice, with either MRI or CT show a systematic lower result of 8.9° in MRI with differences up to 37° left)or up to 13.6° variation between measures in another study , right). It has been suggested that the reason for this systematic error might be the long scanning time of the MRI, which causes the subject to relax and change positions between the scanning of the proximal and distal femur. Alternatively, differences in the appearance of bone tissue between MRI and CT may lead to differences in the positioning of identified landmarks.
Comparisons of US with different imaging techniques such as radiography, MRI and CT have shown mean differences of FNA of 0.5° to more than 10°. US has been shown to have lower inter-and intra-observer reliability than MRI or CT for the 2D methodsbut appears more reliable in dried bones. This may be due to the ability to detect landmarks directly on the dried bone visually, rather than relating them to images on screen through the soft tissue.
US has been considered a good screening method for FNA because of lower inter-and intra-reliability or only moderate correlation (r = .57-.87) with other imaging techniqueseven though these discrepancies also result from errors in the compared methods. US results are closer to MRI than to CT, which may be because the landmarks can be obtained in line with the inclination of the femoral neck. However, the freehand US methods appear more repeatable compared with regular US, with average errors as low as 1.8°and MRI and ICC of 0.95.
We can find a total average difference for biplanar radiography compared with dried femoral measurements of 2.6°-3.6°
with good correlation (r = .82-.91). However, in living subjects, the error could be as high as 20° due to positioning errors. The EOS system yields total average differences compared to CT of 0°-5°. Inter-reader agreement is high for the lowdose EOS system (ICC 0.95) with an average difference of 0.1°and 3.4°. Error sources might include inaccurate positioning due to physical impairments (such F I G U R E 11 Current methods to measure femoral neck anteversion (FNA), with short explanation of landmarks used as spasticity, pain, skeletal deformities or obesity) and inaccurate location of the axes on the roentgenogrambecause of lack of clear guidelines. Additionally, soft tissue can obscure the bone outline, making detection of the bony structure more difficult in obese populations.have shown that even though the total mean difference of functional methods is below 5°, in 45% of the sample the error is more than 10° compared with CT. However, an internal rotation test or TPAT is good at predicting a low, normal or high anteversion. Correlations with CT methods are highly variable, with regression coefficients ranging from less than 0.25 to 0.79for the internal rotation test, and from 0.12to 0.86for TPAT. The reliability of the method is good: the interobserver ICC value for internal ROM is 0.89 and for the TPAT it is 0.81.
## | con clus ions
Abnormal FNA changes the biomechanics of the hip, altering muscular lever arms, hip contact forces and femoral neck shear forces, which may contribute to development of a wide range of skeletal disorders, such as osteoarthritis and alter the kinematics of the lower limbs. FNA grows in line with the growth plate, which seems to adjust according to mechanical forces acting on the proximal femur during movement, increasing the FNA during gestation to more than 30° and thereafter decreasing steadily until completion of growth.
This decrease is less pronounced or absent in individuals with conditions causing neuromuscular and movement impairment such as cerebral palsy. Interestingly, FNA in older adults is consistently lower than in younger adults, raising the question of mechanisms behind the modelling of the femur at a mature age following growth plate closure. Treatment for altered FNA is usually de-rotational osteotomy, suggested only in the case of disabling conditions; passive, non-operative methods such as braces or wearable cables do not have any effect. Despite observational evidence for the effects of muscular activity on FNA development during growth, the efficacy of targeted physical activity remains unexplored. Large variations between methods evaluating the FNA limit the ability to synthesise the large number of studies on the topic, as normative values must be set, relative to the method. It is not possible to draw conclusions on the choice of which femoral axis to utilise, and further studies are needed to determine the relevant forces shaping it. However, the authors endorse identification of landmarks which consider the femoral head and trochanter part of the femoral neck axis. Further studies are needed to explore the distal axis and whether the usual posterior condylar axis is the most relevant one. As for the imaging technique, it is situation-driven, with MRI and CT giving the best images and measurement precision; MRI is superior to CT in terms of radiation hazard and CT is quicker and cheaper. The biplanar radiography EOS system seems to be a quick, low-radiation option but it is not as reliable as the aforementioned approaches. However, both within clinical and basic science research, limitations of these methods prevent their broader application in healthy children and population studies.
New US techniques already permit a 3D depiction of the femur with systems of probe localisation. Further improvements in US imaging and data analysis could provide a cheap, quick, non-invasive and more broadly applicable alternative to MRI and CT.
## Co n fli c t o f i nte r e s t
Matteo Scorcelletti, Neil Reeves, Jörn Rittweger and Alex Ireland declare that they have no conflict of interest.
## Auth o r co ntr i b uti o n s
|
Multiplexed protein profiling by sequential affinity capture
Antibody microarrays enable parallelized and miniaturized analysis of clinical samples, and have proven to provide novel insights for the analysis of different proteomes. However, there are concerns that the performance of such direct labeling and single antibody assays are prone to off-target binding due to the sample context. To improve selectivity and sensitivity while maintaining the possibility to conduct multiplexed protein profiling, we developed a multiplexed and semi-automated sequential capture assay. This novel bead-based procedure encompasses a first antigen capture, labeling of captured protein targets on magnetic particles, combinatorial target elution and a read-out by a secondary capture bead array. We demonstrate in a proof-ofconcept setting that target detection via two sequential affinity interactions reduced off-target contribution, while lowered background and noise levels, improved correlation to clinical values compared to single binder assays. We also compared sensitivity levels with single binder and classical sandwich assays, explored the possibility for DNA-based signal amplification, and demonstrate the applicability of the dual capture bead-based antibody microarray for biomarker analysis. Hence, the described concept enhances the possibilities for antibody array assays to be utilized for protein profiling in body fluids and beyond.
conduct advanced protein profiling in body fluid samples such as serum, plasma, cerebrospinal fluid or urine [bib_ref] Systematic antibody and antigen-based proteomic profiling with microarrays, Ayoglu [/bib_ref].
The discovery-driven antibody array format uses single antibody assays for highly parallelized multiplex analysis and offers a very efficient strategy for antibody-based discoveries in large sample collections [bib_ref] Affinity proteomic profiling of plasma, cerebrospinal fluid, and brain tissue within multiple..., Byström [/bib_ref] [bib_ref] Affinity proteomics within rare diseases: a BIO-NMD study for blood biomarkers of..., Ayoglu [/bib_ref] [bib_ref] Affinity proteomics reveals elevated muscle proteins in plasma of children with cerebral..., Bachmann [/bib_ref]. Single antibody assays consume little amount of reagents and samples but do not match sensitivity and selectivity of preferentially used sandwich immunoassays (Supporting Information [fig_ref] Figure 1: The concept of the dual capture assay [/fig_ref] due to increased background noise and off-target interference of higher abundant proteins [bib_ref] Opportunities for sensitive plasma proteome analysis, Landegren [/bib_ref]. The latter type dual binder assays may though be often limited in flexibility to accommodate and combine assays for many targets as compared to single binder assays with labeled samples [bib_ref] Cross-reactivity in antibody microarrays and multiplexed sandwich assays: shedding light on the..., Juncker [/bib_ref]. Analysis via directly labeled specimen and single antibody interactions remains therefore attractive as a discovery tool, even though such assays may suffer from off-target binding to proteins of higher abundance or epitope similarity. Furthermore, certainty about the validity of the single binder assay data cannot be judged before additional assays using the same sample material and enrichment principles, such as sandwich immunoassays [bib_ref] Analysis of plasma from prostate cancer patients links decreased Carnosine Dipeptidase 1..., Qundos [/bib_ref] or MS readout [bib_ref] Selectivity analysis of single binder assays used in plasma protein profiling, Neiman [/bib_ref] , are applied.
In order to advance current single binder assay designs, off-target binding must be reduced to improve assay selectivity and ultimately sensitivity. To address this, we developed an assay concept, which is based on two sequential affinity binding events using two complementary single binder arrays. The principle of this dual capture assay (DCA) is illustrated in [fig_ref] Figure 1: The concept of the dual capture assay [/fig_ref]. A similar approach was previously described for single analyte assays utilizing microtiter plates as solid support [bib_ref] An immunoassay method for quantitative detection of proteins using single antibodies, Zhou [/bib_ref]. Our proof-of-concept study expands on such efforts and utilizes the flexibility offered by magnetic beads as solid support for multiplexed enrichment and read-out, as well as a combinatorial high and low pH elution. The presented work was built on more than 50 antibodies that are parts of commercially available sandwich assay kits in order to allow comparisons between sandwich, dual-and single-binder assays (Supporting . These sandwich assay antibodies were not validated for any single or DCA use previously.
The DCA protocol is conducted as follows: for the first affinity capture, a neat sample is incubated with the first antibody bead array; beads are then washed and subjected to labeling at a 300-fold molar access of biotin. Thus, captured proteins and the immobilized antibodies are modified, available for later read-out and also QC purposes that determine if captured proteins (and antibodies on beads) have been labeled accordingly. Several conditions such as the effect of detergent during biotin-labeling step (Supporting Information and pH levels for releasing antibody-bound proteins were evaluated (Supporting Information Figs. S3 and S4). This led to subsequent release of proteins in a series of first high and then low pH elution in a small sample volume (15 L), followed by heat treatment (56ЊC). The two eluates were combined and neutralized at pH 7 upon mixing in a TRIS-based assay buffer. For detection, a second bead array is used to capture those labeled target proteins remaining in the neutralized eluate. Hence, a less complex sample environment has been created for read-out analysis by enriching proteins from a neat serum or plasma sample, thus reducing present quantities of both on-and off-target and allowing for an improved assay selectivity -B, Supporting Information .
The DCA concept was further tested in various constellations. At first, we evaluated how the number of beads used in first capture influenced the degree of assay selectivity (Supporting Information . We then tested the use of only a single antibody during first capture to assess the degree of off-target interaction and the potential to measure released proteins by the secondary array. Next, the sensitivity of the assay was compared to both single capture , Supporting Information [fig_ref] Figure 1: The concept of the dual capture assay [/fig_ref] as well as to sandwich assays (Supporting Information [fig_ref] Figure 1: The concept of the dual capture assay [/fig_ref]. When compared to single capture assays, the background observed . For the first affinity capture, a neat sample is incubated with the first set of antibody-coupled beads. The beads are subsequently washed to remove unbound proteins and subjected to labeling with biotin. Captured and labeled target proteins are then eluted first by high pH, followed by low pH treatment, both in a small sample volume. The two eluates are then neutralized to pH 7 and a second set of beads is added to capture and detect the remaining target proteins and analyze the bead array in a flow-cytometer. In comparison to this DCA, single capture assays rely on direct capture of targets biotinylated in solution and subjected to readout. (For simplicity, an ideal assay with no off-target interactions is illustrated.) by antibodies to which no target had been spiked into the sample was low (median fluorescence intensity, MFI < 50) over several sample concentrations and generally not affected by sample complexity www.proteomics-journal.com . Performance comparison on selectivity and sensitivity. (A-B) A ten-fold dilution series of carbonic anhydrase III protein spiked into 0.6 mg/mL BSA was analyzed with a 52-plex antibody array (Supporting Information both in a single-and dual-capture assay format. The yaxes display the MFI values obtained for a subset of the antibodies including the anti-carbonic anhydrase III antibody (red). (C-D) Similarly, a ten-fold dilution series of PSA was spiked into 0.6 mg/mL BSA and analyzed with a 52-plex antibody array both in a single-and dual-capture assay format. The y-axes display the MFI values obtained for the anti-PSA antibody. (E-F) The bacterial protein cholera toxin subunit B was spiked into a 1:30 diluted human plasma or serum sample and analyzed with a 52-plex antibody array both in a single-and dual-capture assay format. The yaxes display the MFI values obtained for the anti-cholera toxin subunit B antibody. In each subfigure, the x-axes display concentrations of the investigated proteins in ng/mL. All measurements were performed in triplicates. Error bars indicate SD. LODs were calculated using a 5-parametric model and the obtained levels for each analysis are indicated as insets.
As described above, DCA showed improved selectivity and low background levels. To make use of the observations and to enhance sensitivity, we applied and evaluated an on-bead signal amplification protocol using rolling-circle amplification for multiplexed protein detection (Supporting Information . As shown in , there was a high concordance between R-phycoerythrin conjugated streptavidin (SAPE) and rolling circle amplification (RCA)-based readout (Pearson's r > 0.9). The average CV obtained from triplicates of a plasma dilution series was slightly higher with RCA detection (CV = 9.8%) compared to SAPE (CV = 4.6%) (Supporting Information , yet the RCA derived background levels (MFI = 9) were lower than for SAPE readout (MFI = 15) when assessing this using beads that were prepared using an antibody-free coupling solution. This proof-of-concept test demonstrated the feasibility of using RCA on beads for DCA analysis but follow-up studies would be needed to further assess the performance characteristics of RCA by using more antibodies.
Finally, we challenged the DCA concept in a biomarker analysis setup by investigating two different plasma sample sets collected in the context of prostate cancer (Supporting [bib_ref] Analysis of plasma from prostate cancer patients links decreased Carnosine Dipeptidase 1..., Qundos [/bib_ref] [bib_ref] Toward next generation plasma profiling via heat-induced epitope retrieval and array-based assays, Schwenk [/bib_ref] [bib_ref] Identification of plasma protein profiles associated with risk groups of prostate cancer..., Nordstrom [/bib_ref]. Using multiple antibodies for multiplexed first capture enrichment, we found differential profiles primarily for prostate specific antigen (PSA, as expected) as well as for insulin-like growth factor binding protein 2 (IGFBP2,. Subsequently, we confirmed on-target detection using only anti-PSA or anti-IGFBP2 antibody beads during first capture enrichment and subsequent multiplexed detection. Compared to previously used single binder assay analysis [bib_ref] Toward next generation plasma profiling via heat-induced epitope retrieval and array-based assays, Schwenk [/bib_ref] , the correlation between clinically determined PSA and DCA-derived levels was very high (Rho > 0.95) and in a linear relation to clinical PSA values < 1 g/mL. At higher PSA values, the DCA assay reached saturation levels (Supporting Information [fig_ref] Figure 1: The concept of the dual capture assay [/fig_ref] and we speculate that using more beads in the first capture step could expand the dynamic range to beyond 1 g/mL. In the current setting and using a 5-parametic fitting function, we estimated a limit of detection for PSA using the DCA assay to 11 pg/mL, thus improving assays sensitivity to direct labeling single capture assays by almost 70-fold . Compared to single binder assays, we also found that the . Selectivity and biomarker analysis in prostate cancer. Two different prostate cancer plasma sample sets, denoted as Study Set 1 and Study 2 as described in Supporting Information Table S1, were analyzed using a 52-plex antibody array. (A-B) For Study Set 1, we compared PSA values determined in the clinic with data generated in both a single-and dual-capture assay format. The y-axes in the scatterplots display the MFI values for the anti-total PSA antibody and the x-axes display the total PSA concentration determined in the clinic. In these scatterplots, MFI values for only those samples with clinical PSA value < 1 g/mL are shown. Scatterplots in Supporting Information display all samples including those with PSA > 1 g/ml. (C-D) A univariate analysis of the protein profiles obtained within Study Set 1 and Study Set 2 using the dual capture assay revealed the most prominent differences for three antibodies targeting total PSA, free PSA and IGFBP2. The boxplots display the MFI values for each sample categorized into the T0/T1 or T3/T4 group within Study Set 1 and into four different groups pre-determined based on clinical total and free PSA values within Study Set 2. A more detailed overview of the differences revealed for all antibodies included in the array is provided in Supporting Information . (E) Six replicates of a sample-free blank and four replicates of a plasma sample pool were included to assess the intra-assay CV in the analysis of Study Set 1 using both a single-and a dual-capture assay format. The density plot displays the distribution of percentage of CV across all antibodies in the technical replicates. (F) Instead of using all 52 antibodies combined for first capture enrichment, single bead populations either coupled to an anti-total PSA, anti-IGFBP2, anti-ALB or without any antibody were used for first capture. The heatmap displays the scaled MFI values obtained for pools of the different sample groups within Study Set 1 and 2 when only a single antibody or no antibody was used in first capture.
correlation between random pairs of antibody profiles was reduced in DCA (Supporting Information [fig_ref] Figure 1: The concept of the dual capture assay [/fig_ref]. While both approaches still benefit from data normalization, the sample dependent background was less influential in DCA, because many more antibodies revealed signals of low intensity levels. We hypothesize that this was reflected by the lowered correlation of random antibody pairs. On top of this, the technical variability for DCA (CV ± 5%) was lower than for single binder assays (%CV ± 15%). Besides these technical aspects, the p-values determined in each of the two different prostate cancer sample sets were computed for single capture and DCA. This revealed that analysis of DCA data generated lower p-values for PSA and suggested a smaller number of tentative candidates (Supporting . We and others have acknowledged that antibody performance is application dependent [bib_ref] Reproducibility crisis: blame it on the antibodies, Baker [/bib_ref]. In this setting, we believe that the observed differences from single binder assay may still be valid if confirmed by other assays, for example the presented DCA concept. As shown in, the use of a single antibody for first capture allows to assess selectivity of the enrichment in relation to the composition of the second capture array. While anti-PSA, anti-IGFBP2 and anti-ALB revealed an on-target enrichment, eluates from a bead carrying no antibody contained proteins that were recognized by an anti-C3a antibody. The latter, presumably off-target detection, would call for further investigations either to confirm whether C3a is enriched due to interactions with antibody-free beads or whether anti-C3a binds to another protein than C3a.
The use of antibody coupled beads for sandwich immunoassay [bib_ref] Sequential multiplex analyte capturing for Phosphoprotein profiling, Poetz [/bib_ref] thus presumably also the direct labelingbased antibody arrays follow the conditions of ambient analyte analysis [bib_ref] Multi-analyte immunoassay, Ekins [/bib_ref]. This implies that the measured intensity levels are dependent on target concentration rather than overall available quantities (mass sensing). At equal concentration, it is therefore likely that on-target interactions are preferred over off-target due to stronger affinities between antibody and target protein. Thus, an environment for a more selective assay can be achieved in DCA's second round of affinity capture, where both on-and off-target levels are expected to be lower than in the neat sample. Even though DCA does not require accommodating two binders on one protein at the same time, a possible limitation of this concept can arise when antibodies reveal affinities that are similar for offtarget and on-target. For such scenarios, complementarity between first and second capture array has to involve additional binders that then bind to different regions of the proteins. The DCA concept even offers to accommodate different types of affinity scaffolds, such as affibody molecules [bib_ref] Affibody molecules in protein capture microarrays: evaluation of multidomain ligands and different..., Renberg [/bib_ref] , aptamers [bib_ref] Aptamer-based multiplexed proteomic technology for biomarker discovery, Gold [/bib_ref] , or DARPins [bib_ref] A designed ankyrin repeat protein evolved to picomolar affinity to Her2, Zahnd [/bib_ref] , for one of the two capture steps. This could address sample components that bind to common regions of those affinity reagents used during first and second capture. A major advantage of the DCA concept is the onbead labeling step, as this allows for removal of buffers that may otherwise be incompatible with NHS esters used for biotin-labeling (e.g. TRIS-HCl) and for avoiding additional steps for sample pre-processing. This feature could open up possibilities for multiplex protein profiling in other sample types such as extracts of tissues and cells. Another option for DCA that can be further explored is the use of different additives and detergents to stabilize or denature targets. Since two incubation steps are used, proteins are currently assumed to be in a native-like form during the first enrichment step, while the read-out is achieved based on denatured proteins. Lastly, labeling an affinity-captured protein may be beneficial and allow for protecting the epitope from the labeling agent. Consequently, capturing such a protein again could be more efficient as compared to capturing a protein labeled in solution.
In summary, the developed sequential affinity capture assay facilitates multiplexed analysis of proteins in plasma.
In the proof-of-concept study, it showed improved performance characteristics compared to classical antibody arrays thus may serve as a first alternative when developing sandwich assays with antibodies found indicative in single binder screening efforts. The results from DCA experiments were in good agreement with clinical analysis. This miniaturized and parallelized concept holds promise to further advance antibody-based discoveries of soluble biomarker molecules in body fluids, and with suitable magnetic bead handling and automation, is ready for larger study sets.
[fig] Figure 1: The concept of the dual capture assay (DCA) [/fig]
[fig] Figure 3: Selectivity and biomarker analysis in prostate cancer. Two different prostate cancer plasma sample sets, denoted as Study Set 1 and Study 2 as described in Supporting Information Table S1, were analyzed using a 52-plex antibody array. (A-B) For Study Set 1, we compared PSA values determined in the clinic with data generated in both a single-and dual-capture assay format. The y-axes in the scatterplots display the MFI values for the anti-total PSA antibody and the x-axes display the total PSA concentration determined in the clinic. In these scatterplots, MFI values for only those samples with clinical PSA value < 1 g/mL are shown. Scatterplots in Supporting Information Fig. S9 display all samples including those with PSA > 1 g/ml. (C-D) A univariate analysis of the protein profiles obtained within Study Set 1 and Study Set 2 using the dual capture assay revealed the most prominent differences for three antibodies targeting total PSA, free PSA and IGFBP2. The boxplots display the MFI values for each sample categorized into the T0/T1 or T3/T4 group within Study Set 1 and into four different groups pre-determined based on clinical total and free PSA values within Study Set 2. A more detailed overview of the differences revealed for all antibodies included in the array is provided in Supporting Information Table S3. (E) Six replicates of a sample-free blank and four replicates of a plasma sample pool were included to assess the intra-assay CV in the analysis of Study Set 1 using both a single-and a dual-capture assay format. The density plot displays the distribution of percentage of CV across all antibodies in the technical replicates. (F) Instead of using all 52 antibodies combined for first capture enrichment, single bead populations either coupled to an anti-total PSA, anti-IGFBP2, anti-ALB or without any antibody were used for first capture. The heatmap displays the scaled MFI values obtained for pools of the different sample groups within Study Set 1 and 2 when only a single antibody or no antibody was used in first capture. [/fig]
|
Pattern of psychiatric in-patient admissions in Al Ain, United Arab Emirates
## Global echoes pattern of psychiatric in-patient admissions in al ain, united arab emirates
Karim Abdel Aziz,Dina Aly El-Gabry,Mouza Al-Sabousi,Ghanem Al-Hassani,Moataz M. Ragheb,Mohamed Elhassan Elamin,Mohamed Abdel-Maksoud,Emmanuel Stip,Aidroos Al-Aidroos,Tareq Al-Shehhi 1,9 and Danilo ArnoneAn understanding of the current state of mental health services in the United Arab Emirates (UAE) from a clinical perspective is an important step in advising government and stakeholders on addressing the mental health needs of the fast-growing population. We conducted a retrospective study of data on all patients admitted to a regional psychiatric in-patient unit between June 2012 and May 2015. More Emiratis (UAE nationals) were admitted compared with expatriates. Emiratis were diagnosed more frequently with substance use disorders and expatriates with stress-related conditions. Psychotic and bipolar disorders were the most common causes for admission and had the longest in-patient stays; advancing age was associated with longer duration of in-patient stay.
Mental illness is a leading cause of suffering and disability in the world. Global estimates predict an increasing impact in the next decades.There is an urgent need for the United Arab Emirates (UAE) to ascertain the current level of mental health services to plan effectively for a rapidly expanding population. Currently, the UAE absorbs 5.54% of the world mental health burden, but the country publishes only 1% of peerreviewed mental health research.The UAE is a multicultural federation of seven emirates, consisting of Abu Dhabi, Ajman, Dubai, Fujairah, Ras Al Khaimah, Sharjah and Umm Al Quwain . UAE psychiatric facilities include 25 psychiatric beds in general hospitals and 80 in specialised units, equivalent to 0.9 hospital beds/ 100 000 people, lower than other Arab countries (e.g. Morocco, 4.1/100 000) and Western countries (e.g. UK, 23.9/100 000; USA, 18.6/100 000; Australia, 7.2/100 000) (www.who.int/en/).
In mid 2016 the population of the emirate of Abu Dhabi was 2.9 million, out of a total population of the UAE of 9.3 million as estimated in 2017 (https://u.ae/en/about-the-uae/fact-sheet). In this paper, we retrospectively reviewed the pattern of admission in the 33-bed psychiatric unit based in Al Ain hospital, the largest psychiatric in-patient unit in the eastern region of the emirate of Abu Dhabi, which absorbs 30% of the national capacity. 3
# Method
We conducted a retrospective study by collecting data on all patients admitted to the psychiatric in-patient unit between June 2012 and May 2015. This in-patient provision is mixed for Emiratis (UAE nationals) and expatriates. Data were obtained from the computerised patient medical record system and included patients of all ages, genders, nationalities, clinical diagnoses according to ICD-10 4 and length of admission. Al Ain hospital's ethics committee approved the study.
Parametric and non-parametric data were analysed using SPSS version 24.0 for Windows. Chi-squared or Fisher's exact tests, independentsample Student's t-tests or Mann-Whitney U-tests were used as appropriate. Pearson Correlation Test (r) and Spearman's rho Correlation Test (r s ) for studying the direction and power of relationships of variables was used for correlational analyses of quantitative data. The statistical threshold for significance was set at α = 0.05.
# Results
Over the 3-year period of the study, 961 patients were admitted; 546 were male (56.8%), 415 were female (43.2%); the mean age was 32.49 years (s.d. = 11.72). The mean number of days in hospital was 14.52 (s.d. = 27.45).shows the frequencies of diagnoses and length of admission. The most frequent diagnoses were psychotic disorders, bipolar disorder and substance use disorders, and these were also associated with the longest duration of admission. There was a significant positive correlation between duration of stay and increasing age (r = 0.09, P = 0.005) but no significant relationship with gender (r = 0.025, P = 0.44).
Of the total, 570 patients (59.3%) were foreign expatriates (from 42 different countries) and 391 (40.7%) were local Emiratis. The most common non-Emirati groups admitted were Ethiopians (9.9%), followed by Pakistanis (7.4%), Omanis (5.6%), Indians (5.4%) and Bengalis (5.3%). Frequency of all nationalities and reasons for admission are presented in supplementary, available at https://doi.org/10.1192/bji. 2020.54. In both Emiratis and expatriates, psychotic disorders were the most common condition. Ethiopians presented with an acute stress reaction in 29% of cases, followed by 26% with psychotic disorders and 17% with bipolar disorder. In the second most frequent group, from Pakistan, psychotic disorders (22%) and bipolar disorder (21%) were the most common diagnoses. Omanis presented most frequently with a psychotic disorder (44%), as did Palestinians (64%). Other interesting trends included major depressive disorder in 47% of Afghani patients and psychotic disorders in 45% of Bangladeshi patients. Indians presented most frequently with psychotic disorders and bipolar disorders (31% each).
A comparison between local and non-local patients indicated that Emiratis were on average 3 years older. Substances use disorders were more frequent in Emiratis and acute stress reactions and adjustment disorders in the expatriates.
# Discussion
To the best of our knowledge, this is the first study investigating the pattern of psychiatric in-patient admissions in the UAE. Psychotic disorders were the most common conditions in the in-patient population, and they involved the longest stay in hospital. Bipolar disorder and substance use disorders followed in terms of frequency and admission length. Increasing age was associated with longer duration of in-patient stay. Expatriates were a heterogeneous group mostly originating from the Middle East, North Africa and Asia. The most common non-Emirati groups were Ethiopians (9.9%), Pakistanis (7.4%), Omanis (5.6%), Indians (5.4%) and Bengalis (5.3%). Expatriate patients were on average 3 years younger than Emiratis. Although more Emiratis had a diagnosis of substances use disorder, expatriates were diagnosed more frequently with conditions associated with stress, namely reactions to acute stress and adjustment disorders.
There were more Emiratis admitted compared with patients from other countries. This might reflect the fact that 19% of the total Emirati population live in the region of Abu Dhabi (41% in Al Ain). Expatriates receive variable levels of health insurance coverage, which may not necessarily include mental healthcare. Expatriates are also screened on entry into the UAE for health problems. The older age of Emiratis at the time of hospital admission may be related to what is reported in the literature as service underutilisation 5 fostered by a tendency to consult traditional healers as a preferred first option to treat mental illness, 6-8 which might delay diagnosis and treatment.Delayed diagnosis in conditions such as schizophrenia and bipolar disorder are associated with increased severity, affecting hospital admissions and outcome.Stigma associated with mental illness is a recognised problem in the Arab world, 11 which must be addressed to minimise suffering, premature death and large economic costs due to non-engagement.The United Nations World Drug Report 2019 suggests that the use of drugs is significant in the Middle East (wdr.unodc.org). Although the National Rehabilitation Centre in Abu Dhabi City is available for substance use disorders in the region,it is located far from Al Ain Hospital, explaining the number of cases admitted locally. Many expatriates originate from countries affected by political unrest, poverty, ethnic/ racial discrimination and military conflicts. Added stressors, such as isolation and societal differences after moving to the UAE, can also occur more frequently, explaining the higher occurrence of stress-related diagnoses in expatriates.
The Global Health Observatory data repository of the World Health Organization indicates that there are only 1.6 psychiatrists, 0.76 psychologists, 0.36 social workers and 4.37 nurses per 100 000 people living in the UAE. This level of workforce is unlikely to provide a level of services comparable to Western countries such as the USA, where there are 10.5 psychiatrists per 100 000. The demand for mental health services in the Abu Dhabi region is likely to be over 100% higher by 2030, according to the Abu Dhabi Department of Health's Healthcare Capacity Masterplan (www.doh.gov.ae/). This forecast, based on the current level of resources, requires careful consideration for planning. A way forward may be to follow the World Psychiatric Association's guidance on community mental healthcare.This approach advocates that in case of limited resources, it is more effective to create a stepped-care approach, with diagnostic evaluation and treatment initiation in primary care under specialist supervision.
# Limitations
Limitations of this study include the retrospective nature of the work, open to recall bias, and the limited availability of clinical data and records of level of insurance cover. We did not systematically record numbers/proportions of patients of Muslim faith and those who consult traditional healers, although 57.4% of the sample included nationalities of Islamic faith. Survey data from Al Ain suggest that only 38% of UAE citizens seek psychiatric help and prefer consulting traditional healers as a first step.The study does not cover the past few years of admissions, although we are not aware of any major societal or structural changes. We were unable to quantify the number of compulsory admissions. Currently in the UAE there is no equivalent to the UK's Mental Health Act. We only collected primary diagnoses. Information regarding secondary diagnoses might have been informative.
## Implications
There is great need to improve knowledge about mental illness in the UAE and to develop ways of increasing access to mental health services. Our analysis provides clinical information that can contribute to promoting knowledge and help to reduce stigma by improving the perception of mental illness. It is likely that a screening programme based in primary care might facilitate earlier detection of mental illness in the community, while a proactive education programme in schools and in the community might prove useful to improve knowledge about mental disorders and how to access services at the earliest convenience. Future service planning would also benefit from creating specialised substance misuse services locally to serve the needs of the community.
# Supplementary material
Supplementary material is available online at https://doi.org/10. 1192/bji.2020.54.
## Data availability
The study data are available on request. |
Isotretinoin Treatment for Autosomal Recessive Congenital Ichthyosis in a Golden Retriever
# Introduction
Ichthyoses represent a large heterogenous group of hereditary cornification disorders that manifest with abnormal differentiation and desquamation of keratinocytes in a form of generalized dry and scaly skin, variable erythroderma and hyperkeratosis [bib_ref] Revised nomenclature and classification of inherited ichthyoses: Results of the First Ichthyosis..., Oji [/bib_ref] [bib_ref] The Clinical and Morphologic Features of Nonepidermolytic Ichthyosis in the Golden Retriever, Mauldin [/bib_ref]. Autosomal recessive congenital ichthyosis (ARCI) has been associated with mutations in the patatinlike phospholipase domain-containing protein 1 (PNPLA 1) gene in dogs and humans [bib_ref] Degorce-Rubiales, F. Clinical, histopathological and genetic data of ichthyosis in the golden..., Guaguere [/bib_ref] [bib_ref] PNPLA1 mutations cause autosomal recessive congenital ichthyosis in golden retriever dogs and..., Grall [/bib_ref]. The PNPLA 1 gene encodes enzymes that are important in lipid metabolism [bib_ref] Proteins sharing PNPLA domain, a new family of enzymes regulating lipid metabolism, Baulande [/bib_ref]. In human medicine, retinoids are frequently used for treatment of ichthyoses because of their ability to modulate epidermal proliferation, differentiation and inflammation [bib_ref] Inherited Nonsyndromic Ichthyoses: An Update on Pathophysiology, Vahlquist [/bib_ref] [bib_ref] Expression of retinoidregulated genes in lamellar ichthyosis vs. healthy control epidermis: Changes..., Loriè [/bib_ref]. Current therapy recommendations for canine ichthyoses are focused on removing the scales and improving the epidermal barrier with topical keratolytic, keratoplastic and/or moisturizing products [bib_ref] Topical polyhydroxy acid treatment for autosomal recessive congenital ichthyosis in the golden..., Puigdemont [/bib_ref] , together with oral fatty acid supplementation [bib_ref] Autosomal recessive congenital ichthyosis due to PNPLA1 mutation in a golden retriever-poodle..., Tamamoto-Mochizuki [/bib_ref] , but unfortunately with variable improvements.
This case report describes clinical and histological improvement following isotretinoin treatment in a golden retriever dog with ARCI.
# Materials and methods
This investigation was conducted at the Veterinary Teaching Hospital (VTH) of the Faculty of Veterinary Medicine at the University of Zagreb with written owner consent.
## Dermatological examination
At day 0, a dermatologist examined the patient to assess the severity of scaling, the size and distribution of scales, and the presence and intensity of pruritus. An eight-yearold spayed female golden retriever was admitted to the VTH due to severe pruritus and severe scaling of the skin that was present from a very young age. Axillar and inguinal hyperpigmentation were present. Based on dermatological examination, differential conditions included ichthyosis, epitheliotropic T-cell lymphoma, severe atopic dermatitis with secondary infection, and dermatophytosis. The mycology culture result was negative. Biopsy under general anesthesia was conducted using a 6 mm punch biopsy machine in the ventro-lateral aspects of the thorax and lumbosacral area.
At day 90, dermatological examination was repeated with an assessment of the severity of scales, their size and distribution, as well as the intensity of pruritus. Biopsy was conducted under general anesthesia using a 6 mm punch biopsy machine in the ventrolateral aspects of the thorax and lumbosacral area, but on the contralateral side of the previous biopsies.
## Treatment protocol
Treatment included oral isotretinoin at a dose of 1 mg/kg twice daily (Roaccutane, La Roche Pharma, Basel, Switzerland). Antibiotics, shampoos, emollients and moisturizing products were not used for a 3-month period.
## Blood analysis and schirmer tear test
Hematology and serum biochemistry were conducted on admission, and after three, six and twelve months. Production of tears was evaluated after two years of isotretinoin usage.
## Histological examination
Biopsy samples were taken at day 0 and 90 days after starting the treatment with isotretinoin. Skin samples were fixed in 10% neutral-buffered formalin for 24 h and then embedded in paraffin wax and processed routinely.
## Genetic testing
Ethylenediaminetetraacetic acid (EDTA)-blood samples were sent to the laboratory (Laboklin GmbH& CoKG, Postfach1810; 97688 Bad Kissingen, Germany) for mutational analysis. DNA extraction and amplification was performed using the PCR technique described in the paper by Grall and colleagues [bib_ref] PNPLA1 mutations cause autosomal recessive congenital ichthyosis in golden retriever dogs and..., Grall [/bib_ref]. It was previously established that wild-type genes are unaffected with no transmission of the variant. Heterozygous dogs were found to be unaffected and able to transmit the PNPLA 1 variant to 50% of their lineage. Homozygous dogs were affected with clinical signs and it is known that they will transmit the PNPLA 1 variant to 100% of their lineage [bib_ref] PNPLA1 mutations cause autosomal recessive congenital ichthyosis in golden retriever dogs and..., Grall [/bib_ref] [bib_ref] Prevalence of PNPLA1 Gene Mutation in 48 Breeding Golden Retriever Dogs, Graziano [/bib_ref].
Analysis confirmed that the patient was homozygous for mutation in the PNPLA 1 gene with an autosomal-recessive trait of inheritance.
# Results
History, clinical signs and histopathological changes led to the diagnosis of ARCI seen in golden retriever dogs. Genetic testing performed on a venous blood sample confirmed that the patient was homozygous for the PNPLA 1 gene mutation.
## Dermatological examination and blood analysis
At initial presentation, severe generalized scaling was present, mostly on the dog's trunk and with less scales on the extremities and head. Mostly medium-to-large, white scales were seen . The scales were loosely adherent to the skin and some embedded hair shafts . Tan-to-yellow scales were noticed in the axillary and inguinal region with moderate hyperpigmentation . Periocular alopecia with moderate lichenification, ceruminous otitis and moderate pododermatitis were also noticed. Based on clinical and histopathological findings, a diagnosis of ichthyosis with a secondary atopic dermatitis was made. Pruritus was graded based on the pruritus visual analog scale (pVAS) as 5/10 [bib_ref] Development of an owner-assessed scale to measure the severity of pruritus in..., Hill [/bib_ref]. The complete blood count (CBC) was within normal range while biochemical analysis revealed mild elevations in triglycerides 3.4 mmol/L (reference range 0.2-1.3) and cholesterol 8.7 mmol/L (reference range 3.5-7.1).
Vet. Sci. 2022, x FOR PEER REVIEW 4 of . Clinical response to isotretinoin therapy in a golden retriever dog with autosomal recessiv congenital ichthyosis (ARCI). Pictures on the left were obtained before and those on the right afte 3 months of treatment at the same body locations but on the contralateral sides. Picture before ( and after treatment (b) on the trunk. Picture before (c) and after treatment (d), with clipped hair fo better visualization of large white scales and latticelike hyperpigmentation (d). Picture before ( and after treatment (f) in axillar region showing tan-to-yellow scales with variable hyperpigment tion.
## Histopathological findings
The histological findings before and after treatment with isotretinoin are illustrate in [fig_ref] Figure 2: Histopathological changes in skin biopsies in a golden retriever with autosomal recessiv... [/fig_ref]. The initial biopsies [fig_ref] Figure 2: Histopathological changes in skin biopsies in a golden retriever with autosomal recessiv... [/fig_ref] ,b,c) revealed moderate-to-severe diffuse lame lar orthokeratotic hyperkeratosis with multifocal black-to-brown granular depositions i . Clinical response to isotretinoin therapy in a golden retriever dog with autosomal recessive congenital ichthyosis (ARCI). Pictures on the left were obtained before and those on the right after 3 months of treatment at the same body locations but on the contralateral sides. Picture before (a) and after treatment (b) on the trunk. Picture before (c) and after treatment (d), with clipped hair for better visualization of large white scales and latticelike hyperpigmentation (d). Picture before (e) and after treatment (f) in axillar region showing tan-to-yellow scales with variable hyperpigmentation.
After 3 months of isotretinoin usage, the clinical examination showed marked improvement. Scales completely disappeared from the whole body and hyperpigmentation decreased on areas where it was originally seen ,d,f). When hair was clipped, latticelike hyperpigmentation was noticed . Pruritus was evaluated again (pVAS 5/10). There were no side effects associated with retinoid usage. Repeated blood analysis revealed normal CBC with normal triglycerides, and increased cholesterol 10.2 (reference range 3.5-7.1).
After 3 months of 1 mg/kg twice-daily dose of isotretinoin, the dose was decreased to the lowest effective amount. The patient was clinically stable at 0.6 mg/kg twice daily for the next 18 months. Repeated CBC and biochemistry analysis results were normal in that time range. After 2 years of isotretinoin usage, therapy was stopped due to drug shortage. Scaling increased after 2-3 weeks of isotretinoin cessation. Three months later, the patient was restarted with isotretinoin therapy 0.6 mg/kg twice daily, which resulted in the disappearance of scales within 3 weeks.
During treatment with oral isotretinoin, the owner was informed about potential side effects and the development of keratoconjunctivitis sicca. After two years of treatment, a Schirmer tear test was conducted. The tear uptake was 25 mm/min in both eyes, which was considered physiological (reference range 29.3 ± 16.9 mm/min) [bib_ref] Analysis of tear uptake by the Schirmer tear test strip in the..., Williams [/bib_ref].
## Histopathological findings
The histological findings before and after treatment with isotretinoin are illustrated in [fig_ref] Figure 2: Histopathological changes in skin biopsies in a golden retriever with autosomal recessiv... [/fig_ref]. The initial biopsies [fig_ref] Figure 2: Histopathological changes in skin biopsies in a golden retriever with autosomal recessiv... [/fig_ref] revealed moderate-to-severe diffuse lamellar orthokeratotic hyperkeratosis with multifocal black-to-brown granular depositions in the lamellar keratin. Epidermis was mostly 2-3 layers thick with multifocal mild acanthosis. The epidermis had a severe diffuse pleated appearance. Mild-to-moderate superficial-tomid-dermal perivascular infiltration of lymphoplasmacytic cells was seen. Moderate-tosevere follicular lamellar hyperkeratosis was noticed in the superficial dermis. A moderate number of vacuolated keratinocytes were scattered in the epidermis [fig_ref] Figure 2: Histopathological changes in skin biopsies in a golden retriever with autosomal recessiv... [/fig_ref].
Vet. Sci. 2022, 9, x FOR PEER REVIEW 5 of the lamellar keratin. Epidermis was mostly 2-3 layers thick with multifocal mild acan thosis. The epidermis had a severe diffuse pleated appearance. Mild-to-moderate super ficial-to-mid-dermal perivascular infiltration of lymphoplasmacytic cells was seen. Mod erate-to-severe follicular lamellar hyperkeratosis was noticed in the superficial dermis. A moderate number of vacuolated keratinocytes were scattered in the epidermis [fig_ref] Figure 2: Histopathological changes in skin biopsies in a golden retriever with autosomal recessiv... [/fig_ref] Histological findings after treatment with isotretinoin [fig_ref] Figure 2: Histopathological changes in skin biopsies in a golden retriever with autosomal recessiv... [/fig_ref] ,e,f) revealed marked improvements. Mild-to-moderate multifocal orthokeratotic hyperkeratosis wa seen, but keratin was more organized in a basket-weave appearance. The epidermis wa mildly acanthotic with a mild-to-moderate multifocal pleated appearance. Mild superfi cial perivascular infiltration of lymphoplasmacytic cells was noticed. Mild superficial fol licular hyperkeratosis was seen. The number of vacuolated keratinocytes decreased.
# Discussion
Our clinical and histopathological findings indicate that treatment with oral isotret inoin was effective in improving ichthyosis without any side-effects. To the best of th authors' knowledge, this is the first clinical and histological investigation of oral isotret Histological findings after treatment with isotretinoin [fig_ref] Figure 2: Histopathological changes in skin biopsies in a golden retriever with autosomal recessiv... [/fig_ref] revealed marked improvements. Mild-to-moderate multifocal orthokeratotic hyperkeratosis was seen, but keratin was more organized in a basket-weave appearance. The epidermis was mildly acanthotic with a mild-to-moderate multifocal pleated appearance. Mild superficial perivascular infiltration of lymphoplasmacytic cells was noticed. Mild superficial follicular hyperkeratosis was seen. The number of vacuolated keratinocytes decreased.
# Discussion
Our clinical and histopathological findings indicate that treatment with oral isotretinoin was effective in improving ichthyosis without any side-effects. To the best of the authors' knowledge, this is the first clinical and histological investigation of oral isotretinoin in ichthyoses in canine patients.
The first clinical signs of ichthyosis can be seen within the first 3 weeks of life but sometimes it can take up to 3 years to be clearly manifested [bib_ref] Degorce-Rubiales, F. Clinical, histopathological and genetic data of ichthyosis in the golden..., Guaguere [/bib_ref]. Scales in golden retrievers with ARCI are usually white-to-grey, loosely adherent and mostly present on the trunk with sparing of the head and extremities. Clipping off the hair allows better visualization of large adherent, fine, polygonal scales [bib_ref] The Clinical and Morphologic Features of Nonepidermolytic Ichthyosis in the Golden Retriever, Mauldin [/bib_ref]. Sparsely haired areas, such as the abdomen and axillae, are hyperpigmented and have a rough, sandpaper-like appearance [bib_ref] Degorce-Rubiales, F. Clinical, histopathological and genetic data of ichthyosis in the golden..., Guaguere [/bib_ref]. Rarely, ceruminous otitis and hyperkeratosis of the paw pads can be seen [bib_ref] Degorce-Rubiales, F. Clinical, histopathological and genetic data of ichthyosis in the golden..., Guaguere [/bib_ref].
Histological changes of ichthyotic skin show laminated or compact orthokeratotic hyperkeratosis with frequent melanic pigmentation. The basal membrane is often undulated and has a "pleated" appearance, which explains the rough, sand-paper-like appearance of the skin [bib_ref] Degorce-Rubiales, F. Clinical, histopathological and genetic data of ichthyosis in the golden..., Guaguere [/bib_ref]. When compared to histological pictures after treatment with isotretinoin, there is a marked improvement in lamellar orthokeratotic hyperkeratosis, which has a more normal, basket-weave-like appearance. In addition, histologic improvement is seen clinically as a loss of white, loosely adherent scales.
Retinoids are structurally and functionally related to vitamin A [bib_ref] Consensus recommendations for the use of retinoids in ichthyosis and other disorders..., Zaenglein [/bib_ref] with proven functions in the skin based on the modulation of epidermal maturation, as well as keratinocyte differentiation, apoptosis, immune function and carcinogenesis [bib_ref] Effects of Topical Retinoids on Cytoskeletal Proteins: Implications for Retinoid Effects on..., Eichner [/bib_ref] [bib_ref] Retinoid chemoprevention in patients at high risk for skin cancer, Digiovanna [/bib_ref]. Moreover, isotretinoin was found to induce apoptosis and cell cycle arrest in human sebocytes [bib_ref] Teratogenic effect of isotretinoin in both fertile females and males (Review), Draghici [/bib_ref]. As a result, retinoids decrease thickening and scaling of the skin, decrease erythema [bib_ref] Consensus recommendations for the use of retinoids in ichthyosis and other disorders..., Zaenglein [/bib_ref] [bib_ref] Alitretinoin reduces erythema in inherited ichthyosis, Onnis [/bib_ref] and facilitate wound healing [bib_ref] Retinoic Acid and Its Derivatives in Skin, Szymański [/bib_ref]. Systemic retinoids have been successfully used for ARCI in human medicine [bib_ref] Severe ectropion in lamellar ichthyosis managed medically with oral acitretin, Singh [/bib_ref] , but until now they have not been used in veterinary medicine for ichthyoses.
Isotretinoin was developed as a synthetic retinoid, but then it was discovered that this 13-cis isomer of naturally occurring tretinoin (trans-retinoic acid) was present in cells as a naturally occurring metabolite [bib_ref] Systemic retinoids in the management of ichthyoses and related skin types, Digiovanna [/bib_ref]. Isotretinoin has a short half-life (~1 h); however, clearance from the organism can take several months [bib_ref] Systemic retinoids in the management of ichthyoses and related skin types, Digiovanna [/bib_ref]. The recommendation is to use the lowest effective dose to decrease side effects [bib_ref] Systemic retinoids in the management of ichthyoses and related skin types, Digiovanna [/bib_ref]. Isotretinoin has a low teratogenic risk but is advised not to be used in pregnant individuals. Monitoring for complete blood count with liver function tests is recommended. The most commonly changed parameters are triglycerides and cholesterol. In human medicine, acute toxicities connected with retinoid usage are cheilitis, xerosis, dry nose, eye irritation and hair loss. Chronic usage of retinoids in human medicine is mostly manifested in the skeletal system (enthesopathy, hyperostosis or spurs along the spine) [bib_ref] Retinoid chemoprevention in patients at high risk for skin cancer, Digiovanna [/bib_ref] [bib_ref] Systemic retinoids in the management of ichthyoses and related skin types, Digiovanna [/bib_ref]. In human and veterinary medicine, retinoids are known to alter the tear composition by decreasing the lipid component in the tears. As a result, increased tear evaporation leads to keratoconjunctivitis sicca [bib_ref] Consensus recommendations for the use of retinoids in ichthyosis and other disorders..., Zaenglein [/bib_ref] [bib_ref] Synthetic Retinoids in Veterinary Dermatology, Power [/bib_ref].
Clinical improvement in the golden retriever was noticed within 3 weeks of treatment, which is compatible with treatment of human ARCIs where thickness and scaling decreases in two weeks from staring the therapy [bib_ref] Systemic retinoids in the management of ichthyoses and related skin types, Digiovanna [/bib_ref]. After discontinuation of isotretinoin in humans, symptoms usually reoccur, which was also seen in our case. Transient elevations of cholesterol and triglycerides were noticed, but were also present before the therapy with isotretinoin. After several months of therapy with isotretinoin and the decreasing of the dose to the lowest effective amount, all biochemical abnormalities resolved and were within reference range. Ichthyosis causes a major epidermal barrier defect and is not so rare that atopic dermatitis (AD) is secondarily manifested. In human patients, certain populations with ichthyosis vulgaris are at increased risk for developing AD due to barrier defects [bib_ref] Skin diseases associated with atopic dermatitis, Fenner [/bib_ref]. Golden retrievers are predisposed to developing AD, with their heritability estimated to be 0.47 [bib_ref] Estimation of heritability of atopic dermatitis in Labrador and Golden Retrievers, Shaw [/bib_ref]. Golden retrievers are also predisposed for ARCI, though the heritability is not known. However, at the present time, no link has been established between the two diseases. Further studies are needed to compare incidence of AD in dogs with or without ARCI in golden retrievers. Isotretinoin did not have any effect on pruritus in our case. The most probable explanation lies in the fact that pruritus develops due to sensitization to allergens as a consequence of a barrier defect and not as a primary symptom of ARCI. As ARCI is a congenital disease, the treatment with isotretinoin could be recommended throughout life; however, long-term side effects have not been deeply investigated and larger studies are needed. Clinicians need to estimate the benefit-risk balance before prescribing an oral retinoid for long-term use. It is advised to monitor liver enzymes, lipids and tear production during isotretinoin treatment.
We could say that ARCI is a cosmetic disease that needs no systemic treatment; however, the subject of our case had a history of constant antibiotic usage due to severe secondary bacterial and fungal infections that were almost impossible to control. After treatment with isotretinoin, the secondary infections almost disappeared completely and the symptoms of AD were easily controlled.
Further controlled studies on larger samples are needed to confirm the beneficial impact of isotretinoin in dogs with ARCI and the general effects of the long-term usage of isotretinoin in this species.
# Conclusions
Clinical and histopathological findings indicate that treatment with oral isotretinoin was effective in improving symptoms of autosomal recessive congenital ichthyosis in a golden retriever. Isotretinoin appeared safe and well tolerated, with no observed side-effects. Institutional Review Board Statement: Ethical review and approval were waived for this study, because standard procedures were followed in accordance with Directive 2010/63/EU and good veterinary practices. Standard procedures were implemented under general anesthesia to avoid pain or distress. Usage of retinoids was not considered experimental due to the EU directive stating that orally given medications are not considered as experimental and due to the fact that retinoids are used in other dermatological and oncological conditions without significant side effects.
# Informed consent statement:
Written informed consent to publish this paper was obtained from the patient's owner.
[fig] Figure 2: Histopathological changes in skin biopsies in a golden retriever with autosomal recessiv congenital ichthyosis (ARCI), before (a-c) and after treatment (d-f) with isotretinoin. Note the di fuse and severe laminar orthokeratotic hyperkeratosis in the stratum corneum in picture (a) (×4), (b (×10) and (c) (×40). After treatment with isotretinoin, keratin was not so abundant, as shown in pic tures (d) (×4), (e) (×10) and (f) (×40), and had a more typical basket-weave appearance (pictures e,f In picture (c) (x40), there is a moderate number of scattered vacuolated keratinocytes (arrows) in th epidermis, while in picture (f) (×40), their number is decreased. [/fig]
[fig] Author: Contributions: Conceptualization, A.P. and N.L.; methodology, A.P.; software, I.-C.Š.-Z.; validation, N.L., M.H. and I.-C.Š.-Z.; formal analysis, A.P.; investigation, A.P.; resources, M.H.; data curation, A.P.; writing-original draft preparation, A.P.; writing-review and editing, N.L., M.H. and I.-C.Š.-Z.; visualization, A.P.; supervision, N.L.; project administration, A.P.; funding acquisition, M.H. All authors have read and agreed to the published version of the manuscript. Funding: This research received no external funding. [/fig]
|
Assessment of disease specific immune responses in enteric diseases using dried blood spot (DBS)
BackgroundBlood collection, transportation and storage remain a problem in countries where infrastructure, laboratory facilities and skilled manpower are scarce. This limits evaluation of immune responses in natural infections and vaccination in field studies. We developed methods to measure antigen specific antibody responses using dried blood spot (DBS) in cholera, ETEC and typhoid fever patients as well as recipients of oral cholera vaccine (OCV).
# Methodology/principle findings
We processed heparinized blood for preparing DBS and plasma specimens from patients with, cholera, ETEC and typhoid as well as OCV recipients. We optimized the conventional vibriocidal method to measure vibriocidal antibody response in DBS eluates. We measured responses in DBS samples and plasma (range of titer of 5 to 10240). Vibriocidal titer showed strong agreement between DBS eluates and plasma in cholera patients (ICC = 0.9) and in OCV recipients (ICC = 0.8) using the Bland-Altman analysis and a positive correlation was seen (r = 0.7, p = 0.02 and r = 0.6, p = 0.006, respectively). We observed a strong agreement of lipopolysaccharide (LPS) and cholera toxin B (CTB)-specific antibody responses between DBS eluates and plasma in cholera patients and OCV recipients. We also found agreement of heat labile toxin B (LTB) and membrane protein (MP)-specific antibody responses in DBS eluates and plasma specimen of ETEC and typhoid patients respectively.
# Conclusion
Our results demonstrate that dried blood specimens can be used as an alternate method for preservation of samples to measure antibody responses in enteric diseases and vaccine trials and can be applied to assessment of responses in humanitarian crisis and other adverse field settings. Introduction Blood collection, processing and quick transportation to laboratories from remote areas in resource poor settings can be limitations in carrying out serological studies especially in developing countries where laboratory facilities and trained manpower as well as cold chain facilities are not optimal. Dried blood spot (DBS) is an alternate method for collecting blood specimens on paper. This lowers biological risks associated with conventional blood transportation, does not require maintaining cold-chain during transportation and DBS card can be shipped at ambient temperature and even by ordinary mail. Blood collection and transportation from resource-limited settings remains an obstacle to follow up large vaccine trials, seroepidemiological studies and diagnosis of infectious diseases as well as large population based estimations of public health interventions. DBS can therefore be a solution to the cumbersome specimen collection and transportation that is generally used. DBS has been used as an alternative method for diagnosis of viral, bacterial, fungaland parasiticinfections. It has also been used to monitor HIV viral load and drug resistance in patients receiving antiretroviral therapy. The DBS analysis platform is routinely used for estimation of DNA, protein and drugswhich have a great impact in clinical practice. DBS has been used for measuring immune responses after hepatitis A virus vaccinationwith a positive agreement with plasma specimens. However, there is lack of information about the sensitivity of DBS to evaluate the immunogenicity in enteric infections.
In this study, we have used the DBS method to collect blood from patients with cholera, enterotoxigenic Escherichia coli (ETEC) diarrhea, typhoid fever as well as recipients of an oral cholera vaccine to evaluate feasibility and accuracy of immune responses. Immune responses measured using the DBS method was compared with results obtained with plasma specimens in the same cohort of patients and vaccinees. There was positive agreement between DBS and plasma specimen in the evaluation of immune responses in cholera, ETEC diarrhea and typhoid patients as well as OCV recipients.
# Methods
## Study participants
Blood specimens were obtained from patients with enteric infections which included cholera, ETEC, typhoid and as well as healthy participants who received two doses of heat-killed oral cholera vaccine (Shanchol, Shantha biotechnics, India). Cholera and ETEC-infected patients were culture confirmed by isolation of Vibrio cholerae O1 and enterotoxigenic Escherichia coli (ETEC) from stool specimens respectively. ETEC was confirmed by the presence of heat-labile toxin (LT) or heat-stable toxin (ST) or both by ELISA and PCR. Typhoid fever was confirmed by isolation of Salmonella Typhi from blood of febrile patients. Patients were given appropriate antibiotics by physicians as was needed (S1 . Specimens were collected between January 2016 and December 2017.
# Ethics statement
Specimen collection from cholera, ETEC and typhoid patients and healthy participants and were approved by the Institutional Research Board (IRB) of icddr,b. Written informed consent was obtained from the study participants , and consent from parents of those aged 5-10 years with assents from the minor participants (11-17 years) were obtained.
## Collection of blood samples in dbs cards
Venous blood was collected (8-10 ml from adults and 3-5 ml from children; only 200 μl was used for preparation of DBS) from the study participants in sodium heparin tube (Beckton Dickinson, CA). Blood was collected from cholera and ETEC diarrheal patients on day 2, 7 and 30 of onset of disease. From typhoid fever patients, blood was collected after culture confirmation (day 1), 2-3 days later (day 2) and 18-22 days later (day 3). Healthy participants were vaccinated with oral cholera vaccine at two time points (days 0 and 14). Blood was collected prior to vaccination on day 0 and 14 (prior to second vaccination) and two more blood samples were collected at day 28 and 42 after the first vaccination. Approximately 50 μl of blood was adsorbed on each spot (200 μl for four spots) of DBS card (Whatman 903, GE Healthcare) and air dried at ambient temperature (15-25˚C) for two hours by placing on a clean paper towel. The card was then transferred to a black card holder (Advantus 4X6 index card holder, Jacksonville, FL, USA) box and kept overnight at RT for adequate drying. The DBS card was placed in a single, gas-impermeable zipper bag containing desiccant sachets (2 desiccants per bag) to protect the cards from moisture. A humidity indicator card was added to monitor the moisture level in the zipper bag and it was checked monthly. If the indicator turned purple then the desiccant sachets were replaced with new desiccant sachets. The dried blood spotted cards were stored at -80˚C until tested (1-6 months).
## Dbs elution
A 3.2 mm spot was punched with a DBS puncher (PerkinElmer, USA) from each blood-soaked circle of the DBS card. All punched spots from a single patient were transferred to one eppendorf tube and 0.5% bovine serum albumin (BSA) in phosphate buffered saline (PBS, 10 mM, pH 7.2) was added according to the desired dilution (18 punch in 270 μl; 1:10 for vibriocidal and 1 punch in 300 μl; 1:200 for ELISAs) and kept at RT overnight. The tube was centrifuged (2500 x g for 7 min) and eluate was collected in a fresh eppendorf tube.
## Plasma separation
Plasma specimens were also separated from the same cohort of patients and vaccinees from the heparinized blood. Blood specimens were centrifuged, aliquoted (200 μl in eppondorf tube) and stored at -20˚C until tested.
## Vibriocidal assay
Vibriocidal antibody titer was measured according to the procedure described earlier. Briefly, 25 μl of cold physiological saline (0.154 M) was dispensed in all wells except column #2 of 96-well microtiter plates (Nunc F, Denmark). Plasma and DBS eluates were heat inactivated at 56˚C for 30 min. For plasma, 45 μl of cold saline and 5 μl of plasma and for DBS eluates 50 μl eluates were dispensed in corresponding wells in column #2. Both plasma and DBS eluates were serially diluted (initial dilution 1:10) in two fold dilutions. The dilution was accomplished by mixing the solution in column #2, aspirating 25 μl and dispensing and mixing the sample in column #3 and so on, until column #12 (1:10240 dilution). At the end, 25 μl was discarded from the last well in each row. V. cholerae O1 Inaba strain (T-19479) was cultured on blood agar plates at 37˚C overnight. This strain was isolated from stool of cholera patientin Bangladesh and well characterizedas well as used to measure vibriocidal antibody titer in plasma. A loopful of bacteria was inoculated in Brain Heart Infusion (BHI) media for 2-3 hours at 37˚C with shaking to mid-log phase of growth (OD to 0.4). Bacterial culture was centrifuged and resuspended in sterile saline (0.154 M) and the OD was adjusted to 0.3 at 600 nm using a spectrophotometer (Beckman DU 530 UV vis). For vibriocidal assay using plasma, 25 μl of bacterial suspension (21.25 μl normal saline, 1.25 μl bacteria and 2.5 μl guinea pig sera) was added to all wells except wells #E1, F1, G1 and H1 in column 1 of 96-well plates and 25 μl of cold saline was added to wells # E1, F1, G1 and H1 . For vibriocidal assay using DBS, 25 μl of bacterial suspension (21.25 μl normal saline, 1.25 μl bacteria and 2.5 μl guinea pig sera) was added to all wells except wells row #A2-A12, B2-B12 wells and column #E1, F1, G1 & H1 wells and 25 μl of cold saline was added to wells # E1, F1, G1 and H1 . The OD value of wells in row #A2-A12, B2-B12 was used as the color background control for DBS . The plates were incubated at 37˚C for 1 hour and 150 μl of BHI media was added to all wells and incubated again at 37˚C. The plates were read spectrophotometrically (Eon, BioTek, Vermont, USA) until the positive control well reached OD around 0.20 to 0.28 at 595 nm.
## V. cholerae lipopolysaccharide (lps) extraction
LPS was purified from V. cholerae strains as discussed previously. Briefly, LPS was purified by the hot phenol-water procedure, ultracentrifugation and treatment with enzymes (proteinase K, DNase and RNase) from V. cholerae O1 Inaba (T-19479) strain.
## V. cholerae lps and ctb specific antibody responses in cholera patients and vaccines
For this assay, 96-well ELISA (Nunc F, Denmark) plates were coated with LPS of V. cholerae O1 in PBS (100 μl/well) at a concentration of 2.5 μg/ml and incubated at RT overnight. Conventional vibriocidal plate DBS eluates vibriocidal plate
## Fig 1. conventional vibriocidal assay and modified vibriocidal assay for dbs eluates.
A) The schematic diagram shows the conventional vibriocidal assay with plasma collected from cholera patients and cholera vaccinees using the 96-well plates (row A-F). Pooled sera obtained from cholera patients was used as positive control on each plate. B). The vibriocidal antibody assay was modified when DBS eluates were used to measure vibriocidal response in cholera patients/vaccinees. The first two rows (rows A-B) were used without adding bacterial inocula to calculate color background of heme in DBS eluates. Rows C to F were used for DBS eluates similar to the conventional vibriocidal assay and the last two rows (G-H) were used as positive control.
## Membrane protein antigen of s. typhi specific antibody responses in typhoid fever patients
MP was prepared from S. Typhi by using a previously described procedure. Briefly, S. Typhi strain (Ty21a) was cultured on sheep blood agar plates and harvested in Tris buffer (10 mM Tris [pH 8.0], 5 mM MgCl 2 ). Mixture was sonicated and centrifuged at 1400 x g for 10 min, supernatant transferred to fresh tubes, centrifuged again at 14900 x g for 30 min. The pellet was suspended in 10 ml of Tris buffer, and the protein content was determind. Antibody responses in plasma as well as DBS of typhoid fever patients against MPwere measured as described previously. was added to each well and OD was measured at 450 nm using the kinetic method.
## Antibody responses against ltb in etec-infected patients
Plasma antibodies specific to LTB were measured using a previously described ELISA method. Briefly, 96-well ELISA (Nunc F, Denmark) plates were coated (100 μl/well) with ganglioside GM1 (0.3 nM/ml) and incubated overnight at ambient temperature (15-25˚C). The plates were washed with PBS and recombinant LTB (0.5 μg/ml) (gift of A.M Svennerholm, University of Gothenburg) added to the wells (100 μl/well) and incubated for 1 h at 37˚C. The plates were washed with PBS-T (0.05%) for three times and once with PBS. The plasma samples (1:200 dilution) was added and incubated for 90 min at 37˚C. The plates were washed three times with PBS-T(0.05%) and once with PBS. HRP-conjugated anti-human antibody was added to wells and incubated for 90 min at 37˚C. The plates were washed three times with PBS-T (0.05%) and once with PBS. Substrate was added to the wells and plates were read at 450 nm using the kinetic method (Eon, BioTek, Vermont, USA).
## Statistical analyses
We assessed differences in the magnitude of the responses using the Mann-Whitney U test using GraphPad prism 6 (GraphPad Software, Inc., La Jolla, CA, USA). Two-sided P values <0.05 were considered significant. Correlations were assessed using Spearman's rank correlation. The Bland-Altman plots with 95% limits of agreement (LoA) were used to evaluate the agreement between results obtained with plasma and DBS. The 95% LoA was calculated as the average difference between the plasma and DBS ± 1.96SD (upper limit and lower limit) in SPSS Statistics version 20.0 (IBM Corp., Armonk, NY, USA). The narrower the 95% LoA, the better is the agreement. In addition, the reliability of the DBS method was tested by intraclass correlation coefficient (ICC); if the ICCs were closer to 1, the reliability was higher. Figures were generated in GraphPad prism 6.
# Results
## Study participants
The patients and vaccines were followed up to day 30 and day 42 respectively after enrollment and were included in the analysis. Demographic features of the patients and vaccinees are provided in.
## Optimization of vibriocidal assay for dbs eluates
We set up the vibriocidal assay procedure using DBS eluates so that the assay could be measured in an ELISA Reader (Eon, BioTek, Vermont, USA) similar to the way the conventional and existing vibriocidal assay was being used . The heme color was the main obstacle in setting up the vibriocidal assay to directly measure the antibody titer. To alleviate this problem, the color effect seen in the DBS specimens from each participant were used as DBS control to which no bacterial inoculum was added and this value was subtracted from the OD value of DBS vibriocidal row using an average value for negative controls. We defined the DBS blank as serial dilution (1:2) of DBS eluates and added physiological saline (0.154 M) instead of bacterial inoculum as described above. The optical density from the DBS blank was considered for the heme color and was deducted from DBS plus indicator wells. We defined the final OD of DBS plus indicator wells according the following formula: OD value of DBS plus indicator well = (OD value of DBS plus indicator-OD value of DBS blank) + Average value of negative controls If this final OD value was less than or equal to half of the average OD value of positive controls then it was considered as 50% killing of the V. cholerae O1 strain.
## Vibriocidal responses in cholera patients and vaccines
We analyzed the vibriocidal responses in DBS eluates, plasma of cholera patients (n = 17) and vaccinees (n = 17). We found similar magnitude of vibriocidal antibody titer in the corresponding day points of DBS and plasma of patients and vaccinees. The responder frequency of vibriocidal titer (four-fold increase) in DBS eluates and plasma (twofold increase) were similar (90%). There was a strong agreement (Bland-Altman agreement) of vibriocidal responses between DBS eluates and plasma in both patients and vaccinees (ICC = 0.9 and 0.8 respectively;.There was also a positive correlation of vibriocidal antibody responses at day 7 for patients (r = 0.73, p = 0.02;and at day 14 for
## Antibody responses in cholera patients and vaccines
We investigated LPS and CTB-specific IgA, IgG and IgM antibody responses in the DBS eluates and plasma specimens of cholera patients and vaccinees. There was a strong agreement of LPS-specific antibody responses between DBS eluates and plasma of cholera patients (Intraclass correlation coefficient, ICC: IgA = 0.9, IgG = 0.7 and IgM = 0.7;and vaccinees (ICC: IgA = 0.6, IgG = 0.7 and IgM = 0.8;. We also observed a strong agreement of CTB-specific antibody responses between DBS eluates and plasma of cholera
## Antibody responses in typhoid fever patients
We used S. Typhi membrane preparation (MP)to measure antibody responses in DBS eluates and plasma specimens of typhoid fever patients. There was a strong agreement of MPspecific antibody responses between DBS eluates and plasma of typhoid fever patients (ICC: IgA = 0.8, IgG = 0.7 and IgM = 0.8;. There was also a positive correlation of MPspecific antibody responses between DBS eluates and plasma of typhoid fever patients at day 1 (IgA: r = 0.7, p = 0.01; IgG: r = 0.76, p = 0.001, and IgM: r = 0.8, p = 0.001;.
# Discussion
Collection and transportation of blood specimen from field settings to laboratories is a problem in resource poor settings and especially in unpredicted epidemics and outbreaks, which often occur in places that have limited laboratory facilities. DBS can also be prepared using finger prick blood since about 50 μl of blood is only required per DBS spot. It is also a less invasive way of blood collection method that this does not necessitate the presence of skilled staff to perform phlebotomy. Thus, the diagnosis and prognosis of diseases and evaluation of disease specific immune responses after natural infection and vaccination can become less Immune responses in enteric disease using DBS difficult to assess and monitor using DBS as an alternative procedure for specimen collection. This procedure as used in this evaluation of enteric diseases and the cholera vaccine brings in an alternative specimen collection and storage method. Although the DBS method has existed for long for sampling for diagnosis of genetic disorders, recently there has been a resurgence of interest in the use of this procedure for carrying out immunological studies in vaccine trials and infectious diseases diagnosis. Enteric infections such as diarrheal diseases (cholera and ETEC diarrhea) as well as typhoid fever are endemic in Africa and Asia. Vaccination is a preventive measure to protect the vulnerable population in these regions. However, to evaluate immunological response to vaccines and natural infection requires collection of blood samples, and transportation to well-equipped laboratories in the cold. We therefore tested the possibility of the use of DBS blood collection method to evaluate immune responses in cholera, ETEC and typhoid patients as well as vaccinees in a country with high rates of enteric infections. Vibriocidal antibody response is considered one of the key surrogate markers for both natural cholera infection and vaccination in the field of cholera immunology. For the vibriocidal assay, blood needs to be collected to obtain serum or plasma. The procedure requires blood collection and then separation using centrifuges and storage and transportation under cold conditions. All these requires such facilities to be present in remote field settings and peripheral laboratories which are mostly non-existent. In this study, we have demonstrated that the vibriocidal assay can be carried out using dried blood spot, DBS specimens which provide comparable responses to that obtained with plasma in the conventional method. Vibriocidal assay requires measurement of OD values to investigate the growth of V. cholerae strain. As the DBS eluates are reddish in color because of the presence of heme makes it impossible for measuring vibriocidal antibody in conventional assay. We were able to circumvent the effect of heme in the DBS eluates by subtracting the OD values of the DBS eluates from the color of the background from test wells. This allows us to measure vibriocidal antibody responses in the 96-well plates similar to that used for the conventional vibriocidal assay. This method does not require additional determination of bacterial CFU measurement checking in microbiological culture plates and this saves time, resources and also makes the procedure less costly. The DBS based optimized vibriocidal technique that we developed does not require DBS serum separator card and it is based on a one-step procedure with measurement of optical density of bacterial growth only. However, like the conventional assay it is programmable in 96-well plate ELISA plate reader (Gene 5). Recently Iyer et. al. used DBS serum separator card to separate serum and then performed vibriocidal assay using the drop-plate technique. The dropplate technique is based on the assumption of determining bactericidal growth or killing using measurement of clumped colonies or serrated margins of growth. However, the method present here does not require additional determination of CFU of bacteria on agar-based culture plate that saves time, resources and money. The method presented here does not require serum separator cards to separate serum for performing the vibriocidal assay. In our opinion, the method described here can replace conventional assay method when only small amount of blood can be obtained from large population and also in field settings with limited resources.
Cholera is water borne disease that can be a cause of death within a few hours if not treated immediately. It has been shown that the oral cholera vaccines can protect people from three to five years from another episode. By collecting blood samples from cholera patients and oral cholera vaccinees using the DBS procedure, we have shown that immune responses measured were similar to that seen using plasma. With the DBS usage using finger prick blood collection by venipuncture can be avoided. We also showed that responses to different diseases and antigens could be carried out using DBS specimens in ELISA procedures. In addition, the antibody responses measured in ETEC-infected diarrheal patients to LTB were also comparable when either DBS or plasma was used. Given that we observed a very positive correlation of immune responses in DBS compared to plasma specimen, we extended this observation to typhoid fever patients. The immune responses against membrane preparation (MP) in DBS eluates were also similar to that seen in plasma samples.
DBS has earlier been found to be useful for measuring antibody responses in viral infections e.g. hepatitis C, HIV and Measles. The DBS specimens have also been used for analyzing antibody responses in bacterial, fungal and parasite infections and to our knowledge, there is no published data using DBS for assessing immune response in patients with cholera, ETEC diarrhea and typhoid fever. Measuring antibody responses using DBS was found to strongly correlate with plasma antibodies. We show here that the vibriocidal antibody assay which is an indirect correlate of protection in choleracan be measured using the DBS eluates and we hope that this will make it easier to measure immune responses in mass vaccination campaigns as well as in serosurveys for estimating disease burden globally. Likewise large population studies on natural infection or vaccination in areas of typhoid and ETEC diarrhea both natural infection and vaccination as well as other infectious diseases will benefit with the DBS method of sample collection and testing described in this communication. Organism isolation and antibiotic uses in this study.
## Supporting information
|
Heavily Armed Ancestors: CRISPR Immunity and Applications in Archaea with a Comparative Analysis of CRISPR Types in Sulfolobales
Prokaryotes are constantly coping with attacks by viruses in their natural environments and therefore have evolved an impressive array of defense systems. Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR) is an adaptive immune system found in the majority of archaea and about half of bacteria which stores pieces of infecting viral DNA as spacers in genomic CRISPR arrays to reuse them for specific virus destruction upon a second wave of infection. In detail, small CRISPR RNAs (crRNAs) are transcribed from CRISPR arrays and incorporated into type-specific CRISPR effector complexes which further degrade foreign nucleic acids complementary to the crRNA. This review gives an overview of CRISPR immunity to newcomers in the field and an update on CRISPR literature in archaea by comparing the functional mechanisms and abundances of the diverse CRISPR types. A bigger fraction is dedicated to the versatile and prevalent CRISPR type III systems, as tremendous progress has been made recently using archaeal models in discerning the controlled molecular mechanisms of their unique tripartite mode of action including RNA interference, DNA interference and the unique cyclic-oligoadenylate signaling that induces promiscuous RNA shredding by CARF-domain ribonucleases. The second half of the review spotlights CRISPR in archaea outlining seminal in vivo and in vitro studies in model organisms of the euryarchaeal and crenarchaeal phyla, including the application of CRISPR-Cas for genome editing and gene silencing. In the last section, a special focus is laid on members of the crenarchaeal hyperthermophilic order Sulfolobales by presenting a thorough comparative analysis about the distribution and abundance of CRISPR-Cas systems, including arrays and spacers as well as CRISPR-accessory proteins in all 53 genomes available to date. Interestingly, we find that CRISPR type III and the DNA-degrading CRISPR type I complexes co-exist in more than two thirds of these genomes. Furthermore, we identified ring nuclease candidates in all but two genomes and found that they generally co-exist with the above-mentioned CARF domain ribonucleases Csx1/Csm6. These observations, together with published literature allowed us to draft a working model of how CRISPR-Cas systems and accessory proteins cross talk to establish native CRISPR anti-virus immunity in a Sulfolobales cell. [fig_ref] Figure 1: Figure 1 [/fig_ref]. CRISPR types in prokaryotes and their mechanisms of action. Class 1 systems (including types I, III, and IV) are prevalent in both archaea and bacteria, whereas Class 2 systems (including types II, V and VI) are found almost exclusively in bacteria. Mechanisms and genes in the overlapping region are shared by the two classes. The different CRISPR steps are indicated in red and explained in the main text. Both classes contain effector complexes that can perform all three types of interference: virus DNA interference (indicated by pink scissors), virus RNA interference (indicated by green scissors) and collateral damage (i.e., promiscuous cleavage of host and virus DNA/RNA, indicated by purple scissors). "S" refers to "spacer", "R" refers to "repeat". * Asterisks refer to enzymes that are not involved in processing of all classes: Cas5d replaces Cas6 function in type I-C [bib_ref] Cas5d protein processes pre-crRNA and assembles into a cascade-like interference complex in..., Nam [/bib_ref] ; RNaseE is involved in processing in a type III-B system [bib_ref] The host-encoded RNase E endonuclease as the crRNA maturation enzyme in a..., Behler [/bib_ref] ; RNase III processes pre-crRNA in type II; tracrRNA is needed in type II and certain subtypes of type V (see text). Type I-D targets ssDNA and dsDNA (not shown) [bib_ref] DNA targeting by subtype I-D CRISPR-Cas shows type I and type III..., Lin [/bib_ref] ; In type II systems, Csn2 is also involved in spacer acquisition (not shown); spacer acquisition from RNA via RTs is not shown; ssRNA targeting of the type V-G subtype is not shown [bib_ref] Functionally diverse type V CRISPR-Cas systems, Yan [/bib_ref] (see text). CRISPR types in prokaryotes and their mechanisms of action. Class 1 systems (including types I, III, and IV) are prevalent in both archaea and bacteria, whereas Class 2 systems (including types II, V and VI) are found almost exclusively in bacteria. Mechanisms and genes in the overlapping region are shared by the two classes. The different CRISPR steps are indicated in red and explained in the main text. Both classes contain effector complexes that can perform all three types of interference: virus DNA interference (indicated by pink scissors), virus RNA interference (indicated by green scissors) and collateral damage (i.e., promiscuous cleavage of host and virus DNA/RNA, indicated by purple scissors). "S" refers to "spacer", "R" refers to "repeat". * Asterisks refer to enzymes that are not involved in processing of all classes: Cas5d replaces Cas6 function in type I-C [bib_ref] Cas5d protein processes pre-crRNA and assembles into a cascade-like interference complex in..., Nam [/bib_ref] ; RNaseE is involved in processing in a type III-B system [bib_ref] The host-encoded RNase E endonuclease as the crRNA maturation enzyme in a..., Behler [/bib_ref] ; RNase III processes pre-crRNA in type II; tracrRNA is needed in type II and certain subtypes of type V (see text). Type I-D targets ssDNA and dsDNA (not shown) [bib_ref] DNA targeting by subtype I-D CRISPR-Cas shows type I and type III..., Lin [/bib_ref] ; In type II systems, Csn2 is also involved in spacer acquisition (not shown); spacer acquisition from RNA via RTs is not shown; ssRNA targeting of the type V-G subtype is not shown [bib_ref] Functionally diverse type V CRISPR-Cas systems, Yan [/bib_ref] (see text).
Biomolecules 2020, 10, 1523 4 of 38 CRISPR immunity can be divided into three main steps, known as "Adaptation", "Processing" and "Interference" [fig_ref] Figure 1: Figure 1 [/fig_ref]. In the adaptation step, the infecting virus is recognized as intruder and a short piece of its DNA is incorporated as spacer into the CRISPR array in order to establish an adaptive immunological memory of the "enemy". Next, the spacer is transcribed and further cleaved into a CRISPR RNA (crRNA) in the "Processing" step [fig_ref] Figure 1: Figure 1 [/fig_ref]. The mature crRNA incorporates into CRISPR effector proteins, generating virus-specific "weapons" that constitutively patrol the cell to be ready for a counterstrike. The "Interference" step is initiated upon a second infection of a cognate virus DNA or RNA which is recognized and bound by a crRNA and further degraded by the catalytic action of the assembled effector proteins [fig_ref] Figure 1: Figure 1 [/fig_ref]. Notably, besides target cleavage, some CRISPR effector proteins can be activated to induce promiscuous cleavage of host and invader nucleic acids, known as collateral damage [bib_ref] C2c2 is a single-component programmable RNA-guided RNA-targeting CRISPR effector, Abudayyeh [/bib_ref]. The individual steps of CRISPR immunity with a focus on interference are discussed below.
## Crispr adaptation-know your enemy
Acquiring spacers from a freshly infecting virus represents the first crucial step defining CRISPR adaptive immunity. De novo spacer acquisition was observed predominantly within regions of free DNA ends provided on the linearized virus DNA upon injection, or upon formation of double strand breaks [bib_ref] CRISPR-Cas systems exploit viral DNA injection to establish and maintain adaptive immunity, Modell [/bib_ref] [bib_ref] CRISPR adaptation biases explain preference for acquisition of foreign DNA, Levy [/bib_ref] [bib_ref] Role of free DNA ends and protospacer adjacent motifs for CRISPR DNA..., Shiimori [/bib_ref]. From these regions, the CRISPR adaptation complex specifically recognizes and excises a pre-spacer sequence which, in most cases, is specifically selected by the presence of a subtype (or sometimes even species)-specific recognition motif termed PAM (protospacer adjacent motif). As the PAM is a prerequisite for recognizing and discriminating between foreign and host double strand DNA (dsDNA) later in the CRISPR interference step (see below), PAM-oriented acquisition ensures selection of functional spacers only [bib_ref] Short motif sequences determine the targets of the prokaryotic CRISPR defence system, Mojica [/bib_ref] [bib_ref] CRISPR interference directs strand specific spacer acquisition, Swarts [/bib_ref]. Following further processing, the novel spacer is integrated next to the leader sequence into the existing CRISPR array. Such polarized spacer acquisition allows to establish a chronological memory of infections, with the first spacer generally originating from the most recent invader [fig_ref] Figure 1: Figure 1 [/fig_ref]. Mechanistically, the integration of novel spacers involves two cleavage-ligation reactions at the leader and spacer end of the first repeat (reviewed in [bib_ref] Molecular mechanisms of CRISPR-Cas spacer acquisition, Mcginn [/bib_ref] , executed by the heterohexameric CRISPR adaptation complex consisting of the Cas1 integrase and Cas2 nuclease [bib_ref] Structural and Mechanistic Basis of PAM-Dependent Spacer Acquisition in CRISPR-Cas Systems, Wang [/bib_ref] [bib_ref] Cas1-Cas2 complex formation mediates spacer acquisition during CRISPR-Cas adaptive immunity, Nuñez [/bib_ref] [fig_ref] Figure 1: Figure 1 [/fig_ref]. Cas1 constitutes the most highly conserved cas gene associated with the majority of CRISPR types of both classes and therefore serves as marker to identify active CRISPR arrays in prokaryotic genomes [bib_ref] Evolutionary classification of CRISPR-Cas systems: A burst of class 2 and derived..., Makarova [/bib_ref] [bib_ref] The CRISPRdb database and tools to display CRISPRs and to generate dictionaries..., Grissa [/bib_ref]. Apart from Cas1-Cas2, de novo spacer acquisition is aided by other accessory proteins in some organisms, such as non-CRISPR related host factors [bib_ref] CRISPR Immunological Memory Requires a Host Factor for Specificity, Nuñez [/bib_ref] [bib_ref] Asymmetric positioning of Cas1-2 complex and Integration Host Factor induced DNA bending..., Yoganand [/bib_ref] [bib_ref] Prespacer processing and specific integration in a type I-A CRISPR system, Rollie [/bib_ref] , CRISPR interference enzymes such as the bacterial Cas9 [bib_ref] Cas9 specifies functional viral targets during CRISPR-Cas adaptation, Heler [/bib_ref] or endonucleases of the Cas4 family [bib_ref] Prespacer processing and specific integration in a type I-A CRISPR system, Rollie [/bib_ref] [bib_ref] Cas4-Dependent Prespacer Processing Ensures High-Fidelity Programming of CRISPR Arrays, Lee [/bib_ref] [bib_ref] Cas4 Nucleases Define the PAM, Length, and Orientation of DNA Fragments Integrated..., Shiimori [/bib_ref] [bib_ref] Cas4 Facilitates PAM-Compatible Spacer Selection during CRISPR Adaptation, Kieper [/bib_ref] [bib_ref] Cas4 nucleases can effect specific integration of CRISPR spacers, Zhang [/bib_ref] [bib_ref] Coupling transcriptional activation of CRISPR-Cas system and DNA repair genes by Csa3a..., Liu [/bib_ref] [bib_ref] Adaptation of the Haloarcula hispanica CRISPR-Cas system to a purified virus strictly..., Li [/bib_ref]. Recently, the latter has gained increased attention in the archaeal CRISPR field, as Cas4 was found to play significant roles in PAM recognition and length determination of pre-spacers during the adaptation process in some archaeal models [bib_ref] Prespacer processing and specific integration in a type I-A CRISPR system, Rollie [/bib_ref] [bib_ref] Cas4 Nucleases Define the PAM, Length, and Orientation of DNA Fragments Integrated..., Shiimori [/bib_ref] [bib_ref] Cas4 nucleases can effect specific integration of CRISPR spacers, Zhang [/bib_ref] [bib_ref] Coupling transcriptional activation of CRISPR-Cas system and DNA repair genes by Csa3a..., Liu [/bib_ref]. In addition, Cas4 overexpression inhibited spacer acquisition in Saccharolobus (formerly: Sulfolobus) islandicus [bib_ref] Life in hot acid: A genome-based reassessment of the archaeal order Sulfolobales, Counts [/bib_ref] , suggesting its potential to be exploited by viruses as an anti-CRISPR factor to inhibit CRISPR immunity [bib_ref] Cas4 nucleases can effect specific integration of CRISPR spacers, Zhang [/bib_ref].
Besides de novo spacer acquisition, a pre-existing spacer in the CRISPR array that fully or partially matches an incoming virus during the CRISPR type I interference reaction, can "prime" acquisition leading to a boosted immune response, as more spacers against the same intruder are collected (reviewed in [bib_ref] Molecular mechanisms of CRISPR-Cas spacer acquisition, Mcginn [/bib_ref] [bib_ref] Assembly and Translocation of a CRISPR-Cas Primed Acquisition Complex, Dillard [/bib_ref]. Interference-driven acquisition is prevalent in subtype I-F systems, where Cas2 and the interference-nuclease Cas3 (see below) are fused [bib_ref] Interference-driven spacer acquisition is dominant over naive and primed adaptation in a..., Staals [/bib_ref]. Most interestingly, there is also evidence that mRNA of DNA viruses and the genomes of RNA viruses could be sampled for spacers, which are further reverse transcribed by a Cas1-reverse transcriptase (RT) fusion enzyme (sometimes harboring an additional Cas6 domain), and are subsequently integrated as DNA spacers into the CRISPR array [bib_ref] A Reverse Transcriptase-Cas1 Fusion Protein Contains a Cas6 Domain Required for Both..., Mohr [/bib_ref] [bib_ref] Spacer acquisition from RNA mediated by a natural reverse transcriptase-Cas1 fusion protein..., González-Delgado [/bib_ref] [bib_ref] Recruitment of Reverse Transcriptase-Cas1 Fusion Proteins by Type VI-A CRISPR-Cas Systems, Toro [/bib_ref]. Even if such fusion proteins have only been detected in some bacteria so far [bib_ref] Evolutionary classification of CRISPR-Cas systems: A burst of class 2 and derived..., Makarova [/bib_ref] , bona fide RTs might as well work in concert with Cas1 in order to establish immunity against virus RNA. Indeed, CRISPR-associated RTs are located proximal to type III effector complexes capable Biomolecules 2020, 10, 1523 5 of 38 of RNA degradation (see below) in representatives of the methanogenic archaeal genus Methanosarcina as well as in Methanomethylovorans hollandica (according to own observations based on [bib_ref] Evolutionary classification of CRISPR-Cas systems: A burst of class 2 and derived..., Makarova [/bib_ref].
## Processing of crrnas-forging weapons
Upon transcription of a CRISPR array from the leader, a long precursor crRNA (CRISPR RNA) is formed and subsequently truncated within repeats into small effector crRNAs. In class I systems, Cas6 ribonucleases [bib_ref] Type IV CRISPR RNA processing and effector complex formation in Aromatoleum aromaticum, Özcan [/bib_ref] [bib_ref] Structural basis of Type IV CRISPR RNA biogenesis by a Cas6 endoribonuclease, Taylor [/bib_ref] [bib_ref] Cas6 is an endoribonuclease that generates guide RNAs for invader defense in..., Carte [/bib_ref] [bib_ref] Sequence-and structure-specific RNA processing by a CRISPR endonuclease, Haurwitz [/bib_ref] [bib_ref] Csy4 is responsible for CRISPR RNA processing in Pectobacterium atrosepticum, Przybilski [/bib_ref] [bib_ref] Cas6 specificity and CRISPR RNA loading in a complex CRISPR-Cas system, Sokolowski [/bib_ref] , Cas5d [bib_ref] Cas5d protein processes pre-crRNA and assembles into a cascade-like interference complex in..., Nam [/bib_ref] [bib_ref] Cas5d processes pre-crRNA and is a member of a larger family of..., Garside [/bib_ref] or a host RNase E [bib_ref] The host-encoded RNase E endonuclease as the crRNA maturation enzyme in a..., Behler [/bib_ref] are involved in crRNA maturation and cleave the repeats within specific regions that are often found structured into stem loops [bib_ref] Recognition and cleavage of a nonstructured CRISPR RNA by its processing endoribonuclease..., Shao [/bib_ref] [bib_ref] Approaches to study CRISPR RNA biogenesis and the key players involved, Behler [/bib_ref] (cf. [fig_ref] Figure 1: Figure 1 [/fig_ref]. In class 2 systems however, the interference effector proteins (see below) are involved in the maturation step and either execute this reaction intrinsically [bib_ref] C2c2 is a single-component programmable RNA-guided RNA-targeting CRISPR effector, Abudayyeh [/bib_ref] [bib_ref] The CRISPR-associated DNA-cleaving enzyme Cpf1 also processes precursor CRISPR RNA, Fonfara [/bib_ref] [bib_ref] Two distinct RNase activities of CRISPR-C2c2 enable guide-RNA processing and RNA detection, East-Seletsky [/bib_ref] , or in combination with a trans-activating CRISPR RNA (tracrRNA) and host RNase III, as in Cas9 systems [bib_ref] CRISPR RNA maturation by trans-encoded small RNA and host factor RNase III, Deltcheva [/bib_ref] (cf. [fig_ref] Figure 1: Figure 1 [/fig_ref]. In the latter, the tracrRNAs hybridizes to the partially complementary repeat sequences on the pre-crRNA, thereby licensing its cleavage solely within repeat regions by the dsRNA-specific ribonuclease RNase III [bib_ref] CRISPR RNA maturation by trans-encoded small RNA and host factor RNase III, Deltcheva [/bib_ref]. Different from class II systems which can vary in length of the flanking repeat residues [bib_ref] Approaches to study CRISPR RNA biogenesis and the key players involved, Behler [/bib_ref] , the generic class I mature crRNA carries a defined 5 handle derived from the preceding repeat and a 3 residue of the downstream repeat bordering the spacer [fig_ref] Figure 1: Figure 1 [/fig_ref].
## Crispr interference-counterstrike
In the CRISPR interference step, ribonucleoprotein complexes assemble with effector Cas proteins and are guided by the crRNA to complementary DNA or RNA sequences (i.e., protospacer) of a virus invader upon a "second" infection wave [fig_ref] Figure 1: Figure 1 [/fig_ref]. Through specific base pairing between crRNA and the virus protospacer, the interference machinery cleaves and eliminates the target. The interference process, target recognition and target specificity depend on the effector proteins of the specific CRISPR types, with CRISPR type I, II, V and potentially also the still ambiguous type IV primarily recognizing virus DNA substrates, while both, type III and VI identify virus RNA [fig_ref] Figure 1: Figure 1 [/fig_ref]. Based on different genomic and phylogenetic criteria [bib_ref] Evolutionary classification of CRISPR-Cas systems: A burst of class 2 and derived..., Makarova [/bib_ref] , the individual types are further divided into subtypes which often vary in the their gene and catalytic domain architectures [bib_ref] Evolutionary classification of CRISPR-Cas systems: A burst of class 2 and derived..., Makarova [/bib_ref] [bib_ref] An updated evolutionary classification of CRISPR-Cas systems, Makarova [/bib_ref]. For instance, while the signature gene cas3 is represented as a single gene in most type I subtypes, it is fused to cas2 in subtype I-F (see above) or its catalytic domains are partitioned into two genes (Cas3 and Cas3") in subtype I-A [bib_ref] Evolutionary classification of CRISPR-Cas systems: A burst of class 2 and derived..., Makarova [/bib_ref]. Furthermore in subtype I-D, the type I-specific large subunit cas8 is replaced by a type III-like large subunit cas10d fused to a Cas3"domain, while the other Cas3 domain is encoded by a separate gene, therefore representing a chimera between type I and type III systems [bib_ref] Evolutionary classification of CRISPR-Cas systems: A burst of class 2 and derived..., Makarova [/bib_ref] [bib_ref] DNA targeting by subtype I-D CRISPR-Cas shows type I and type III..., Lin [/bib_ref]. Excitingly, a recent study revealed that cas10d internally codes for another, small subunit protein (termed Cas11d) that is required for DNA binding of the interference complex [bib_ref] DNA targeting by subtype I-D CRISPR-Cas shows type I and type III..., Lin [/bib_ref]. These extra codons were also found in cas8 genes of type I-B and type I-C systems, indicating that internal translation from large subunits might be a widespread feature among certain type I subtypes.
In rare cases, subtypes even differ in their targeting specificities, as for instance, subtype V-G is so far the only experimentally verified type V representative found to specifically recognize RNA instead of DNA (see below and ref. [bib_ref] Functionally diverse type V CRISPR-Cas systems, Yan [/bib_ref]. Similarly, type I-D is so far the only known type I member exhibiting a dual, dsDNA and ssDNA targeting mode (see below and ref. [bib_ref] DNA targeting by subtype I-D CRISPR-Cas shows type I and type III..., Lin [/bib_ref]. However, as many subtypes of the different CRISPR types have not been experimentally investigated so far, such differences in target specificities remain to be examined more closely. small subunits. The latter seem to interact with the target RNA which is hybridized to the crRNA [bib_ref] Structure of an RNA silencing complex of the CRISPR-Cas immune system, Spilman [/bib_ref] [bib_ref] Structure and Activity of the RNA-Targeting Type III-B CRISPR-Cas Complex of Thermus..., Staals [/bib_ref] [bib_ref] Crystal Structure of the CRISPR-Cas RNA Silencing Cmr Complex Bound to a..., Osawa [/bib_ref] [fig_ref] Figure 2: Activated and deactivated immune response of CRISPR type III complex [/fig_ref]. The three phases of type III-mediated immunity are activated when a nascent mRNA is recognized by a crRNA in a type III complex. (1) Specific RNA cleavage: The Cas7 backbone cleaves the mRNA specifically at the protospacer region. (2) Unspecific ssDNA cleavage: When crRNA 5′ handle and PAS are unpaired (red arrow), the HD domain (pink) of the large subunit Cas10 is activated and mediates sequence-unspecific ssDNA cleavage of nearby DNA bubbles. (3) Collateral RNA shredding: As in the case for ssDNA activation, the PALM domain (violet) in the large subunit Cas10 is activated if the 5′ handle is unpaired to the PAS (red arrow). The PALM domain converts ATP into cyclic oligoadenylates which bind to the CARF domain of RNases, such as Csm6 and Csx1, thereby activating nonspecific RNA shredding. (B) Base pairing between the 5′ handle of the crRNA and the 3′ PAS. Complementarity in regions -3, -4, -5 deactivates HD and PALM domain activity, thereby only allowing specific RNA cleavage mediated by the backbone of the complex. Nucleotides in position -1, -6, -7, -8 do not contribute to base pairing, as they are distorted (-1), or tightly bound into specific pockets (-6, -7, -8). The three phases of type III-mediated immunity are activated when a nascent mRNA is recognized by a crRNA in a type III complex. (1) Specific RNA cleavage: The Cas7 backbone cleaves the mRNA specifically at the protospacer region. (2) Unspecific ssDNA cleavage: When crRNA 5 handle and PAS are unpaired (red arrow), the HD domain (pink) of the large subunit Cas10 is activated and mediates sequence-unspecific ssDNA cleavage of nearby DNA bubbles. (3) Collateral RNA shredding: As in the case for ssDNA activation, the PALM domain (violet) in the large subunit Cas10 is activated if the 5 handle is unpaired to the PAS (red arrow). The PALM domain converts ATP into cyclic oligoadenylates which bind to the CARF domain of RNases, such as Csm6 and Csx1, thereby activating nonspecific RNA shredding. (B) Base pairing between the 5 handle of the crRNA and the 3 PAS. Complementarity in regions -3, -4, -5 deactivates HD and PALM domain activity, thereby only allowing specific RNA cleavage mediated by the backbone of the complex. Nucleotides in position -1, -6, -7, -8 do not contribute to base pairing, as they are distorted (-1), or tightly bound into specific pockets (-6, -7, -8).
## Crispr dna interference
For CRISPR systems that specifically cleave dsDNA, primary target recognition relies on the presence of a specific 2-6 bp PAM flanking the matching protospacer (reviewed in [bib_ref] PAM identification by CRISPR-Cas effector complexes: Diversified mechanisms and structures, Gleditzsch [/bib_ref]. Once a cognate PAM is sensed during interrogation of the DNA by the effector proteins, the DNA is unwound, Biomolecules 2020, 10, 1523 7 of 38 allowing for subsequent hybridization of the loaded crRNA to the protospacer sequence which subsequently licenses target cleavage [bib_ref] Communicating, and Engineering the CRISPR PAM, Leenay [/bib_ref]. Notably, PAM dependent-interference protects the cells from self-targeting because CRISPR arrays, naturally providing a match to every spacer expressed, are devoid of a functional PAM in flanking repeat sequences [bib_ref] Type I-E CRISPR-Cas Systems Discriminate Target from Non-Target DNA through Base Pairing-Independent..., Westra [/bib_ref]. CRISPR type I systems in general employ a multi-subunit effector complex termed CRISPR-associated complex for antiviral defense (Cascade) which, upon PAM recognition and crRNA hybridization catalyzes protospacer degradation by the action of the Cas3 enzyme endowed with ssDNA-endonuclease (provided by an HD domain) and helicase activity. Cas3 nicks the strand opposite to the crRNA-protospacer heteroduplex and is then activated to processively cleave the upstream region [bib_ref] CRISPR Immunity Relies on the Consecutive Binding and Degradation of Negatively Supercoiled..., Westra [/bib_ref] [bib_ref] Surveillance and Processing of Foreign DNA by the Escherichia coli CRISPR-Cas System, Redding [/bib_ref] [bib_ref] Structural basis for CRISPR RNA-guided DNA recognition by Cascade, Jore [/bib_ref] [bib_ref] Cas3 is a single-stranded DNA nuclease and ATP-dependent helicase in the CRISPR/Cas..., Sinkunas [/bib_ref] [bib_ref] Structural and biochemical analysis of nuclease domain of clustered regularly interspaced short..., Mulepati [/bib_ref] [bib_ref] Small CRISPR RNAs guide antiviral defense in prokaryotes, Brouns [/bib_ref] [bib_ref] Structure basis for RNA-guided DNA degradation by Cascade and Cas3, Xiao [/bib_ref] [bib_ref] CRISPR RNA-guided DNA cleavage by reconstituted Type I-A immune effector complexes, Majumdar [/bib_ref] (c.f. [fig_ref] Figure 1: Figure 1 [/fig_ref].
Type I immunity was among the first studied CRISPR interference reactions [bib_ref] Small CRISPR RNAs guide antiviral defense in prokaryotes, Brouns [/bib_ref] with early structural and mechanistical insights for both, bacterial and archaeal Cascades [bib_ref] Structural basis for CRISPR RNA-guided DNA recognition by Cascade, Jore [/bib_ref] [bib_ref] Structural and functional characterization of an archaeal clustered regularly interspaced short palindromic..., Lintner [/bib_ref]. Intriguingly, only recently, the aforementioned type I-D Cascade complex of the archaeaon Saccharolobus islandicus LAL 14/1 was biochemically characterized which revealed a unique chimeric cleavage mode, resembling type I and type III-targeting (see below) [bib_ref] DNA targeting by subtype I-D CRISPR-Cas shows type I and type III..., Lin [/bib_ref]. Besides type I-bona fide PAM-dependent cleavage of dsDNA executed by the action of a Cas3 helicase and Cas10d endonuclease (see above), the I-D Cascade backbone specifically degraded ssDNA protospacers in a ruler-like and PAM independent manner.
Type II systems act via a dual RNA guide composed of crRNA and tracrRNA together with the single-component effector Cas9 against DNA, cleaving both strands of a PAM-flanked protospacer mediated by the opposing RuvC-like and a HNH-nuclease domains [bib_ref] A programmable dual-RNA-guided DNA endonuclease in adaptive bacterial immunity, Jinek [/bib_ref] (cf. [fig_ref] Figure 1: Figure 1 [/fig_ref]. Similarly, the type V specific signature protein Cas12 carries a RuvC domain executing dsDNA cleavage of the protospacer in dependence of a PAM (cf. [fig_ref] Figure 1: Figure 1 [/fig_ref]. Different to type II systems however, the majority of the various type V subtypes do not require a tracrRNA [bib_ref] Functionally diverse type V CRISPR-Cas systems, Yan [/bib_ref] [bib_ref] Cpf1 Is a Single RNA-Guided Endonuclease of a Class 2 CRISPR-Cas System, Zetsche [/bib_ref] and some were shown to cleave protospacers placed on M13 phage-ssDNA in a PAM-independent manner [bib_ref] CRISPR-Cas12a target binding unleashes indiscriminate single-stranded DNase activity, Chen [/bib_ref] (cf. [fig_ref] Figure 1: Figure 1 [/fig_ref]. Remarkably, once activated by hybridization to a protospacer on ds or ssDNA, some Type V subtypes, including the well-studied effectors Cas12a and Cas12b1, executed collateral damage by non-specifically shredding ssDNA [bib_ref] Functionally diverse type V CRISPR-Cas systems, Yan [/bib_ref] (c.f. [fig_ref] Figure 1: Figure 1 [/fig_ref]. Such collateral damage might represent a robust way to quickly eradicate the infecting virus following the tabula rasa principle. However, indiscriminate nucleic acid cleavage also harms the host and might induce dormancy or even suicide of the infected cell. Thus, reminiscent of abortive infection, collateral damage represents an altruistic facet of the CRISPR immunity, sacrificing the infected cell for the sake of the population [bib_ref] Coupling immunity and programmed cell suicide in prokaryotes: Life-or-death choices, Koonin [/bib_ref].
In comparison to the above-mentioned systems, the function and role of the CRISPR type IV systems is still enigmatic, as these systems are predominantly carried with bacterial plasmids and lack a nuclease domain [bib_ref] Evolutionary classification of CRISPR-Cas systems: A burst of class 2 and derived..., Makarova [/bib_ref]. Instead, they often encode a helicase, which was recently shown to be involved in conferring immunity against plasmids in interference assays [bib_ref] A Type IV-A CRISPR-Cas System in Pseudomonas aeruginosa Mediates RNA-Guided Plasmid Interference..., Crowley [/bib_ref]. In line with this, spacers of adjacent CRISPR arrays were found to predominantly match other plasmids, suggesting that the type IV system might specifically serve to interfere with competing plasmids [bib_ref] Type IV CRISPR-Cas systems are highly diverse and involved in competition between..., Pinilla-Redondo [/bib_ref].
60% of both, CRISPR-equipped archaea and bacteria, destroy virus DNA using Type I Cascade [bib_ref] Evolutionary classification of CRISPR-Cas systems: A burst of class 2 and derived..., Makarova [/bib_ref] , including representatives of famous archaeal and bacterial (CRISPR) model organisms such as the archaeon Haloferax volcanii and Escherichia coli. Thus, type I systems are the most prevalent CRISPR-Cas systems in prokaryotes, whereas type II and V systems are generally underrepresented and rare in archaea. The absence of class 2 systems (including type II, V, VI) in archaea might go along with the absence of the bacteria-specific RNase III ribonuclease which is required for crRNA maturations in most of these systems (see above). Only recently, type V-F CRISPR-Cas systems were detected in uncultivated archaea of the DPANN superphylum [bib_ref] Programmed DNA destruction by miniature CRISPR-Cas14 enzymes, Harrington [/bib_ref] and type II-Cas9 systems were found in two representatives of the nanoarchaeota [bib_ref] New CRISPR-Cas systems from uncultivated microbes, Burstein [/bib_ref]. However, in case of the latter, the activity remains to be investigated in the native host, as neither in vitro nor in vivo cleavage was detected when purified from or examined in heterologous models [bib_ref] New CRISPR-Cas systems from uncultivated microbes, Burstein [/bib_ref].
## "rnattack"
Different from the above-mentioned types, all subtypes of CRISPR type III and type VI systems recognize the invader s RNA. Class II-specific type VI CRISPR-Cas systems are exclusively found in bacteria and are absent from archaea [bib_ref] Evolutionary classification of CRISPR-Cas systems: A burst of class 2 and derived..., Makarova [/bib_ref]. They consist of the single-component effector protein Cas13, endowed with RNase activity provided by two conserved higher eukaryotes and prokaryotes, nucleotide binding (HEPN) domains, and confer immunity against a ssRNA phage carrying a cognate protospacer [bib_ref] C2c2 is a single-component programmable RNA-guided RNA-targeting CRISPR effector, Abudayyeh [/bib_ref] [bib_ref] Cas13b Is a Type VI-B CRISPR-Associated RNA-Guided RNase Differentially Regulated by Accessory..., Smargon [/bib_ref] [bib_ref] Cas13d Is a Compact RNA-Targeting Type VI CRISPR Effector Positively Modulated by..., Yan [/bib_ref] [bib_ref] Two HEPN domains dictate CRISPR RNA maturation and target cleavage in Cas13d, Zhang [/bib_ref]. For some, but not all subtypes, target recognition seemed to be dependent on a protospacer flanking site (PFS) located either downstream or upstream of the protospacer [bib_ref] C2c2 is a single-component programmable RNA-guided RNA-targeting CRISPR effector, Abudayyeh [/bib_ref] [bib_ref] Cas13b Is a Type VI-B CRISPR-Associated RNA-Guided RNase Differentially Regulated by Accessory..., Smargon [/bib_ref] [bib_ref] RNA targeting with CRISPR-Cas13, Abudayyeh [/bib_ref]. Initial target recognition was reported to be highly specific, as already two mismatches in the central region of the crRNA, abolished target identification [bib_ref] C2c2 is a single-component programmable RNA-guided RNA-targeting CRISPR effector, Abudayyeh [/bib_ref] [bib_ref] RNA targeting with CRISPR-Cas13, Abudayyeh [/bib_ref]. This sensitivity has encouraged scientists to exploit Cas13 as tool for tracking of pathogens such as Zika virus, and it might be even applicable for quick detection of Covid-19 RNA [bib_ref] Clinical validation of a Cas13-based assay for the detection of SARS-CoV-2 RNA, Patchsung [/bib_ref]. Similarly to the collateral damage against ssDNA observed in type V enzymes, once activated by binding to the protospacer on the RNA, Cas13 cleaves ssRNA non-specifically [bib_ref] C2c2 is a single-component programmable RNA-guided RNA-targeting CRISPR effector, Abudayyeh [/bib_ref] (cf. [fig_ref] Figure 1: Figure 1 [/fig_ref]. This seems to be initiated by the dramatic conformational change of the protein upon target binding, which approximates the two HEPN domains and exposes them to the environment, generating a new catalytic site indiscriminately cleaving any proximate RNA substrate. Thus, upon specific detection of probably even lowly abundant numbers of ssRNA phages or virus transcripts, type VI effectors transform into promiscuous RNases that eradicate the virus but cause fitness loss of the cell [bib_ref] C2c2 is a single-component programmable RNA-guided RNA-targeting CRISPR effector, Abudayyeh [/bib_ref] [bib_ref] Cas13-induced cellular dormancy prevents the rise of CRISPR-resistant bacteriophage, Meeske [/bib_ref]. A similar immune response was recently reported for a newly characterized type V effector Cas12g, extracted from a hot spring metagenome. When heterologously expressed in E. coli, Cas12g collaterally cleaved both, ssRNA and ssDNA following specific recognition of a protospacer on virus ssRNA [bib_ref] Functionally diverse type V CRISPR-Cas systems, Yan [/bib_ref].
Class I-specific type III systems form multiprotein complexes, which are structurally similar to the DNA-targeting type I complex Cascade, suggesting a common multi-subunit ancestor [fig_ref] Figure 1: Figure 1 [/fig_ref]. Type III systems are often found in archaea (~35%) but are also common in bacteria (~25%), and therefore constitute the second-most abundant CRISPR type [bib_ref] Evolutionary classification of CRISPR-Cas systems: A burst of class 2 and derived..., Makarova [/bib_ref]. Interestingly, hyperthermophilic archaea of the phylum crenarchaeota are considerably enriched for CRISPR type III effector complexes, which often coexist with CRISPR type I systems, sometimes even sharing crRNAs from the same CRISPR arrays in an organism (see below) [bib_ref] Structure of the CRISPR interference complex CSM reveals key similarities with cascade, Rouillon [/bib_ref] [bib_ref] Structure and mechanism of the CMR complex for CRISPR-mediated antiviral immunity, Zhang [/bib_ref]. Besides RNA targeting and cleavage, type III systems also degrade DNA and, similar to type VI and V systems, can induce collateral cell damage through promiscuous RNA shredding activated by a novel signaling pathway. Thus, type III systems have three different ways to confer immunity to a cell, representing the Swiss army knife of all CRISPR.
## All good things come in (type) threes-a tripartite immune response of crispr type iii systems
Type III systems are further subdivided into subtypes A to F [bib_ref] Evolutionary classification of CRISPR-Cas systems: A burst of class 2 and derived..., Makarova [/bib_ref] [bib_ref] CRISPR adaptive immune systems of Archaea, Vestergaard [/bib_ref] , of which types III-A through D are the most abundant subtypes with experimentally described variants. Contrarily, the recently defined subtypes III-E and III-F are only found in few microorganisms and remain to be experimentally investigated [bib_ref] Evolutionary classification of CRISPR-Cas systems: A burst of class 2 and derived..., Makarova [/bib_ref] [bib_ref] CRISPR adaptive immune systems of Archaea, Vestergaard [/bib_ref]. The most intensely studied type III complexes belong to types III-A, III-D (collectively referred to as Csm) and type III-B (referred to as Cmr). The large subunit Cas10 is generally found in types A-D as well as F and is endowed with cyclase (PALM) and nuclease (HD) domains (sometimes inactivated or missing in certain III-C or III-D variants) (see below, [fig_ref] Figure 2: Activated and deactivated immune response of CRISPR type III complex [/fig_ref]. Furthermore, type III-B and type III-A subtypes can be further differentiated by the presence of certain signature genes encoding for the small subunit of the effector complex, namely cmr5 in III-B, and csm2 in III-A [bib_ref] An updated evolutionary classification of CRISPR-Cas systems, Makarova [/bib_ref]. Subtype III-D CRISPR-Cas loci share the general effector complex gene composition with those of subtype III-A, however, they encode a distinct cas5 variant, termed csx10, which is a signature gene for this subtype [bib_ref] Evolutionary classification of CRISPR-Cas systems: A burst of class 2 and derived..., Makarova [/bib_ref] [bib_ref] An updated evolutionary classification of CRISPR-Cas systems, Makarova [/bib_ref]. Type III-B complexes often carry the additional subunits Cmr1 and Cmr7, implicated in capping of the target 3 end and possibly allosteric regulation of RNA cleavage efficiency, respectively [bib_ref] A seed motif for target RNA capture enables efficient immune defence by..., Pan [/bib_ref] [bib_ref] Structures of the Cmr-β Complex Reveal the Regulation of the Immunity Mechanism..., Sofos [/bib_ref].
Despite those differences, however, type III-A/D and type III-B ribonucleolytic complexes exhibit similar architecture and both bind a class I-specific crRNA (see above and [fig_ref] Figure 1: Figure 1 [/fig_ref]. The crRNA is bound by the backbone consisting of varying numbers of Cas7-subunits (i.e., Cmr4/Csm3), reaching into the Cas5 (Cmr3/Csm2) protein with its 5 handle. The base of the complex is formed by the Cas10 family protein which interacts with a scaffold consisting of an assembly of small subunits. The latter seem to interact with the target RNA which is hybridized to the crRNA [bib_ref] Structure of an RNA silencing complex of the CRISPR-Cas immune system, Spilman [/bib_ref] [bib_ref] Structure and Activity of the RNA-Targeting Type III-B CRISPR-Cas Complex of Thermus..., Staals [/bib_ref] [bib_ref] Crystal Structure of the CRISPR-Cas RNA Silencing Cmr Complex Bound to a..., Osawa [/bib_ref] [fig_ref] Figure 2: Activated and deactivated immune response of CRISPR type III complex [/fig_ref].
## Specific rna cleavage
The initial opinion about the target specificity and function of type III systems was split among researchers. Initial in vivo studies of a type III-A system in the mesophilic bacterium Staphylococcus epidermidis showed that the system conferred immunity against a plasmid DNA carrying a matching protospacer [bib_ref] Self versus non-self discrimination during CRISPR RNA-directed immunity, Marraffini [/bib_ref] [bib_ref] CRISPR Interference Limits Horizontal Targeting DNA, Marraffini [/bib_ref] , whereas in vitro studies of type III-B systems of the hyperthermophilic archaea Pyrococcus furiosus and Saccharolobus (previously known as Sulfolobus) solfataricus [bib_ref] Life in hot acid: A genome-based reassessment of the archaeal order Sulfolobales, Counts [/bib_ref] and the thermophilic bacterium Thermus thermophilus, all demonstrated strong degradation of ssRNA upon crRNA hybridization [bib_ref] Structure and mechanism of the CMR complex for CRISPR-mediated antiviral immunity, Zhang [/bib_ref] [bib_ref] Structure and Activity of the RNA-Targeting Type III-B CRISPR-Cas Complex of Thermus..., Staals [/bib_ref] [bib_ref] RNA-guided RNA cleavage by a CRISPR RNA-Cas protein complex, Hale [/bib_ref]. Therefore, despite the striking similarity between Csm and Cmr, they were attributed different target specificities in the early days. Only later, biochemical analysis revealed that type III-A systems universally cleave crRNA-recognized target ssRNA into specific fragment lengths, revealing a 6 nt interval cleavage pattern [bib_ref] Structure and Activity of the RNA-Targeting Type III-B CRISPR-Cas Complex of Thermus..., Staals [/bib_ref] [bib_ref] RNA Targeting by the Type III-A CRISPR-Cas Csm Complex of Thermus thermophilus, Staals [/bib_ref] [bib_ref] Programmable RNA Shredding by the Type III-A CRISPR-Cas System of Streptococcus thermophilus, Tamulaitis [/bib_ref]. The 6 nt spaced cleavage was later confirmed for the above-mentioned type III-B systems and additionally, for the type III-B of the hyperthermophilic bacterium Thermotoga maritima [bib_ref] RNA-activated DNA cleavage by the Type III-B CRISPR-Cas effector complex, Estrella [/bib_ref] [bib_ref] Multiple nucleic acid cleavage modes in divergent type III CRISPR systems, Zhang [/bib_ref] [bib_ref] Structural Model of a CRISPR RNA-Silencing Complex Reveals the RNA-Target Cleavage Activity..., Benda [/bib_ref] [bib_ref] Target RNA capture and cleavage by the Cmr type III-B CRISPR-Cas effector..., Hale [/bib_ref] [bib_ref] Cmr4 is the slicer in the RNA-targeting Cmr CRISPR complex, Zhu [/bib_ref]. The endoribonucleolytic reaction was found to be executed by Cmr4/Csm3 subunits constituting the middle units of the Cas7 backbone [bib_ref] Programmable RNA Shredding by the Type III-A CRISPR-Cas System of Streptococcus thermophilus, Tamulaitis [/bib_ref] [bib_ref] Structural Model of a CRISPR RNA-Silencing Complex Reveals the RNA-Target Cleavage Activity..., Benda [/bib_ref] [bib_ref] Cmr4 is the slicer in the RNA-targeting Cmr CRISPR complex, Zhu [/bib_ref]. A crystal structure of a chimeric Type III complex assembled from subunits of the two archaea Archaeoglobus fulgidus and P. furiosus bound to a target RNA revealed the reason for this periodic cleavage pattern: every Cmr4 backbone subunit penetrates the crRNA-target RNA duplex with a β-hairpin residue, leading to a distortion at every 6th nucleotide (cf. [fig_ref] Figure 2: Activated and deactivated immune response of CRISPR type III complex [/fig_ref] [bib_ref] Crystal Structure of the CRISPR-Cas RNA Silencing Cmr Complex Bound to a..., Osawa [/bib_ref]. Only recently, a very detailed structural investigation of a type III-B system of the archaeon Sulfolobus islandicus bound to different target RNAs confirmed these earlier observations, and suggested that additionally to the Cmr4, also the opposing small subunit Cmr5 might play a role in the cleavage reaction [bib_ref] Structures of the Cmr-β Complex Reveal the Regulation of the Immunity Mechanism..., Sofos [/bib_ref].
Collectively, these studies clearly confirmed that RNase activity was a universal feature of CRISPR type III complexes and in vivo investigations have proven this to hold true in the living cell, where virus mRNA carrying a protospacer complementary to a host crRNA was efficiently degraded [bib_ref] An archaeal CRISPR type III-B system exhibiting distinctive RNA targeting features and..., Peng [/bib_ref] [bib_ref] CRISPR-mediated targeted mRNA degradation in the archaeon Sulfolobus solfataricus, Zebec [/bib_ref] (see below).
## Unspecific ssdna cleavage
The question regarding a potential type III-encoded DNase activity remained, as later in vivo studies observed transcription-dependent DNA cleavage of type III systems [bib_ref] A novel interference mechanism by a type IIIB CRISPR-Cmr module in Sulfolobus, Deng [/bib_ref] [bib_ref] A Conditional tolerance of temperate phages via transcription-dependent CRISPR-Cas targeting, Goldberg [/bib_ref]. Biochemical analysis finally solved the mystery: type III systems cleaved ssDNA promiscuously, once activated by subtle conformational change upon binding of the type III effector complex to a target RNA [bib_ref] Structure and mechanism of the CMR complex for CRISPR-mediated antiviral immunity, Zhang [/bib_ref] [bib_ref] Structures of the Cmr-β Complex Reveal the Regulation of the Immunity Mechanism..., Sofos [/bib_ref] [bib_ref] RNA-activated DNA cleavage by the Type III-B CRISPR-Cas effector complex, Estrella [/bib_ref] [bib_ref] Multiple nucleic acid cleavage modes in divergent type III CRISPR systems, Zhang [/bib_ref] [bib_ref] Bipartite recognition of target RNAs activates DNA cleavage by the Type III-B..., Elmore [/bib_ref] [bib_ref] Co-transcriptional DNA and RNA Cleavage during Type III CRISPR-Cas Immunity, Samai [/bib_ref] [bib_ref] A type III-B CRISPR-Cas effector complex mediating massive target DNA destruction, Han [/bib_ref] [bib_ref] Structure Studies of the CRISPR-Csm Complex Reveal Mechanism of Co-transcriptional Interference, You [/bib_ref]. Cleavage was shown to be mediated by the large subunit Cas10 which is endowed with ssDNA activity conferred by its HD domain [bib_ref] Bipartite recognition of target RNAs activates DNA cleavage by the Type III-B..., Elmore [/bib_ref] [bib_ref] Crystal structure of the Csm1 subunit of the Csm complex and its..., Jung [/bib_ref] [bib_ref] Spatiotemporal Control of Type III-A CRISPR-Cas Immunity: Coupling DNA Degradation with the..., Kazlauskiene [/bib_ref] [fig_ref] Figure 2: Activated and deactivated immune response of CRISPR type III complex [/fig_ref]. Binding of the target mRNA allosterically activates the nuclease domain, as ssDNA cleavage was significantly reduced [bib_ref] Spatiotemporal Control of Type III-A CRISPR-Cas Immunity: Coupling DNA Degradation with the..., Kazlauskiene [/bib_ref] or totally abolished [bib_ref] Co-transcriptional DNA and RNA Cleavage during Type III CRISPR-Cas Immunity, Samai [/bib_ref] when the complex was devoid of a target RNA, or when the target RNA disassociated upon cleavage. Thus, the in vitro findings aligned well with the concept of transcription-dependent DNA degradation drawn from in vivo studies: In the cell, the crRNA binds the nascent virus mRNA and similar to a leash, approximates the type III complex to the transcription bubble or any other R-loop nearby [bib_ref] Target preference of Type III-A CRISPR-Cas complexes at the transcription bubble, Liu [/bib_ref] , which consequently offers a ssDNA substrate for cleavage by Cas10 [fig_ref] Figure 2: Activated and deactivated immune response of CRISPR type III complex [/fig_ref]. Notably, mechanistic insights of a type III-B complex from S. islandicus, that differs from other type III systems by the presence of additional Cmr7 subunits (referred to as Cmr-β), showed that the HD domain also catalyzed unspecific ssRNA cleavage [bib_ref] Structures of the Cmr-β Complex Reveal the Regulation of the Immunity Mechanism..., Sofos [/bib_ref] , at least in this complex. Upon degradation of the target RNA by the Cas7 backbone, the complex is released and the DNase domain inactivated. Importantly, ssDNA degradation is fully abolished, when the 3 end of the target RNA was complementary to the repeat-derived 5 handle of the crRNA [fig_ref] Figure 2: Activated and deactivated immune response of CRISPR type III complex [/fig_ref] [bib_ref] Structures of the Cmr-β Complex Reveal the Regulation of the Immunity Mechanism..., Sofos [/bib_ref] [bib_ref] RNA-activated DNA cleavage by the Type III-B CRISPR-Cas effector complex, Estrella [/bib_ref] [bib_ref] Spatiotemporal Control of Type III-A CRISPR-Cas Immunity: Coupling DNA Degradation with the..., Kazlauskiene [/bib_ref]. Mutational studies revealed that matches in positions -3, -4, -5 sufficed to inhibit promiscuous ssDNA cleavage by the HD domain in vivo [bib_ref] Unexpectedly broad target recognition of the CRISPR-mediated virus defence system in the..., Manica [/bib_ref] which was confirmed by structural studies showing that the other nucleotides (except for the -2 positions) were inaccessible for Watson-Crick base pairing, as they were either distorted or tightly bound into specific pockets of Cas5 [fig_ref] Figure 2: Activated and deactivated immune response of CRISPR type III complex [/fig_ref] [bib_ref] Structures of the Cmr-β Complex Reveal the Regulation of the Immunity Mechanism..., Sofos [/bib_ref] [bib_ref] Crystal Structure of the CRISPR-Cas RNA Silencing Cmr Complex Bound to a..., Osawa [/bib_ref] [bib_ref] Structure Studies of the CRISPR-Csm Complex Reveal Mechanism of Co-transcriptional Interference, You [/bib_ref]. The 3 region on the target mRNA is here referred to as protospacer adjacent sequence (PAS) (also termed "PFS" for protospacer flanking sequence or "auto-immunity tag") and, similar to the above-mentioned PAM, inhibits type III-mediated targeting of the chromosomal CRISPR array as the repeats, being the origin of the 5 handle, represent a matching PAS. Therefore, even in the rare cases of antisense transcription of a CRISPR array (providing a RNA complementary to the crRNA) [bib_ref] CRISPR families of the crenarchaeal genus Sulfolobus: Bidirectional transcription and dynamic properties, Lillestøl [/bib_ref] [bib_ref] RNA-guided RNA cleavage by a CRISPR RNA-Cas protein complex, Hale [/bib_ref] , the HD domain of the type III complex would not be activated owing to the presence of a PAS in the repeat.
## Collateral ssrna cleavage via coa signaling
Type III genes are often found located proximal to genes coding for proteins endowed with a specific ligand-binding CARF domain (CRISPR -associated Rossmann fold) and an effector domain, most commonly a RNase-specific HEPN domain [bib_ref] Ligand-binding regulators of prokaryotic defense systems, Makarova [/bib_ref]. While one member of the CARF protein family, Csa3, was shown to regulate spacer acquisition and crRNA synthesis on the transcriptional level in type I-A [bib_ref] Coupling transcriptional activation of CRISPR-Cas system and DNA repair genes by Csa3a..., Liu [/bib_ref] , Csm6 and Csx1 were found to degrade RNA in vitro [bib_ref] Structural basis for the endoribonuclease activity of the type III-A CRISPR-associated protein..., Niewoehner [/bib_ref] [bib_ref] The CRISPR-associated Csx1 protein of Pyrococcus furiosus is an adenosine-specific endoribonuclease, Sheppard [/bib_ref] [bib_ref] Allosteric regulation of Csx1, a type IIIB-associated CARF domain ribonuclease by RNAs..., Han [/bib_ref] and their genetic disruption seemed to decrease the type III immune reaction in vivo [bib_ref] A novel interference mechanism by a type IIIB CRISPR-Cmr module in Sulfolobus, Deng [/bib_ref] [bib_ref] Degradation of Phage Transcripts by CRISPR-Associated RNases Enables Type III CRISPR-Cas Immunity, Jiang [/bib_ref]. However, as neither of these enzymes seemed to physically interact with the type III complex [bib_ref] Structure and Activity of the RNA-Targeting Type III-B CRISPR-Cas Complex of Thermus..., Staals [/bib_ref] [bib_ref] Crystal Structure of the CRISPR-Cas RNA Silencing Cmr Complex Bound to a..., Osawa [/bib_ref] [bib_ref] Programmable RNA Shredding by the Type III-A CRISPR-Cas System of Streptococcus thermophilus, Tamulaitis [/bib_ref] , their interplay and specific role in type III immunity have long remained elusive. Three years ago, two breakthrough studies revealed that CARF-domain enzymes were specifically activated by the type III complex to initiate massive RNA shredding [bib_ref] Type III CRISPR-Cas systems produce cyclic oligoadenylate second messengers, Niewoehner [/bib_ref] [bib_ref] A cyclic oligonucleotide signaling pathway in type III CRISPR-Cas systems, Kazlauskiene [/bib_ref]. The activation was mediated by the large subunit Cas10 which, additionally to the above-mentioned HD domain, harbors two PALM domains, one carrying a typical GGDD motif, resembling the core domain generally found in nucleotidyl cyclases [bib_ref] Evolution and classification of the CRISPR-Cas systems, Makarova [/bib_ref]. Upon crRNA-target RNA binding, the conformational change of the type III effector complex activates the cyclase of this PALM domain which starts to polymerize available ATP into cyclic oligoadenylates (cOA), newly discovered second messenger molecules consisting of three to six AMP monomers (cA3 to cA6). These cyclic molecules in turn allosterically activate the HEPN-ribonucleases Csm6 and Csx1 by binding to their CARF domains [bib_ref] Type III CRISPR-Cas systems produce cyclic oligoadenylate second messengers, Niewoehner [/bib_ref] [bib_ref] A cyclic oligonucleotide signaling pathway in type III CRISPR-Cas systems, Kazlauskiene [/bib_ref] [bib_ref] Structure of Csx1-cOA4 complex reveals the basis of RNA decay in Type..., Molina [/bib_ref] [bib_ref] Cyclic oligoadenylate signalling mediates Mycobacterium tuberculosis CRISPR defence, Grüschow [/bib_ref] [bib_ref] A Type III-B Cmr effector complex catalyzes the synthesis of cyclic oligoadenylate..., Han [/bib_ref] , leading to collateral shredding of RNA substrates in the cell. Furthermore, recently the DNA nickase Can1 of Thermus thermophilus was also found to be activated by cA4, indicating that cOA-signaling can also impact the DNA level [bib_ref] Structure and mechanism of a Type III CRISPR defence DNA nuclease activated..., Mcmahon [/bib_ref]. Similar to type VI systems (see above), cOA-induced RNA shredding was shown to lead to growth arrest which, if continued, could potentially result in cell dormancy or death [bib_ref] Non-specific degradation of transcripts promotes plasmid clearance during type III-A CRISPR-Cas immunity, Rostøl [/bib_ref]. Even if the cOA synthesis is deactivated upon degradation of the target mRNA [bib_ref] Structures of the Cmr-β Complex Reveal the Regulation of the Immunity Mechanism..., Sofos [/bib_ref] , already produced cOA can remain in the cell for a longer period, thus keeping HEPN-RNases activated. Only recently, a new protein family, the ring nucleases Crn1 (CRISPR-associated ring nuclease 1), was discovered in the archaeon Saccharolobus solfataricus [bib_ref] Ring nucleases deactivate type III CRISPR ribonucleases by degrading cyclic oligoadenylate, Athukoralage [/bib_ref] and shown to specifically linearize and therefore inactivate cOA molecules, resetting the cell to a ground state once the virus is defeated by type III systems. While Crn1 ring nucleases seem to be limited to crenarchaeota, a new type, Crn3 (former Csx3) was recently biochemically characterized in the euryarchaeon A. fulgidus and found to be widespread in prokaryotes. Furthermore, Csx1/Csm6 ribonucleases, or even Crn-Csx fusion proteins exist which intrinsically degrade the cOA substrates themselves, thus representing self-limiting enzymes independent of a trans-acting ring nuclease [bib_ref] CRISPR-Cas III-A Csm6 CARF Domain Is a Ring Nuclease Triggering Stepwise cA4..., Jia [/bib_ref] [bib_ref] Regulation of the RNA and DNA nuclease activities required for Pyrococcus furiosus..., Foster [/bib_ref] [bib_ref] Activation and self-inactivation mechanisms of the cyclic oligoadenylate-dependent CRISPR ribonuclease Csm6, Garcia-Doval [/bib_ref] [bib_ref] A Type III CRISPR Ancillary Ribonuclease Degrades Its Cyclic Oligoadenylate Activator, Athukoralage [/bib_ref]. Moreover, in S. islandicus REY15A, a membrane-associated DHH-DHHA1 family nuclease (MAD) was recently shown to degrade cOA in vitro and is hypothesized to aid the main cellular ring nuclease in controlling type III immunity by possibly degrading diffused cOA. Intriguingly, specialized forms of ring nucleases were recently found encoded on genomes of archaeal viruses and bacteriophages and were shown to function as anti-CRISPRs to counteract the cOA-induced immune response [bib_ref] An anti-CRISPR viral ring nuclease subverts type III CRISPR immunity, Athukoralage [/bib_ref]. Thus, collateral damage by cOA-signaling might constitute the most effective stage of virus defense by type III systems, which is probably especially important when the protospacer is situated in a late-expressed or lowly-expressed viral gene [bib_ref] Degradation of Phage Transcripts by CRISPR-Associated RNases Enables Type III CRISPR-Cas Immunity, Jiang [/bib_ref] [bib_ref] Non-specific degradation of transcripts promotes plasmid clearance during type III-A CRISPR-Cas immunity, Rostøl [/bib_ref]. In such scenarios, transcription-dependent ssDNA cleavage of type III is inefficient due to the low transcript number available, leading to an accumulation of the virus in the cell. Yet, cOAs might still be sufficiently produced, causing an interim cell shut down due to promiscuous RNA shredding, thereby preventing completion of the lytic virus cycle and buying time for type III-mediated virus DNA clearance [bib_ref] Degradation of Phage Transcripts by CRISPR-Associated RNases Enables Type III CRISPR-Cas Immunity, Jiang [/bib_ref] [bib_ref] Non-specific degradation of transcripts promotes plasmid clearance during type III-A CRISPR-Cas immunity, Rostøl [/bib_ref]. In line with this, a recently discovered anti-CRISPR protein seemed to physically interfere with cOA-mediated virus defense by binding to the type III complex when middle/late viral transcripts were targeted (see below and ref..
Importantly, just as the HD domain (see above), the PALM domain and therefore cOA signaling remained inactive when the bound target mRNA contained a PAS hybridizing to the 5 handle [bib_ref] Type III CRISPR-Cas systems produce cyclic oligoadenylate second messengers, Niewoehner [/bib_ref] [bib_ref] A cyclic oligonucleotide signaling pathway in type III CRISPR-Cas systems, Kazlauskiene [/bib_ref] , as it was shown that the induced conformational change leads to blockage of the entrance channel of the cOA substrates [bib_ref] Structures of the Cmr-β Complex Reveal the Regulation of the Immunity Mechanism..., Sofos [/bib_ref] [bib_ref] Structure Studies of the CRISPR-Csm Complex Reveal Mechanism of Co-transcriptional Interference, You [/bib_ref] [bib_ref] Type III-A CRISPR-Cas Csm Complexes: Assembly, Periodic RNA Cleavage, Jia [/bib_ref].
In summary, one can say that type III complexes pull out all the stops to efficiently curtail a virus spread: First, the virus transcript is recognized and eventually sliced within the protospacer region by the action of the backbone endoribonucleases. Second, when the complex is bound to a target, ssDNA is cleaved in DNA bubbles which is catalyzed by the HD domain of the Cas10 subunit. Third, unspecific RNA shredding by CARF-domain nucleases is activated via secondary molecules synthesized by the PALM domain of Cas10 [fig_ref] Figure 2: Activated and deactivated immune response of CRISPR type III complex [/fig_ref]. However, all but the first reaction, are allosterically blocked when the 3 end of the target (PAS) binds the 5 handle [fig_ref] Figure 2: Activated and deactivated immune response of CRISPR type III complex [/fig_ref]. Thus, PAS-handle complementarity constitutes an "off switch" for all secondary immune reactions of type III systems, permitting backbone cleavage of the target RNA only.
## Crispr research and application in archaeal model organisms
Shortly after the first discovery of the -back then still enigmatic -regularly spaced CRISPR repeats in the E. coli genome in 1987 [bib_ref] Nucleotide sequence of the iap gene, responsible for alkaline phosphatase isozyme conversion..., Ishino [/bib_ref] , these arrays were also identified within early sequencing studies of halophilic euryarchaea [bib_ref] Transcription at different salinities of Haloferax mediterranei sequences adjacent to partially modified..., Mojica [/bib_ref] [bib_ref] Long stretches of short tandem repeats are present in the largest replicons..., Mojica [/bib_ref]. Whereas initially not experimentally investigated in any bacterium, in vivo studies specifically dedicated to unravelling the physiological role of those repeats were first conducted in Haloferax volcanii, where a plasmid engineered with CRISPR repeats was used in transformation assays [bib_ref] Long stretches of short tandem repeats are present in the largest replicons..., Mojica [/bib_ref]. Back then, twelve years before the function of CRISPR as an immune system had been resolved, the thereby observed reduction of cell viability and chromosomal content of the polyploid organism was interpreted as a role of the repeats in replicon partitioning [bib_ref] Long stretches of short tandem repeats are present in the largest replicons..., Mojica [/bib_ref]. However, from today s view, Mojica and colleagues might have witnessed CRISPR-mediated self-targeting, triggered by an increased acquisition of chromosomal spacers into the extra CRISPR array [157]-a nowadays well-known phenomenon in halophilic archaea (see below). Thus, even if misinterpreted at the time, one can argue that this study represented the first CRISPR-directed in vivo experiment in a prokaryote.
The following chapter delineates prominent archaeal model organisms that were used to study and characterize CRISPR mechanisms in archaea. We give an overview about seminal experiments and the current research performed in those organisms highlighting special features of the individual lab strains. A short section is dedicated to the CRISPR applications in archaeal models. Furthermore, a list of the most prominent archaeal lab strains referring to notable publications of CRISPR research performed in the individual organism is presented in [fig_ref] Table 1: Widely used archaeal model organisms for CRISPR in vitro and in vivo... [/fig_ref].
## Experimental crispr research in archaeal model strains
## Haloarchaea-attack one's own kind
As genetically tractable and relatively easily cultivatable archaea, halophiles have been used as model organisms in different research fields. However, after the above-mentioned study, there was a long break of using them for CRISPR research, probably because with no identifiable spacer match, halophiles missed out on the boom when CRISPR-targeted viruses were identified [bib_ref] Intervening sequences of regularly spaced prokaryotic repeats derive from foreign genetic elements, Mojica [/bib_ref]. After 17 years of a (non-funded) dry spell, halophilic CRISPR research was revived by characterizing crRNA processing in engineered Haloferax CRISPR mutant strains [bib_ref] A complex of cas proteins 5, 6, and 7 is required for..., Brendel [/bib_ref] [bib_ref] Characterization of CRISPR RNA biogenesis and Cas6 cleavage-mediated inhibition of a provirus..., Li [/bib_ref] , as well as CRISPR-mediated DNA interference using plasmid-based invader assays [bib_ref] An archaeal immune system can detect multiple protospacer adjacent motifs (PAMs) to..., Fischer [/bib_ref] [bib_ref] Essential requirements for the detection and degradation of invaders by the Haloferax..., Maier [/bib_ref] [bib_ref] The role of Cas8 in type I CRISPR interference, Cass [/bib_ref]. Thereby, researchers observed that six different PAMs efficiently triggered an immune response by the type I-B complex (the only effector complex in halophiles), marking the H. volcanii Cascade as one of the most versatile type I complexes described [bib_ref] An archaeal immune system can detect multiple protospacer adjacent motifs (PAMs) to..., Fischer [/bib_ref]. Contrarily, in Haloarcula hispanica four PAMs, three of which were distinct from those found in H. volcanii, were needed for type I-B plasmid degradation [bib_ref] Haloarcula hispanica CRISPR authenticates PAM of a target sequence to prime discriminative..., Li [/bib_ref]. Besides interference, also spacer acquisition from a halovirus (HHPV-2) has been experimentally shown in infection assays in H. hispanica, indicating that it was primed, as additionally to the adaptation cassette, also Cas3 and a partially matching native spacer were required [bib_ref] Adaptation of the Haloarcula hispanica CRISPR-Cas system to a purified virus strictly..., Li [/bib_ref]. The thereby established virus-based acquisition assay prompted follow-up studies in this model organism investigating repeat duplication [bib_ref] DNA motifs determining the accuracy of repeat duplication during CRISPR adaptation in..., Wang [/bib_ref] , spacer size [bib_ref] The spacer size of I-B CRISPR is modulated by the terminal sequence..., Li [/bib_ref] , and crRNA requirements for proper acquisition [bib_ref] Primed adaptation tolerates extensive structural and size variations of the CRISPR RNA..., Gong [/bib_ref].
By expressing a chromosome-targeting crRNA, it was shown that H. volcanii is one of the few prokaryotes that tolerates CRISPR-Cas self-targeting potentially because of a potent microhomology-repair pathway [bib_ref] High tolerance to self-targeting of the genome by the endogenous CRISPR-Cas system..., Stachler [/bib_ref] , marking it as a distinguished model organism to study CRISPR autoimmunity. For instance, a very recent study demonstrated that overexpression of a self-targeting spacer triggered adaptation of novel spacers collected from the vicinity of the originally targeted chromosomal locus [bib_ref] Adaptation induced by self-targeting in a type I-B CRISPR-Cas system, Stachler [/bib_ref]. Apart from the own chromosome, H. volcanii and H. mediterranei were recently shown to acquire spacers from each other s chromosomes in mating assays [bib_ref] Pervasive acquisition of CRISPR memory driven by inter-species mating of archaea can..., Turgeman-Grott [/bib_ref] , where cells of both species fuse by forming cytoplasmatic bridges [bib_ref] Low species barriers in halophilic archaea and the formation of recombinant hybrids, Naor [/bib_ref]. Moreover, when the H. mediterranei genome was engineered to be recognized by a native H. volcanii spacer in such a crossing experiment, the mating efficiency was decreased. Given the many spacers matching other haloarchaea species found in haloarchaeal genomes [bib_ref] Pervasive acquisition of CRISPR memory driven by inter-species mating of archaea can..., Turgeman-Grott [/bib_ref] , this study delivered the experimental proof that CRISPR-mediated cross targeting can shape the gene flux between haloarchaeal species which can be studied in a laboratory set-up.
## Pyrococcus-shaping crispr crystals
Advantaged with hyperthermal stability (100 - C) facilitating mechanistic and structural studies of proteins, the hyperthermophilic euryarchaeon Pyrococcus furiosus was a pioneer archaeal CRISPR model regarding biochemistry of Cas proteins. Within the early quest to unrevealing processing and architecture of spacer-derived crRNAs (earlier called psiRNA for prokaryotic silencing) [bib_ref] RNA-guided RNA cleavage by a CRISPR RNA-Cas protein complex, Hale [/bib_ref] [bib_ref] Prokaryotic silencing (psi)RNAs in Pyrococcus furiosus, Hale [/bib_ref] , Cas6 was purified and crystallized from P. furiosus [bib_ref] Cas6 is an endoribonuclease that generates guide RNAs for invader defense in..., Carte [/bib_ref] , leading to its biochemical characterization and the detailed investigation of pre-crRNA binding and cleavage [bib_ref] Binding and cleavage of CRISPR RNA by Cas6, Carte [/bib_ref] [bib_ref] Interaction of the Cas6 riboendonuclease with CRISPR RNAs: Recognition and cleavage, Wang [/bib_ref]. Shortly after that, the first complete prokaryotic type III-B system was isolated from P. furiosus, leading to the characterization of the complex composition and the bound crRNAs as well as the first experimental evidence in a prokaryote that type III cleaves RNA in vitro [bib_ref] RNA-guided RNA cleavage by a CRISPR RNA-Cas protein complex, Hale [/bib_ref]. Additionally, cleavage products of an antisense transcript of a crRNA detected in Northern blots and co-purification of the type III complex verified in vivo activity [bib_ref] Essential features and rational design of CRISPR RNAs that function with the..., Hale [/bib_ref]. These studies prompted the resolution of numerous structures of the different subunits, subcomplexes and entire complexes of the P. furiosus type III-B effector [bib_ref] Structure of an RNA silencing complex of the CRISPR-Cas immune system, Spilman [/bib_ref] [bib_ref] Crystal Structure of the CRISPR-Cas RNA Silencing Cmr Complex Bound to a..., Osawa [/bib_ref] and reviewed in [bib_ref] The RNA-and DNA-targeting CRISPR-Cas immune systems of Pyrococcus furiosus, Terns [/bib_ref] which helped to reveal the molecular details of the ruler-like RNA cleavage mechanism (see above and [bib_ref] Crystal Structure of the CRISPR-Cas RNA Silencing Cmr Complex Bound to a..., Osawa [/bib_ref] [bib_ref] Structural Model of a CRISPR RNA-Silencing Complex Reveals the RNA-Target Cleavage Activity..., Benda [/bib_ref] [bib_ref] Target RNA capture and cleavage by the Cmr type III-B CRISPR-Cas effector..., Hale [/bib_ref] [bib_ref] Essential Structural and Functional Roles of the Cmr4 Subunit in RNA Cleavage..., Ramia [/bib_ref]. Before the type III cOA-signaling pathway was discovered (see above), the accessory protein Csx1 of P. furiosus was crystallized [bib_ref] Crystal structure and nucleic acid-binding activity of the CRISPR-Associated protein Csx1 of..., Kim [/bib_ref] and identified to be an adenosine-specific ribonuclease [bib_ref] The CRISPR-associated Csx1 protein of Pyrococcus furiosus is an adenosine-specific endoribonuclease, Sheppard [/bib_ref]. Recently, it was also demonstrated that like some other CARF-domain nucleases (see above), also Csx1 of P. furiosus was endowed with ring nuclease activity, self-inactivating its cOA activators [bib_ref] Regulation of the RNA and DNA nuclease activities required for Pyrococcus furiosus..., Foster [/bib_ref]. Apart from additional biochemical studies focusing on the two other type I effector complexes in P. furiosus [bib_ref] CRISPR RNA-guided DNA cleavage by reconstituted Type I-A immune effector complexes, Majumdar [/bib_ref] , in vivo immunity against engineered plasmid invaders was shown for all effector complexes independently by analyzing a plethora of cas (and accessory genes) mutants [bib_ref] Bipartite recognition of target RNAs activates DNA cleavage by the Type III-B..., Elmore [/bib_ref] [bib_ref] Regulation of the RNA and DNA nuclease activities required for Pyrococcus furiosus..., Foster [/bib_ref] [bib_ref] DNA targeting by the type I-G and type I-A CRISPR-Cas systems of..., Elmore [/bib_ref]. Recently P. furious has also become a model to study spacer acquisition, revealing that new extrachromosomal spacers are preferentially acquired from broken DNA ends, supplied by plasmids with rolling-circle rather than theta replication [bib_ref] Role of free DNA ends and protospacer adjacent motifs for CRISPR DNA..., Shiimori [/bib_ref]. It was further shown that Cas1 and Cas2 alone could acquire spacers in vitro [bib_ref] CRISPR repeat sequences and relative spacing specify DNA integration by Pyrococcus furiosus..., Grainy [/bib_ref] , but that Cas4 proteins are essential for acquisition of functional spacers in vivo [bib_ref] Cas4 Nucleases Define the PAM, Length, and Orientation of DNA Fragments Integrated..., Shiimori [/bib_ref]. Furthermore, by supplying plasmids with partially matching protospacers, primed acquisition could be triggered in P. furiosus which was dependent on Cas3 and the type I-B effector complex [bib_ref] Primed CRISPR DNA uptake in Pyrococcus furiosus, Garrett [/bib_ref].
## Methanoarchaea-crispr models on the fast lane?
Similar to P. furiosus, hyperthermophilic methanogenic lab strains have served to purify and crystallize diverse CRISPR proteins, enabling early biochemical characterization of the type I nuclease Cas3 from Methanocaldococcus jannaschii [bib_ref] Structure and activity of the Cas3 HD nuclease MJ0384, an effector enzyme..., Beloglazova [/bib_ref] or type III-A nuclease Csm3 of Methanopyrus kandleri alone or in a subcomplex with Csm4 [bib_ref] Structure and RNA-binding properties of the type III-A CRISPR-associated protein Csm3, Hrle [/bib_ref] [bib_ref] Crystal structure of the Csm3-Csm4 subcomplex in the type III-A CRISPR-Cas interference..., Numata [/bib_ref]. Furthermore, the type I backbone subunit Cas8 from Methanothermobacter thermoautotrophicus was biochemically characterized and identified to be the PAM recognition factor, essential for interference [bib_ref] The role of Cas8 in type I CRISPR interference, Cass [/bib_ref] [bib_ref] A novel nuclease-ATPase (Nar71) from archaea is part of a proposed thermophilic..., Guy [/bib_ref]. Array transcription and crRNA processing were studied in the mesophilic models Methanococcus maripaludis and Methanosarcina mazei from which Cas6 enzymes were purified and characterized in vitro [bib_ref] Characterization of CRISPR RNA processing in Clostridium thermocellum and Methanococcus maripaludis, Richter [/bib_ref] [bib_ref] Comparative analysis of Cas6b processing and CRISPR RNA stability, Richter [/bib_ref] [bib_ref] Two CRISPR-Cas systems in Methanosarcina mazei strain Gö1 display common processing features..., Nickel [/bib_ref] [bib_ref] A Non-Stem-Loop CRISPR RNA Is Processed by Dual Binding Cas6, Shao [/bib_ref]. Furthermore, recent in vivo investigations of M. mazei Cas6 mutant strains revealed that only one of the two Cas6 endonucleases executes pre-crRNA maturation of both, type I and type III-adjacent arrays. In vivo studies of CRISPR interference could not be easily performed in native hosts, probably due to the lack of appropriate virus-host systems and assays for methanogens. However, the type I-B system of M. maripaludis together with artificial crRNAs was heterologously expressed in E. coli demonstrating that, dependent on the flanking PAM, phage lambda infection was reduced to different levels and that efficient interference required Cas8 and Cas3 subunits [bib_ref] Fragmentation of the CRISPR-Cas Type I-B signature protein Cas8b, Richter [/bib_ref].
Most interestingly, methanogens have recently gained attention regarding CRISPR evolution, as Casposons, a sparsely distributed new class of putatively self-synthesizing DNA transposons, were found to be abundant and presumably mobile in M. mazei genomes [bib_ref] Recent mobility of casposons, self-synthesizing transposons at the origin of the CRISPR-cas..., Krupovic [/bib_ref]. Recent phylogenetic and biochemical analysis suggest that the transposase (i.e., Casposase) might represent the ancestor of the spacer-integrase Cas1 [bib_ref] Casposons: A new superfamily of self-synthesizing DNA transposons at the origin of..., Krupovic [/bib_ref] [bib_ref] The casposon-encoded Cas1 protein from Aciduliprofundum boonei is a DNA integrase that..., Hickman [/bib_ref] , leading to the assumption that CRISPR had evolved from Casposons. Thus, the Casposase of M. mazei was recently biochemically and structurally characterized, revealing that it tetramerizes upon target binding and, reminiscent of spacer-acquisition, actively integrated substrates into a preferred target site [bib_ref] Casposase structure and the mechanistic link between DNA transposition and spacer acquisition..., Hickman [/bib_ref]. Notably, a putative regulator of the Casposase expression was identified in M. mazei just now and is published within this special issue. Hence, Methanosarcina sp. hold great potential for studying the evolution and mechanisms of CRISPR adaptation. This is not only because they comprise genetically tractable laboratory strains that carry Casposons, but also because they harbor type III-adjacent reverse transcriptases that hypothetically could be involved in spacer acquisition from RNA substrates (see above and ref.. Such RNA-derived spacers could be used to probe for potential archaeal RNA viruses, which haven t been identified in archaea yet [bib_ref] Identification of novel positive-strand RNA viruses by metagenomic analysis of archaea-dominated Yellowstone..., Bolduc [/bib_ref]. Furthermore, a recently isolated DNA virus infecting Methanosarcina sp. might facilitate CRISPR studies in vivo in native hosts [bib_ref] Methanosarcina Spherical Virus, a Novel Archaeal Lytic Virus Targeting Methanosarcina Strains, Weidenbach [/bib_ref].
## Sulfolobales-the virus fighters
The hyperthermophilic archaea of the order Sulfolobales are the most intensely studied archaea regarding the CRISPR system. Many different computational analyses of viral distribution based on spacer tracking have been conducted, array transcription analyzed, and many Cas proteins and CRISPR-related proteins have been biochemically characterized (reviewed in [bib_ref] Hot and crispy: CRISPR-Cas systems in the hyperthermophile Sulfolobus solfataricus, Zhang [/bib_ref] [bib_ref] CRISPR-mediated defense mechanisms in the hyperthermophilic archaeal genus Sulfolobus, Manica [/bib_ref] [bib_ref] CRISPR-based immune systems of the Sulfolobales: Complexity and diversity, Garrett [/bib_ref] [bib_ref] Electron microscopy studies of Type III CRISPR machines in Sulfolobus solfataricus, Cannone [/bib_ref] and notable references in [fig_ref] Table 1: Widely used archaeal model organisms for CRISPR in vitro and in vivo... [/fig_ref]. Furthermore, owing to the great number of purified viruses and plasmids [bib_ref] The enigmatic archaeal virosphere, Prangishvili [/bib_ref] and available assays to study virus-host interactions in the culture flask, Sulfolobales lab strains have become distinguished pioneer models for studying CRISPR interference in vivo, some of which will be briefly introduced below. Due to the large number of CRISPR-Cas systems in their genomes, Sulfolobales are also particularly suited to characterize the effects and interactions of multiple CRISPR-Cas systems within one cell.
CRISPR-mediated DNA interference in archaea was first studied in S. solfataricus and S. islandicus where either a plasmid or a virus was engineered with a cognate protospacer, respectively [bib_ref] Dynamic properties of the Sulfolobus CRISPR/Cas and CRISPR/Cmr systems when challenged with..., Gudbergsdottir [/bib_ref] [bib_ref] In vivo activity of CRISPR-mediated virus defence in a hyperthermophilic archaeon, Manica [/bib_ref]. In plasmid invader assays, where a metabolic gene needed for cell survival was supplied by the protospacer-carrying plasmid, cells only survived upon partial deletion of the native CRISPR locus including the targeting spacer. Thus, by mapping and identifying escape mutations, efficiency of CRISPR-mediated interference could be indirectly determined in this study [bib_ref] Dynamic properties of the Sulfolobus CRISPR/Cas and CRISPR/Cmr systems when challenged with..., Gudbergsdottir [/bib_ref]. In the selection-independent virus approach employing shuttle vectors based on the lysogenic virus SSV1 [bib_ref] A reporter gene system for the hyperthermophilic archaeon Sulfolobus solfataricus based on..., Jonuscheit [/bib_ref] , interference efficiency was directly quantified by counting transfected cells in plaque assays [bib_ref] In vivo activity of CRISPR-mediated virus defence in a hyperthermophilic archaeon, Manica [/bib_ref]. Both of these strategies were applied in various follow-up in vivo studies to further characterize PAM requirements for type I targeting [bib_ref] Genetic determinants of PAM-dependent DNA targeting and pre-crRNA processing in Sulfolobus islandicus, Peng [/bib_ref] , protospacer-crRNA matches needed for proper interference [bib_ref] Unexpectedly broad target recognition of the CRISPR-mediated virus defence system in the..., Manica [/bib_ref] [bib_ref] Major and minor crRNA annealing sites facilitate low stringency DNA protospacer binding..., Mousaei [/bib_ref] , crRNA processing and transcription regulation [bib_ref] Genetic determinants of PAM-dependent DNA targeting and pre-crRNA processing in Sulfolobus islandicus, Peng [/bib_ref] [bib_ref] a Modulation of CRISPR locus transcription by the repeat-binding protein Cbp1 in..., Deng [/bib_ref] and type III-mediated RNA recognition and interference [bib_ref] CRISPR-mediated targeted mRNA degradation in the archaeon Sulfolobus solfataricus, Zebec [/bib_ref] [bib_ref] A novel interference mechanism by a type IIIB CRISPR-Cmr module in Sulfolobus, Deng [/bib_ref]. A significant in vivo finding, already foreshadowing a link between CARF-domain nucleases and the type III-B immune response, was the demonstration of transcription-dependent DNA interference against a plasmid to be dependent on csx1/csm6 locus and an intact type III-B system in S. islandicus [bib_ref] A novel interference mechanism by a type IIIB CRISPR-Cmr module in Sulfolobus, Deng [/bib_ref]. Cells escaping plasmid-targeting arose upon spontaneous deletion of the csx1/csm6 gene locus, which restored plasmid interference when reintroduced into the mutant cells [bib_ref] A novel interference mechanism by a type IIIB CRISPR-Cmr module in Sulfolobus, Deng [/bib_ref]. Soon after this study, a mutational analysis of protospacer-crRNA hybrids revealed that a match of three distinct base pairs between the PAS-5 handle sufficiently abolished virus DNA degradation in S. solfataricus in vivo [bib_ref] Unexpectedly broad target recognition of the CRISPR-mediated virus defence system in the..., Manica [/bib_ref]. Remarkably, prior to any knowledge of type III collateral damage, this study disclosed the (minimal) sequence requirements for inhibiting secondary type III immune responses in vivo, years before the biochemical determinants were resolved in vitro (see above and ref. [bib_ref] Structure of the CRISPR interference complex CSM reveals key similarities with cascade, Rouillon [/bib_ref] [bib_ref] Structures of the Cmr-β Complex Reveal the Regulation of the Immunity Mechanism..., Sofos [/bib_ref]. Later, researchers found that mutated Cas10-HD-domain type III variants from S. islandicus lost their DNA cleavage activity in vitro, but astonishingly, cognate plasmids were still degraded in the respective mutants in vivo, suggesting another type III-mediated interference activity to be in place [bib_ref] A type III-B CRISPR-Cas effector complex mediating massive target DNA destruction, Han [/bib_ref].
Type III-mediated degradation of a viral mRNA in vivo in an archaeon was first shown in S. solfataricus, where RNA cleavage efficiency could be quantified when the transcribed protospacer (engineered to match a native crRNA) was flanked by a PAS, thereby inhibiting DNA degradation and type III-collateral damage [bib_ref] CRISPR-mediated targeted mRNA degradation in the archaeon Sulfolobus solfataricus, Zebec [/bib_ref]. 40% reduction of protospacer mRNA in the cell was measured using quantitative PCR and Northern blot, and cleavage by the purified S. solfataricus type III-B complex was verified in vitro [bib_ref] CRISPR-mediated targeted mRNA degradation in the archaeon Sulfolobus solfataricus, Zebec [/bib_ref]. Analysis of two co-existing type III complexes in S. islandicus REY15A using miniCRISPR-based silencing assays in respective mutant strains (see Section 4.2), revealed both complexes to confer differently strong degradation of RNA [bib_ref] An archaeal CRISPR type III-B system exhibiting distinctive RNA targeting features and..., Peng [/bib_ref]. Furthermore, a type III complex deficient in the Cmr1 subunit showed decreased RNA and DNA interference in vivo, attributed to decreased target capture efficiency [bib_ref] A seed motif for target RNA capture enables efficient immune defence by..., Pan [/bib_ref] [bib_ref] She, Q. Cmr1 enables efficient RNA and DNA interference of a III-B..., Li [/bib_ref]. Shortly after the elucidation of the cOA-induced type III signaling pathway in bacteria [bib_ref] Type III CRISPR-Cas systems produce cyclic oligoadenylate second messengers, Niewoehner [/bib_ref] [bib_ref] A cyclic oligonucleotide signaling pathway in type III CRISPR-Cas systems, Kazlauskiene [/bib_ref] , Csx1 of S. islandicus was shown to be activated upon binding of its CARF domain to an mRNA adenosine tail [bib_ref] Allosteric regulation of Csx1, a type IIIB-associated CARF domain ribonuclease by RNAs..., Han [/bib_ref]. Furthermore, cOA production was characterized for type III-D complexes of S. solfataricusand III-B complexes of S. islandicus [bib_ref] Structures of the Cmr-β Complex Reveal the Regulation of the Immunity Mechanism..., Sofos [/bib_ref] [bib_ref] A Type III-B Cmr effector complex catalyzes the synthesis of cyclic oligoadenylate..., Han [/bib_ref]. The breakthrough finding of cOA-mopping ring nucleases in S. solfataricus probably further fueled the public interest in these model organisms [bib_ref] Ring nucleases deactivate type III CRISPR ribonucleases by degrading cyclic oligoadenylate, Athukoralage [/bib_ref].
Interestingly, recent in vitro studies of a CRISPR type I-D Cascade from S. islandicus LAL14/1 revealed type I-specific dsDNA cleavage as well as ssDNA degradation (see above). Reminiscent of type III-mediated RNA degradation, ssDNA cleavage was found to be ruler-like as governed by the periodically allocated Cas7 subunits of the Cascade backbone. Thus, evolutionary traces of type III systems can be found in the cleavage mechanism of type I-D, suggesting that it constitutes an intermediate between type I and type III systems [bib_ref] DNA targeting by subtype I-D CRISPR-Cas shows type I and type III..., Lin [/bib_ref]. This system will be exciting to study in the future, perhaps also because a bacterial type I-D Cascade has recently been successfully applied for targeted mutagenesis in human cells.
Apart from studying molecular details of CRISPR interference, infection studies using free virions or environmental virus/plasmid mixes allowed real-time monitoring of the temporal regulation of cas genes and CRISPR arrays and the spatiotemporal emergence of viral countermeasures [bib_ref] Transcriptome analysis of Sulfolobus solfataricus infected with two related fuselloviruses reveals novel..., Fusco [/bib_ref] [bib_ref] Massive Activation of Archaeal Defense Genes during Viral Infection, Quax [/bib_ref] [bib_ref] Stable maintenance of the rudivirus SIRV3 in a carrier state in Sulfolobus..., Papathanasiou [/bib_ref] [bib_ref] Tolerance of Sulfolobus SMV1 virus to the immunity of I-A and III-B..., Guo [/bib_ref] [bib_ref] Virus-induced dormancy in the archaeon Sulfolobus islandicus, Bautista [/bib_ref]. Within such experiments, the first two archaeal anti-CRISPR-Cas systems (Acr) encoded by the lytic virus SIRV2 infecting S. islandicus were discovered. These were found to either inhibit type I-D or type III-B immunity, respectively, by binding to catalytic complex subunits [bib_ref] Peng, X. viruses inhibit subtype I-D immunity, He [/bib_ref]. As mentioned above (see Section 3.3), the functional characterization of Acr-IIIB1 in vivo was particularly important to also understanding the impact of the different mechanisms of type IIIB-immunity on lytic virus infection in S. islandicus. Acr-IIIB1 blocked type III cOA signaling only when the protospacer was located on middle/late virus genes, suggesting that collateral damage might be the prevalent immune response acting during the late virus life cycle. Thus, HD-domain-mediated virus DNA cleavage might be predominant when targeting early virus genes, where a higher amount of protospacer transcript is available.
Very recently, a newly identified ring nuclease was shown to function as Acr in vivo by challenging S. islandicus M.16.4, solely carrying the type III system, with a lytic phage (see above). Normally degraded owing to a matching native spacer, the virus could stably infect S. islandicus when the Acr was heterologously expressed [bib_ref] An anti-CRISPR viral ring nuclease subverts type III CRISPR immunity, Athukoralage [/bib_ref]. Acrs represent exciting subjects to future studies in those model organisms.
In early studies, spacer acquisition in Sulfolobales models could only be induced when applying environmental virus mixtures [bib_ref] Selective and hyperactive uptake of foreign DNA by adaptive immune systems of..., Erdmann [/bib_ref] or specific other co-infecting viruses [bib_ref] Inter-viral conflicts that exploit host CRISPR immune systems of Sulfolobus, Erdmann [/bib_ref] [bib_ref] SMV1 virus-induced CRISPR spacer acquisition from the conjugative plasmid pMGB1 in Sulfolobus..., Erdmann [/bib_ref] , as the presence of particular viruses could specifically trigger spacer acquisition from another component in the mix. For instance, incubation with SMV1 triggered highly selective uptake of spacers exclusively from a conjugative plasmid in S. solfataricus or from a co-infecting STSV2 virus in S. islandicus [bib_ref] Inter-viral conflicts that exploit host CRISPR immune systems of Sulfolobus, Erdmann [/bib_ref] [bib_ref] SMV1 virus-induced CRISPR spacer acquisition from the conjugative plasmid pMGB1 in Sulfolobus..., Erdmann [/bib_ref]. In later studies, the transcription factor Csa3a was shown to enhance adaptation in S. islandicus [bib_ref] Coupling transcriptional activation of CRISPR-Cas system and DNA repair genes by Csa3a..., Liu [/bib_ref] [bib_ref] Transcriptional regulator-mediated activation of adaptation genes triggers CRISPR de novo spacer acquisition, Liu [/bib_ref] and spacer acquisition from a single extrachromosomal element could be triggered if it was present in a higher copy number [bib_ref] Transcriptome changes in STSV2-infected Sulfolobus islandicusREY15A undergoing continuous CRISPR spacer acquisition, León-Sobrino [/bib_ref] [bib_ref] Diverse CRISPR-Cas responses and dramatic cellular DNA changes and cell death in..., Liu [/bib_ref]. Besides Cas1 and Cas2, Cas4 was shown to regulate spacer acquisition in vivo in S. islandicus (see Section 2.1 and ref. [bib_ref] Cas4 nucleases can effect specific integration of CRISPR spacers, Zhang [/bib_ref] and in vitro in S. solfataricus [bib_ref] Prespacer processing and specific integration in a type I-A CRISPR system, Rollie [/bib_ref]. The in vivo dynamics, potentially shaped by Acrs-like mechanisms inactivating spacer acquisition [bib_ref] Cas4 nucleases can effect specific integration of CRISPR spacers, Zhang [/bib_ref] , are an interesting field of research, especially with the broad virus selection available for Sulfolobales, and await future studies.
## Crispr application in archaeal models
Shortly after the elucidation of the CRISPR interference mechanisms, many researches focused on exploiting the CRISPR system for genomic engineering. Especially, CRISPR Class II systems have gained tremendous attention and have been engineered as a genome editing tool for setting mutations in bacteria [bib_ref] Programmable repression and activation of bacterial gene expression using an engineered CRISPR-Cas..., Bikard [/bib_ref] and virtually all eukaryotic model systems including human cell lines (reviewed in. Recently, Cas9 as well as its engineered nuclease-deficient version dCas9 were successfully heterologously expressed in the mesophilic archaeon Methanosarcina acetivorans for gene editing and silencing, respectively [fig_ref] Table 1: Widely used archaeal model organisms for CRISPR in vitro and in vivo... [/fig_ref]. Upon supplying a repair template that triggered homologous recombination and a crRNA complementary to the target gene, the crRNA-guided Cas9 could induce insertions and deletions in the chromosome of M. acetivorans. Repair of the Cas9-induced double strand break without a repair template was only possible when co-expressing a nonhomologous end-joining pathway [bib_ref] Cas9-mediated genome editing in the methanogenic archaeon Methanosarcina acetivorans, Nayak [/bib_ref]. The nuclease deficient variant of Cas9 efficiently blocked transcription of desired genes by crRNA-guided binding to the DNA, achieving a silencing efficiency of up to 90% of the nif genes involved in nitrogen fixation [bib_ref] A CRISPRi-dCas9 system for archaea and its use to examine gene function..., Dhamad [/bib_ref].
Contrarily to M. acetivorans, stable expression of the commonly used Cas9 of the mesophilic bacterium Streoptococcus pyogenes [bib_ref] A programmable dual-RNA-guided DNA endonuclease in adaptive bacterial immunity, Jinek [/bib_ref] in hyperthermophilic or halophilic archaea could not be achieved due to instability of the enzyme. However, in those archaeal models, endogenous type I and type III systems can be efficiently hijacked for gene silencing and genome editing [fig_ref] Table 1: Widely used archaeal model organisms for CRISPR in vitro and in vivo... [/fig_ref]. H. volcanii mutants, that carry an endogenous DNA-targeting Type I-B complex deficient in Cas3 nuclease activity can be used for gene silencing via transcription blocking (i.e., CRISPR interference), similarly to dCas9 [bib_ref] Gene Repression in Haloarchaea Using the CRISPR (Clustered Regularly Interspaced Short Palindromic..., Stachler [/bib_ref]. CRISPRi efficiency can be increased by preventing the occupation of the available dCascade complexes by endogenous crRNAs which can be achieved by deleting either the native CRISPR arrays or the processing gene cas6, respectively. To ensure proper processing of the artificial crRNA in the latter approach, the crRNA must be flanked by t-elements that are recognized and processed by endogenous tRNases, generating a mature crRNA [bib_ref] Gene Repression in Haloarchaea Using the CRISPR (Clustered Regularly Interspaced Short Palindromic..., Stachler [/bib_ref]. CRISPRi was successfully used in H. volcanii to silence non-essential and essential genes, achieving 78% knockdown of the essential RNase P [bib_ref] Gene Repression in Haloarchaea Using the CRISPR (Clustered Regularly Interspaced Short Palindromic..., Stachler [/bib_ref]. Besides gene silencing, in S. islandicus as well as H. hispanica, the native intact type I system was successfully exploited for genome editing by supplying a helper plasmid and an artificial crRNA targeting the chromosomal locus to be edited [bib_ref] Harnessing the native type I-B CRISPR-Cas for genome editing in a polyploid..., Cheng [/bib_ref] [bib_ref] Harnessing Type I and Type III CRISPR-Cas systems for genome editing, Li [/bib_ref]. As for Cas9, silencing or editing via a type I system requires a PAM in the target sequence.
Interestingly, an innovative virus editing technology makes use of an anti-CRISPR system, conferring immunity to I-D-mediated DNA interference in S. islandicus LAL 14/1, as selection marker for targeted gene knockouts in virus derivates (deficient of an Acr) in vivo [bib_ref] Anti-crispr-based and crispr-based genome editing of sulfolobus islandicus rod-shaped virus 2, Mayo-Muñoz [/bib_ref].
A native CRISPR type III system can be repurposed for posttranscriptional silencing of host genes in S. solfataricus [bib_ref] CRISPR-mediated targeted mRNA degradation in the archaeon Sulfolobus solfataricus, Zebec [/bib_ref] [bib_ref] Efficient CRISPR-Mediated Post-transcriptional Gene Silencing in a Hyperthermophilic Archaeon Using Multiplexed crRNA..., Zebec [/bib_ref] , S. islandicus [bib_ref] An archaeal CRISPR type III-B system exhibiting distinctive RNA targeting features and..., Peng [/bib_ref] [bib_ref] Reverse gyrase functions in genome integrity maintenance by protecting DNA breaks in..., Han [/bib_ref] and S. acidocaldarius [bib_ref] Comparative CRISPR type III-based knockdown of essential genes in hyperthermophilic Sulfolobales and..., Zink [/bib_ref]. For type III-mediated silencing, crRNAs are heterologously expressed from a miniCR vector, incorporated into a native CRISPR type III endonuclease and guided to complementary loci on the mRNA of a desired gene which is subsequently cleaved [bib_ref] An archaeal CRISPR type III-B system exhibiting distinctive RNA targeting features and..., Peng [/bib_ref] [bib_ref] CRISPR-mediated targeted mRNA degradation in the archaeon Sulfolobus solfataricus, Zebec [/bib_ref]. The protospacer on the mRNA requires a PAS in order keep the collateral damage immune response and unspecific ssDNA cleavage of type III systems inactivated (see above and [bib_ref] CRISPR-mediated targeted mRNA degradation in the archaeon Sulfolobus solfataricus, Zebec [/bib_ref]. By increasing the number of expressed crRNAs, we could gradually increase the knockdown levels, which were stably maintained over the course of growth in S. solfataricus [bib_ref] Efficient CRISPR-Mediated Post-transcriptional Gene Silencing in a Hyperthermophilic Archaeon Using Multiplexed crRNA..., Zebec [/bib_ref] [bib_ref] CRISPR-mediated gene silencing reveals involvement of the archaeal S-layer in cell division..., Zink [/bib_ref]. In theory, this technology can readily be applied in any genetically accessible organism carrying a type III system, as no genetic manipulation of the complex is required beforehand. Besides some non-essential genes that could be silenced to almost 100% [bib_ref] An archaeal CRISPR type III-B system exhibiting distinctive RNA targeting features and..., Peng [/bib_ref] [bib_ref] Efficient CRISPR-Mediated Post-transcriptional Gene Silencing in a Hyperthermophilic Archaeon Using Multiplexed crRNA..., Zebec [/bib_ref] [bib_ref] Reverse gyrase functions in genome integrity maintenance by protecting DNA breaks in..., Han [/bib_ref] , we have recently applied the type III-mediated knockdown on essential genes of different functional categories, including cell division, transcription, cell wall biogenesis and translation [bib_ref] Comparative CRISPR type III-based knockdown of essential genes in hyperthermophilic Sulfolobales and..., Zink [/bib_ref] [bib_ref] CRISPR-mediated gene silencing reveals involvement of the archaeal S-layer in cell division..., Zink [/bib_ref] [bib_ref] Indications for a moonlighting function of translation factor aIF5A in the crenarchaeum..., Bassani [/bib_ref]. We found that dependent on the gene, a maximum silencing efficiency of 40-75% was achieved and could not be exceeded [bib_ref] Comparative CRISPR type III-based knockdown of essential genes in hyperthermophilic Sulfolobales and..., Zink [/bib_ref]. Within this range, the silencing effect was stable and specific phenotypes could be analyzed in vivo, allowing functional characterization of the respective gene [bib_ref] CRISPR-mediated gene silencing reveals involvement of the archaeal S-layer in cell division..., Zink [/bib_ref]. Higher silencing levels conferred by stronger spacers (or an increased numbers of otherwise stable spacers) were not tolerated, leading to precise excision of those spacers from the miniCRISPR array. Thus, this suggests the presence of a probably CRISPR-linked mechanism to eradicate deleterious spacers [bib_ref] Comparative CRISPR type III-based knockdown of essential genes in hyperthermophilic Sulfolobales and..., Zink [/bib_ref].
## A hot fuzz: crispr immunity in sulfolobales
Members of the order Sulfolobales are thermoacidophilic crenarchaea, belonging to the TACK superphylum within the archaea. Almost all thrive at around 80 - C and a pH of 3 on organic carbon sources under aerobic conditions and are found in high temperature, mostly terrestrial environments, such as solfataric and volcanic hot springs in Iceland, Kamchatka or Yellowstone National Park. Owing to the extremophilic life style, Sulfolobales have become literally "hot" objects of study regarding industrial applications, biochemistry and also evolution as they potentially could have withstood the harsh environments on an early Earth [bib_ref] Hyperthermophiles in the history of life, Stetter [/bib_ref]. Moreover, Sulfolobales are known to share their environments with many viruses [bib_ref] Viruses of archaea: Structural, functional, environmental and evolutionary genomics, Krupovic [/bib_ref]. The Sulfolobales viruses that have been described so far belong to six different archaeal virus families and are characterized by an impressive diversity regarding their morphologies, as well as genome structures and life cycles [bib_ref] The enigmatic archaeal virosphere, Prangishvili [/bib_ref] [bib_ref] Viruses of archaea: Structural, functional, environmental and evolutionary genomics, Krupovic [/bib_ref]. While viruses of bacteria are mostly lytic, the majority of archaeal viruses rather seem to persist in their host cells in a stable carrier state [bib_ref] Archaeal viruses at the cell envelope: Entry and egress, Quemin [/bib_ref] , which sometimes is beneficial for the host [bib_ref] Killer archaea: Virus-mediated antagonism to CRISPR-immune populations results in emergent virus-host mutualism, Dewerff [/bib_ref]. Similar to prophage lambda, some temperate archaeal viruses can be activated by various stimuli, such as the well described Fuselloviridae infecting Sulfolobales, however they do not always cause cell lysis after induction of virion production. A recent study on host-virus interactions of geographically separated, natural S. islandicus populations with SSVs and SIRVs suggests that the CRISPR-Cas immunity is more diversified in response to lytic viruses and free virions compared to non-lytic and integrated viruses [bib_ref] Diversified local CRISPR-Cas immunity to viruses of Sulfolobus islandicus, Pauly [/bib_ref]. In light of the enormous diversity of viruses, but also of large numbers of IS elements in Sulfolobales, it is not astonishing that they are equipped with extensive CRISPR-Cas systems, which makes them particularly interesting study objects for this field. We here present a thorough comparative genomic analysis on the distribution and abundance of CRISPR-Cas systems in all sequenced genomes available to date.
## Distribution of crispr types in sulfolobales
Altogether, the 38 fully sequenced representative Sulfolobales genomes currently available harbor 124 individual CRISPR-Cas loci assigned to two different CRISPR-Cas types, namely type I (52 loci, 42%) and type III (72 loci, 58%) [fig_ref] Figure 3: Presence of specific CRISPR-Cas [/fig_ref]. The derivative strains of S. solfataricus SULA and Metallosphaera sedula DSM 5348 (15 strains in total) generated via adaptive laboratory evolution were excluded from the dataset to not skew the analysis, as they exhibit similar CRISPR-Cas systems and spacers as the parental strains. Amongst all type I CRISPR-Cas loci present in Sulfolobales, I-A is by far the most prevalent subtype (33x), followed by I-D (14x) and I-B (5x) [fig_ref] Figure 3: Presence of specific CRISPR-Cas [/fig_ref] , Lower panel). While type I-B systems have been shown to target DNA in several other archaea [bib_ref] An archaeal immune system can detect multiple protospacer adjacent motifs (PAMs) to..., Fischer [/bib_ref] [bib_ref] DNA targeting by the type I-G and type I-A CRISPR-Cas systems of..., Elmore [/bib_ref] , type I-A and I-D mediated DNA targeting was experimentally verified in members of the Sulfolobales [bib_ref] DNA targeting by subtype I-D CRISPR-Cas shows type I and type III..., Lin [/bib_ref] [bib_ref] Dynamic properties of the Sulfolobus CRISPR/Cas and CRISPR/Cmr systems when challenged with..., Gudbergsdottir [/bib_ref] [bib_ref] Genetic determinants of PAM-dependent DNA targeting and pre-crRNA processing in Sulfolobus islandicus, Peng [/bib_ref] [bib_ref] Peng, X. viruses inhibit subtype I-D immunity, He [/bib_ref].
The majority of type III CRISPR-Cas loci in Sulfolobales consists of subtypes III-D (32x) and III-B (22x), whereas in comparison the third designated subtype III-A represents a rather small fraction (3x). Interestingly, a considerable fraction (15x) of the detected CRISPR-Cas loci constitute a distinct variant of type III systems exclusively found in Sulfolobales, predominantly in different strains of S. acidocaldarius and S. islandicus (see [fig_ref] Figure 3: Presence of specific CRISPR-Cas [/fig_ref] lower panel, [fig_ref] Figure 4: Distribution of CRISPR-Cas [/fig_ref] , and see [bib_ref] Evolutionary classification of CRISPR-Cas systems: A burst of class 2 and derived..., Makarova [/bib_ref]. These complexes comprise divergent Cas10 and Cas5 subunits as well as a putative additional subunit of unknown function termed Csx26 and need yet to be experimentally investigated [bib_ref] Evolutionary classification of CRISPR-Cas systems: A burst of class 2 and derived..., Makarova [/bib_ref].
All members of the Sulfolobales possess at least one complete type I or type III CRISPR-Cas system, the sole exception being Stygiolobus azoricus FC6, which only contains an incomplete subtype III-D locus [fig_ref] Figure 1: Figure 1 [/fig_ref]. While S. solfataricus SULA and its derivative strains only harbor a single subtype I-A system, S. acidocaldarius Y14_18-5 and Acidianus ambivalens LEI10, in contrast, both exclusively harbor one CRISPR-Cas system of subtype III-D. Apart from the exceptions mentioned above, all other Sulfolobales genomes encode more than one CRISPR-Cas system, and in those genomes type I and type III systems always co-exist [fig_ref] Figure 1: Figure 1 [/fig_ref]. S. solfataricus P2 is the current record holder with a total of six complete CRISPR-Cas loci, more precisely, three loci of subtype I-A, two of subtype III-B and one locus of subtype III-D [fig_ref] Figure 1: Figure 1 [/fig_ref]. The majority of type III CRISPR-Cas loci in Sulfolobales consists of subtypes III-D (32x) and III-B (22x), whereas in comparison the third designated subtype III-A represents a rather small fraction (3x). Interestingly, a considerable fraction (15x) of the detected CRISPR-Cas loci constitute a distinct variant of type III systems exclusively found in Sulfolobales, predominantly in different strains of S. acidocaldarius and S. islandicus (see [fig_ref] Figure 3: Presence of specific CRISPR-Cas [/fig_ref] lower panel, [fig_ref] Figure 4: Distribution of CRISPR-Cas [/fig_ref] , and see [bib_ref] Evolutionary classification of CRISPR-Cas systems: A burst of class 2 and derived..., Makarova [/bib_ref]. These complexes comprise divergent Cas10 and Cas5 subunits as well as a putative additional subunit of unknown function termed Csx26 and need yet to be experimentally investigated [bib_ref] Evolutionary classification of CRISPR-Cas systems: A burst of class 2 and derived..., Makarova [/bib_ref].
All members of the Sulfolobales possess at least one complete type I or type III CRISPR-Cas system, the sole exception being Stygiolobus azoricus FC6, which only contains an incomplete subtype III-D locus [fig_ref] Figure 1: Figure 1 [/fig_ref]. While S. solfataricus SULA and its derivative strains only harbor a single subtype I-A system, S. acidocaldarius Y14_18-5 and Acidianus ambivalens LEI10, in contrast, both exclusively harbor one CRISPR-Cas system of subtype III-D. Apart from the exceptions mentioned above, all other Sulfolobales genomes encode more than one CRISPR-Cas system, and in those genomes type I and type III systems always co-exist [fig_ref] Figure 4: Distribution of CRISPR-Cas [/fig_ref] , Supplementary Table S1). S. solfataricus P2 is the current record holder with a total of six complete CRISPR-Cas loci, more precisely, three loci of subtype I-A, two of subtype III-B and one locus of subtype III-D [fig_ref] Figure 4: Distribution of CRISPR-Cas [/fig_ref] , Supplementary. [fig_ref] Table 1: Widely used archaeal model organisms for CRISPR in vitro and in vivo... [/fig_ref].
Notably, S. azoricus FC6 and A. ambivalens LEI10 represent the only two genomes which do not contain any or a seemingly non-functional adaptation cassette (internal stop in cas1, see Sulfurisphaera (2) Sulfuracidifex (2) Sulfolobus (9) Sulfodiicoccus (1) S. islandicus (10) Saccharolobus (4) Metallosphaera (5) Acidianus (5) [fig_ref] Figure 3: Presence of specific CRISPR-Cas [/fig_ref]. Presence of specific CRISPR-Cas (sub)types and their abundances among Sulfolobales genera.
Overall abundances of CRISPR-Cas types and the specific subtypes within the general types I and III found in the order Sulfolobales (Upper panel). The bar chart shows the distribution of the CRISPR-Cas subtypes throughout the genera within Sulfolobales (Lower panel). The number of genomes included in the analysis for each genus is given in brackets. The analysis is based on data published in ref. [bib_ref] Evolutionary classification of CRISPR-Cas systems: A burst of class 2 and derived..., Makarova [/bib_ref] and/or obtained by using programs CRISPRminer (version 1, [bib_ref] CRISPRminer is a knowledge base for exploring CRISPR-Cas systems in microbe and..., Zhang [/bib_ref] , and CRISPRCasFinder (version CRISPR-Cas++ 1.1.2, [bib_ref] CRISPRCasFinder, an update of CRISRFinder, includes a portable version, enhanced performance and..., Couvin [/bib_ref].
Notably, S. azoricus FC6 and A. ambivalens LEI10 represent the only two genomes which do not contain any or a seemingly non-functional adaptation cassette (internal stop in cas1, see [fig_ref] Table 1: Widely used archaeal model organisms for CRISPR in vitro and in vivo... [/fig_ref] , respectively, and hence in those organisms, spacer acquisition is probably impaired [fig_ref] Figure 4: Distribution of CRISPR-Cas [/fig_ref] , [fig_ref] Table 1: Widely used archaeal model organisms for CRISPR in vitro and in vivo... [/fig_ref]. Furthermore, we could not identify a cas6 in the S. azoricus FC6 genome, indicating that this organism might not be able to process pre-crRNAs. Thus, despite harboring three CRISPR arrays [fig_ref] Figure 5: CRISPR array, spacers and virus matches in Sulfolobales genera [/fig_ref] , S. azoricus FC6 might be the only representative of the Sulfolobales without functional CRISPR-Cas immunity, as it lacks all crucial genes for spacer acquisition, maturation as well as a complete interference module [fig_ref] Figure 4: Distribution of CRISPR-Cas [/fig_ref]. All other members of the Sulfolobales, additionally to complete effector complexes (see above), possess at least one functional gene set for spacer acquisition as well as at least one cas6 and therefore should be capable of performing all steps of CRISPR immunity [fig_ref] Figure 4: Distribution of CRISPR-Cas [/fig_ref] , Supplementary Table S1, cf. [fig_ref] Figure 1: Figure 1 [/fig_ref].
The accessory genes csx1, which encode indiscriminatory ssRNA nucleases playing a crucial role in the CRISPR type III immune response (see above), could be readily identified in almost all Sulfolobales genomes. As described above, cOAs produced by type III effector complexes serve as activators for Csx1/Csm6, and in turn those signaling molecules are degraded by ring nucleases Crn allowing the CRISPR immune response to return to ground state (see above). Thus, the type III effector complex, Csx1/Csm6 and Crn should co-occur in order to govern the controlled catalytic cycle of the type III immune response [bib_ref] The dynamic interplay of host and viral enzymes in type iii crispr-mediated..., Athukoralage [/bib_ref]. Indeed, we generally find all three components in one genome, however, there are exceptions to this rule [fig_ref] Figure 4: Distribution of CRISPR-Cas [/fig_ref]. S. solfataricus SULA (and derivative strains) contain both csx1/csm6 ribonucleases and ring nucleases, but lack a type III system indicating that these might have functions beyond type III immunity [bib_ref] Fuse to defuse: A self-limiting ribonuclease-ring nuclease fusion for type III CRISPR..., Samolygo [/bib_ref]. Contrarily, S. islandicus Y.G.57.14 which encodes four type III systems (3x subtype III-B, 1x type III unclassified), thereby representing the highest detected number of type III systems within our dataset, marks the only genome without an identifiable csx1/csm6 gene, suggesting no cOA-driven collateral damage to be in place. Furthermore, for S. islandicus strains (M.16.4, and Y.G.57.14) we could not identify any ring nuclease, although both are equipped with type III effectors. Although S. islandicus M.16.4 encodes for the promiscuous RNA shredding Csx1, its antagonistic Crn seems to be dysfunctional as it is disrupted by an internal stop codon [fig_ref] Figure 4: Distribution of CRISPR-Cas [/fig_ref] and [fig_ref] Table 1: Widely used archaeal model organisms for CRISPR in vitro and in vivo... [/fig_ref]. Thus, if type III effectors as well as CARF-domain nucleases are active in those strains, it remains unclear how cOAs levels are controlled. One explanation could be, that some CARF-domain nucleases are self-limiting and degrade cOAs by themselves, as has been observed for other organisms (see above and refs. [bib_ref] CRISPR-Cas III-A Csm6 CARF Domain Is a Ring Nuclease Triggering Stepwise cA4..., Jia [/bib_ref] [bib_ref] Activation and self-inactivation mechanisms of the cyclic oligoadenylate-dependent CRISPR ribonuclease Csm6, Garcia-Doval [/bib_ref] [bib_ref] A Type III CRISPR Ancillary Ribonuclease Degrades Its Cyclic Oligoadenylate Activator, Athukoralage [/bib_ref]. It could also be that sometimes collateral damage is regulated in trans, through other genetic elements like Crn2 ring nucleases encoded by archaeal viruses, such as STIV which is abundant in Sulfolobales (see above and ref. [bib_ref] An anti-CRISPR viral ring nuclease subverts type III CRISPR immunity, Athukoralage [/bib_ref]. For two other S. islandicus strains (M.14.25 and M. only two ring nuclease gene copies but no csx1 gene could be identified [fig_ref] Figure 4: Distribution of CRISPR-Cas [/fig_ref]. It should be noted that some ring nucleases in our analysis were previously assigned to the Csx1/Csm6 family [bib_ref] Evolutionary classification of CRISPR-Cas systems: A burst of class 2 and derived..., Makarova [/bib_ref] , however they do not seem to contain a bona fide HEPN domain. Moreover, biochemically characterized representatives encoded by S. solfataricus P2 did not show cA4-stimulated RNase activity however did exhibit cA4 degradation activity in vitro [bib_ref] Ring nucleases deactivate type III CRISPR ribonucleases by degrading cyclic oligoadenylate, Athukoralage [/bib_ref].
CRISPR regulators (casR) shown to induce (Csa3a) or repress (Csa3b) expression of adaptation cassettes/CRISPR arrays and effector complexes, respectively [bib_ref] Coupling transcriptional activation of CRISPR-Cas system and DNA repair genes by Csa3a..., Liu [/bib_ref] [bib_ref] CRISPR-Cas type I-A Cascade complex couples viral infection surveillance to host transcriptional..., He [/bib_ref] , were found in all genomes, except S. solfataricus SULA and derivates, Sulfuracidifex tepidarius as well as S. azoricus FC6 [fig_ref] Figure 4: Distribution of CRISPR-Cas [/fig_ref].
## Crispr arrays and virus matches in sulfolobales
In general, Sulfolobales harbor several CRISPR arrays, ranging from only two in M. cuprina Ar-4, Sulfodiicoccus acidophilus HS-1, S. acidocladarius Y14 18-5 and some S. islandicus strains, to up to ten in A. sulfidivorans JP7 and S. tepidarius IC-006 and IC-007 [fig_ref] Figure 5: CRISPR array, spacers and virus matches in Sulfolobales genera [/fig_ref] , [fig_ref] Table 1: Widely used archaeal model organisms for CRISPR in vitro and in vivo... [/fig_ref]. Although containing only two CRISPR arrays, S. acidophilus HS-1 exhibits the third highest total number of spacers (447), as both arrays encompass more than 200 spacers each, thereby also representing the by far longest CRISPR arrays amongst the Sulfolobales. This large total number of spacers is only exceeded by unclassified Saccharolobus sp. A20 and Sulfurisphaera tokodaii 7, harboring 463 and 454 spacers, respectively, partitioned between 6 CRISPR arrays each. In comparison, S. acidocaldarius Y14 18-5 only harbors 14 spacers within its two CRISPR arrays altogether, and hence clearly contains the lowest total number of spacers [fig_ref] Figure 5: CRISPR array, spacers and virus matches in Sulfolobales genera [/fig_ref] , [fig_ref] Table 1: Widely used archaeal model organisms for CRISPR in vitro and in vivo... [/fig_ref].
Collectively, the 38 Sulfolobales genomes accommodate over 10,400 spacers, and for approximately 6.7% of those spacers (partially) matching protospacers could be identified in the genomes of known Sulfolobales viruses (see [fig_ref] Table 1: Widely used archaeal model organisms for CRISPR in vitro and in vivo... [/fig_ref]. This value is comparable to previous analyses which were based on the entirety of spacers from all publicly available prokaryotic genomes or the CRISPRome of a natural population of Sulfolobales (7 and 6% in ref [bib_ref] Virus-borne mini-CRISPR arrays are involved in interviral conflicts, Medvedeva [/bib_ref] , respectively). In general, hyperthermophilic archaea are especially enriched in CRISPR-Cas systems, however readily identifiable spacer matches to viral genomes are comparably scarce [bib_ref] Evolutionary classification of CRISPR-Cas systems: A burst of class 2 and derived..., Makarova [/bib_ref]. As mentioned above, this observation possibly reflects the enormous variety of viruses which yet remain to be discovered. While S. solfataricus P1 and P2 exhibit the highest absolute numbers of virus matching spacers (namely 83 and 91 spacers, respectively), two S. islandicus strains, Y.G.57.14 and Y.N.15.51, exhibit the highest relative proportion of spacers (around 24 and 30%, respectively) with identifiable matching protospacers in viral genomes [fig_ref] Figure 1: Figure 1 [/fig_ref]. Generally, the top three targeted virus species (highest numbers of protospacers found in their genomes) are Sulfolobus islandicus rod-shaped virus (SIRV), Sulfolobus monocaudavirus (SMV) and Acidianus two-tailed virus (ATV) [fig_ref] Figure 5: CRISPR array, spacers and virus matches in Sulfolobales genera [/fig_ref]. altogether, and hence clearly contains the lowest total number of spacers [fig_ref] Figure 5: CRISPR array, spacers and virus matches in Sulfolobales genera [/fig_ref] , [fig_ref] Table 1: Widely used archaeal model organisms for CRISPR in vitro and in vivo... [/fig_ref]. [fig_ref] Table 1: Widely used archaeal model organisms for CRISPR in vitro and in vivo... [/fig_ref]. Additionally, abundances of solo cas4 (not encoded within 15 ORFs with respect to other adaptation genes) and accessory genes csx1/csm6, crn (ring nuclease), casR (specific transcriptional regulator) are illustrated by bubbles in corresponding sizes. In cases with great overlap between different strains of the same species, only one genome is shown as a representative for all similar strains (e.g., S. solfataricus SULA and derivative strains). The analysis is based on data published in [bib_ref] Evolutionary classification of CRISPR-Cas systems: A burst of class 2 and derived..., Makarova [/bib_ref] , and/or obtained by using programs [fig_ref] Figure 4: Distribution of CRISPR-Cas [/fig_ref]. Distribution of CRISPR-Cas (sub) types and accessory genes in Sulfolobales genomes. The bubble plot shows the distribution and abundance of genes encoding proteins and protein complexes partaking in the CRISPR-Cas immune response in Sulfolobales genomes. The abundances of adaptation cassettes (cas1-cas2, frequently also cas4), cas6 (crRNA processing) and CRISPR-Cas effectors (interference) are depicted by size; abundances of the different subtypes of CRISPR-Cas effectors are indicated by different color shadings. Some adaptation cassettes contain either a cas1 or cas2 gene with internal stop codon or frameshift, highlighted in [fig_ref] Table 1: Widely used archaeal model organisms for CRISPR in vitro and in vivo... [/fig_ref]. Additionally, abundances of solo cas4 (not encoded within 15 ORFs with respect to other adaptation genes) and accessory genes csx1/csm6, crn (ring nuclease), casR (specific transcriptional regulator) are illustrated by bubbles in corresponding sizes. In cases with great overlap between different strains of the same species, only one genome is shown as a representative for all similar strains (e.g., S. solfataricus SULA and derivative strains). The analysis is based on data published in [bib_ref] Evolutionary classification of CRISPR-Cas systems: A burst of class 2 and derived..., Makarova [/bib_ref] , and/or obtained by using programs CRISPRminer (version 1, [bib_ref] CRISPRminer is a knowledge base for exploring CRISPR-Cas systems in microbe and..., Zhang [/bib_ref] , CRISPRCasFinder (version CRISPR-Cas++ 1.1.2, [bib_ref] CRISPRCasFinder, an update of CRISRFinder, includes a portable version, enhanced performance and..., Couvin [/bib_ref] , and BLAST [bib_ref] Gapped BLAST and PSI-BLAST: A new generation of protein database search programs, Altschul [/bib_ref] [bib_ref] Protein database searches using compositionally adjusted substitution matrices, Altschul [/bib_ref] ; BLAST analysis of Crn and Csx1 was performed based on amino acid sequences of the respective biochemically and structurally characterized proteins [bib_ref] Structure of Csx1-cOA4 complex reveals the basis of RNA decay in Type..., Molina [/bib_ref] [bib_ref] Ring nucleases deactivate type III CRISPR ribonucleases by degrading cyclic oligoadenylate, Athukoralage [/bib_ref].
## A general scenario for crispr interference in a sulfolobales cell
According to our analysis and the current literature, we can draw simplified scenarios with three possible levels of how CRISPR interference against a cognate virus (i.e., carrying a protospacer that perfectly matches a native CRISPR spacer) could be achieved in a Sulfolobales cell. These scenarios do not consider variables, such as viral toxins, that might have an additional impact on virus-host interactions in nature. We reason that the efficiency of CRISPR interference largely depends on protospacer flanking motifs, such as PAM or PAS, and the transcription of the protospacer. The stacked bar plot shows the total amount of spacers identified on the genus level that (partially) match protospacers carried on genomes of viruses associated with the Sulfolobales. The data (CRISPR arrays and spacer) for the analyses were retrieved using programs CRISPRCasFinder (CRISPR-Cas++ 1.1.2, [bib_ref] CRISPRCasFinder, an update of CRISRFinder, includes a portable version, enhanced performance and..., Couvin [/bib_ref] and orientation of CRISPR arrays was determined using CRISPRstrand (implemented in CRISPRmap v1.3.0-2013, [bib_ref] Predicting repeat orientations to determine the crRNA-encoding strand at CRISPR loci, Alkhnbashi [/bib_ref] [bib_ref] CRISPRmap: An automated classification of repeat conservation in prokaryotic adaptive immune systems, Lange [/bib_ref] Acidianus (5) Metallosphaera (5) Saccharolobus solfataricus (4)
## Saccharolobus islandicus (10)
Sulfodiicoccus (1) Sulfolobus (8) Sulfuracidifex (2) Sulfurisphaera ( The stacked bar plot shows the total amount of spacers identified on the genus level that (partially) match protospacers carried on genomes of viruses associated with the Sulfolobales. The data (CRISPR arrays and spacer) for the analyses were retrieved using programs CRISPRCasFinder (CRISPR-Cas++ 1.1.2, [bib_ref] CRISPRCasFinder, an update of CRISRFinder, includes a portable version, enhanced performance and..., Couvin [/bib_ref] and orientation of CRISPR arrays was determined using CRISPRstrand (implemented in CRISPRmap v1.3.0-2013, [bib_ref] Predicting repeat orientations to determine the crRNA-encoding strand at CRISPR loci, Alkhnbashi [/bib_ref] [bib_ref] CRISPRmap: An automated classification of repeat conservation in prokaryotic adaptive immune systems, Lange [/bib_ref]. Virus matches were identified using BLAST+ (version 2.10.0,, spacers were blasted against the NCBI viral genomic RefSeq database [bib_ref] Reference sequence (RefSeq) database at NCBI: Current status, taxonomic expansion, and functional..., O'leary [/bib_ref] specifying the following parameters: word size = 8, e-value ≤ 0.01. Results were filtered for a query coverage ≥ 85% (calculated by dividing the alignment length by the query length) and allowing for a maximum number of 5 mismatches (see [fig_ref] Table 1: Widely used archaeal model organisms for CRISPR in vitro and in vivo... [/fig_ref] , duplicate hits of a spacer to the same virus were removed.
In the presence of a functional PAM sequence, the type I Cascade can recognize and degrade the DNA of the infecting virus [fig_ref] Figure 6: Three scenarios of anti-virus immunity in a Sulfolobales cell [/fig_ref] , upper panel "efficient"). Previous interference studies in S. islandicus mutants only carrying a type I-A system as sole effector module have shown that such type I-A-mediated interference can be sufficient to efficiently eradicate an invading plasmid [bib_ref] Genetic determinants of PAM-dependent DNA targeting and pre-crRNA processing in Sulfolobus islandicus, Peng [/bib_ref]. However, in a theoretical scenario where the type I system alone could not provide efficient DNA interference, possibly due to invasion of an overbearing amount of virus copies, masking of the protospacer/PAM or due to type I-A anti-CRISPRs [bib_ref] Tolerance of Sulfolobus SMV1 virus to the immunity of I-A and III-B..., Guo [/bib_ref] , the type III complex could function as a "backup" system to confer immunity [fig_ref] Figure 6: Three scenarios of anti-virus immunity in a Sulfolobales cell [/fig_ref] , upper panel, "inefficient"). crRNAs transcribed from different CRISPR arrays were shown to incorporate into both, type I and type III systems in S. solfataricus, substantiating that type III can utilize the same crRNAs as type I [bib_ref] Structural and functional characterization of an archaeal clustered regularly interspaced short palindromic..., Lintner [/bib_ref] [bib_ref] Structure and mechanism of the CMR complex for CRISPR-mediated antiviral immunity, Zhang [/bib_ref]. This however could only occur if the protospacer is transcribed and its mRNA (or antisense RNA) matches the cognate crRNA [fig_ref] Figure 6: Three scenarios of anti-virus immunity in a Sulfolobales cell [/fig_ref] , upper panel "inefficient"). As a functional PAM always mismatches the 5 handle of the crRNA, all three immune functions of the type III system (RNA degradation, HD-mediated unspecific ssDNA cleavage and cOA-signaling-dependent collateral RNA damage) would then be activated, efficiently eradicating the infecting virus. The fate of the cell would depend on the efficiency of the ring nuclease resetting the cell to a ground state after the virus has been defeated (see above and refs [bib_ref] Ring nucleases deactivate type III CRISPR ribonucleases by degrading cyclic oligoadenylate, Athukoralage [/bib_ref].
In a second scenario, a protospacer is flanked by a PAS, matching the 5 handle of the native crRNA. As the protospacer would not be recognized by the type I complex on DNA level due to the missing PAM, the virus can efficiently infect the cell and propagate, leading to virus spreading in the population [fig_ref] Figure 6: Three scenarios of anti-virus immunity in a Sulfolobales cell [/fig_ref] , lower right panel "untranscribed PS"). However, if the protospacer is transcribed and the transcript can hybridize to the native crRNA, the type III complex can be activated to degrade the virus RNA [fig_ref] Figure 6: Three scenarios of anti-virus immunity in a Sulfolobales cell [/fig_ref] , lower right panel "PS transcript"). As the handle -PAS match allosterically inhibits activation of secondary immune responses, the type III system would only silence the virus by specifically degrading its RNA. This could lead to a stabilization of the intracellular virus number (i.e., prophage stage) while inhibiting the propagation of the virus [bib_ref] A Conditional tolerance of temperate phages via transcription-dependent CRISPR-Cas targeting, Goldberg [/bib_ref]. Indeed, expression of a virus mRNA carrying a cognate protospacer flanked by a PAS in S. solfataricus led to stable virus DNA copy numbers, whereas the targeted mRNA levels were reduced [bib_ref] Efficient CRISPR-Mediated Post-transcriptional Gene Silencing in a Hyperthermophilic Archaeon Using Multiplexed crRNA..., Zebec [/bib_ref]. Thus, in scenario 2, the chance is highest for the virus to persist in the population.
If an infecting virus carries a protospacer without a flanking region that is similar to a PAM/PAS, the type I system would not be triggered [fig_ref] Figure 6: Three scenarios of anti-virus immunity in a Sulfolobales cell [/fig_ref] , lower left panel "untranscribed PS"). If the protospacer is transcribed, survival of the cell would depend solely on the proper activity of the type III complex for eradication of the virus, and the efficiency of the ring nuclease in degrading residual cOAs [fig_ref] Figure 6: Three scenarios of anti-virus immunity in a Sulfolobales cell [/fig_ref] , lower left panel "PS transcript").
We conclude that in Sulfolobales harboring type I and type III systems, viruses carrying a protospacer with a PAM would most likely be efficiently degraded and eradicated from the population, whereas protospacers flanked with a PAS would have the highest probability to persist in a carrier or prophage state. However, as mentioned above, these dynamics can change under different environmental conditions and probably largely depend on the nature of the infecting element-factors we did not consider here. For instance, a silenced virus could theoretically still kill the host upon expression of a harmful protein.
Although it is quite probable that not all functions and benefits of CRISPR immunity have been discovered yet, it becomes clear that the diversity of viruses is reflected in an impressive diversity of CRISPR-Cas systems in the Sulfolobales. The sophisticated immune system in turn does not necessarily always result in eradication of viruses but may also foster co-existences that are known to be beneficial to natural microbial populations, as they increase genetic diversity and exchange. We conclude that in Sulfolobales harboring type I and type III systems, viruses carrying a protospacer with a PAM would most likely be efficiently degraded and eradicated from the population, whereas protospacers flanked with a PAS would have the highest probability to persist in a carrier or prophage state. However, as mentioned above, these dynamics can change under different environmental conditions and probably largely depend on the nature of the infecting element-factors we did not consider here. For instance, a silenced virus could theoretically still kill the host upon expression of a harmful protein.
Although it is quite probable that not all functions and benefits of CRISPR immunity have been discovered yet, it becomes clear that the diversity of viruses is reflected in an impressive diversity of CRISPR-Cas systems in the Sulfolobales. The sophisticated immune system in turn does not [fig_ref] Figure 6: Three scenarios of anti-virus immunity in a Sulfolobales cell [/fig_ref]. Three scenarios of anti-virus immunity in a Sulfolobales cell. Scenario 1 (PAM) would be the most efficient to degrade a virus in a Sulfolobales cell, whereas scenario 3 (no PAM, no PAS) could lead to virus spreading and/or cell death. Scenario 2 (PAS) is the only scenario that could lead to virus silencing on RNA level and, consequently, persistence of the virus in the population.
# Supplementary materials:
The following are available online at http://www.mdpi.com/2218-273X/10/11/1523/s1. Supplementary Table S1 contains all source data for the comparative analysis of CRISPR-Cas systems in Sulfolobales.
Author Contributions: I.A.Z. conceived the study, did literature research and wrote the manuscript, E.W. performed comparative analysis of Sulfolobales and wrote the respective text part; C.S. edited the text and gave scientific supervision and input. All authors discussed data and revised the manuscript. All authors have read and agreed to the published version of the manuscript.
[fig] Figure 1: Figure 1. CRISPR types in prokaryotes and their mechanisms of action. Class 1 systems (including types I, III, and IV) are prevalent in both archaea and bacteria, whereas Class 2 systems (including types II, V and VI) are found almost exclusively in bacteria. Mechanisms and genes in the overlapping region are shared by the two classes. The different CRISPR steps are indicated in red and explained in the main text. Both classes contain effector complexes that can perform all three types of interference: virus DNA interference (indicated by pink scissors), virus RNA interference (indicated by green scissors) and collateral damage (i.e., promiscuous cleavage of host and virus DNA/RNA, indicated by purple scissors). "S" refers to "spacer", "R" refers to "repeat". * Asterisks refer to enzymes that are not involved in processing of all classes: Cas5d replaces Cas6 function in type I-C [25]; RNaseE is involved in processing in a type III-B system [26]; RNase III processes pre-crRNA in type II; tracrRNA is needed in type II and certain subtypes of type V (see text). Type I-D targets ssDNA and dsDNA (not shown) [27]; In type II systems, Csn2 is also involved in spacer acquisition (not shown); spacer acquisition from RNA via RTs is not shown [28]; ssRNA targeting of the type V-G subtype is not shown [29] (see text). [/fig]
[fig] Figure 2: Activated and deactivated immune response of CRISPR type III complex. (A) [/fig]
[fig] Figure 3: Presence of specific CRISPR-Cas (sub)types and their abundances among Sulfolobales genera. Overall abundances of CRISPR-Cas types and the specific subtypes within the general types I and III found in the order Sulfolobales (Upper panel). The bar chart shows the distribution of the CRISPR-Cas subtypes throughout the genera within Sulfolobales (Lower panel). The number of genomes included in the analysis for each genus is given in brackets. The analysis is based on data published in ref.[19] and/or obtained by using programs CRISPRminer (version 1,[258]), and CRISPRCasFinder (version CRISPR-Cas++ 1.1.2,[241]). [/fig]
[fig] Figure 4: Distribution of CRISPR-Cas (sub)types and accessory genes in Sulfolobales genomes. The bubble plot shows the distribution and abundance of genes encoding proteins and protein complexes partaking in the CRISPR-Cas immune response in Sulfolobales genomes. The abundances of adaptation cassettes (cas1-cas2, frequently also cas4), cas6 (crRNA processing) and CRISPR-Cas effectors (interference) are depicted by size; abundances of the different subtypes of CRISPR-Cas effectors are indicated by different color shadings. Some adaptation cassettes contain either a cas1 or cas2 gene with internal stop codon or frameshift, highlighted in [/fig]
[fig] Figure 5: CRISPR array, spacers and virus matches in Sulfolobales genera. (A) The number of arrays (depicted in different color shadings) and the number of spacers per array are shown for representative genomes of the Sulfolobales. (B) [/fig]
[fig] Figure 6: Three scenarios of anti-virus immunity in a Sulfolobales cell. Scenario 1 (PAM) would be the most efficient to degrade a virus in a Sulfolobales cell, whereas scenario 3 (no PAM, no PAS) could lead to virus spreading and/or cell death. Scenario 2 (PAS) is the only scenario that could lead to virus silencing on RNA level and, consequently, persistence of the virus in the population. [/fig]
[fig] Funding: This work was financed through a DOC-fellowship of the Austrian Academy of Sciences (OEAW) and a PhD completion grant of the University of Vienna to I.A.Z. as well as an ERC-Adv grant TACKLE (No. 695192) and FWF grant (W1257) to C.S. Open Access Funding by the Austrian Science Fund (FWF). [/fig]
[table] Table 1: Widely used archaeal model organisms for CRISPR in vitro and in vivo studies. Pioneer studies regarding the respective CRISPR step performed in each model organism are cited.Table 1. Cont. [/table]
[bib_ref] Evolutionary classification of CRISPR-Cas systems: A burst of class 2 and derived..., Makarova [/bib_ref] [bib_ref] An updated evolutionary classification of CRISPR-Cas systems, Makarova [/bib_ref] [bib_ref] CRISPRCasFinder, an update of CRISRFinder, includes a portable version, enhanced performance and..., Couvin [/bib_ref] [bib_ref] Type III CRISPR-Cas systems produce cyclic oligoadenylate second messengers, Niewoehner [/bib_ref] [bib_ref] A cyclic oligonucleotide signaling pathway in type III CRISPR-Cas systems, Kazlauskiene [/bib_ref] |
Health and nutritional aspects of sustainable diet strategies and their association with environmental impacts: a global modelling analysis with country-level detail
Background Sustainable diets are intended to address the increasing health and environmental concerns related to food production and consumption. Although many candidates for sustainable diets have emerged, a consistent and joint environmental and health analysis of these diets has not been done at a regional level. Using an integrated health and environmental modelling framework for more than 150 countries, we examined three different approaches to sustainable diets motivated by environmental, food security, and public health objectives.MethodsIn this global modelling analysis, we combined analyses of nutrient levels, diet-related and weight-related chronic disease mortality, and environmental impacts for more than 150 countries in three sets of diet scenarios. The first set, based on environmental objectives, replaced 25-100% of animal-source foods with plant-based foods. The second set, based on food security objectives, reduced levels of underweight, overweight, and obesity by 25-100%. The third set, based on public health objectives, consisted of four energy-balanced dietary patterns: flexitarian, pescatarian, vegetarian, and vegan. In the nutrient analysis, we calculated nutrient intake and changes in adequacy based on international recommendations and a global dataset of nutrient content and supply. In the health analysis, we estimated changes in mortality using a comparative risk assessment with nine diet and weightrelated risk factors. In the environmental analysis, we combined country-specific and food group-specific footprints for greenhouse gas emissions, cropland use, freshwater use, nitrogen application, and phosphorus application to analyse the relationship between the health and environmental impacts of dietary change.Findings Following environmental objectives by replacing animal-source foods with plant-based ones was particularly effective in high-income countries for improving nutrient levels, lowering premature mortality (reduction of up to 12% [95% CI 10-13] with complete replacement), and reducing some environmental impacts, in particular greenhouse gas emissions (reductions of up to 84%). However, it also increased freshwater use (increases of up to 16%) and had little effectiveness in countries with low or moderate consumption of animal-source foods. Following food-security objectives by reducing underweight and overweight led to similar reductions in premature mortality (reduction of up to 10% [95% CI 9-11]), and moderately improved nutrient levels. However, it led to only small reductions in environmental impacts at the global level (all impacts changed by <15%), with reduced impacts in high-income and middle-income countries, and increased resource use in low-income countries. Following public health objectives by adopting energy-balanced, low-meat dietary patterns that are in line with available evidence on healthy eating led to an adequate nutrient supply for most nutrients, and large reductions in premature mortality (reduction of 19% [95% CI 18-20] for the flexitarian diet to 22% [18-24] for the vegan diet). It also markedly reduced environmental impacts globally (reducing greenhouse gas emissions by 54-87%, nitrogen application by 23-25%, phosphorus application by 18-21%, cropland use by 8-11%, and freshwater use by 2-11%) and in most regions, except for some environmental domains (cropland use, freshwater use, and phosphorus application) in low-income countries.Interpretation Approaches for sustainable diets are context specific and can result in concurrent reductions in environmental and health impacts globally and in most regions, particularly in high-income and middle-income countries, but they can also increase resource use in low-income countries when diets diversify. A public health strategy focused on improving energy balance and dietary changes towards predominantly plant-based diets that are in line with evidence on healthy eating is a suitable approach for sustainable diets. Updating national dietary guidelines to reflect the latest evidence on healthy eating can by itself be important for improving health and reducing environmental impacts and can complement broader and more explicit criteria of sustainability.Funding Wellcome Trust, EAT, CGIAR, and British Heart Foundation.
# Introduction
The food people eat impacts their health and the health of the environment. Imbalanced diets low in fruits, vegetables, nuts, and whole grains and high in red and processed meat are responsible for the greatest health burden worldwide and in most regions. [bib_ref] Global, regional, and national comparative risk assessment of 79 behavioural, environmental and..., Forouzanfar [/bib_ref] In addition to imbalanced diets, about 2 billion people are overweight and obese, 2 billion have nutritional deficiencies, and about 800 million are still suffering from hunger due to poverty and poorly developed food systems.As the dietary transition towards more processed and highvalue food products (in terms of cost and perceived value) continues in many regions of the world, these dietary health risks are expected to worsen. [bib_ref] Analysis and valuation of the health and climate change cobenefits of dietary..., Springmann [/bib_ref] The environmental impacts of food production are similarly daunting. Agriculture is responsible for about a quarter of all greenhouse gas emissions, [bib_ref] Climate change and food systems, Vermeulen [/bib_ref] it occupies about 40% of the Earth's surface [bib_ref] Farming the planet: 1. Geographic distribution of global agricultural lands in the..., Ramankutty [/bib_ref] and uses 70% of all freshwater resources, [bib_ref] World water resources at the beginning of the twenty-first century, Shiklomanov [/bib_ref] and the overapplication of fertilisers in some regions has led to pollution of surface water and groundwater and created dead zones in oceans. [bib_ref] Spreading dead zones and consequences for marine ecosystems, Diaz [/bib_ref] As a result, the global food system has contributed to the crossing of several of the proposed planetary boundaries that attempt to define a safe operating space for humanity on a stable Earth system. [bib_ref] Agriculture production as a major driver of the Earth system exceeding planetary..., Campbell [/bib_ref] In the absence of dedicated mitigation strategies or changes in demand, many of these environmental impacts are expected to intensify as demand for foods with greater environmental impact, such as meat and dairy, increases and the global population grows from 7 billion to a predicted 10 billion in the next 30 years. [bib_ref] Options for keeping the food system within environmental limits, Springmann [/bib_ref] The concept of sustainable diets combines the challenges of creating a food system that supplies healthy diets for a growing population while reducing its environmental impacts and staying within planetary boundaries.The literature on sustainable diets has grown substantially in the past decade, [bib_ref] The impacts of dietary change on greenhouse gas emissions, land use, water..., Aleksandrowicz [/bib_ref] [bib_ref] alignment of healthy dietary patterns and environmental sustainability: a systematic review, Nelson [/bib_ref] [bib_ref] Environmental impact of dietary change: a systematic review, Hallström [/bib_ref] [bib_ref] The impact of nutritional choices on global warming and policy implications: examining..., Joyce [/bib_ref] and the concept has been expanded to economic, ethical, and cultural aspects of diets.However, consistent health analyses of commonly proposed diets are scarce, [bib_ref] Do low-carbon-emission diets lead to higher nutritional quality and positive health outcomes?..., Payne [/bib_ref] and approaches that are based primarily on health rather than environmental objectives are, with few exceptions, 3,17 rarely considered.
Here, we use a comprehensive modelling framework for more than 150 countries and regions to analyse the environmental and health impacts of three different
## Research in context
Evidence before this study Several systematic reviews of the sustainable diet literature have been published since 2014. These reviews have suggested that reductions in environmental impacts of food production are generally proportional to reductions in animal-source foods. However, most studies included in the reviews were national case studies from high-income countries with a predominant focus on greenhouse gas emissions as environmental impact, and the studies generally used different reference diets, environmental footprints, and scenario designs, all of which complicates comparisons between studies. With a few exceptions, the health impacts of the sustainable-diet scenarios were often not explicitly analysed beyond adherence to national dietary guidelines or directional changes in nutrient levels. One systematic review of the health impacts of diets with reduced greenhouse gas emissions found no consistent relationship, but some association of such diets with decreased micronutrient content.
## Added value of this study
Here, we increase the evidence base beyond summaries of national case studies by using a study design that allows for a consistent and region-specific assessment of the health and environmental impacts of dietary changes for all world regions and more than 150 individual countries. Our study includes a full nutritional analysis, a comparative risk analysis with nine dietary and weight-related risk factors, and an environmental analysis with five domains, including greenhouse gas emissions, cropland use, freshwater use, nitrogen application, and phosphorus application. Our results indicate that energy-balanced, predominantly plant-based dietary patterns that are in line with the current evidence on healthy eating can lead to reductions in environmental impacts in high-income and middle-income countries, while improving nutrient levels and reducing diet-related premature mortality in all regions. In low-income countries, adoption of healthier diets can increase demand for environmental resources, in part due to inefficient production systems and baseline diets high in staple crops. Across regions, the associations between health and environmental benefits are strongest for greenhouse gas emissions, moderate for the demand for cropland, nitrogen, and phosphorus, and small for freshwater use. Our analysis contributes to several open questions in the literature regarding the association between environmental and health impacts, the importance of improving energy imbalances and changing dietary composition, and the generalisability of national case studies.
## Implications of all the available evidence
Sustainable diets are context specific. They attain nutritional adequacy and reduce diet-related mortality by addressing both dietary composition and energy balance, and they balance environmental impacts between global and regional scales. Our study shows that a synergistic perspective on sustainable diets would include technological and management-related aspects, in addition to dietary and environmental aspects. Updating national dietary guidelines to reflect the latest evidence on healthy eating is important for improving health and reducing environmental impacts, and can complement broader and more explicit criteria of sustainability.
approaches to sustainable diets, with a focus on the nutritional implications and impacts on chronic, noncommunicable disease mortality, and how they relate to changes in greenhouse gas emissions, cropland use, freshwater use, nitrogen application, and phosphorus application. We distinguish between three different approaches: those that follow the environmental objectives of reducing the impacts of animal-source foods; those that follow the food-security objectives of addressing energy imbalances; and those that follow the public health objectives of encouraging nutritionally balanced dietary patterns based on available evidence on healthy eating.
# Methods
## Scenario design
In this global modelling analysis, we assessed the effects of three sets of dietary-change strategies (panel; appendix pp 7-9) on health and environmental factors in more than 150 countries and compare these effects across different regions worldwide.
Our baseline data consists of current and projected levels of food consumption and weight distributions. In our main analysis, we focused on the year 2010 for analysing nutrient adequacy and on the year 2030 for the mortality and environmental analyses to allow for a transition time for dietary and technological changes. Additionally, we used the years 2010 and 2050 in sensitivity analyses to study the effects of technological and socioeconomic changes.
We estimated baseline and projected food intake for more than 150 countries by adapting food demand projections from the International Model for Policy Analysis of Agricultural Commodities and Trade (IMPACT) that were based on a harmonised dataset of country-specific food availability data, and we adjusted those for food waste at the household level (appendix pp 2-4).For estimating the prevalence of underweight (body-mass index [BMI] <18 kg/m²), overweight (BMI ≥25 to <30 kg/m²), and obesity (BMI ≥30 kg/m²) in each country, we fitted log-normal distributions to WHO estimates of mean BMI and the prevalence of overweight and obesity using a cross-entropy method that jointly mini mised the deviation of the prevalence data, [bib_ref] Global and regional health effects of future food production under climate change:..., Springmann [/bib_ref] and we projected weight changes using correlations between changes in mean BMI and changes in food availability (appendix pp [bib_ref] Farming the planet: 1. Geographic distribution of global agricultural lands in the..., Ramankutty [/bib_ref] [bib_ref] World water resources at the beginning of the twenty-first century, Shiklomanov [/bib_ref]. [bib_ref] Global and regional health effects of future food production under climate change:..., Springmann [/bib_ref] In the first diet pattern set that we assessed, which was based on environmental concerns by reducing animalsource products, we progressively reduced the amount of animal-source foods in each country's diet by 25% (ani-25), 50% (ani-50), 75% (ani-75), and 100% (ani-100) and replaced it with plant-based foods. On the basis of observational data on food-group substitution across dietary patterns, we chose a substitution rule by which two-thirds of animal-source foods were replaced by legumes and a third by fruits and vegetables. We kept the total energy content of the diets constant to isolate the effects of changes in dietary composition in this set of scenarios. In the second set, based on food security and improving energy balance, we progressively reduced levels of underweight, over weight, and obesity in a simultaneous fashion by 25% (kcal-25), 50% (kcal-50), 75% (kcal-75), and 100% (kcal-100). For adjusting total energy intake, we applied scaling factors to baseline diets that preserved their composition. In the third set, based on public health priorities, we constructed four nutritionally balanced dietary patterns that are in line with evidence on healthy eating. [bib_ref] Current evidence on healthy eating, Willett [/bib_ref] For that purpose, we used energy-balanced varieties of the flexitarian, pescatarian, vegetarian, and vegan dietary patterns defined by the EAT-Lancet Commission on Healthy Diets from Sustainable Food Systems (appendix pp 7-9). The flexitarian dietary patterns contain no processed meat, low amounts of red meat (including beef, lamb, and pork) and sugar, moderate amounts of poultry, dairy, and fish, and generous amounts of fruits, vegetables, legumes, and nuts. The other three dietary patterns replace meat (pescatarian or vegetarian) or all animalsource foods (vegan) with two-thirds either fish and seafood (pescatarian diets) or legumes (vegetarian and vegan diets) and a third fruits and vegetables. We regionalised the public health dietary patterns for each country by preserving the national preferences for types Panel: Dietary-change strategies for sustainable diets Reduction of animal-source foods following environmental objectives Replacement of 25-100% of animal-source foods with plantbased ones at constant total calorie intake (ani-25, ani-50, ani-75, and ani-100); plant-based replacements consist of 75% legumes and 25% fruits and vegetables Improving calorie intake and weight levels following food-security objectives Improvement of 25-100% in energy imbalances (kcal-25, kcal-50, kcal-75, and kcal-100) with simultaneous reductions in underweight, overweight, and obesity Using balanced diet patterns following public health objectives Nutritionally balanced diet patterns in line with available evidence on healthy eating: - Flexitarian: no processed meat, small amounts of red meat (one serving per week), moderate amounts of other animal-source foods (poultry, fish, and dairy), and generous amounts of plant-based foods (fruits, vegetables, legumes, and nuts) - Pescatarian: replaces meat with two-thirds fish and seafood and a third fruits and vegetables - Vegetarian: replaces meat with two-thirds legumes and a third fruits and vegetables - Vegan: replaces all animal-source foods with two-thirds legumes and a third fruits and vegetables See Online for appendix of grains, fruits, red meat, and fish. The diet patterns were all compared with a benchmark diet based on our estimates of current and future food consumption and weight distributions, including increased consump tion of high-value products (animal-source foods), in line with income and population changes.
# Nutrient analysis
We analysed the nutrient adequacy of the diet scenarios by calculating their nutrient content and comparing these values to international recommendations. For calculating the nutrient content, we paired the consumption of each food group with its nutrient density as reported in the Global Expanded Nutrient Supply dataset, 21 a global dataset of nutrient supply of 23 nutrients across 225 food categories for more than 150 countries, supplemented by nutritional data on pantothenate and vitamin B12 from the nutrient databases maintained by Harvard University and the US Department of Agriculture. For our analysis, we aggregated the nutrient dataset to the commodity and regional detail of our consumption data, and we normalised calorie densities to those of the UN Food and Agriculture Organization for consistency with our diet scenarios (appendix pp 10-11). We then compared the calculated nutrient content of the diet scenarios to recommendations by WHO. Because the recommendations differ by age and sex, we calculated population-level average values for each nutrient by using the age and sex structure for the year of analysis based on data by the Global Burden of Disease project and forward projections by the UN Population Division. Our estimates of recommended energy intake account for the age-specific and sex-specific energy needs for a moderately active population with US height as an upper bound and include the energy costs of pregnancy and lactation.Our estimates of calcium intake account for the average calcium content of drinking water in line with previous assessments. [bib_ref] Global trends in dietary micronutrient supplies and estimated prevalence of inadequate intakes, Beal [/bib_ref] Because WHO did not set guidelines for phosphorus and copper, we used recommended intakes for these nutrients from the US Institute of Medicine.
# Mortality analysis
To analyse the implications of dietary change for chronic disease mortality, we constructed a comparative risk assessment framework with nine risk factors and five disease endpoints. The risk factors included high con sumption of red meat, low consumption of fruits, vegetables, nuts and seeds, fish, and legumes, as well as being underweight (BMI <18·5 kg/m²), overweight (BMI ≥25 to <30 kg/m²), or obese (BMI ≥30 kg/m²). The disease endpoints were coronary heart disease, stroke, type 2 diabetes mellitus, cancer (in aggregate and as sitespecific ones, such as colon and rectum cancers), and an aggregate of other causes that are associated with changes in weight. The disease endpoints accounted for about half of all deaths in 2015, [bib_ref] Global, regional, and national life expectancy, all-cause mortality, and cause-specific mortality for..., Wang [/bib_ref] and the risk factors were responsible for two-thirds of deaths attributable to dietary risk factors, and for a third of all attributable deaths. [bib_ref] Global, regional, and national comparative risk assessment of 79 behavioural, environmental and..., Forouzanfar [/bib_ref] We estimated the mortality and disease burden attributable to dietary risk factors by calculating population impact fractions and applying those to age-specific and country-specific mortality rates (appendix pp 12-13). [bib_ref] Vander Hoorn S. Comparative quantification of health risks: conceptual framework and methodological..., Murray [/bib_ref] Population impact fractions are the proportions of cases of disease that would be avoided when the risk exposure was changed from a baseline situation (the benchmark diet) to a counterfactual situation (the dietary scenarios). We used relative risk estimates from meta-analyses of prospective cohort studies for dietary risks, which relate the risk factors to the disease endpoints, and pooled cohort studies for weight-related risks (appendix pp [bib_ref] Environmental impact of dietary change: a systematic review, Hallström [/bib_ref] [bib_ref] The impact of nutritional choices on global warming and policy implications: examining..., Joyce [/bib_ref] [bib_ref] Do low-carbon-emission diets lead to higher nutritional quality and positive health outcomes?..., Payne [/bib_ref] [bib_ref] Global diets link environmental sustainability and human health, Tilman [/bib_ref] [bib_ref] Current evidence on healthy eating, Willett [/bib_ref] [bib_ref] Global and regional health effects of future food production under climate change:..., Springmann [/bib_ref]. In line with the meta-analyses, we included non-linear dose-response relationships for fruits and vegetables, nuts and seeds, and fish, and assumed linear dose-response relationships for the remaining risk factors. Because our analysis was primarily focused on mortality from chronic diseases, we focused on adults aged 20 years or older, and we adjusted the relative risk estimates for attenuation with age on the basis of a pooled analysis of cohort studies focused on metabolic risk factors [bib_ref] The age-specific quantitative effects of metabolic risk factors on cardiovascular diseases and..., Singh [/bib_ref] in line with other assessments. [bib_ref] Global, regional, and national life expectancy, all-cause mortality, and cause-specific mortality for..., Wang [/bib_ref] In addition to changes in total mortality, we calculated years of life lost (appendix p 20), but focus on changes in premature mortality among people aged 30-70 years in the main analysis.
# Environmental analysis
For analysing the environmental impacts of the diet scenarios, we used a food-systems model that connects food consumption and production across regions (appendix pp [bib_ref] Global Expanded Nutrient Supply (GENuS) Model: a new method for estimating the..., Smith [/bib_ref] , and we paired the production estimates with country-specific environmental footprints for greenhouse gas emissions, cropland use, freshwater use, and nitrogen and phosphorus application (appendix pp 22-24). [bib_ref] Options for keeping the food system within environmental limits, Springmann [/bib_ref] The food-systems model accounts for trade, feed, and processing of primary commodities and is calibrated with data from the IMPACT agricultureeconomic model.The greenhouse gas emissions associated with agriculture included methane and nitrous oxide emissions, but they exclude carbon dioxide emissions which, following the methods of the International Panel on Climate Change, are allocated to the energy sector or others. Freshwater use, as assessed here, denotes the consumption of surface water and groundwater, and nitrogen and phosphorus application are associated with fertiliser use. The footprints for animal-source foods include the indirect impacts associated with feed production and, for greenhouse gas emissions, direct impacts associated with methane emissions. The projec tion of environmental footprints includes improve ments in technology and management practices along different socioeconomic development pathways (appendix pp [bib_ref] Global trends in dietary micronutrient supplies and estimated prevalence of inadequate intakes, Beal [/bib_ref] [bib_ref] Global, regional, and national life expectancy, all-cause mortality, and cause-specific mortality for..., Wang [/bib_ref] [bib_ref] Vander Hoorn S. Comparative quantification of health risks: conceptual framework and methodological..., Murray [/bib_ref]. [bib_ref] Options for keeping the food system within environmental limits, Springmann [/bib_ref] In this study, we report on the environmental analysis for contextualisation and focus on the comparison between the health and environmental impacts. For that purpose, we focus on the directional changes in the environmental parameters, which provide a good indication for increased environmental pressures in most regions and globally. [bib_ref] Planetary boundaries: guiding human development on a changing planet, Steffen [/bib_ref] Some exceptions exist, especially for fertiliser use where increased application in low-applying regions can lead to increased agricultural yields without major environmental impacts. [bib_ref] Closing yield gaps through nutrient and water management, Mueller [/bib_ref] [bib_ref] Life's bottleneck: sustaining the world's phosphorus for a food secure future, Cordell [/bib_ref] [bib_ref] The global nitrogen cycle in the twenty-first century, Fowler [/bib_ref] Our socioeconomic development trajectories include a rebalancing of fertiliser application between overapplying and underapplying regions by 2050, [bib_ref] Closing yield gaps through nutrient and water management, Mueller [/bib_ref] which reduces the potential for such exceptions.
# Uncertainty analysis
We accounted for the major uncertainties in each analysis. In the comparative risk analysis, we calculated uncertainty intervals associated with changes in mortality using error propagation and the CIs of the relative risk parameters (appendix pp [bib_ref] alignment of healthy dietary patterns and environmental sustainability: a systematic review, Nelson [/bib_ref] [bib_ref] Closing yield gaps through nutrient and water management, Mueller [/bib_ref] [bib_ref] Life's bottleneck: sustaining the world's phosphorus for a food secure future, Cordell [/bib_ref]. In the nutritional analysis, we explicitly calculated low and high supply values of each nutrient on the basis of the reported CIs (appendix p 26). And in the environmental analysis, we assessed un certainty by considering different population and income projections, which change the absolute amount and the dietary composition of foods demanded (appendix p 25).
## Role of the funding source
The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. All authors had full access to all the data in the study and the corresponding author had final responsibility for the decision to submit for publication.
# Results
Across the dietary-change scenarios, nutrient availability was generally improved for nutrients that were at low levels at baseline (2010), but with large differences between the scenarios (figure 1; CIs and regional results are reported in the appendix pp [bib_ref] The age-specific quantitative effects of metabolic risk factors on cardiovascular diseases and..., Singh [/bib_ref] [bib_ref] Planetary boundaries: guiding human development on a changing planet, Steffen [/bib_ref]. In the scenarios that replace animal-source foods (ani-25, ani-50, ani-75, and ani-100), the macronutrient content of diets changed towards lower protein and fat content, with large reductions in saturated fatty acids. Protein intake remained adequate in high-income and middle-income countries, but they decreased to lower than recommended amounts in low-income countries. Micronutrient intake improved, particularly in high-income and middle-income countries, where large amounts of animal-source foods could be replaced by plant-based ones. In the high-income and middle-income countries, the baseline low levels of vitamin A, folate, iron, potassium, and fibre increased to greater than recommended values, but calcium, pantothenate (B5), and vitamin B12 decreased to less than recommended levels under full substitution. In lowincome countries, the small amounts of animal-source foods that were replaced were not enough to sufficiently increase vitamin A and potassium, and calcium and riboflavin also did not achieve recommended values.
In the scenarios focused on improving energy balance (kcal-25, kcal-50, kcal-75, and kcal-100), total energy intake was reduced to achieve recommended values in highincome and middle-income countries, whereas it was increased to achieve recommendations in low-income countries (appendix p 27). Increased energy intake improved micronutrient intake in low-income countries, but vitamin A, folate, calcium, potassium, and riboflavin remained below recommended values. In high-income and middle-income countries, reductions in energy intake did not improve the baseline low values of folate, iron, potassium, and fibre. The scenarios based on balanced dietary patterns (flexitarian, pescatarian, vegetarian, and vegan) combined the nutritional impacts of improving energy balance with food-based dietary guidelines for all regions. As a result, most of the nutrients that are at low levels in the baseline diets (vitamin A, folate, iron, potassium, and fibre), increased to recommended values in all four patterns. However, as in the other scenarios based on energy balance or animal-source food substitution, ribo flavin remained low, and calcium and vitamin B12 fell below recommended values in the vegetarian or vegan scenarios, or both (appendix p 27). These nutrients would have to be supplemented to attain the recommended value.
In the comparative risk assessment, premature mortality decreased both with reductions in animalsource foods and with improvements in the energy balance of diets [fig_ref] Figure 2: Premature mortality and environmental impacts of diet scenarios in 2030 [/fig_ref]. Progressively replacing animal-source foods with plant-based foods led to progressive reductions in premature mortality of 4% (95% CI 4-4) in the ani-25 scenario up to 12% [bib_ref] The impacts of dietary change on greenhouse gas emissions, land use, water..., Aleksandrowicz [/bib_ref] [bib_ref] alignment of healthy dietary patterns and environmental sustainability: a systematic review, Nelson [/bib_ref] in the ani-100 scenario in 2030 (absolute values are in the appendix pp [bib_ref] Closing yield gaps through nutrient and water management, Mueller [/bib_ref] [bib_ref] Life's bottleneck: sustaining the world's phosphorus for a food secure future, Cordell [/bib_ref]. More than half of the total percentage reduction was due to increased vegetable consumption (51-58% across the scenarios), a third of the difference was due to increased fruit consumption (29-31%), a fifth due to increased legume consumption (18-23%), and a tenth due to reductions in red meat consumption (8-11%; [fig_ref] Figure 2: Premature mortality and environmental impacts of diet scenarios in 2030 [/fig_ref]. Reduction in fish intake led to a 0·3-1% increase in premature mortality across the scenarios. Coronary heart disease (35-36%), stroke (32-34%), and cancer (29-30%) each accounted for about a third of deaths averted, and a small number were from type 2 diabetes (2-3%; appendix p 30). The reductions in premature mortality were two to three times greater in high-income and middle-income countries (12-14% in the ani-100 scenario), where a larger portion of animalsource foods can be substituted, than in low-income countries (5%; [fig_ref] Figure 3: Regional changes in premature mortality and environmental impacts of dietary changeThe scenarios... [/fig_ref].
Progressively improving the energy balance of diets by reducing both underconsumption and overconsumption, while preserving the overall composition of diets (kcal-25, kcal-50, kcal-75, and kcal-100 scenarios), led to reductions in premature mortality from 2% (95% CI 2-3) in the kcal-25 scenario to 10% (9-11) in the kcal-100 scenario [fig_ref] Figure 2: Premature mortality and environmental impacts of diet scenarios in 2030 [/fig_ref]. Across risk factors, reductions in obesity contributed the most to the overall reduction in premature mortality (61-65% across the scenarios), followed by reductions in underweight (42-43%) and reductions in overweight (excluding obesity; 15-17%). As energy intake was adjusted upwards or downwards, the intake of specific foods increased or decreased, and in aggregate led to a 2% increase in mortality, in particular from reductions in fruit and vegetable consumption. Adjusting weight levels without affecting nutrition-sensitive food groups (eg, by adjusting staple foods) would increase the overall reduction in premature mortality by 19% (to a total reduction in premature mortality of 11%). Across disease endpoints, most averted premature deaths were from non-cardiovascular causes related to weight (about 60% with each of the four scenarios), followed by type 2 diabetes (17%), cancer (9-12%), coronary heart disease (9-11%), and stroke (2-3%; appendix p 30). The reductions in premature mortality were generally evenly distributed across regions (8-14% across income groups in the kcal-100 scenario), with the greatest reduction in the upper-middle-income (14%) and high-income (12%) countries, which have high burdens of obesity, followed by low-income countries, (12%) which have high burdens of underweight, and lower-middle-income countries (8%) with an intermediate weight profile [fig_ref] Figure 3: Regional changes in premature mortality and environmental impacts of dietary changeThe scenarios... [/fig_ref].
Combining changes in dietary composition with changes in energy balance substantially increased the reductions in premature mortality that each strategy can achieve. Dietary changes to balanced flexitarian, pescatarian, vegetarian, and vegan diets led to reductions in premature mortality of 19% (95% CI [bib_ref] Current evidence on healthy eating, Willett [/bib_ref] [bib_ref] Global and regional health effects of future food production under climate change:..., Springmann [/bib_ref] for the flexitarian scenario to 22% [bib_ref] Current evidence on healthy eating, Willett [/bib_ref] [bib_ref] Global and regional health effects of future food production under climate change:..., Springmann [/bib_ref] [bib_ref] Global Expanded Nutrient Supply (GENuS) Model: a new method for estimating the..., Smith [/bib_ref] [bib_ref] Global trends in dietary micronutrient supplies and estimated prevalence of inadequate intakes, Beal [/bib_ref] [bib_ref] Global, regional, and national life expectancy, all-cause mortality, and cause-specific mortality for..., Wang [/bib_ref] for the vegan scenario [fig_ref] Figure 2: Premature mortality and environmental impacts of diet scenarios in 2030 [/fig_ref]. Reductions in underweight, overweight, and obesity contributed similar proportions to reductions in F le x i t a r i a n V e g e t a r i a n P e s c a t a r i a n V e g a n a n i -2 5 a n i -5 0 a n i -7 5 a n i premature mortality (11% across the four diets), as did change in diet composition (9-12%). In the vegetarian and vegan scenarios, small increases in premature mortality (about 1%) from reduced fish intake were compensated (in part in the vegetarian scenario, and in full in the vegan scenario) by reductions in premature mortality from additional intake of legumes, fruits, and vegetables. Increases in nut consumption in the four scenarios, an aspect not included in the other dietary-change strategies, led to reductions in premature mortality (about 3%) in addition to those associated with replacing animal-source foods. Across disease endpoints, the averted premature deaths were about a quarter each from coronary heart disease (25-29%) and non-cardiovascular causes (25-30%), a fifth (19-21%) from cancer, followed by stroke (14-18%) and type 2 diabetes (8-10%; appendix p 30). The reductions in premature mortality were generally evenly distributed across regions (eg, 19-24% across income groups in the flexitarian scenario), with the greatest reductions in upper-middle-income countries, where diets and energy intake were most imbalanced [fig_ref] Figure 3: Regional changes in premature mortality and environmental impacts of dietary changeThe scenarios... [/fig_ref].
For all 12 of the dietary-change approaches, the changes in environmental impacts differed by enviro nmental domain and region. Progressively replac ing animal products with plant-based foods led to large reductions in greenhouse gas emissions (from 20% in the ani-25 scenario to 84% in the ani-100 scenario) because the demand for emissions-intensive animal-source foods was reduced [fig_ref] Figure 2: Premature mortality and environmental impacts of diet scenarios in 2030 [/fig_ref]. However, fresh water use was increased (from 4% in the ani-25 scenario to 16% in the ani-100 scenario) because the demand for waterdemanding crops, such as legumes, vegetables, and fruits was increased (figure 2B; appendix pp 31-32). Regional analyses showed further differences. In upper-middleincome and high-income countries, replacement of animal-source foods led to reduced cropland use (12% in upper-middle-income countries and 29% in high-income countries), nitrogen application (22% and 38%), and phosphorus application (25% and 35%), in line with reductions in the demand for livestock-related intensive feed production and fertilisation (figure 3). By contrast, these impacts increased in low-income and lower-middleincome countries (cropland 1% in lower-middle-income countries and 15% in low-income countries; nitrogen application 7% and 1%; and phosphorus application 7% and 3%), which use less intensive feeds and fertilisers and had generally lower yields, so that the impacts of increased demand for legumes and vegetables outweighed feedrelated re ductions [fig_ref] Figure 3: Regional changes in premature mortality and environmental impacts of dietary changeThe scenarios... [/fig_ref].
Improving the energy balance of diets led to moderate reductions across environmental impacts globally in the kcal-100 scenario; [fig_ref] Figure 2: Premature mortality and environmental impacts of diet scenarios in 2030 [/fig_ref] ; appendix pp 31-32). Environmental impacts were reduced in high-income and middle-income countries (8-18% in the kcal-100 scenario), where levels of overweight and obesity required a reduction in energy intake. By contrast, impacts were increased (3-8% in the kcal-100 scenario) for greenhouse gas emissions, cropland use, and freshwater use in lowincome countries, where levels of underweight required an increase in energy intake [fig_ref] Figure 3: Regional changes in premature mortality and environmental impacts of dietary changeThe scenarios... [/fig_ref].
Globally, moving to the balanced dietary patterns resulted in large reductions in greenhouse gas emissions (54-87% across the scenarios), medium-level reductions in nitrogen application (23-25%) and phosphorus application (18-21%), and small to moderate reductions in cropland (8-11%) and freshwater use (2-11%; [bib_ref] Systematic review of greenhouse gas emissions for different fresh food categories, Clune [/bib_ref]. Greenhouse gas emissions and nitrogen application were reduced in all regions, but the changes in cropland use, freshwater use, and phosphorus application were split into reductions in high-income and middle-income countries and increases in low-income countries (figure 3; appendix p 31), in line with regional differences in yields, water use, and fertilisation intensity.
Several scenarios showed alignment between the health and environmental benefits of dietary change, but large differences existed between regions and dietarychange approaches [fig_ref] Figure 3: Regional changes in premature mortality and environmental impacts of dietary changeThe scenarios... [/fig_ref]. For the scenarios reducing animal-source foods (ani scenarios), reductions in premature mortality were positively associated with reductions in greenhouse gas emissions and negatively associated with freshwater use. Other environmental domains had mostly positive relationships with mortality in high-income and upper-middle-income countries, and negative relation ships in low-income and lower-middleincome countries. For the scenarios aimed at improving energy balance (kcal scenarios), the changes in environmental impacts were positively associated with changes in premature mortality in high-income and middle-income countries, but they were negatively associated in low-income countries [fig_ref] Figure 4: Coefficients of association between health and environmental impactsCoefficients were calculated by dividing... [/fig_ref]. The balanced dietary-pattern scenarios (flexitarian, pescatarian, vegetarian, and vegan) showed the greatest alignment of health and environ mental impacts, with positive association for almost all environmental domains in high-income and middle-income countries, but negative associations for cropland use, freshwater use, and phosphorus application in low-income countries.
In a sensitivity analysis, we analysed the influence the time period has on the health and environmental impacts of dietary change. Analysing dietary changes in 2050 instead of 2030 resulted in greater alignment of the health and environmental impacts as technologies and manage ment practices improved, particularly for fresh water and cropland use (appendix p 33). Conversely, analysing dietary changes in 2010 resulted in greater relative increases in cropland and freshwater use than in 2030. Considering data of the emissions of carbon dioxide (based on a global meta-analysis of lifecycle analyses 31 ) instead of the data limited to methane and nitrous oxide emissions as in the main analysis, increased greenhouse gas emissions, particularly from fishing, and as a result widened the difference between the pescatarian and vegetarian scenarios (appendix p 34).
# Discussion
The concept of sustainable diets combines health and environmental concerns. Although many candidates for sustainable diets have emerged, no consistent and joint environmental and health analysis of these diets has been done at a regionally comparative level. Here, we examined three stylised approaches to sustainable diets using an integrated health and environmental modelling framework for more than 150 countries. Our results show that a public health approach focused on dietary
[formula] ani-25 5·19 -0·11 -1·05 0·15 0·17 4·18 1·40 -1·57 1·82 1·67 4·66 0·59 -1·43 1·09 1·24 4·98 0·08 -0·69 -0·44 -0·46 8·26 -6·63 -4·02 -1·76 -2·12 6·15 0·17 -1·20 0·22 0·24 5·13 1·72 -1·93 2·23 2·05 5·46 0·69 -1·68 1·28 1·45 5·68 0·09 -0·78 -0·50 -0·53 11·70 -4·12 -4·33 -0·78 -1·12 6·83 0·30 -1·32 0·26 0·29 5·83 1·95 -2·19 2·54 2·33 6·22 0·79 -1·91 1·46 1·65 6·13 0·10 -0·85 -0·54 -0·57 14·16 -3·30 -4·80 -0·39 -0·76 -0·54 0·89 0·37 0·85 1·33 1·33 0·78 0·76 0·83 0·75 0·67 1·25 1·17 1·22 1·20 1·34 1·98 1·92 1·24 2·19 2·14 -1·98 -3·46 -0·51 -0·83 -0·91 1·08 0·81 0·88 1·40 1·40 0·76 0·75 0·82 0·73 0·66 1·15 1·07 1·12 1·11 1·24 1·98 1·92 1·24 2·19 2·14 -1·04 -1·64 -0·32 -0·17 -0·21 1·13 0·95 0·88 1·41 1·41 0·77 0·76 0·83 0·74 0·67 1·21 1·13 1·18 1·16 1·30 1·86 1·80 1·16 2·06 2·01 -0·72 -1·00 -0·25 0·07 0·05 1·14 1·00 0·87 1·40 1·40 0·74 0·72 0·79 0·71 0·64 1·21 1·13 1·18 1·17 1·30 1·80 1·75 1·13 2·00 1·95 -0·54 -0·65 -0·21 0·19 0·18 2·85 0·45 0·57 1·22 0·96 3·51 1·30 0·58 1·70 1·57 3·01 0·87 0·70 1·22 1·04 2·59 0·32 0·87 1·23 0·93 2·03 -0·58 -1·42 0·24 -0·49 Pescatarian 3·71 0·55 0·50 1·18 0·92 3·69 1·45 0·51 1·85 1·68 3·31 1·01 0·60 1·33 1·13 3·75 0·37 0·78 1·08 0·80 3·59 -0·45 -1·49 0·31 -0·43 3·82 0·49 0·40 1·27 1·02 3·84 1·45 0·35 1·93 1·74 3·44 0·95 0·46 1·39 1·20 3·87 0·39 0·72 1·21 0·94 3·59 -0·69 -1·73 0·30 -0·41 3·91 0·48 0·08 1·14 0·96 3·84 1·57 -0·01 1·97 1·85 3·46 0·91 0·09 1·36 1·27 3·89 0·33 0·38 0·97 0·77 4·30 -0·69 -2·18 0·37 -0·33 [/formula]
Upper-middle-income countries
Lower-middle-income countries
## Low-income countries
## Region d omain d iet scenario
High-income countries Global changes towards predominantly plant-based diets that are in line with evidence on healthy eating performs better in reducing environmental pressures, potential nutrient deficiencies, and diet-related mortality than approaches motivated only by environmental and foodsecurity concerns. Our analysis has several strengths. It significantly improves on the methods used in existing global assessments by doubling the number of risk factors and environmental indicators covered. [bib_ref] Analysis and valuation of the health and climate change cobenefits of dietary..., Springmann [/bib_ref] [bib_ref] Global diets link environmental sustainability and human health, Tilman [/bib_ref] It improves the detail of dietary scenarios by including, in the balanced dietarypattern scenarios, recommended ranges for all major food groups based on the available evidence on healthy eating as reviewed by the EAT-Lancet Commission on Healthy Diets from Sustainable Food Systems. And it covers all major countries and regions in a comparative fashion, which allows for the identification of regional differences that extend the evidence base beyond national case studies. [bib_ref] The impacts of dietary change on greenhouse gas emissions, land use, water..., Aleksandrowicz [/bib_ref] [bib_ref] Environmental impact of dietary change: a systematic review, Hallström [/bib_ref] [bib_ref] The impact of nutritional choices on global warming and policy implications: examining..., Joyce [/bib_ref] Where comparisons are possible, our results are consistent with empirical data and other estimates with similar coverage. The nutritional estimates are based on existing datasets, and the assessment of nutritional levels and potential deficiencies is in line with existing analyses for most nutrients, but can differ for calcium for which we used global recommended intake values instead of the sometimes higher national values,and for zinc, for which we assumed medium bioavailability rather than using a function of absorption that is not yet validated. [bib_ref] Global trends in dietary micronutrient supplies and estimated prevalence of inadequate intakes, Beal [/bib_ref] Our mortality estimates from changes in dietary risk factors are similar to those of comparable dietary patterns in cohort studies. [bib_ref] Beyond meatless, the health effects of vegan diets: findings from the Adventist..., Le [/bib_ref] [bib_ref] Vegetarian dietary patterns and mortality in Adventist Health Study 2, Orlich [/bib_ref] However, by analysing the disease burden of dietary patterns in equilibrium, we abstracted from many real-world complexities, such as time lags between adoption of diets and changes in mortality. Although most of the relative risk factors used have been adjusted for major confounding factors, such as smoking, bodyweight, and other dietary risks, residual confounding with other parameters cannot be ruled out completely. In line with our joint focus on the health and environmental impacts of dietary change, we focused on those risk factors that we could include on the basis of foodavailability data, and we did not include risk factors, such as processed meat and whole grains, that would have required processing factors on the basis of data derived by a different method. [bib_ref] Global, regional and national consumption of major food groups in 1990 and..., Micha [/bib_ref] Including those risk factors as part of recommendations to reduce processed meat consumption and increase whole grain con sumption would further increase the estimated benefits of dietary change in the dietary-pattern and substitution scenarios (appendix pp [bib_ref] Environmental impact of dietary change: a systematic review, Hallström [/bib_ref] [bib_ref] The impact of nutritional choices on global warming and policy implications: examining..., Joyce [/bib_ref] [bib_ref] Do low-carbon-emission diets lead to higher nutritional quality and positive health outcomes?..., Payne [/bib_ref] [bib_ref] Global diets link environmental sustainability and human health, Tilman [/bib_ref] [bib_ref] Current evidence on healthy eating, Willett [/bib_ref] [bib_ref] Global and regional health effects of future food production under climate change:..., Springmann [/bib_ref]. The environmental assessment is based on data that reflects current environmental impacts at the country level (appendix pp [bib_ref] Global trends in dietary micronutrient supplies and estimated prevalence of inadequate intakes, Beal [/bib_ref] , and the scenario changes are in line with other assessments with similar detail. [bib_ref] Options for keeping the food system within environmental limits, Springmann [/bib_ref] Our results highlight the regional differences in the health and environmental impacts of dietary-change strategies. Following an environmental strategy by substituting animal-source foods can be particularly effective in high-income countries for improving nutrient levels, lowering premature mortality, and reducing some environ mental impacts, in particular greenhouse gas emissions. However, it can also lead to increased freshwater use, and has little effectiveness in countries with low or moderate consumption of animal-source foods. Following a food-security strategy by improving the energy balance of diets can lead to similar reductions in premature mortality, but in our model scenarios it only moderately improved nutrient levels and led to small reductions in environ mental impacts at the global level, with reduced impacts in high-income and middle-income countries, and increased resource use in low-income countries. Following a public health strategy by adopting energy-balanced, low-meat dietary patterns that are in line with available evidence on healthy eating addressed the problems of high regional variability of the animalsubstitution scenarios and the low nutritional and environmental effectiveness of the weight scenarios. Adopting the balanced and predominantly plant-based dietary patterns led to an adequate nutrient supply, except for a small number of nutrients (riboflavin, calcium, and vitamin B12), which might have to be supplemented, large reductions in premature mortality, and significant reductions in environmental impacts globally and in most regions, except for some environ mental domains (cropland use, freshwater use, and phosphorus application) in low-income countries.
Our analysis has several implications for the study of sustainable diets. First, qualitative differences exist between the health and environmental benefits that can be attained by dietary changes. Our analysis suggests that although a comprehensive and context-specific public health strategy for dietary change can lead to healthier diets in all regions, differences between the environmental impacts and regions are large. Dietary changes towards healthy, low-meat diets can be effective in reducing greenhouse gas emissions, moderately effective in reducing cropland use and fertiliser application in highincome and middle-income countries, but less effective for reducing freshwater use, particularly in low-income countries. Changes in low-income countries depend more strongly on technological improvements and changes in manage ment, [bib_ref] Options for keeping the food system within environmental limits, Springmann [/bib_ref] suggesting that a synergistic perspect ive on sustainable diets should include both technological and dietary aspects. Although reducing greenhouse gas emissions is important at the global level for mitigating climate change, changes in the other domains relate to predominantly local environ mental impacts. This highlights the need for context-specific strategies that balance environmental impacts between global and regional scales.
Second, addressing dietary composition and energy intake as part of food-based dietary guidelines could be a comprehensive strategy for achieving sustainable diets.
Whether food-based dietary guidelines should include sustainability criteria has been a major issue of discussion in several countries. [bib_ref] alignment of healthy dietary patterns and environmental sustainability: a systematic review, Nelson [/bib_ref] In this study, we find that when food-based dietary recommendations reflect available evidence on healthy eating, including balanced energy intake, low amounts of red meat and sugar, low to moderate amounts of other animal-source foods, and generous amounts of fruits, vegetables, legumes, and nuts, then the resulting diets would be in line with sustainability criteria of reducing environmental impacts in most regions, and they would still improve dietary health in the remaining ones. However, many national dietary guidelines do not reflect this evidence on healthy eating and include no or too lax limits for animal-source foods, particularly meat and dairy, [bib_ref] The impact of global dietary guidelines on climate change, Ritchie [/bib_ref] despite an opposing evidence base (appendix pp [bib_ref] Environmental impact of dietary change: a systematic review, Hallström [/bib_ref] [bib_ref] The impact of nutritional choices on global warming and policy implications: examining..., Joyce [/bib_ref] [bib_ref] Do low-carbon-emission diets lead to higher nutritional quality and positive health outcomes?..., Payne [/bib_ref] [bib_ref] Global diets link environmental sustainability and human health, Tilman [/bib_ref] [bib_ref] Current evidence on healthy eating, Willett [/bib_ref] [bib_ref] Global and regional health effects of future food production under climate change:..., Springmann [/bib_ref]. [bib_ref] Three daily servings of reduced-fat milk: an evidence-based recommendation?, Ludwig [/bib_ref] [bib_ref] Calcium intake and risk of fracture: systematic review, Bolland [/bib_ref] [bib_ref] Milk intake and risk of hip fracture in men and women: a..., Bischoff-Ferrari [/bib_ref] A general problem is that many nutrient recommendations used to formulate food-based dietary guidelines are based on a few shortterm studies with few participants that measure nutrient pass-through in high-consuming individuals instead of lower limits. [bib_ref] Dietary reference intakes: resuscitate or let die?, Bier [/bib_ref] By contrast, most of the available evidence on healthy eating comes from long-term and large-scale epidemiological cohort studies. [bib_ref] Current evidence on healthy eating, Willett [/bib_ref] Our results show that updating national dietary guidelines to reflect the latest evidence on healthy eating can by itself be important for improving health and environmental sustainability, and can complement broader and more explicit criteria of sustainability.
Some unanswered questions remain. We were not able to include all aspects of importance to sustainable diets, including biodiversity impacts and economic aspects.Although biodiversity impacts are related to land use in a given region, impacts can be expected to differ on the basis of a region's biodiversity richness and how land use is managed. With respect to economic aspects, large changes in food demand and supply can be expected to affect production methods, technologies, and commodity prices, which in turn would impact environmental footprints, agricultural incomes, the affordability of diets, and purchasing behaviour. Identifying concrete policy options that could support the dietary changes modelled here is similarly important. Globally, overweight and obesity, as well as consumption of red meat and dairy are projected to increase in the future without dedicated policy approaches. [bib_ref] National, regional, and global trends in adult overweight and obesity prevalences, Stevens [/bib_ref] Although informational campaigns and voluntary actions by industry can be important, the literature on behavioural change suggests that they are unlikely to be effective on their own. [bib_ref] Dietary and policy priorities for cardiovascular disease, diabetes, and obesity: a comprehensive..., Mozaffarian [/bib_ref] [bib_ref] Population approaches to improve diet, physical activity, and smoking habits a scientific..., Mozaffarian [/bib_ref] Instead, crosscutting regulatory approaches that focus on the whole food environment, [bib_ref] Health in all policies-the Finnish initiative: background, principles, and current issues, Puska [/bib_ref] combine multiple incentives including fiscal ones, [bib_ref] Beverage purchases from stores in Mexico under the excise tax on sugar..., Colchero [/bib_ref] [bib_ref] The effects of the Danish saturated fat tax on food and nutrient..., Smed [/bib_ref] and offer support and positive reenforcement for individuals [bib_ref] Extended and standard duration weight-loss programme referrals for adults in primary care..., Ahern [/bib_ref] [bib_ref] Behavioral counseling to promote a healthful diet and physical activity for cardiovascular..., Patnode [/bib_ref] have been successful in specific contexts, but would need to be upscaled to lead to substantial dietary changes at the population level. Finding effective combinations of policies and approaches that consider local characteristics will be essential for success fully upscaling initiatives and achieving reductions in the health and environmental burden at the population level and globally.
## Contributors
MS designed the study, compiled the models, conducted the analysis, interpreted the results, and wrote the manuscript. KW, DM-D, and TBS contributed model components for the environmental analysis. All authors commented on the manuscript draft and approved the final submission.
# Declaration of interests
We declare no competing interests.
## Data sharing
The country-level results generated for the study are available on the Oxford University Research Archive. Additional data available on request to MS.
[fig] Figure 1: Nutrient supply by diet scenario in 2010Red values are those that are lower than minimum recommendations or higher than maximum recommendations. [/fig]
[fig] Figure 2: Premature mortality and environmental impacts of diet scenarios in 2030 (A) Diamonds show reductions in premature mortality due to diet patterns and bars show proportion contributions of individual risk factors to the reduction. Total percentage contributions can exceed 100% because individual risks are attenuated when combined and can be compensated by opposing risk factors. [/fig]
[fig] Figure 3: Regional changes in premature mortality and environmental impacts of dietary changeThe scenarios include diets in which all animal-source foods have been replaced by plant-based ones (ani-100; A), diets with optimal energy intake and weight levels (kcal-100; B), and flexitarian diets that are energy balanced and contain small amounts of animal-source foods (flexitarian; [/fig]
[fig] Figure 4: Coefficients of association between health and environmental impactsCoefficients were calculated by dividing the percentage changes in environmental impacts by the percentage changes in premature mortality. Positive values (green) indicate that health and environmental changes are aligned (larger values show stronger positive associations), whereas negative values (red) indicate opposing changes (more-negative values show stronger negative associations). Darker shades show stronger associations, whereas lighter shades show weaker associations. [/fig]
|
A Case Report of Infective Endocarditis with Failure of the Empirical Treatment—Q Fever Endocarditis Diagnosed by Metagenomic Next-Generation Sequencing
# Introduction
Q fever is a naturally occurring epidemic disease caused by the Coxiella burnetii (C. burnetii) infection. The hosts of C. burnetii include mammals, birds, arthropods, etc. It is found in most countries around the world and is mainly transmitted through aerosols. [bib_ref] The sero epidemiology of Coxiella burnetii (Q fever) across livestock species and..., Larson [/bib_ref] The clinical manifestations of patients with Q fever who usually have contact with animals are diverse and Q fever can be divided into acute and chronic infections, with chronic infections being more common. Q fever endocarditis is the most common manifestation of chronic infection and usually occurs in patients with previous valvular heart disease, impaired immune function, and during pregnancy. [bib_ref] Surveillance for Q Fever Endocarditis in the United States, Straily [/bib_ref] However, Q fever endocarditis is difficult to diagnose clinically and may lead to very serious or even life-threatening outcomes if not diagnosed promptly. In the present study, a case of Q fever endocarditis that occurred based on valvular heart disease was reported. Accurate diagnosis and treatment were achieved by metagenomic next-generation sequencing (mNGS) combined with detection of the Q fever serological antibodies.
## Case report
A 43-year-old male patient who had suffered from chest tightness and breath holding with each episode of strenuous exercise since the age of 15 was admitted; his symptoms were relieved after rest. Transthoracic echocardiography (TTE) was conducted, and the results suggested malformation of the bicuspid aortic valve (BAV) with stenosis; no specific treatment was performed. In August 2021, the patient visited our hospital after feeling chest tightness and breath holding with slight activity. These symptoms were accompanied by a fever with the highest temperature being 38.5°C. There were no chills and chills, no cough and sputum. The results of a physical examination of his heart at admission were as follows: The heart rhythm was uniform, a soft grade II/6 systolic murmur could be heard in the second intercostal space on the left edge of the sternum, and there were no pericardial friction sounds. The results of laboratory examinations were as follows: The white blood cell count was 9.61 × 10^9/L, neutrophil count was 6.25 × 10^9/L, hemoglobin was 157 g/L, platelet count was 274* × 10^9/L, and ultrasensitive C-reactive protein was 27.5 mg/ L. The calcitoninogen was 0.06 ng/mL, hematocrit was 27 mm/h, and rheumatoid factor was 16 IU/mL. The hepatic and renal function, as well as the urine routine test, were normal. The results of TTE were as follows: BAV with stenosis, septal hypertrophy, and widening of the aortic ascending segment existed. Strong echogenicity was visible at the root of the aortic valve. The results of the transesophageal echocardiography (TEE) showed BAV with severe stenosis and strong echogenicity at the aortic valve root; the possibility of an inflammatory mass could not be excluded [fig_ref] Figure 1: The results of transesophageal echocardiography [/fig_ref]. There was no significant abnormality in the electrocardiogram (ECG), chest computed tomography (CT), abdominal ultrasound, and cranial magnetic resonance imaging (MRI). The diagnosis of infective endocarditis (IE), severe aortic stenosis, and dilatation of the ascending aorta was considered with the combination of the clinical manifestations, physical examination, and laboratory tests. Vancomycin (1 g, q12h) plus ceftriaxone (2 g, qd) were administered empirically as intravenous anti-infection therapy. The patient still had recurrent fever after more than 10 days of treatment, and the blood culture (the bilateral double vials) was negative four consecutive times. The presence of uncommon pathogens was considered highly likely, thus mNGS in the blood sample was conducted to assist the diagnosis. Two days after sending for the mNGS, the results suggested the existence of C. burnetii with a blood cell layer sequence number of 134 and a plasma layer sequence number of 1 [fig_ref] Figure 3: The results of metagenomic next-generation sequencing in blood [/fig_ref]. The diagnosis was considered to be Q fever endocarditis, and therefore, the Q fever serological antibody assay was conducted (positive for the Q fever Phase II IgG). Doxycycline (100 mg, every 12 hours) was then administered intravenously in combination with hydroxychloroquine (200 mg, three times a day) orally, and the temperature of the patient returned to normal after one week of treatment. In October 2021, the patient underwent a mechanical aortic valve replacement with an ascending aortic replacement under general anesthesia. The intraoperative observations were as follows: The ascending aorta was significantly thickened, with a diameter of approximately 5.0 cm, together with the thin aortic wall. BAV, severe calcification, and thickening of the valve leaflets existed, resulting in valve stenosis with an incomplete closure, [fig_ref] Figure 4: The results of metagenomic next-generation sequencing in cardiac tissues [/fig_ref]. The combination therapy of doxycycline with hydroxychloroquine was continued postoperatively, and the patient was followed up for more than one month without the occurrence of fever and without symptoms of chest This study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of Yantai Yuhuangding Hospital. The consent from the patient for the publication of the case was obtained.
# Discussion
Q fever is a globally spread zoonotic disease caused by the C. burnetii infection. The clinical manifestations of Q fever are diverse; while some patients have severe manifestations of acute or chronic infection, others have mild clinical signs and symptoms, often without, or only low, fever. 3 Q fever often causes endocarditis, however, this diagnosis is currently challenging and requires a combination of clinical, microbiological, and/or imaging findings. In the present study, a patient was diagnosed with Q fever endocarditis combined with valvular heart disease, which is a high risk for the development of Q fever. 2 A heart valve vegetation was revealed in the echocardiogram conducted after admission and the possibility of IE was considered. However, the empirical anti-infection therapy was ineffective, with four consecutive negative results in the blood cultures. Therefore, samples were sent for mNGS in the blood to clarify the pathogenic diagnosis. Unbiased detection of a wide range of pathogenic microorganisms can be achieved through mNGS (including viruses, bacteria, fungi, and parasites) through shotgun sequencing of DNA or RNA from clinical samples, regardless of the success culture of the clinical sample, as long as the samples contain detectable DNA or RNA.The results of the mNGS provided a direction for pathogenic diagnosis. Samples were then sent for the Q fever serological antibody assay, which showed the existence of a positive Q fever phase II antibody.
Although vegetation was found in the heart valves during the TTE and TEE in the present case, vegetation was rarely found in most patients with Q fever endocarditis using cardiac ultrasonography.This was because the vegetation in Q fever endocarditis, unlike those in bacterial endocarditis, manifested as endothelial-covered nodules on the valves, which were often small and difficult to be detected and were often missed in diagnosis. As with the diagnosis of endocarditis caused by other pathogens, the sensitivity of TEE in the diagnosis of Q fever endocarditis is higher than that of TTE. [bib_ref] Diagnosis and management of Q fever-United States, 2013: recommendations from CDC and..., Anderson [/bib_ref] Surgical replacement of the injured valve is often necessary for patients with Q fever endocarditis due to hemodynamic reasons. However, the histologic manifestations of the tissue obtained intraoperatively from patients with Q fever endocarditis are nonspecific with the inclusion of significant fibrosis and calcification, mild inflammation and angiogenesis, and the absence or presence of only small vegetation. These pathological features are easily confused with noninfected degenerative valve lesions. [bib_ref] Cardiac valves in patients with Q fever endocarditis: microbiological, molecular, and histologic..., Lepidi [/bib_ref] While the DNA of the C. burnetii could be detected in tissue by mNGS.
Combined therapy of hydroxychloroquine (hydroxychloroquine 600 mg orally, once a day or 200 mg, three times a day) and doxycycline (100 mg orally, twice a day) is preferred in patients with Q fever endocarditis for a minimum duration of 18 months. Minocycline may be applied in patients who cannot tolerate doxycycline (eg, the occurrence of nausea). Some case reports have found that the combination of doxycycline with rifampin, doxycycline, and quinolones is also effective. A retrospective study compared the therapeutic effects of hydroxychloroquine plus doxycycline with doxycycline plus a quinolone and showed that the former had better clinical efficacy. [bib_ref] Treatment of Q fever endocarditis: comparison of 2 regimens containing doxycycline and..., Raoult [/bib_ref] In the treatment of Q fever endocarditis, the dose of hydroxychloroquine is relatively high and is prone to result in ocular toxicity. The retina should be examined once before administration and every six months thereafter to determine the occurrence of early signs of toxicity associated with hydroxychloroquine, such as corneal deposition and retinopathy.
# Conclusion
In recent years, with the increase in the immune-compromised population, it is more difficult to clarify the etiology of IE by traditional methods, especially in patients with negative results in blood cultures, in whom Q fever endocarditis is common. The application of mNGS could provide a rapid and accurate assay in clinical diagnosis and play a decisive role in the pathogenetic diagnosis of some infectious diseases. 9
# Funding
This study was supported by effect of TLR2 on dectim-1 expression in human bronchial epithelial cells infected with Mycoplasma pneumoniae (No. 2017WS165).
## Infection and drug resistance
## Dovepress
## Publish your work in this journal
Infection and Drug Resistance is an international, peer-reviewed open-access journal that focuses on the optimal treatment of infection (bacterial, fungal and viral) and the development and institution of preventive strategies to minimize the development and spread of resistance. The journal is specifically concerned with the epidemiology of antibiotic resistance and the mechanisms of resistance development and diffusion in both hospitals and the community. The manuscript management system is completely online and includes a very quick and fair peer-review system, which is all easy to use. Visit http://www.dovepress.com/testimonials.php to read real quotes from published authors.
[fig] Figure 1: The results of transesophageal echocardiography. https://doi.org/10.2147/IDR.S361969 DovePress Infection and Drug Resistance 2022:15 and punctate vegetation was visible. The resected valve with lesions was sent for pathological examination and mNGS. The results of mNGS in the cardiac tissue revealed the existence of C. burnetii with the sequence number of 214,080 ( [/fig]
[fig] Figure 2: The results of transesophageal echocardiography. [/fig]
[fig] Figure 3: The results of metagenomic next-generation sequencing in blood. Infection and Drug Resistance 2022:15 https://doi.org/10.2147/IDR.S361969 DovePress 2547 tightness and breath holding. The TTE examination suggested good prosthetic valve function and smooth blood flow in the prosthetic vessels. [/fig]
[fig] Figure 4: The results of metagenomic next-generation sequencing in cardiac tissues. https://doi.org/10.2147/IDR.S361969 DovePress Infection and Drug Resistance 2022:15 [/fig]
|
DegNorm: normalization of generalized transcript degradation improves accuracy in RNA-seq analysis
RNA degradation affects RNA-seq quality when profiling transcriptional activities in cells. Here, we show that transcript degradation is both gene-and sample-specific and is a common and significant factor that may bias the results in RNA-seq analysis. Most existing global normalization approaches are ineffective to correct for degradation bias. We propose a novel pipeline named DegNorm to adjust the read counts for transcript degradation heterogeneity on a gene-by-gene basis while simultaneously controlling for the sequencing depth. The robust and effective performance of this method is demonstrated in an extensive set of simulated and real RNA-seq data.
Background RNA-seq is currently the most prevailing method for profiling transcriptional activities using high-throughput sequencing technology [bib_ref] RNA-Seq: a revolutionary tool for transcriptomics, Wang [/bib_ref]. The sequencing tag count per unit of transcript length is used to measure the relative abundance of the transcript [bib_ref] Transcript assembly and quantification by RNA-Seq reveals unannotated transcripts and isoform switching..., Trapnell [/bib_ref]. Various factors exist that may affect the faithful representation of transcript abundance by RNA-seq read counts. Normalization is a crucial step in post-experiment data processing to ensure a fair comparison of gene expression in RNA-seq analysis [bib_ref] Evaluation of statistical methods for normalization and differential expression in mRNA-Seq experiments, Bullard [/bib_ref] [bib_ref] A comprehensive evaluation of normalization methods for Illumina high-throughput RNA sequencing data..., Dillies [/bib_ref]. The most commonly used approach is to normalize the read counts globally by a sample-specific scale factor to adjust the sequencing depth. Choices of the scale factor include the total number of reads (or mean), median, trimmed mean of M values [bib_ref] A scaling normalization method for differential expression analysis of RNA-seq data, Robinson [/bib_ref] , and upper quartile [bib_ref] Evaluation of statistical methods for normalization and differential expression in mRNA-Seq experiments, Bullard [/bib_ref]. The second type of normalization aims to remove the read count bias due to physical or chemical features of RNA sequences or uncontrollable technical aspects. The GC content is known to affect the read counts in a nonlinear way [bib_ref] Substantial biases in ultrashort read data sets from high-throughput DNA sequencing, Dohm [/bib_ref] [bib_ref] Modeling non-uniformity in short-read rates in RNA-Seq data, Li [/bib_ref] , and this effect can be sample specific under different culture or library preparation protocols [bib_ref] GC-content normalization for RNA-Seq data, Risso [/bib_ref]. Systematic bias may also arise due to technical effects such as library preparation and sequencing batches. Such systematic biases can be quantified and removed using factor analysis provided that the unwanted variation is uncorrelated with the covariates of interest [bib_ref] Capturing heterogeneity in gene expression studies by surrogate variable analysis, Leek [/bib_ref] [bib_ref] Normalization of RNA-seq data using factor analysis of control genes or samples, Risso [/bib_ref].
Another type of bias arises from cDNA fragmentation and mRNA degradation. The RNA-seq assay requires fragmenting the cDNA (reversely transcribed from mRNA) or mRNA for high-throughput sequencing. Ideally, for a complete, non-degraded transcript, if the fragmentation is completely random, we expect to see reads uniformly distributed along the transcript. Nevertheless, the fragmentation by random priming is not truly random due to primer specificity [bib_ref] Biases in Illumina transcriptome sequencing caused by random hexamer priming, Hansen [/bib_ref] [bib_ref] Improving RNA-Seq expression estimates by correcting for fragment bias, Roberts [/bib_ref] [bib_ref] Measure transcript integrity using RNA-seq data, Wang [/bib_ref]. Consequently, read count per unit length of a transcript may not strictly reflect the transcript abundance when comparing the expression of different genes. For the same gene, assuming the same protocol is applied to different samples, the bias attributable to fragmentation across samples should be similar. Thus, fragmentation bias is less problematic in a gene-by-gene differential expression (DE) analysis. In contrast, mRNA degradation can vary substantially in both extent and pattern between genes and between samples [bib_ref] Precision and functional specificity in mRNA decay, Wang [/bib_ref] [bib_ref] Decay rates of human mRNAs: correlation with functional characteristics and sequence attributes, Yang [/bib_ref]. The mRNA degradation has different pathways and can happen in any region of a transcript [bib_ref] The many pathways of RNA degradation, Houseley [/bib_ref]. Perfect control of sample degradation during the experiment is difficult, particularly when the samples are collected from field studies or clinical samples. More importantly, different genes may degrade at different rates [bib_ref] RNA-seq: impact of RNA degradation on transcript quantification, Romero [/bib_ref] , which makes it impossible to remove this bias by normalizing the read counts of all genes in the same sample by the same constant.
While the major impact of RNA degradation on gene expression analysis has been well recognized [bib_ref] RNA-seq: impact of RNA degradation on transcript quantification, Romero [/bib_ref] [bib_ref] Impact of RNA degradation on gene expression profiles: assessment of different methods..., Copois [/bib_ref] , methods for correcting the degradation bias have not been fully explored in the literature. A few methods have been proposed to quantify the RNA integrity including RNA integrity numbers (RIN) [bib_ref] The RIN: an RNA integrity number for assigning integrity values to RNA..., Schroeder [/bib_ref] , mRIN [bib_ref] mRIN for direct assessment of genome-wide and gene-specific mRNA integrity from large-scale..., Feng [/bib_ref] , and transcript integrity number (TIN) [bib_ref] Measure transcript integrity using RNA-seq data, Wang [/bib_ref]. The RIN gives a sample-specific overall RNA quality measure, but not at the gene level. In practice, a sample with RIN ≥ 7 (on a scale of 0 to 10) is often regarded as having good quality. The mRIN and TIN measures were both defined in the gene level by comparing the sample read distribution with reference to the hypothetical uniform distribution. In real data, due to GC content bias, primer specificity, and other complexities, the read count may substantially deviate from the uniform distribution along the transcript [bib_ref] Modeling non-uniformity in short-read rates in RNA-Seq data, Li [/bib_ref].
To reduce the degradation effect, Finotello et al. proposed to quantify the exon-level expression by the maximum of its per-base counts instead of the raw read counts [bib_ref] Reducing bias in RNA sequencing data: a novel approach to compute counts, Finotello [/bib_ref]. If a given exon is in a more degraded region, the local maximum may still be an underestimate of true abundance. On the other hand, the larger variance associated with the local maximum (e.g., spikes) may result in instability in DE analysis. Based on the TIN measure, Wang et al. proposed a degradation normalization method based on loess regression of read counts on the TIN measure for genes within the same sample [bib_ref] Measure transcript integrity using RNA-seq data, Wang [/bib_ref]. However, the uniform baseline assumption and the failure to compare gene-specific degradation across samples appear to be the two major limitations, which may lead to extreme variability and bias in DE analysis (to be shown below). Jaffe et al. proposed a quality surrogate variable analysis (qSVA) to remove the confounding effect of RNA quality in DE analysis [bib_ref] qSVA framework for RNA quality correction in differential expression analysis, Jaffe [/bib_ref]. They investigated the degradation of RNA-seq data from dorsolateral prefrontal cortex (DLPFC) tissue under two different RNA-seq protocols, namely, poly(A)+ (mRNA-seq) vs. ribosomal depletion (Ribo-Zero-seq). Thousands of features significantly associated with degradation were identified under either protocol separately, while no overlap was found between the two protocols. Furthermore, comparing the DLPFC samples and the peripheral blood mononuclear cell (PBMC) samples [bib_ref] RNA-seq: impact of RNA degradation on transcript quantification, Romero [/bib_ref] both sequenced under the same poly(A)+ protocol, they found only four shared features. It is unclear how the sequence features identified in this study can be generalized for degradation bias correction in other RNA-seq data in practice.
Alternative splicing is frequently observed in higher organisms, and it further complicates the gene expression estimation in RNA-seq [bib_ref] Deep surveying of alternative splicing complexity in the human transcriptome by high-throughput..., Pan [/bib_ref]. In the gene-level DE analysis, we test the equivalence of relative abundance of transcripts in copy numbers between samples or conditions. If the two samples have differential exon usage, read counts need to be adjusted accordingly to better represent the transcript relative abundance in the respective samples. Currently, most existing statistical packages for RNA-seq analysis (e.g., DESeq [bib_ref] Differential expression analysis for sequence count data, Anders [/bib_ref] and edgeR [bib_ref] edgeR: a Bioconductor package for differential expression analysis of digital gene expression..., Robinson [/bib_ref] all take the raw read counts as input, while such complexities are completely ignored in practice.
In this paper, we propose a novel data-driven method to quantify the transcript degradation in a generalized sense for each gene within each sample. Using the estimated degradation index scores, we build a normalization pipeline named DegNorm to correct for degradation bias on a gene-by-gene basis while simultaneously controlling the sequencing depth. The performance of the proposed pipeline is investigated using simulated data, and an extensive set of real data that came from both cell line and clinical samples sequenced in poly(A)+ or Ribo-Zero protocol.
# Results
## Data sets
We consider six RNA-seq data sets that were generated from cell lines or clinical samples under either the mainstream poly(A) enrichment (mRNA-seq) or ribosomal RNA depletion protocol (Ribo-Zero-seq). The first one was from a brain glioblastoma (GBM) cell line study of a human for the impact of RNA degradation on gene expression analysis [bib_ref] Sequencing degraded RNA addressed by 3′ tag counting, Sigurgeirsson [/bib_ref]. Technical replicates of RNA samples were fragmented under different incubation time and temperature using the NEBNext Magnesium RNA fragmentation module. We chose to analyze nine mRNA-seq samples in three groups of three, corresponding to three average RNA integrity number (RIN) = 10, 6, and 4, respectively (to be referred to as R10, R6, and R4 for simplicity). We will perform DE analysis for R10 vs. R4 and R6 vs. R4.
The second set contained 32 single-end mRNA-seq samples from human peripheral blood mononuclear cells (PBMC) of 4 different subjects: S00, S01, S02, and S03 [bib_ref] RNA-seq: impact of RNA degradation on transcript quantification, Romero [/bib_ref]. The extracted RNA sample from each subject was kept in room temperature for 0, 12, 24, 36, 48, 60, 72, and 84 h, respectively, to approximate the natural degradation process. We choose S01 as an illustrating example and will perform DE analysis for 0 + 12 h vs. 24 + 48 h (results for other subjects are similar).
The third set was from Sequencing Quality Control (SEQC) Consortium [bib_ref] A comprehensive assessment of RNA-seq accuracy, reproducibility and information content by the..., Consortium [/bib_ref] [bib_ref] Comprehensive evaluation of differential gene expression analysis methods for RNA-seq data, Rapaport [/bib_ref] and contained two subsets of mRNA-seq data, namely SEQC-AA and SEQC-AB. The SEQC-AA subset consisted of 16 technical replicates from Stratagene's universal human reference (UHR) RNA library with two runs for eight lanes each. We will run DE analysis of the first run vs. the second run. The second subset contained two biological conditions: condition A of five samples from the same Stratagene's UHR RNA, and condition B of five samples from Ambion's human brain reference RNA. The first four replicates from both conditions were prepared by the same technician while the fifth was by Illumina. We excluded the fifth sample from both conditions because they showed a dramatic difference in coverage curves compared to the rest.
The fourth data set contained RNA-seq data from dorsolateral prefrontal cortex (DLPFC) tissue of five brainsthree controls and two schizophrenia cases [bib_ref] qSVA framework for RNA quality correction in differential expression analysis, Jaffe [/bib_ref]. Each tissue was left in room temperature (off of ice) for 0, 15, 30, and 60 min for degradation. The RNA sample was extracted and prepared for both mRNA-seq and Ribo-Zero-seq. We chose to analyze one schizophrenia case (Br1729, results for other subjects are similar). We will perform DE analysis T0+T15 min vs. T30+T60 min under the same protocol, i.e., mRNA-seq or Ribo-Zero-seq, and then cross-platform DE analysis between the two protocols.
The fifth data set originated from three pairs of matched fresh-frozen (FF) and formalin-fixed paraffinembedded (FFPE) tissues of three breast tumor patients (namely T1, T2, T3) with a moderate archival time of about 4-5 years [bib_ref] mRNA-seq whole transcriptome profiling of fresh frozen versus archived fixed tissues, Ben-Moshe [/bib_ref]. The FFPE samples are typically partially degraded. We will analyze the mRNA-seq data of FF (500 ng) and FFPE (100 ng) to investigate whether the degradation normalization can help improve the DE analysis in fragmented clinical RNA samples.
The last data set arose from a clinical study on how AMP kinase (AMPK) promotes glioblastoma bioenergetics and tumor growth [bib_ref] AMP kinase promotes glioblastoma bioenergetics and tumour growth, Chhipa [/bib_ref]. It was shown that cancer cells can activate AMPK and highjack the stress-regulating pathway in cells. Thus, inhibiting AMPK in cancer cells may lead to treatment of GBM. RNA-seq data was collected from two patient-derived GBM stem cell (GSC) lines (GBM9 and GBM10) between control and AMPK knockout to identify differentially expressed genes. We will perform DE analysis between the three control and three knockout samples in GBM10 cell line. Unlike the first five sets where the ground truth of gene expression was known or mostly verified, the AMPK knockout data represents a typical case in clinical studies where only a handful genes of interest, most often suspected as differentially expressed, were PCR verified. For this reason, in the following analysis, we will compare the different normalization methods by benchmarking our analyses using the first five sets and then present the AMPK data as a case study in the last.
## Non-uniformity and heterogeneity in read distribution pattern
We define a total transcript as the concatenation of all annotated exons from the same gene. The read coverage score at a given location within the transcript is defined as the total number of reads (single-end) or DNA fragments (paired-end) that cover this position (Additional file 1). If mRNA transcripts are complete and the fragmentation is random, we expect to see a flat coverage curve in the entire transcript except in the head and tail region . Nevertheless, in real data, the read coverage curves rarely display a uniform pattern; instead, dramatic and gene-specific differences are often observed across samples . The non-uniformity itself is less concerning as long as the coverage pattern is consistent across samples such that the read counts can still faithfully represent the relative abundance of transcripts. In contrast, heterogeneous coverage patterns are often observed where some samples show significantly decayed read counts in some regions . One major cause of this heterogeneity is mRNA degradation, which is clearly shown in the case of ACTN4 gene of R4 samples from the GBM data . Different sample preparation methods may lead to distinct read distributions. For example, FFPE samples may show a highly localized discrete read distribution pattern in contrast to a continuous distribution typically observed in FF samples . Alternative splicing may also result in depleted read count in the entire region of an exon, as exemplified in the ST7 gene of B samples (region 1900-2900 bp) from the SEQC-AB data . For the gene-level DE analysis, loss of read count due to such complexities needs to be compensated to ensure unbiased quantification of gene expression.
## Generalized degradation and degradation normalization algorithm
The degradation we target to normalize is defined in a generalized sense. Any systematic decay of read count in any region of a transcript in one or more samples compared to the rest in the same study is regarded as degradation. Clearly, mRNA degradation is one main cause, but alternative splicing and other factors may be the confounders that are difficult to deconvolute. To avoid confusion, in the following context, we will reserve the term "mRNA degradation" for the physical degradation of mRNA sequences, and "degradation" or "transcript degradation" for the generalized degradation without specification.
We propose DegNorm, a degradation normalization pipeline based on non-negative matrix factorization over-approximation (NMF-OA, see the "Methods" section and Additional file 1). We assume there is a gene-specific ideal shape of coverage curve, called an "envelope" function, identical across the samples in the given study. Each envelope function is scaled by a sample-and gene-specific abundance parameter to represent the expected coverage curve for the given gene within each sample if no degradation occurs. Degradation may occur in any region of the transcript to cause negative bias in the observed read counts. To illustrate this, we generated four expected coverage curves of identical shape but with different abundance levels , among which samples S1 and S2 are subject to degradation in the 5′ end with different patterns. Based on the expected curves, we further simulated a random realization of four complete curves with sampling error imposed , to be referred to as latent curves) and two degraded for sample S1 and S2, respectively . The NMF-OA algorithm takes the four observed coverage curves (i.e., two non-degraded (S3 and S4) and two degraded (S1 and S2)) as input and estimates the latent curves by minimizing the squared distance between the observed and latent, subject to the constraint that the latent curves must dominate their respective observed curves at all positions (Figs. 2d, e; the "Methods" section). We define the degradation index (DI) score for each gene within each sample, as the fraction of area covered by the estimated latent curve, but above the observed curve . It measures the proportion of missing read count due to degradation given the current sequencing depth.
The DegNorm pipeline iteratively corrects for degradation bias while simultaneously normalizing sequencing depth. First, the NMF-OA algorithm is applied to sequencing depth-normalized read counts for all genes one by one to estimate the DI scores. Second, the resulting DI scores are then used to adjust the read counts by extrapolation for each gene. The adjusted read counts are used to normalize the raw data for sequencing depth. These two steps are repeated until the algorithm converges (the "Methods" section).
## Di score as sample quality diagnostics
The estimated DI scores provide an overview of the within-sample, between-sample, and between-condition variation of degradation extent and patterns. We plotted the DI scores in three ways: a box plot of DI scores for each sample [fig_ref] Figure 3: Degradation index [/fig_ref] , a heatmap of the DI scores a b d c e RNA-seq read coverage score shows between-sample heterogeneity in the pattern along transcripts. a The read coverage score, defined as the number of reads that cover each base pair, is expected to have a trapezoidal shape along the transcript if the read start position is uniformly distributed. b An example from SEQC-AB data shows a non-uniform but consistent read coverage pattern, where the average magnitude of coverage score for each sample may faithfully represent the transcript abundance given the sequencing depth is normalized. The diagram in red under the coverage plot shows the total transcript with exon boundaries from genome annotations (same for c, d and e below). c The TMEM229B gene from the breast tumor data shows differential coverage score patterns between FF and FFPE samples. Reads from FF samples are continuously distributed across the entire transcript while those from FFPE samples are highly enriched in a few disjoint blocks or fragments. d The ACTN4 gene from the GBM data with RIN number = 10 vs. RIN = 4 shows clear degraded coverage score towards the 5′ end of the transcripts in the latter group. e An example from the SEQC-AB data shows that alternative splicing likely causes sharply decayed coverage score across the entire alternatively spliced exon region sorted in ascending order of the average scores of the first condition defined in the DE analysis [fig_ref] Figure 3: Degradation index [/fig_ref] , and a pairwise correlation matrix of DI scores between samples (Additional file 2: -f ). For SEQC-AA data with 16 technical replicates, the median of DI scores is~0.35, consistent across samples [fig_ref] Figure 3: Degradation index [/fig_ref]. While the degradation pattern varies between genes, no systematic between-condition difference is observed [fig_ref] Figure 3: Degradation index [/fig_ref] , Additional file 2: . In contrast, for the SEQC-AB data, the 4 samples from condition B have relatively lower and more homogeneous degradation than that from A samples [fig_ref] Figure 3: Degradation index [/fig_ref]. Many genes show a condition-specific clustered pattern in DI scores [fig_ref] Figure 3: Degradation index [/fig_ref] , resulting in a high within-condition correlation (Additional file 2: .
The PBMC and GBM data are known to have differential mRNA degradation. The DI scores of PBMC S01 data confirm a progressive deterioration of average degradation when samples underwent degradation in room temperature for 0, 12, 24, and 48 h, respectively [fig_ref] Figure 3: Degradation index [/fig_ref] , i, Additional file 2: . The degradation from 24 to 48 h was particularly accelerated compared to the first 24 h. The nine GBM samples were previously classified into three groups according to the RNA integrity number (RIN), R = 10, 6, and 4, respectively. The DI scores show a clear escalating pattern of degradation severity across the three groups [fig_ref] Figure 3: Degradation index [/fig_ref] , j) with two strong clusters, i.e., R10 vs. R6+R4 (Additional file 2: . The scatter plots of DI scores further exemplify a higher correlation between samples within the same RIN group than across different RIN groups (Additional file 2: g-i).
For the DLPFC Br1729 Ribo-Zero-seq data, DegNorm recovered an increasing pattern of degradation from time 0 to 60 min as expected [fig_ref] Figure 3: Degradation index [/fig_ref] , k, Additional file 2: . The three pairs of breast tumor samples were prepared in two different ways-fresh frozen (FF) and formalin-fixed paraffin-embedded (FFPE)-but both sequenced under the mRNA-seq protocol. The DI scores confirm that the mRNA transcripts in FFPE samples tend to be highly degraded compared to the paired FF samples [fig_ref] Figure 3: Degradation index [/fig_ref] , and degradation patterns are strongly clustered within the same FF or FFPE group (Additional file 2: .
In summary, the DI scores from the DegNorm provide meaningful quantification of gene-level degradation between samples for both cell line and clinical samples under both mRNA-seq and Ribo-Zero-seq protocols. observed normalized S1 expected S1 degraded S2 expected S2 degraded S3 expected S4 expected A proof-of-concept example of the proposed method for normalizing degradation patterns. a Expected or theoretical read coverage curves of one gene from four samples of identical shape (solid lines, without degradation), two of which (S1 and S2) are subject to degradation according to a rate indicated by the dashed lines in the 5′ end. b A realization of the four coverage curves without degradation randomly simulated according to the expected curves in a. These curves are regarded as the latent and unobserved data. c Observed coverage curves after imposing random degradation to S1 and S2 (S3 and S4 stay intact). d Estimates of the non-degraded latent curves from the proposed algorithm solely based on the observed coverage curves in c. e A sample-by-sample comparison between the observed and estimated latent coverage curves
The degradation pattern is gene-specific, and the degradation extent may vary substantially between samples or conditions.
## Degnorm improves accuracy in gene expression analysis
We set out to evaluate how the proposed DegNorm pipeline may improve differential expression analysis by comparing it with other seven normalization methods including UQ [bib_ref] Evaluation of statistical methods for normalization and differential expression in mRNA-Seq experiments, Bullard [/bib_ref] , TIN [bib_ref] Measure transcript integrity using RNA-seq data, Wang [/bib_ref] , RUVr, RUVg [bib_ref] Normalization of RNA-seq data using factor analysis of control genes or samples, Risso [/bib_ref] , trimmed mean of M values (TMM) [bib_ref] A scaling normalization method for differential expression analysis of RNA-seq data, Robinson [/bib_ref] , relative log expression (RLE) [bib_ref] Differential expression analysis for sequence count data, Anders [/bib_ref] , and total read count (TC) [bib_ref] A comprehensive evaluation of normalization methods for Illumina high-throughput RNA sequencing data..., Dillies [/bib_ref]. The RUV methods were designed to remove unwanted variation, but it is unclear whether it is effective for correcting degradation bias. We dropped the RUVg method from the main text for its performance can be very sensitive to the choice of empirical control genes (Additional file 2: -f ) or the choice of factor(s) from the factor analysis in the estimation of unwanted variation (Additional file 2: We first examine the five data sets that originated from the samples that had no true biological difference between conditions under test (i.e., SEQC-AA, PBMC-S01, GBM, DLPFC, and breast tumor). RNA degradation induces bias and thus may cause extra variance. Severe degradation may even result in a difference of transcript abundance for some genes when they are prepared for RNA-seq. Thus, we investigate how different methods may reduce variance by plotting the coefficient of variation (CV) of normalized read count vs. mean read count in log scale (the logged mean was linearly transformed to 0-1 range, [fig_ref] Figure 4: Differential expression [/fig_ref]. Overall, the TIN method gives a relatively larger CV than the other three methods. When RNA degradation is a major concern such as in GBM-R10vsR4, and breast tumor FF vs. compared to other methods except in the very lower or upper end where the CV was inflated due to outliers [fig_ref] Figure 4: Differential expression [/fig_ref]. The RUVr approach applies UQ normalization first and then further removes the additional variation estimated from the factor analysis. It always reduces CV over the UQ method. We normalized the raw read count by dividing it by 1 -DI score for each gene within each sample. The adjusted read counts (rounded) were input into the edgeR package [bib_ref] edgeR: a Bioconductor package for differential expression analysis of digital gene expression..., Robinson [/bib_ref] for DE analysis. When all genes are true nulls, the empirical cumulative distribution function (ECDF) of p value tends to be a diagonal line. Thus, an ECDF curve closer to the diagonal line indicates better performance of the normalization method in correcting the degradation bias. For SEQC-AA data with 16 technical replicates, all 4 methods resulted in expected ECDF curves close to the diagonal line [fig_ref] Figure 4: Differential expression [/fig_ref]. For PBMC-S01 and GBM-R6vsR4 comparisons with known modest between-condition difference in mRNA degradation, the ECDF curves from different methods were all well above the diagonal line (except TIN for PBMC data), suggesting that differential degradation probably has caused a difference in gene abundance level when the samples were sequenced [fig_ref] Figure 4: Differential expression [/fig_ref]. Both DegNorm and TIN methods brought the ECDF curve down towards the diagonal line compared to UQ and RUVr, indicating that correction for degradation bias helps reduce potential false positives. Nevertheless, the TIN curve in PBMC-S01 data was well below the diagonal line [fig_ref] Figure 4: Differential expression [/fig_ref] , which may indicate a loss of statistical power due to the large variance of the normalized read counts [fig_ref] Figure 4: Differential expression [/fig_ref]. In contrast, the RUVr ECDF curves in both comparisons are significantly higher than that in UQ (regardless that RUVr had lower CV than UQ), The mean counts for each data were scaled to 0-1 range by a linear transformation (i.e., (X i − min j (X j ))/max j (X j ) where X i is the log of the mean count for gene i). The CV curve was generated using the R built-in function smooth.spline. The beanplot under the CV plot shows the density of log of mean read counts from DegNorm (the densities of read counts from other normalization methods are similar and not shown). g-l Empirical cumulative distribution function (ECDF) of the p value from DE analysis. The RUVr results were generated using the RUV-seq package. For TIN method, we followed Wang et al. [bib_ref] Measure transcript integrity using RNA-seq data, Wang [/bib_ref] with details described in the uploaded R Markdown file suggesting an ineffective correction of degradation bias or even an adverse effect to cause extra false positives. For the GBM-R10vsR4 comparison, ECDF curves are all far above the diagonal line, likely indicating a substantial change of transcript abundance level for many genes due to drastic degradation in R4 samples [fig_ref] Figure 4: Differential expression [/fig_ref]. The DLPFC and breast tumor FF-FFPE RNA-seq data were both generated from clinical tissue samples. For the DLPFC Br1729 T0+T15 vs. T30+T60 comparison, the DegNorm resulted in a slightly lower p value than UQ in Ribo-Zero [fig_ref] Figure 4: Differential expression [/fig_ref] but much lower than UQ in mRNA-seq (Additional file 2: [fig_ref] Figure 4: Differential expression [/fig_ref] data. For the breast tumor data, FFPE samples were shown substantially fragmented and degraded than FF samples [fig_ref] Figure 3: Degradation index [/fig_ref]. DegNorm resulted in a lower p value curve than all other methods [fig_ref] Figure 4: Differential expression [/fig_ref]. We also did cross-protocol DE analysis by comparing the four Br1729 mRNA-seq with four Ribo-Zero samples. All p value curves were way above the diagonal line (Additional file 2: [fig_ref] Figure 4: Differential expression [/fig_ref] , suggesting DE analysis across different sequencing protocols should not be recommended.
[formula] a b g h c d i j e f k l [/formula]
The p value ECDF curve provides a global picture of a false-positive rate at different type-I error rate thresholds when all null hypotheses are true. In practice, as the ground truth is unknown, one typically claims the DE by controlling the false discovery rate (FDR) to correct multiple comparison errors. Thus, we further compared the false-positive rate of different methods by controlling the nominal FDR under the criterion of q value ≤ 0.05 using q-value package [bib_ref] A direct approach to false discovery rates, Storey [/bib_ref] [bib_ref] The positive false discovery rate: a Bayesian interpretation and the q-value, Storey [/bib_ref] [bib_ref] Statistical significance for genomewide studies, Storey [/bib_ref]. For SEQC-AA data with a little degradation difference between samples, all four methods resulted in very few claimed positives (Additional file 3: , consistent with the close-to-diagonal-line nature of p value ECDF curves in [fig_ref] Figure 4: Differential expression [/fig_ref]. For the rest data with differential degradation, all methods yielded a substantial number of false positives. For comparison, we plotted the ratio of the false-positive rate of UQ, RUVr, and TIN over DegNorm in log2 scale [fig_ref] Figure 5: Differential expression analysis results [/fig_ref]. The UQ and RUVr methods consistently yielded more false positives than DegNorm, by a factor ranging from 1 to 3. [bib_ref] Evaluation of DNA microarray results with quantitative gene expression platforms, Canales [/bib_ref]. Genes with absolute log2 fold change value ≥ 2 were defined as true positives, and absolute log2 fold change value ≤ 0.1 was defined as true negatives. Genes with log2 fold change in between were disregarded. e ROC calculated based on~17,304 PCR-verified genes published in a separate study [bib_ref] A comprehensive assessment of RNA-seq accuracy, reproducibility and information content by the..., Consortium [/bib_ref] for the same SEQC AB samples. The same threshold values of log2 fold change as in c were used in defining the positives and negatives. f The area under the ROC curve (AUC) statistic as a function of the threshold value of absolute log2 fold change used to define the true positives based on the PCR data from e Nevertheless, we will show below that this relatively lower false-positive rate from the TIN method is an indication of undermined power due to the excessively inflated variance [fig_ref] Figure 4: Differential expression [/fig_ref].
The SEQC-AB data presents an atypical example containing an unusually large number of differentially expressed genes [bib_ref] A comprehensive assessment of RNA-seq accuracy, reproducibility and information content by the..., Consortium [/bib_ref] [bib_ref] Evaluation of DNA microarray results with quantitative gene expression platforms, Canales [/bib_ref]. DegNorm produced an overall lower CV than UQ, RUVr, and TIN methods [fig_ref] Figure 5: Differential expression analysis results [/fig_ref]. The DegNorm ECDF lies below those of UQ and RUVr methods while above TIN [fig_ref] Figure 5: Differential expression analysis results [/fig_ref]. With the presence of truly differentially expressed genes, the lower ECDF can be interpreted as a tendency to result in fewer false positives (good) or more false negatives (bad, less power) or a mix in the DE analysis. To investigate this, we utilized 2 sets of PCR-verified genes, 1 from the original MAQC study of 843 genes [bib_ref] Evaluation of DNA microarray results with quantitative gene expression platforms, Canales [/bib_ref] and the other from SEQC study of 20,801 genes [bib_ref] A comprehensive assessment of RNA-seq accuracy, reproducibility and information content by the..., Consortium [/bib_ref] , as the ground truth to construct receiver operating characteristic curves (ROC) [fig_ref] Figure 5: Differential expression analysis results [/fig_ref]. Similar to Rissio et al. [bib_ref] Normalization of RNA-seq data using factor analysis of control genes or samples, Risso [/bib_ref] , we defined a gene as a positive if the absolute value of log2 fold change ≥ 2, a negative if ≤ 0.1, and undefined otherwise. For both sets, the ROC curves suggest that DegNorm achieved better true-positive rate (sensitivity) than UQ, RUVr, and TIN while controlling the false-positive rate (1-specificity). For example, at FPR = 0.05 (specificity = 0.95), the larger PCR set suggested a 78.1% true-positive rate for Deg-Norm in comparison with 73.7%, 67.3%, and 48.3% for UQ, RUVr, and TIN methods, respectively. When we varied the of log2 fold change threshold value from 1 to 5 to define the positives, the area under the ROC curve (AUC) from all methods (except TIN) increased as expected, while the AUC from DegNorm remained the largest and the gap between DegNorm and other methods enlarge [fig_ref] Figure 5: Differential expression analysis results [/fig_ref]. As the true-negative set was fixed in this experiment, the true-positive rate drove the change of AUC as the threshold value increases. This suggests DegNorm improves the power to identify highly differentially expressed genes over other normalization methods while controlling the false-positive rate. Therefore, we conclude that the lower ECDF curve from DegNorm method [fig_ref] Figure 5: Differential expression analysis results [/fig_ref] manifests a good tendency to reduce false-positive rate or increase specificity without sacrifice of statistical power or sensitivity.
The RUVr and TIN methods both showed pronouncedly lower power than DegNorm and UQ, but due to different reasons [fig_ref] Figure 5: Differential expression analysis results [/fig_ref]. The RUVr method is guaranteed to reduce more variation based on the UQ-normalized data. Nevertheless, the removed variation may contain true biological difference if it is confounded with unwanted variation, which will result in more false negatives. On the other hand, the large variance incurred by TIN method appears to severely undermine the power in the DE analysis [fig_ref] Figure 5: Differential expression analysis results [/fig_ref].
We present three examples that exemplify how DegNorm may improve the accuracy in the DE analysis. We used local false discovery rate (lfdr) from q-value package [bib_ref] A direct approach to false discovery rates, Storey [/bib_ref] [bib_ref] The positive false discovery rate: a Bayesian interpretation and the q-value, Storey [/bib_ref] [bib_ref] Statistical significance for genomewide studies, Storey [/bib_ref] to quantify the significance of DE analysis. Compared to q value, which quantifies the average FDR for all genes with smaller p value than the given one, lfdr is more appropriate to quantify the false discovery rate associated with any individual p value. The MMP14 gene in the GBM-R10vsR4 comparison displayed a clear degradation in the 5′ end in R4 samples and tested positive using UQ method (p value = 1.31e−8, lfdr = 1.1e−8). DegNorm compensated for the degraded portion of R4 samples and returned negative test result (p value = 0.91, lfdr = 0.32, . The second example is the PIK3C2A gene from SEQC-AB comparison, a PCR-verified negative with log2 fold change = 0.06. It tested positive under UQ (p value = 0.005, lfdr = 0.02) while negative under DegNorm with degradation correction in the 5′ end of A samples (p value = 0.77, lfdr = 0.97, . The third example is the NDUFV3 gene from the SEQC-AB data, showing nearly depleted coverage in the entire third exon region from~223 to 1327 nt for B samples likely due to alternative splicing. It tested positive under UQ (p value = 1.94e−37, lfdr = 3.39e−09, . DegNorm returned a negative result (p value = 0.808, lfdr = 0.942, , consistent with negative PCR verification (log2 fold change = − 0.114).
## Gbm ampk knockout data
GBM AMPK data presents a typical case in a clinical study where RNA-seq is used to survey a transcriptome for differential expression between control and treatment samples. DegNorm uncovered higher degradation in the control than the AMPK knockout samples [fig_ref] Figure 7: d, Additional file 2 [/fig_ref] and heterogeneity in degradation pattern between the two conditions [fig_ref] Figure 7: d, Additional file 2 [/fig_ref]. DegNorm also resulted in a lower CV curve and a lower ECDF of p value than UQ and RUVr methods [fig_ref] Figure 7: d, Additional file 2 [/fig_ref]. We claimed differential expression under q value ≤ 0.05 and compared the test-positive sets between DegNorm, UQ, and RUVr [fig_ref] Figure 7: d, Additional file 2 [/fig_ref]. There were 2798 positive genes shared by all 3 methods, while UQ and RUVr produced 935 and 775 more positives than DegNorm, respectively. We suspect degradation may cause an excess of false positives (as suggested by similar plots in for SEQC-AB data); nevertheless, without a large set of PCR-verified genes, it is impossible to rigorously assess the sensitivity and specificity. Chhipa et al. [bib_ref] AMP kinase promotes glioblastoma bioenergetics and tumour growth, Chhipa [/bib_ref] analyzed the RNA-seq data and concluded that the bioenergetics of cellular metabolism was the most significantly downregulated pathway in AMPK-depleted samples. They applied RT-qPCR and verified a small set of downregulated genes using independent GBM cell lines from 3 other patients. Among the 12 genes verified as downregulated, including HIF1a, LDHA, SLC2A1, HK1, GPI, ALDOA, TPI1, PFKM, ENO1, GABPA, TFAM, and COX20, DegNorm, UQ, and RUVr all successfully identified 8 except HK1, PFKM, GABPA, and COX20, whereas TIN method missed 2 additional genes, GPI and ALDOA (Additional file 4: . The discrepancy between the RNA-seq set and PCR results could be due to cell line-specific variation, lack of power due to small sample size or sample quality (personal communication with Dr. Dasgupta). Clearly, such a small-scale verification is insufficient to conclude about the sensitivity, neither can we evaluate the specificity without verification of the negatives. The discrepancy between DegNorm and other methods does provide alerts to users of possible false positives caused by degradation when interpreting the results of such studies.
## Simulation study
To further systematically investigate the performance of DegNorm, we conducted a simulation study in a two-condition comparison: 4 control (A) vs. and NDUFV3 (e, f) genes, respectively, in the SEQC-AB comparison. All three genes changed from positive to negative after degradation normalization in DE analysis using edgeR.
In (e, f), annotated exon boundaries are plotted to show possible alternative splicing in the exon from 223 to 1327 nt genes had no degradation whereas in the rest 3 settings, 80% of the genes were randomly selected for degradation. In the second setting, for a gene selected for degradation, either 3, 4, or 5 samples out of the 8 were randomly chosen for degradation, whereas in the third, either all 4 control samples or 4 treatment samples were randomly chosen for degradation. In both second and third settings, the degradation extent for each degraded gene was random but following the same distribution. In the fourth setting, for each gene selected to degrade, 2 control samples were randomly selected for degradation with the same expected severity, while all treatment samples had a sample-specific systematic difference in expected severity. The simulation details are described in the "Methods" section and Additional file 1. Simulation I presents a scenario where samples have no degradation bias or any other bias but only between-sample variation in sequencing depth. In the following, we shall refer the latent count as the true read count for a gene before degradation is imposed. Deg-Norm is data driven and always returns non-negative DI scores by design. The estimated DI scores have a mediañ 0.11 in all samples [fig_ref] Figure 8: Results for simulation I [/fig_ref] , demonstrating the absence of between-condition heterogeneity [fig_ref] Figure 8: Results for simulation I [/fig_ref]. To investigate how the positive bias in DI scores may impact DE analysis, we first plotted the normalized vs. latent read count from different methods [fig_ref] Figure 8: Results for simulation I [/fig_ref] (the latent read count in simulation I is just the raw read count).
Unsurprisingly, the UQ method perfectly normalizes the sequencing depth [fig_ref] Figure 8: Results for simulation I [/fig_ref] , while all other three methods caused bias or extra variance to different extents [fig_ref] Figure 8: Results for simulation I [/fig_ref]. The positive bias from DegNorm is more pronounced when read counts are low such that read coverage curve cannot be well estimated [fig_ref] Figure 8: Results for simulation I [/fig_ref]. As for any gene, all samples are subject to this over-estimation bias; this bias may partially cancel off in DE analysis. As a result, the p value ECDF and ROC curves of DegNorm almost perfectly overlap with the latent and UQ curves [fig_ref] Figure 8: Results for simulation I [/fig_ref]. At FPR = 0.05, DegNorm, RUVr, and TIN had sensitivity decay of 0.6%, 1.1%, and 8.6%, respectively, compared to using the latent counts or UQ method (Additional file 5: .
When degradation is an issue as in simulations II-IV, the estimated DI scores provide an informative characterization of the overall degradation severity of each sample (Additional file 2: [fig_ref] Figure 5: Differential expression analysis results [/fig_ref] -f ) as well as the similarity of gene-specific degradation pattern between samples (Additional file 2: [fig_ref] Figure 5: Differential expression analysis results [/fig_ref]. DegNorm demonstrates consistently better correction of degradation bias than the UQ, TIN, and RUVr methods, as evidenced by higher regression coefficient of determination (R 2 ), and a nearly symmetric distribution of data points around the diagonal line in the scatter plot of normalized vs. latent read counts normalizing sequencing depth , Additional file 2: , e). Similar extreme variance issue is also observed in the TIN method in simulations II-IV , Additional file 2: , f ).
To gain more insights into the CV plots, we singled out the 18,000 truly non-differentially expressed genes from simulations II-IV and plotted the CV against the mean of normalized count . Without the confounding effect of biological difference, the CV of these genes tends to be inflated due to degradation-caused loss of read count. Thus, the CV is an informative measure to assess the effectiveness of degradation normalization. Indeed, the UQ and RUVr curves are both well above the latent curve in all simulations, suggesting degradation bias was not or inadequately corrected. Similar to what we observed in the real data, the TIN CV curve dominates other methods in each setting, echoing the excess of variance observed in the scatter plot , Additional file 2: , f ). In contrast, the DegNorm curves were the closest to the respective latent curves but with a slight underestimation of CV in the lower half.
DegNorm improves the accuracy in DE analyses. In both ECDF and ROC plots [fig_ref] Figure 8: Results for simulation I [/fig_ref] , the DegNorm curve was the closest to the latent one, demonstrating improvement over other normalization methods to different extents. We tabulated the sensitivity of each method at FPR = 0.05 (Additional file 5: and plotted it in [fig_ref] Figure 10: Bar chart of sensitivity [/fig_ref]. In settings II and III, where degradation was randomly chosen among samples (II) or conditions (III), the UQ and RUVr methods were both ineffective to correct this non-systematic but gene-specific bias (Figs. 9k, l and 10). In particular, in simulation III as degradation was applied to one condition of random choice for a given gene, the degradation bias was completely confounded with the covariate of interest and cannot be removed by UQ or RUVr method. Consequently, many false positives were called due to degradation bias . At FPR = 0.05 threshold, DegNorm improved the sensitivity by a factor of 1.28, 1.30, and 2.01 compared to UQ, RUVr, and TIN methods, respectively [fig_ref] Figure 10: Bar chart of sensitivity [/fig_ref] , Additional file 5: . In contrast, the treatment samples in setting IV had a systematic difference in average degradation (Additional file 2: [fig_ref] Figure 5: Differential expression analysis results [/fig_ref] , both UQ and RUVr performed reasonably well [fig_ref] Figure 10: Bar chart of sensitivity [/fig_ref]. In all four degradation settings, the TIN method showed inferior power to detect true DE genes.
# Discussion and conclusions
In this paper, we showed that RNA degradation pattern and severity are not only sample specific, but also gene-specific, and thus commonly used global normalized mean read count (in log scale) for 18,000 true-negative genes (out of a total of 20,000 genes). The bean pot under the CV plot shows the density of log of mean read counts from DegNorm. h-j Empirical CDF of p value from edgeR DE analysis. The latent curve corresponds to the results using the true read counts before the degradation was imposed. k-m Receiver operating characteristic curves of DE analysis normalization methods that impose a sample-specific constant adjustment to all genes within the same sample are ineffective to correct for this bias. The RUVr approach is guaranteed to reduce variation, while they failed to show pronounced improvement over the UQ method in the DE analysis in all data sets considered in this study (even got worse in all data under consideration). One complexity is that the true biological difference is often confounded with degradation (e.g., SEQC-AB data) and other unwanted variation. The factor analysis cannot well separate the unwanted from the wanted variation, and it may even remove the true biological difference of interest. This confounding issue was illustrated in the SEQC-AB data by the high within-condition and low between-condition correlation of DI scores (Additional file 2: . In particular, the RUVg method was sensitive to the selection of empirical control genes or the factor(s) used to estimate the unwanted variation (Additional file 2: . More objective criteria in this regard need to be developed for the RUVg method.
Although motivated by mRNA degradation, we defined the degradation in this paper in a generalized and relative sense. The quantified DI scores may reflect confounding effects from mRNA degradation, alternative splicing, and other factors. Risso et al. [bib_ref] Normalization of RNA-seq data using factor analysis of control genes or samples, Risso [/bib_ref] showed that in the SEQC-AB data, samples were clustered due to the difference of sample preparation, experiment protocol, sequencing run batches and flow cell, etc. Such factors could impact the RNA samples by changing the read coverage curves. Thus, normalizing heterogeneity in coverage curves may help reduce bias due to such factors. Indeed, in all five benchmark data sets considered in this paper, DegNorm performed consistently better compared to other methods regardless of whether mRNA degradation was a known concerning issue. Unlike mRIN and TIN measures where degradation was defined as the deviation from hypothesized uniform coverage curve, the DI score from DegNorm is defined with reference to an adaptively estimated latent coverage curve that minimizes the distance to the observed coverage curves. If a gene has 50% degradation but having consistent coverage curves across samples, the estimated DI scores will all be nearly 0. In this case, the read count in each sample can still accurately reflect the relative abundance between samples, and degradation correction is unnecessary. From this perspective, the DI score is defined in a relative sense. Normalizing the read counts for degradation bias using DI scores is hoped to minimize the extrapolation needed, thus avoids an excess of variance. The advantage of this strategy was exemplified in all real and simulated data sets in contrast to the TIN method.
There are a few limitations of the DegNorm method. First, like any other normalization method, DegNorm is a post hoc approach that is designed to alleviate the issues due to degradation heterogeneity between samples/ genes and thus improve the accuracy of DE analysis. It cannot completely remove the bias for every single gene. In particular, cautions must be taken when testing DE for samples that have a dramatic difference in degradation (e.g., GBM R10 vs. R4), as it may change the true abundance level of transcripts of interest and lead to the excess of false positive or false negatives. High-quality RNA samples are always desirable in RNA-seq. Furthermore, cross-platform DE analysis is not recommendable. Second, the core component of DegNorm is a matrix factorization over-approximation algorithm aiming to correct for the degradation bias that commonly exists in RNA samples, even for the high-quality SEQC data. DegNorm tends to result in a pronounced over-estimation bias in DI scores for genes with low read counts regardless of whether degradation is present (simulations II-IV) or absent (simulation I). We showed in simulation I that this bias is not a big concern in DE analysis as it tends to be homogeneous across all samples for all non-degraded genes. Third, DegNorm has only been tested in this paper on the bulk RNA-seq data generated from a cell line or clinical samples (FF or FFPE) under mRNA-seq or Ribo-Zero-seq protocol. The effectiveness of DegNorm for RNA-seq data from 3′ end sequencing (3seq) or other variants needs to be further investigated in the future. Lastly, DegNorm is computing intensive due to parsing read alignment results, calculation of read coverage curves, and repeated non-negative matrix factorization of large matrices. We have implemented DegNorm in a Python package available at https://nustatbioinfo.github.io/DegNorm/. Currently, it took about 9 h to run the entire pipeline on the FF vs. FFPE comparison (6 samples) on a 22-core node on the Linux cluster. We have implemented an MPI release through parallel computing that can reduce the time by a factor of n where n is the number of nodes used.
In summary, we conclude DegNorm provides a pipeline for informative quantification of gene-/sample-specific transcript degradation pattern and for effective correction of degradation bias in RNA-seq. We intend for DegNorm to serve as a general normalization method to improve the accuracy in the gene expression differentiation analysis.
# Methods
## Degnorm algorithm
Suppose we have p samples and n genes, X ij is the read count for gene i in sample j. For simplicity of notation, we first illustrate the proposed method by focusing on one gene. Let f ij (x), x = 1, … , L i ; i = 1, … , n; j = 1, … , p be the read coverage score for transcript i of length L i from sample j. When different isoforms are present, the L i positions represent the assembly of all expressed exons in the sequential order. We assume there is an envelope function e i (x) that defines the ideal shape of read coverage curve for gene i if no degradation exists. The actual ideal coverage curve for the given gene in the jth sample is k ij e i (x), where k ij denotes the confounded effect of sequencing depth and relative abundance of gene i in sample j.
Degradation causes f ij (x) to deviate downward from k ij e i (x) in degraded region(s). Clearly, sampling error can cause random fluctuation of f ij (x) from the ideal curve k ij e i (x) . We assume the random error is negligible compared to the major bias arising from degradation. Thus we require k ij e i (x) ≥ f ij (x) for all j and x. The difference between k ij e i (x) and f ij (x) provides an estimate of degraded portion of read count. We propose a method that allows to estimate k ij and e i (x) to quantify the degradation extent of each gene within each sample while simultaneously controlling the sequencing depth.
Estimating degradation via non-negative matrix overapproximation , … , e i (L i )) T . We propose to estimate K i and E i by minimizing the following quadratic loss function subject to some constraint:
[formula] Let f ij = (f ij (1), … , f ij (L i )) T , j = 1, … , p and F i = (f i1 , … , f ip ) T . Let K i = (k i1 , … , k ip ) T , E i = (e iQ K i ; E i ð Þ¼ X L i x¼1 X p j¼1 k ij e i x ð Þ− f ij x ð Þ h i 2 s:t:k ij e i x ð Þ − f ij x ð Þ≥ 0; k ij ; e i x ð Þ > 0; ∀ j; ∀x: [/formula]
We can configure this problem into a non-negative matrix factorization problem [bib_ref] Learning the parts of objects by non-negative matrix factorization, Lee [/bib_ref] as follows:
[formula] min K i ;E i ‖K i E i T −F i ‖ 2 s:t: F i ≤ K i E i T ; K i ≥ 0; E i ≥ 0; [/formula]
where ‖- ‖ 2 stands for the element-wise quadratic norm (i.e., sum of squared elements), and ≤ and ≥ for elementwise logical comparison. We call this a rank-one non-negative matrix factorization over-approximation (NMF-OA) problem as K i and E i both have rank 1 and K i E i T ≥ F i . The proposed iterative algorithm is described in details in the Additional file 1.
## Refinement of nmf-oa algorithm
The NMF-OA optimization algorithm provides an approximate solution to this problem. However, in the RNA-seq data, the performance of the solution can be affected by two confounding factors, the degradation extent and the sequencing depth. In our quadratic objective function Q(K i , E i ), an f ij of larger magnitude tends to have more influence on the estimation of envelope function. A dominant scale in f ij may force the algorithm to fit an envelope function that resembles f ij to minimize the loss. Thus, a good scale normalizing factor for sequencing depth is important to yield a good estimate of the envelope function e i (x). With gene-specific and sample-specific degradation, the total number of reads may not provide a reliable measure of sequencing depth.
Second, given F i that is appropriately normalized for sequencing depth, the scale factor K i should reflect the relative abundance of the gene in the non-degraded region of each sample (to be referred to as the baseline region below) . The non-degraded region must preserve a similar shape in gene's coverage curves from different samples. If one can first estimate K i from the identified baseline region, then NMF-OA algorithm will lead to a better estimate of the envelope function E i , particularly in the situation when the degradation extent is severe as the GBM data. We account for these two considerations by proposing an iterative degradation normalization pipeline (DegNorm) as follows:
1. Sequencing depth adjustment: given the current estimate of (K i ,E i ) for i = 1, … , n, define the degradation index (DI) score as:
[formula] ρ ij ¼ 1− P L i x¼1 f ij x ð Þ P L i x¼1 k ij e i x ð Þ : [/formula]
Graphically, ρ ij stands for the fraction of the total area under the curve k ij e i (x) but above f ij (x) . The DI score is used to calculate an adjusted read count by extrapolation:
[formula] X ij ¼ X ij 1−ρ ij : [/formula]
Next, we calculate the sequencing depth scaling factor using the degradation-corrected total number of read count:
[formula] s j ¼ P n i¼1X ij Median j P n i¼1X ij È É: [/formula]
In this paper, the results presented were all based on this scale normalization. Alternatively, we can use other normalization methods like TMM or UQ to calculate the normalizing constant s j based on the degradation-adjusted read countX ij if the presence of extreme outliers is a concern.
2. Degradation estimation: given the estimated sequencing depth s j from step 1, adjust the coverage curves as follows:
f ij ← f ij s j :
[formula] Let F i = (f i1 , … , f ip ) T . [/formula]
Run NMF-OA for each gene on updated coverage curves F i and obtain the estimate of (K i ,E i ). Divide each gene into 20 bins. Define the residual matrix as R i = K i E i T − F i . We identify a subset of bins on which F i preserves the most similar shape as the envelope function E i across all samples by progressively dropping bins that have the largest sum of squares of normalized residuals (R i /F i ) and repeatedly applying NMF-OA to the remaining bins. This step stops if the maximum DI score obtained from the remaining bins is ≤ 0.1 or if 80% bins have been dropped (see details in Additional file 1). The remaining transcript regions on selected bins are regarded as the baseline. Denote the read coverage curve on baseline region as F Ã i . Run NMF-OA on F Ã i . The resulting K Ã i is a refined estimate of K i , given which the envelope function can be obtained as:
[formula] e i ðxÞ ¼ max j f i j ðxÞ k à i j ( ) ; x ¼ 1; …; L i : [/formula]
3. Steps 1 and 2 are repeated until the algorithm converges.
The DE analysis using edgeR was carried out based on the degradation normalized read countX ij (rounded) at convergence with upper quartile (UQ) normalization for sequencing depth.
## Simulations
We first simulated the latent read count for each gene within each sample. For a given gene without DE, the latent counts (without degradation) in both control and treatment samples were randomly simulated from the negative binomial distribution with the same mean and same dispersion parameters that were randomly chosen from the fitted values of SEQC-AB data. For a gene with DE, the latent counts for the control samples were first simulated as above, and those for the treatment samples were simulated from the negative binomial with mean parameter increased/decreased by a factor of (1.5 + γ) for up-or downregulated genes, respectively, where γ was simulated randomly from an exponential distribution with mean = 1. In the second step, we simulated the degradation given the latent read count (simulations II-IV). For a given gene, we first chose a Gaussian mixture distribution that covers the entire range of total transcript to model the read start position distribution. For simplicity, we only consider 5′ end degradation. Each read from a gene that was pre-selected for degradation degraded (or was disregarded) with the probability that depended on the start position of the read, defined by a cumulative distribution function (CDF) of a lognormal distribution. The degradation pattern and extent can be tuned by varying the parameters in the lognormal distribution (see more details in Additional file 1).
[fig] Figure 3: Degradation index (DI) scores show gene-/sample-/condition-specific degradation heterogeneity. a-f Box plots of DI scores presented in a between-group comparison defined for the differential expression analysis as follows. a SEQC-AA data (16,670 genes): the 8 (1-8) technical replicates from the first run vs. the 8 (9-16) from the second run. b SEQC-AB data (19,061 genes): 4 biological replicates from A condition vs. 4 from B condition. c PBMC data for subject S01 (14,051 genes): 2 samples exposed at room temperature for 0 and 12 h (S01_T1 and S01_T2) vs. 2 for 24 and 48 h (S01_T3 and S01_T4), respectively. d GBM data (14,298 genes): 3 replicates each for RIN number = 10, 6, and 4, respectively. e DLPFC data for subject Br1729 from Ribo-Zero-seq (18,634 genes): 2 samples exposed to room temperature for 0 and 15 min (T0, T15) vs. 2 for 30 and 60 min (T30, T60). f Breast tumor data of 3 matched pair (T1, T2, T3) prepared from FF and FFPE methods, respectively (10,996 genes). g-l For each data set presented in a-f, the heatmap presents the DI scores of genes sorted in the ascending order of the average DI score of the first condition (in the GBM case, the R10 samples), where each row corresponds to the same gene across samples [/fig]
[fig] Figure 4: Differential expression (DE) analysis in data sets that had no true differential expression. Results are shown for SEQC-AA, PBMC-S01, GBM R10 + R6, GBM R10 + R4, DLPFC Br1729 Ribo-Zero, and breast tumor data. a-f Coefficient of variation (CV) vs. mean read counts (in log scale): compared are results from proposed DegNorm pipeline and other methods including upper quartile (UQ), RUVr, and TIN. [/fig]
[fig] 7: and 1.4-36.8, respectively. The TIN method reduced the false-positive rate over Deg-Norm in the PBMC-S01 with a factor of 1.37. [/fig]
[fig] Figure 5: Differential expression analysis results. a Log2 ratio of false-positive rate of UQ/RUVr/TIN relative to DegNorm at q value = 0.05 criterion for data sets where genes have no differential expression. b-f DE results for SEQC-AB data. b Coefficient of variation (CV) vs. mean normalized read counts (in log scale). c Empirical cumulative distribution function (ECDF) of the p value from DE analysis. d Receiving operating characteristic curve (ROC) calculated from~500 PCR-verified genes [/fig]
[fig] 4: treatment (B) samples in 4 different degradation settings. Each sample of 20,000 genes was simulated with a random sequencing depth of 40-60 million reads, 5% of which were chosen to be upregulated and another 5% for downregulated in expression. In the first setting, all [/fig]
[fig] Figure 8: Results for simulation I. a Box plot of estimated DI scores where conditions A and B stand for control and treatment samples, respectively. b Heatmap of DI scores of genes sorted in the ascending order of the average DI score of condition A (control), showing no between-condition clustered pattern. c-f Scatter plot of normalized read count vs. latent read count in log2 scale with diagonal line imposed for simulation I. Compared are results from proposed DegNorm pipeline and other methods including upper quartile (UQ), RUVr, and TIN. g, h ECDF plots of p value and ROC curves for simulation I under different normalization methods. All curves except TIN were well overlapping with each other, suggesting close performance in DE analysisFig. 9 Comparison of different normalization methods in simulations II, III, and IV. a-d Scatter plot of normalized read count vs. latent read count in log2 scale with diagonal line imposed from DegNorm, UQ, RUVg, RUVr, and TIN methods for simulation II. e-g Coefficient of variation (CV) vs. [/fig]
[fig] Figure 10: Bar chart of sensitivity (TPR) of different methods in simulations II, III, and IV. True-positive rate (sensitivity) is compared at false-positive rate threshold = 0.05 for each method in each simulation [/fig]
|
Dark Plasmonic Breathing Modes in Silver Nanodisks
The experimental EEL maps shown inFigure 1bof the main part were acquired with the STEM-EELS technique (STEM scanning transmission electron microscope, EELS electron energy loss spectroscopy), scanning a 200 keV electron beam over the region of interest and taking EEL spectra with an energy-resolution of 0.12 eV at each pixel (SupportingFigure 1). Due to stable experimental conditions during the measurement we used an exposure time per pixel of 0.2 seconds.To account for the lower signal of elastically scattered electrons when the electron beam transits the silver disk (position 1 in SupportingFigure 2a) as compared to trajectories through the silicon nitride substrate only (position 2 in SupportingFigure 2a), spectra were normalized to their according zero-loss intensity (inset in SupportingFigure 2). As illustrated in SupportingFigure 2b, the breathing mode peak around 2.5 eV (position 1) without normalization is of comparable intensity as the tail of zero loss peak measured on the substrate (position 2). Only by normalizing,
The experimental EEL maps shown in [fig_ref] Supplementary, Figure 1: Schematic view of the used measurement technique [/fig_ref] of the main part were acquired with the STEM-EELS technique (STEM scanning transmission electron microscope, EELS electron energy loss spectroscopy), scanning a 200 keV electron beam over the region of interest and taking EEL spectra with an energy-resolution of 0.12 eV at each pixel (Supporting [fig_ref] Supplementary, Figure 1: Schematic view of the used measurement technique [/fig_ref]. Due to stable experimental conditions during the measurement we used an exposure time per pixel of 0.2 seconds.
To account for the lower signal of elastically scattered electrons when the electron beam transits the silver disk (position 1 in Supporting as compared to trajectories through the silicon nitride substrate only (position 2 in Supporting , spectra were normalized to their according zero-loss intensity (inset in Supporting . As illustrated in Supporting , the breathing mode peak around 2.5 eV (position 1) without normalization is of comparable intensity as the tail of zero loss peak measured on the substrate (position 2). Only by normalizing, * To whom correspondence should be addressed † Graz University of Technology ‡ University of Graz the correct relative signal strengths are recovered. In addition, the zero-loss signal was subtracted from each individual spectrum using a fitted logarithmic function ("fitted logarithm tail" model within Digital Micrograph software [Gatan, USA]). All EEL maps in [fig_ref] Supplementary, Figure 1: Schematic view of the used measurement technique [/fig_ref] were measured on the same silver nanodisk of 200 nm in diameter, while the single spectra in were acquired in the middle of nanodisks with diameters ranging from 130 to 1000 nm. The single spectra were also zero-loss subtracted using the same subtraction procedure as mentioned above and smoothed by averaging EEL-counts over an energy-width of 0.1 eV.
## Surface charge distribution (simulation)
To simulate EEL maps of the silver nanodisk in
## Numercial calculation of the surface plasmon dispersion relation
For calculating the surface plasmon dispersion we applied the optical transfer matrix method which allows to calculate the amplitudes of the plane waves reflected and transmitted by a multilayer system in relation to an incoming plane wave. As surface plasmons are bound modes the incoming wave is assumed evanescent, akin of the Otto excitation scheme but with the coupling prism considered at infinite distance. To obtain the dispersion of the plasmon mode of the vacuum-Ag-silicon nitride-vacuum system with a field maximum at the silver-silicon nitride interface, the exciting evanescent wave was assumed to be incident from the vacuum-side of the silver film. For a given wavenumber, the frequency was varied to find the (local) maximum amplitude of the transmitted (evanescent) wave, which indicates the optimum condition for exciting this surface plasmon mode with real wavenumber (i.e., no spatial decay) and real frequency (no temporal decay). This leads to the dispersion relation, which is probed by the experiments where the wavenumber of the breathing mode is defined by the disk diameter and the frequency of maximum excitation probability is selected from the EEL spectra. Top: EEL intensity before normalization. For a loss energy of around 2.5 eV, positions 1 and 2 show similar EEL intensities, although no breathing mode is excited at position 2 (Inset: zero-loss intensity on and beside the particle). Bottom: EEL signal after normalization.
[fig] 1: in the main part and to calculate surface charge distributions on this particle as a function of loss energy and position, the MNPBEP matlab toolbox was used (seeRef. 21 in main article). For the computation of surface charge distributions, the electron beam was set to 3 different positions on the silver particle (in the center, at half radial distance (R/2) and on the edge), while the energy loss was varied for each position between 2 to 4 eV (center position) and 1 to 4 eV (R/2 and edge) with an energy step size of 0.01 eV. Surface charge maps as a function of loss energy are summarized in three Supporting Videos. Video 1 shows the spectral evolution of the surface charge distribution for the center position of the electron beam, while Video 2 and 3 refer to the R/2 and edge positions, respectively. [/fig]
|
Chronic Thromboembolic Pulmonary Hypertension: A Comprehensive Review and Multidisciplinary Approach to Surgical Treatment
Chronic thromboembolic pulmonary hypertension (CTEPH) is an underdiagnosed and undertreated sequelae of acute pulmonary embolism. In this comprehensive review, we provide an introductory overview of CTEPH, highlight recent advances in its diagnostic imaging, and describe the surgical technique for pulmonary thromboendarterectomy (PTE), the only established curative treatment for CTEPH. We also discuss the emerging role of balloon pulmonary angioplasty, both independently and combined with PTE, for patients with inoperable, residual, or refractory pulmonary hypertension post PTE. Finally, we stress the importance of a specialized multidisciplinary team approach to CTEPH patient care and share our approach to optimizing care for these patients.
This review article provides a clinical guide to managing CTEPH, from diagnostic workup and treatment options to our institutional experience in surgical decision making and optimizing CTEPH patient care through multidisciplinary collaboration. CTEPH is a condition in which PH arises from unresolved, precapillary thromboembolic disease.The natural history of acute pulmonary embolism (PE) usually results in the complete resolution of pulmonary thrombi with return to baseline pulmonary hemodynamics. In rare cases, chronic and incomplete resolution of these thrombi induces vascular remodeling and formation of fibrotic scar tissue in central pulmonary vessels.Although most of the clot burden typically resolves in these cases, a significant number of patients had more than 30% of segments still obstructed at 3 months, as shown on perfusion scan.Consequently, the persistence of these flow-limiting thrombi at various levels of the pulmonary vascular bed increases PVR, leading to increased pulmonary pressure that, left untreated, causes right heart failure and ultimately death.
## Epidemiology and risk factors
While many studies have examined the incidence and risk factors contributing to CTEPH, its true incidence is hard to pinpoint. Estimates vary from between 1% and 5% and depend on the patient population studied, study design, and CTEPH definition. Retrospective studies have estimated CTEPH incidence at approximately 1% among patients with PE.Prospective studies that examine these patients at annual follow-up estimate a CTEPH incidence between Group 4 pulmonary hypertension (PH) is defined by an elevated pulmonary arterial pressure of > 20 mm Hg at rest, a pulmonary vascular resistance (PVR) of > 3 Wood units, and a wedge pressure < 15 mm Hg. The unifying physiologic mechanism of the numerous causes of PH is increased PVR. Thus, the World Health Organization has classified PH into five groups by etiology.- Group 1: Pulmonary arterial hypertension (PAH)
- Group 2: PH due to left heart disease - Group 3: PH due to lung disease and/or hypoxia - Group 4: PH due to pulmonary artery obstructions - Group 5: PH due to unclear and/or multifactorial mechanisms The subject of this review article is chronic thromboembolic hypertension (CTEPH), and while group 4 PH does not strictly require the presence of a chronic clot, this is the primary etiology and manifestation of PH in group 4. Among the five groups of PH, only CTEPH is potentially curable, and it is particularly important to identify because its management strategy is unique compared with other forms of PH. Early identification and treatment of CTEPH is critical for successful patient outcomes. However, it is challenging to diagnose clinically because of its nonspecific symptoms with a wide differential. Of note, prospective trials are needed to determine whether group 4 PH from thromboembolic disease would benefit from specific management. 1 REVIEW JOURNAL.HOUSTONMETHODIST.ORG e19 3% and 5%.A recent meta-analysis of 16 studies estimates a prevalence of 0.5% in the general population and 3% in PE survivors.Given that the estimate of PE incidence in the United States is 300,000 per year, even the most conservative estimate projects at least 3,000 new cases of CTEPH per year. Even so, this projection does not account for 25% of CTEPH patients with no prior history of PE.One prospective Chinese study following 239 first-time PE patients over 5 years revealed a CTEPH incidence of 11.2% and 4.5% from 3 months to 3 years, suggesting that the actual incidence of PE may be much higher than previously reported.Together, these data show that there is a very large unmet need in CTEPH diagnosis and treatment. Since CTEPH often goes undetected and underdiagnosed, it is up to the astute clinician to perform a thorough evaluation to detect and treat the disease in a timely fashion.
CTEPH is a relatively rare outcome in PE, suggesting that other risk factors play a role in its development. For example, several coagulopathies and hypercoagulable states are associated with CTEPH, including elevated antiphospholipid antibodies and factor VIII.However, there is no difference in levels of antithrombin III, protein C and S, and factor V Leiden among the general population and patients with CTEPH.Other less common factors thought to be associated with CTEPH development include history of splenectomy, presence of ventriculoarterial shunts, chronic systemic inflammatory disorders, malignancy, and iatrogenic infections such as pacemaker leads or dialysis catheters.While numerous risk factors have been associated with CTEPH, none are sensitive or specific enough to independently warrant CTEPH evaluation in patients with a history of PE. One clinical prediction model suggests that unprovoked PE and a > 2 week delay in diagnosis from symptom onset increases the risk for developing CTEPH.Research is ongoing to determine a risk classification for CTEPH at the time of acute PE diagnosis.
## Clinical presentation
Making a diagnosis of CTEPH can be challenging because it has no specific signs or symptoms. Patients with CTEPH most commonly present with progressive dyspnea and fatigue, which can be mistaken for other diagnoses. Furthermore, a history of PE is often absent. As the disease progresses, the increasing PH eventually leads to nonspecific physical findings of right ventricular failure, including peripheral edema, jugular venous distension, ascites, and hepatomegaly. Due to its nonspecific clinical picture and lack of awareness of the disease, CTEPH is usually diagnosed late in the disease, when right heart failure has already occurred. Given this insidious natural history of disease, any patient who is functionally limited or symptomatic 3 to 6 months post-acute PE or who is undergoing workup for cause of PH should be referred for evaluation for CTEPH.
In a patient with documented PE and proper anticoagulation management, resolution of clots and symptomatic improvement usually occurs within the first 3 to 6 months.According to the European CTEPH registry, the median time from initial onset of symptoms to CTEPH diagnosis is 14 months, reflecting the insidious progressive nature of the disease.Thus, persistent symptoms with relevant history should prompt the clinician to pursue additional CTEPH workup.
## Team management
The diagnosis and management of CTEPH is a complex task requiring referral to a specialized center that can provide optimized care through multidisciplinary collaboration. The team should include all those who are actively involved in assessing the patient, formulating the workup plan, and performing and interpreting diagnostic tests. For longitudinal care, the same team should carry full responsibility for both medical and surgical management of CTEPH. At Houston Methodist Hospital, we have established a dedicated CTEPH team that includes cardiac surgery, pulmonology, interventional cardiology, anesthesia, critical care, radiology, and a nursing coordinator. Together, the team runs a CTEPH clinic once a month for new patient evaluation and established patient follow-up, and they meet the first Monday of each month to discuss all patient updates and test results. In this meeting, the team also plans and coordinates management of patients scheduled for surgical pulmonary thromboendarterectomy. We prioritize scheduling of these surgeries for the beginning of the week, which allows the entire CTEPH team to round daily on each patient in the early postoperative period, which is typically the most eventful time for this complex condition.
## Clinical workup
As previously mentioned, diagnosing CTEPH from a history and physical exam can be challenging due to its nonspecific clinical presentation and the coexistence of other comorbidities. CTEPH should remain on the differential for any patient with progressive dyspnea and other related cardiopulmonary symptoms with unknown origin, regardless of history of PE. Leading up to CTEPH diagnosis, a general workup for PH must be performed to exclude other causes of PH and to assess for other comorbidities.Whenever CTEPH is suspected, diagnostic tools should be considered, and early referral to an experienced center REVIEW JOURNAL.HOUSTONMETHODIST.ORG e20 should be initiated. Depending on the level of expertise and resources available, diagnostic workup may include echocardiography, lung ventilation-perfusion scintigraphy, computed tomography (CT), and right heart catheterization (RHC).
At our own institution, we prefer RHC, left heart catheterization (LHC), and pulmonary arteriogram. For more on this topic, see "Evaluation, Diagnosis, and Classification of Pulmonary Hypertension" by Beshay et al. in this issue.
## Imaging
## Echocardiography
Echocardiography is a readily available initial diagnostic tool. While it alone cannot diagnose CTEPH, it can indirectly measure pulmonary artery pressure (PAP). It is also useful in visualizing other cardiac signs of PH, including dilated right heart chambers, right ventricular systolic dysfunction, and paradoxical movement of the septum. Echocardiography is also essential to rule out other causes of PH, including intracardiac shunting and underlying left heart dysfunction. Importantly, contrast echocardiography, also known as a bubble study, must be ordered to assess intracardiac shunting. Finally, echocardiography is an important tool for monitoring improvement in patients with acute PE who develop PH and reduced right heart function.
## Ventilation-perfusion lung scan
A normal ventilation-perfusion (V/Q) scan is sufficient to exclude CTEPH, with a sensitivity of 90% to 100% and specificity of 94% to 100%.On V/Q scan, CTEPH presents as normal ventilation with a wedge-shaped perfusion deficit. This deficit is usually seen bilaterally, and unilateral absence of perfusion should increase suspicion of other conditions, such as malignancy, lung fibrosis, or vasculitis. Data from the international CTEPH registry indicates that 98% of CTEPH patients have abnormal perfusion scans while around 20% have abnormal ventilation scans.With improvements in CT technology, the advantage of V/Q over CT in detecting CTEPH has narrowed dramatically. However, V/Q scanning remains the preferred method for CTEPH screening as it requires less radiation and has less contrast-related adverse events. The main disadvantage of V/Q scanning is that it can underestimate the effect of an incomplete occlusion.
## Ct pulmonary angiogram
CT pulmonary angiogram (CTPA) is the gold standard of diagnostic imaging for acute PE. It is also quite reliable in detecting and evaluating the extent of thromboembolic disease for assessing operability in CTEPH. Even so, a negative CT scan does not entirely rule out the diagnosis of CTEPH. However, CTPA has excellent sensitivity and specificity for detecting thromboembolic disease at lobar (97-100% and 95-100%, respectively) and segmental levels (86-100% and 93-99%, respectively) 19-21 when read by an experienced cardiothoracic radiologist. CTEPH on CTPA has distinct features that can help discern the chronicity of thrombi compared to acute PE. The visualized filling defect is usually eccentric and web-like, which can occur after partial recanalization of the thrombosis.
Unlike acute PE, where the occluded vessel is expanded around the clot, the surrounding vessel in CTEPH is retracted and atrophied. Additionally, partially stenotic or occluded pulmonary vasculature may demonstrate adjacent poststenotic dilation while total occlusion results in "pouching defects," a convex ballooning where the distal vessel is cut off.Other radiologic signs of CTEPH on CTPA include healed lung infarct, cylindrical bronchiectasis, and calcified thrombi.Enlarged pulmonary arteries, mosaicism, and bronchial collaterals are other signs consistent with CTEPH and can help rule out mimickers of CTPEH (eg, angiosarcoma). The cardiothoracic radiology team at our institution has proposed a standardized protocol to combine perfusion scan with CT in the workup of all suspected CTEPH patients.
## Dual-energy computed tomography
Dual-energy computed tomography (DECT) is an emerging tool in the workup of CTEPH patients. Its basic principle is to capture iodine attenuation by simultaneous low (70-100 kV) and high (140-150 kV) tube voltage without additional radiation or contrast, and this data is used to generate a perfused blood volume (PBV) map. PBV maps provide both qualitative (morphologic) and quantitative (hemodynamic) assessment of disease burden, combining information that would usually be gleaned separately from a V/Q scan and RHC. DECT can be an excellent tool to corroborate V/Q scan findings and accurately diagnose CTEPH.The hemodynamic measurements derived from DECT also correlate well to those derived from RHC.Establishing a role for DECT in CTEPH workup is a work in progress. While it is more commonly used in the UK, most US clinicians still rely on angiography.
## Angiography and right heart catheterization
The role of angiography in CTEPH workup is threefold: digital subtraction angiography, RHC, and LHC to detect incidental coronary artery disease. Occasionally, LHC may also reveal coronary-to-bronchial collateralization, but this is an incidental finding and does not affect management.
Digital subtraction angiography used to be the gold standard to diagnose and assess operability in CTEPH but has fallen out of favor with advancements and improvement in CT imaging. Additionally, CTPA is superior in detailing distal vascular anatomy and for operative planning. While RHC directly measures hemodynamic status, this can now be measured noninvasively with DECT. Despite the diminishing role of angiography in diagnosing CTEPH, it remains an important tool for treatment of CTEPH with balloon pulmonary angioplasty.
## Surgical candidacy
## Pulmonary thromboendarterectomy
Riociguat is the only drug that has been approved by the US Food and Drug Administration to improve pulmonary blood flow and functional symptoms in CTEPH patientsand is an option for those who are not surgical candidates. However, the only curative treatment for CTEPH is pulmonary thromboendarterectomy (PTE), which surgically removes the chronic and organized thromboembolic lesions that are obstructing the pulmonary vasculature.
The survival benefit of PTE is estimated to be 90% and 75% at 3 and 10 years, respectively. With advances in surgical techniques, PTE can be performed more aggressively to tackle more complex and distal disease.Despite PTE being potentially curative, the operation is performed only half as often in the United States compared with Europe 28 and often depends on patient preference. PTE is often overlooked as a treatment option due to lack of referral to an expert center that can offer the surgery.
PTE is a technically demanding procedure, but surgical outcomes are still strongly influenced by operable patient selection. Age is not a strict contraindication to PTE, with patients in their 70s and 80s achieving results comparable to younger patients.The most important factors in assessing surgical risk are extent and location of chronic thromboembolic disease and degree of resultant pulmonary hemodynamic dysfunction. Other cardiac comorbidities are less impactful. Concomitant cardiac surgery for coronary artery and valvular heart disease can be performed safely without any added perioperative risk.The only true contraindications to PTE are patients with concomitant parenchymal lung disease who typically benefit little from surgery. The improvement in perfusion may not effectively improve symptoms if ventilation is still severely compromised. These patients also face significant risk for respiratory failure after surgery.The decision to operate is still a gray zone and relies mainly on CTEPH team experience. Experienced centers have proposed that the potential benefit of PTE for CTEPH be extended to all patients, including those with distal disease.Candidacy for PTE remains a subjective decision, but a second opinion from another expert center can be helpful for borderline cases or cases deemed to be inoperable.
The most important factor to consider for PTE is the correlation of patient symptoms to the severity of PH, right heart failure, and extent of disease. The severity of PH and right ventricular dysfunction carry a more complicated postoperative course but should not be a hard exclusion for offering PTE. Thromboembolic disease in the proximal pulmonary vasculature (main, lobar, or segmental arteries) is more amenable to surgical approach and results in better perioperative outcomes. On the other hand, removal of distal disease seems to yield the most symptomatic improvement. In the hands of the most experienced PTE surgeons, even distal disease can be approached surgically.Additional complicating factors for PTE include patients with prior splenectomy and those with ventriculoatrial shunts, venous access catheters, or pacemaker leads. Like any other surgical procedure, patients with severe pulmonary disease, terminal illness, advanced malignancy, and frail status may also be at increased risk of operative morbidity and mortality. In the decision to operate, the CTEPH team should always consider what meaningful benefit is offered for the patient in terms of quality of life and survival advantage.Surgical Technique PTE is a technically demanding and complicated procedure. Bilateral pulmonary arteries are exposed and accessed through median sternotomy. Back bleeding from bronchial collaterals can obscure visibility, thus necessitating cardiopulmonary bypass and circulatory arrest to provide a clear, bloodless field during dissection. Cardiopulmonary bypass (CPB) is achieved through central aortic cannulation and bicaval venous cannulation. The venous bicaval cannula is inserted into the body of the atrium and directed into both cava. This allows for improved retraction of the superior vena cava and exposure of the right pulmonary artery. The dissection is done under deep hypothermic circulatory arrest for 15 to 20 minutes per side to achieve satisfactory removal of distal disease. During rewarming, any additional indicated concomitant cardiac surgery can be performed. CPB is resumed in between while closing the right atriotomy before starting the left side. At our institution, we start with the right side, resume circulation, and finish on the left side. The key to successful PTE is to identify the correct plane between thromboembolism and vessel. Typically, it is the easiest plane to dissect and leaves behind a clear, white, smooth vessel.
Details of the operation, instruments, and surgical classification of disease are detailed by Video presentation and walk through of the operation from our institution can be found online (Video 1).
## Postoperative care and outcomes
PTE performed by an experienced team can be curative and fully restore hemodynamic function to baseline. Hemodynamic changes are usually immediate, but structural changes and remodeling are not. Data from the University of California San Diego and the International CTEPH registry show that PVR can drop up to 65% (PVR 700 to 800 dyn·s·cm -5 to 250 dyn·s·cm -5 ) and mean pulmonary arterial pressure (mPAP) to 43% (46 mm Hg to 26 mm Hg).With return to normal hemodynamics, there is marked improvement in right heart function and tricuspid regurgitation. Clinically, great improvement can be seen in New York Heart Association functional class, 6-minute walk distance, and quality of life.Operative mortality from the University of California San Diego, the highest volume center in the United States, was < 2%. In addition, mortality rate was improved with increased case volume, even when adjusted for acceptance of high-risk PTE candidates. Unsurprisingly, mortality rate was directly proportional to degree of preoperative PVR. Preoperative PVR > 1,000 dyn·s·cm -5 resulted in nearly four times the mortality rate than in patients with preoperative PVR < 1,000 dyn·s·cm -5 .Reperfusion pulmonary edema (RPE) and coronary steal are the most common post-PTE complications. Clinically, RPE is characterized by persistent hypoxemia after initial improvement. In RPE, removal of chronic thromboembolism restores perfusion to the pulmonary arteries, reducing pulmonary resistance. This sudden plunge in pulmonary resistance, when lower than the rest of the patent pulmonary vasculature, can cause relative hyperperfusion of the endarterectomized vessels and "steal" from the previously normal perfused lung, as blood takes the path of least resistance; this can be detected by V/Q scan. This redistribution of flow usually resolves over weeks to months,and thus treatment is supportive, with diuresis to reduce lung water and respiratory support. Severe RPE may be successfully treated with extracorporeal membrane oxygenation.Reperfusion resulting in pulmonary infiltrate distal to the site of PTE is associated with higher morbidity and mortality.The degree of arteriopathy on lung biopsy could be an important determinant in development of RPE post-PEA. Patients with residual PH have a tenfold mortality rate (10.3 vs 1%) compared to patients with no residual PH after PEA.Additionally, low postoperative cardiac index with high mean PAP and PVR and are associated with reduced survival after PTE.Persistent PH can be explained by incomplete PTE performed by an inexperienced surgeon or by residual distal small vessel disease, which cannot be resected surgically. Identifying the latter is important since residual macrovascular disease can be resolved with pulmonary angioplasty and microvascular disease can be improved with vasodilator therapy. Even with successful PTE, adequate anticoagulation is essential to prevent recurrent disease. Additional prospective studies are needed to determine the choice of anticoagulation for CTEPH; however, recent studies suggest that the choice of anticoagulant does not affect functional and hemodynamic post-PTE outcomes.Current guidelines advise CTEPH patients to receive anticoagulation for life. Vitamin K antagonists (coumadin) are most commonly used, but NOACs are increasingly being used with no safety issues. However, antiphospholipid syndrome is a contraindication to novel oral anticoagulants.At our institution, we generally do not place perioperative inferior vena cava filters by default for PTE unless otherwise indicated. For more on this topic, see "Novel Treatment Pathways in Pulmonary Arterial Hypertension" by Tonelli and Qaiser in this issue.
## Balloon pulmonary angioplasty
The role of balloon pulmonary angioplasty (BPA) in CTEPH treatment was first described in 2001 by Feinstein et al., who demonstrated improvement in mPAP and functional status post BPA in inoperable CTEPH patients. It fell from favor mainly due to high rates of post-BPA perfusion edema and high 30-day mortality rates.In contrast to BPA, patients undergoing PTE show immediate improvement in mPAP and PVR, whereas hemodynamic improvement from BPA may not be seen for 3 months. With improved technology and experience at high volume centers, BPA is once again becoming a promising option for CTEPH treatment.BPA utilization remains a viable option to be discussed within the CTEPH team but is mainly helpful for patients with distal disease, severe hemodynamic lesions, unfavorable risk/benefit ratio, recurrent disease, prohibitive risk for PTE, or residual lesions after PTE. Its role continues to be studied, and its exact indication for CTEPH treatment remains controversial.
Unlike conventional angioplasty, BPA uses an undersized balloon over a wire to break lesions in the pulmonary arteries without disturbing the medial layers of the vessel. Pulmonary arteries are dilated after BPA, but this is not an immediate effect. BPA is performed in a staged, stepwise manner starting with smaller balloons, and the response in pulmonary artery size is re-evaluated before moving to a larger balloon. The optimal pressure ratio across each lesion is > 0.8.BPA sessions are divided into 3 to 10 sessions, each a week apart. This approach helps limit radiation exposure, fluoroscopy time, and contrast time. Additionally, spacing Specimen from a 37-year-old male with a history significant for antiphospholipid syndrome presenting with complete occlusion of right pulmonary artery, which appears mushroom-like (asterisk). Left pulmonary endarterectomy specimen shows an ulcerated lesion (arrow). REVIEW JOURNAL.HOUSTONMETHODIST.ORG e25 out BPA sessions was thought to reduce the incidence of reperfusion edema, but some centers have shown that multiple lobes can be tackled in a single session because the mechanism of reperfusion edema may be different for BPA versus PEA.Taken together, BPA can provide hemodynamic advantages for some patients prior to surgical PTE by reducing operative risks. BPA can also be used sequentially to tackle residual lesions and PH after PEA.Thus, BPA can be a useful tool in hybrid surgical therapy for risky or refractory surgical candidates. It is important to note, however, that while these studies have demonstrated good outcomes for BPA, the outcomes have not yet been replicated in other countries. BPA should only be attempted at a CTEPH center of excellence with appropriate experience. PTE remains the gold standard for CTEPH treatment.
# Conclusion
There are still many unresolved questions in the understanding, diagnosis, and treatment of CTEPH. First, the impact of pathobiology and PE-related risk factors on CTEPH progression is not fully understood. Furthermore, the role of newer imaging modalities (magnetic resonance angiography, DECT, optical coherence tomography) in CTEPH evaluation is still being established. Finally, there are no objective operability criteria for PEA, and the role of bridging medical therapy, BPA, and combination treatment strategy requires additional studies. For more on the topic of medical management, see "Chronic Thromboembolic Pulmonary Hypertension Medical Management" by Safdar and Logue in this issue. CTEPH continues to be an underdiagnosed and undertreated sequelae of acute PE. Improving physician awareness about the incidence and treatability of this disease will help reduce rates of missed and delayed diagnosis. The most sensitive screening test is a V/Q scan, which should be followed up by RHC to assess hemodynamics. Finally, high-quality pulmonary angiography is required to confirm disease and assess operability.
All patients with CTEPH should be evaluated for potential PTE surgery by a specialized CTEPH team. All guidelines emphasize the importance of this evaluation because PTE surgery is the only potentially curative option for CTEPH in properly selected patients. Riociguat is the only medical treatment approved for CTEPH in patients deemed technically inoperable or with persistent PH after PTE. Finally, the role of BPA in treating CTEPH is still being defined. Although BPA shares some technical similarities to coronary and peripheral interventions, there are major differences in the type of specialist required, the nature of complications, and treatment goals. The role of BPA and its place amongst surgical and medical therapy is still being defined. This highlights the importance of an integrated, multidisciplinary approach with a dedicated CTEPH team to diagnose, treat, and follow up on CTEPH patients.
## Key points
- All patients with thromboembolic pulmonary hypertension (CTEPH) should be evaluated for operability by an experienced CTEPH team since this diagnosis is often missed or delayed. - Ventilation-perfusion scan is the best initial screening test, followed by right heart catheterization and pulmonary angiography, for diagnosis of CTEPH and operability assessment. - Although pulmonary thromboendarterectomy (PTE) is a technically challenging operation, even when performed and managed at an experienced CTEPH center, it can dramatically cure or improve pulmonary hypertension. - The role of balloon pulmonary angioplasty in CTEPH treatment is still being defined. - Increased awareness of CTEPH, training in PTE surgery, and establishment of CTEPH teams will continue to benefit patients and improve CTEPH outcomes. |
Beyond breast density: a review on the advancing role of parenchymal texture analysis in breast cancer risk assessment
Background: The assessment of a woman's risk for developing breast cancer has become increasingly important for establishing personalized screening recommendations and forming preventive strategies. Studies have consistently shown a strong relationship between breast cancer risk and mammographic parenchymal patterns, typically assessed by percent mammographic density. This paper will review the advancing role of mammographic texture analysis as a potential novel approach to characterize the breast parenchymal tissue to augment conventional density assessment in breast cancer risk estimation. Main text: The analysis of mammographic texture provides refined, localized descriptors of parenchymal tissue complexity. Currently, there is growing evidence in support of textural features having the potential to augment the typically dichotomized descriptors (dense or not dense) of area or volumetric measures of breast density in breast cancer risk assessment. Therefore, a substantial research effort has been devoted to automate mammographic texture analysis, with the aim of ultimately incorporating such quantitative measures into breast cancer risk assessment models. In this paper, we review current and emerging approaches in this field, summarizing key methodological details and related studies using novel computerized approaches. We also discuss research challenges for advancing the role of parenchymal texture analysis in breast cancer risk stratification and accelerating its clinical translation. Conclusions: The objective is to provide a comprehensive reference for researchers in the field of parenchymal pattern analysis in breast cancer risk assessment, while indicating key directions for future research.
# Background
The incidence and mortality rates of breast cancer remain extremely high despite advances in screening and treatment. In the USA, it is estimated that, in 2016, there will be 246,660 new cases of invasive breast cancer and 40,450 breast cancer deaths. Therefore, better strategies are urgently needed to identify women at high risk for developing breast cancer who could benefit the most from supplemental screening and preventive therapies [bib_ref] Breast cancer screening: time to target women at risk, Hall [/bib_ref] [bib_ref] Prevention of breast cancer in the context of a national breast screening..., Howell [/bib_ref].
Unfortunately, to date, the broadly available risk assessment models cannot identify high-risk women reliably within the general population. Current models predict either the risk of carrying a high-risk genetic mutation such as BRCA1/2 (e.g., Claus model, BOADICEA, and BRCAPRO) or the risk of developing breast cancer over time with or without such a mutation (e.g., Gail model, BOADICEA, Rosner-Colditz model, and Tyrer-Cuzick model) [bib_ref] Assessing women at high risk of breast cancer: a review of risk..., Amir [/bib_ref]. These models have only modest discriminatory capacity and continuing efforts are needed to improve these models at the individual level [bib_ref] Comparing breast cancer risk assessment models, Gail [/bib_ref]. In addition, genetic susceptibility models are only useful in the familial setting (where cancer pedigree history is known) and are not of relevance to the general population where the great majority of women have no relevant family history. Therefore, in striving to tailor breast cancer screening recommendations for the individual woman [bib_ref] Breast cancer screening in an era of personalized regimens: a conceptual model..., Onega [/bib_ref] it is crucial to develop more accurate risk assessment models that can be easily adopted in routine clinical practice.
While mammography remains the cornerstone of early breast cancer detection [bib_ref] Clinical diagnosis and management of breast cancer, Mcdonald [/bib_ref] , it also provides a readily accessible method to assess the distribution of fatty and dense, or fibroglandular (stromal and epithelial), tissues in the breast. In x-ray imaging, fatty tissue appears radiographically lucent, or darker, and dense tissue is radio-opaque, or brighter. Mammographic percent density (PD), a measure of the relative amount of fibroglandular tissue within the breast, has been shown to be related to screening sensitivity and specificity and has also been established as a strong independent risk factor for breast cancer [bib_ref] Imaging breast density: established and emerging modalities, Chen [/bib_ref] [bib_ref] Vision 20/20: Mammographic breast density and its clinical applications, Ng [/bib_ref] [bib_ref] Raised mammographic density: causative mechanisms and biological consequences, Sherratt [/bib_ref] [bib_ref] Breast density and parenchymal patterns as markers of breast cancer risk: a..., Mccormack [/bib_ref]. Studies have repeatedly shown significant associations with breast cancer risk for both qualitative and quantitative breast density measures and a potential to improve cancer risk assessment models [bib_ref] Mammographic density adds accuracy to both the Tyrer-Cuzick and Gail breast cancer..., Brentnall [/bib_ref] [bib_ref] Using clinical factors and mammographic breast density to estimate breast cancer risk:..., Tice [/bib_ref]. Recent legislation in several US states mandates notification of breast density, and substantial research continues to be devoted to accurate measurement of this key biomarker and to its incorporation into risk prediction models [bib_ref] Imaging breast density: established and emerging modalities, Chen [/bib_ref] [bib_ref] Utility of relative and absolute measures of mammographic density versus clinical risk..., Abdolell [/bib_ref].
Compared to the global image measure of breast density, parenchymal texture descriptors can provide more refined, localized descriptors to characterize the complexity as well as the morphological distribution of the breast parenchymal patterns. Breast density measures are generally dichotomous, or each area or voxel of breast density measured in the mammogram is compared to a threshold of "dense" or "not dense" without reflecting the broader range and spatial distribution of the various breast parenchymal elements. Parenchymal textural features have been proposed as not only imaging markers that could identify parenchymal changes associated with breast cancer development [bib_ref] Prediction of near-term breast cancer risk based on bilateral mammographic feature asymmetry, Tan [/bib_ref] [bib_ref] Association between changes in mammographic image features and risk for near-term breast..., Tan [/bib_ref] [bib_ref] Computerized detection of breast tissue asymmetry depicted on bilateral mammograms: a preliminary..., Wang [/bib_ref] , but also with subtypes and grading of subsequent breast malignancies [bib_ref] Risk factors and tumor characteristics of interval cancers by mammographic density, Holm [/bib_ref] [bib_ref] Early stage triple-negative breast cancer: imaging and clinical-pathologic factors associated with recurrence, Bae [/bib_ref] [bib_ref] Size, node status and grade of breast tumours: association with mammographic parenchymal..., Sala [/bib_ref]. In addition, there is growing evidence in support of textural features of the breast parenchyma reflecting inherent, independent, biologic risk factors associated with cancer development, and this may thus have the potential to augment breast density in assessing an individual woman's risk of developing cancer [bib_ref] Mammographic parenchymal patterns: a marker of breast cancer risk, Oza [/bib_ref] [bib_ref] Mammographic parenchymal patterns as an imaging marker of endogenous hormonal exposure: a..., Daye [/bib_ref] [bib_ref] Mammographic parenchymal patterns and breast cancer risk, Saftlas [/bib_ref]. Therefore, efforts to incorporate breast parenchymal texture analyses in breast cancer risk assessment have recently also gained substantial momentum.
This article reviews approaches to quantitate mammographic textural features and methods to incorporate these features into breast cancer risk assessment models, focusing primarily on novel computerized approaches. A systematic review of the literature in PubMed was performed to identify all original articles published up to April 2016 that evaluated computational measures of mammographic texture in breast cancer risk assessment. The following keywords were used in combination: "texture" or "parenchymal patterns" or "image features", "mammography" or "mammogram", and "breast cancer risk" or "mammographic risk". To broaden the search, the "related articles" function provided in PubMed was also used, and all articles and citations obtained were reviewed. The references from all the articles identified were also examined for further relevant studies. The last search was conducted on 29 April 2016. Studies not considered relevant to the scope of the review were excluded; other exclusion criteria included: study not published in the English language, full text not available, letter to the editor, and duplicate publication. In the rest of this manuscript, we summarize key methodological details and evaluation results from the 44 research papers identified by the search and discuss future challenges in this promising research field.
## Mammographic texture analysis using automatically extracted features
The value of characterizing the mammographic texture of the breast parenchyma in breast cancer risk estimation was originally demonstrated in the pioneering studies of Wolfe [bib_ref] Breast patterns as an index for developing breast cancer, Wolfe [/bib_ref] [bib_ref] Risk for breast cancer development determined by mammographic parenchymal pattern, Wolfe [/bib_ref] , Boyd et al. [bib_ref] Mammographic signs as risk factors for breast cancer, Boyd [/bib_ref] [bib_ref] Quantitative classification of mammographic densities and breast cancer risk: results from the..., Boyd [/bib_ref] [bib_ref] Relationship between mammographic and histological risk factors for breast cancer, Boyd [/bib_ref] , Gram et al. [bib_ref] The Tabar classification of mammographic parenchymal patterns, Gram [/bib_ref] , and Brisson et al. [bib_ref] Mammographic features of the breast and breast cancer risk, Brisson [/bib_ref] , proposing visually assessed, qualitative or quantitative classifications which were based on the extent and the characteristics of breast densities in a mammogram. These early approaches have been used by several groups, generally reporting elevated risks among women with more complex parenchymal tissue patterns [bib_ref] Breast cancer mammography patterns, Egan [/bib_ref] [bib_ref] Mammographic parenchymal patterns as a risk indicator for prevalent and incident cancer, Krook [/bib_ref] [bib_ref] Association between mammographic parenchymal pattern classification and incidence of breast cancer, Threatt [/bib_ref] [bib_ref] Mammographic parenchymal patterns: risk indicator for breast cancer?, Tabár [/bib_ref] [bib_ref] Mammographic parenchymal patterns and quantitative evaluation of mammographic densities: a case-control study, Wolfe [/bib_ref] [bib_ref] Mammographic parenchymal patterns as indicators of breast cancer risk, Saftlas [/bib_ref] [bib_ref] Reproducibility of mammographic classifications, Myers [/bib_ref] [bib_ref] Reproducibility of Wolfe's classification of mammographic parenchymal patterns, Toniolo [/bib_ref] [bib_ref] Mammographic parenchymal pattern and breast cancer risk: a critical appraisal of the..., Goodwin [/bib_ref] [bib_ref] The risk of developing breast cancer in relation to mammography findings, Witt [/bib_ref] [bib_ref] The risk of breast cancer associated with mammographic parenchymal patterns: a meta-analysis..., Warner [/bib_ref] [bib_ref] BI-RADS and Tabár based mammographic risk assessment, Muhimmah [/bib_ref] [bib_ref] Percentage density, Wolfe's and Tabar's mammographic patterns: agreement and association with risk..., Gram [/bib_ref]. Nevertheless, these studies also observed increased heterogeneity and low reproducibility in corresponding risk estimates due to subjectivity and inter-observer variation in visual appraisal of the mammogram [bib_ref] Breast cancer mammography patterns, Egan [/bib_ref] [bib_ref] Mammographic parenchymal patterns as a risk indicator for prevalent and incident cancer, Krook [/bib_ref] [bib_ref] Association between mammographic parenchymal pattern classification and incidence of breast cancer, Threatt [/bib_ref] [bib_ref] Mammographic parenchymal patterns: risk indicator for breast cancer?, Tabár [/bib_ref] [bib_ref] Mammographic parenchymal patterns and quantitative evaluation of mammographic densities: a case-control study, Wolfe [/bib_ref] [bib_ref] Mammographic parenchymal patterns as indicators of breast cancer risk, Saftlas [/bib_ref] [bib_ref] Reproducibility of mammographic classifications, Myers [/bib_ref] [bib_ref] Reproducibility of Wolfe's classification of mammographic parenchymal patterns, Toniolo [/bib_ref] [bib_ref] Mammographic parenchymal pattern and breast cancer risk: a critical appraisal of the..., Goodwin [/bib_ref] [bib_ref] The risk of developing breast cancer in relation to mammography findings, Witt [/bib_ref] [bib_ref] The risk of breast cancer associated with mammographic parenchymal patterns: a meta-analysis..., Warner [/bib_ref] [bib_ref] BI-RADS and Tabár based mammographic risk assessment, Muhimmah [/bib_ref] [bib_ref] Percentage density, Wolfe's and Tabar's mammographic patterns: agreement and association with risk..., Gram [/bib_ref]. By introducing computerized texture features to automate the characterization of breast parenchymal patterns, later studies addressed the limitations of visual classifications and re-established the potential of texture descriptors in breast cancer risk assessment [bib_ref] Automated analysis of mammographic densities, Byng [/bib_ref] [bib_ref] Characterisation of mammographic parenchymal pattern by fractal dimension, Caldwell [/bib_ref] [bib_ref] Mammographic texture analysis: an evaluation of risk for developing breast cancer, Magnin [/bib_ref] [bib_ref] Computer-assisted diagnosis: the classification of mammographic breast parenchymal patterns, Tahoces [/bib_ref] [bib_ref] Measuring image texture to separate "difficult" from "easy" mammograms, Taylor [/bib_ref]. Since then, this research field has continuously been evolving. A variety of quantitative methodologies have been developed, involving different techniques to sample the breast and multiple texture descriptors to characterize the texture properties of the sampled regions of interest (ROIs) from cranio-caudal (CC) and mediolateral-oblique (MLO) view mammograms [fig_ref] Table 1: Key studies in automated parenchymal texture analysis for breast cancer risk assessment... [/fig_ref].
In most studies, texture analysis has been performed within a single ROI in the breast [fig_ref] Table 1: Key studies in automated parenchymal texture analysis for breast cancer risk assessment... [/fig_ref]. This single ROI is usually placed in the retroareolar breast area, while, in some cases, it can be a larger region corresponding to the entire breast or to the largest rectangular box inscribed within the breast [fig_ref] Figure 1: Regions of interest [/fig_ref]. In an attempt to capture the granularity and heterogeneity of the parenchymal texture within the breast, more recent studies have estimated texture in multiple ROIs throughout the breast [fig_ref] Figure 1: Regions of interest [/fig_ref]. A lattice-based strategy Mammograms: F Digitized screen-film, D Full-field digital, CC cranio-caudal, MLO mediolateral-oblique; Dataset: A cancer cases ( P prior, unaffected, images, C images from the contralateral, unaffected, breast at the time of cancer diagnosis) or other high-risk population (i.e., BRCA1/2 carriers), B controls; Breast sampling: S1 retro-areolar region or the entire breast/dense tissue as a single region of interest (ROI), S2 multiple ROIs covering the entire breast; Types of texture features: T1 gray-level histogram, T2 co-occurrence, T3 run-length, T4 structural/pattern, T5 multi-resolution/spectral which splits the entire breast into multiple square patches was proposed by Zheng et al. [bib_ref] Parenchymal texture analysis in digital mammography: a fully-automated pipeline for breast cancer..., Zheng [/bib_ref] [bib_ref] Kontos D, editors. A fullyautomated software pipeline for integrating breast density and..., Zheng [/bib_ref] showing that, with respect to single ROI methodologies, this breast sampling technique may improve risk assessment, with performance being maximized when smaller patches (6.3 × 6.3 mm 2 ) are used. Multiple ROIs defined at various scales of breast tissue density were used by Sun et al. [bib_ref] Using multiscale texture and density features for near-term breast cancer risk analysis, Sun [/bib_ref] , where it was shown that fusing features from different density scales may prove to be another effective way to enhance the cancer prediction performance. The texture descriptors used in breast cancer risk assessment to date can be broadly classified into five feature groups [fig_ref] Table 2: Parenchymal texture descriptors for breast cancer risk assessment [/fig_ref] , each of which reveals different aspects of the mammographic texture [fig_ref] Figure 2: Characterization of parenchymal patterns using computerized texture analysis [/fig_ref] : 1) grey-level intensity/ histogram features [bib_ref] Computerized analysis of mammographic parenchymal patterns for assessing breast cancer risk: effect..., Li [/bib_ref] [bib_ref] Computerized analysis of digitized mammograms of BRCA1 and BRCA2 gene mutation carriers, Huo [/bib_ref] [bib_ref] Characterizing mammographic images by using generic texture features, Häberle [/bib_ref] ; 2) co-occurrence (Haralick/ Markovian) descriptors [bib_ref] Textural features for image classification. Systems, Man and Cybernetics, Haralick [/bib_ref] ; 3) run-length features [bib_ref] Texture analysis using gray level run lengths, Galloway [/bib_ref] [bib_ref] Use of gray value distribution of run lengths for texture analysis, Chu [/bib_ref] ; 4) structural/pattern measures [bib_ref] Automated analysis of mammographic densities, Byng [/bib_ref] [bib_ref] Automated analysis of mammographic densities and breast carcinoma risk, Byng [/bib_ref] [bib_ref] A novel and automatic mammographic texture resemblance marker is an independent risk..., Nielsen [/bib_ref] [bib_ref] On the orientation of mammographic structure, Reiser [/bib_ref] [bib_ref] Multiresolution local binary pattern texture analysis combined with variable selection for application..., Choi [/bib_ref] [bib_ref] Mammographic texture resemblance generalizes as an independent risk factor for breast cancer, Nielsen [/bib_ref] [bib_ref] Multiresolution gray-scale and rotation invariant texture classification with local binary patterns. Pattern..., Ojala [/bib_ref] ; and 5) multiresolution/spectral features [bib_ref] Using multiscale texture and density features for near-term breast cancer risk analysis, Sun [/bib_ref] [bib_ref] A novel and automatic mammographic texture resemblance marker is an independent risk..., Nielsen [/bib_ref] [bib_ref] Mammographic texture resemblance generalizes as an independent risk factor for breast cancer, Nielsen [/bib_ref] [bib_ref] Texture features from mammographic images and risk of breast cancer, Manduca [/bib_ref] [bib_ref] Multi-scale textural feature extraction and particle swarm optimization based model selection for..., Zyout [/bib_ref]. Gray-level intensity histogram features are common first-order statistics which describe the distribution of gray-level intensity within the breast tissue. The co-occurrence features also consider the spatial relationships of pixel intensities in different directions and are based on the gray-level cooccurrence matrix (GLCM) which encodes the relative Fourier power spectrum wavelet/Gabor Gaussian Kernels power-law spectrum frequency of neighboring intensity values. Run-length features capture the coarseness of texture in specified directions by measuring strings of consecutive pixels (i.e., runs) which have the same gray-level intensity along specific linear orientations. Fine textures tend to contain more short runs with similar gray-level intensities, while coarse textures have longer runs with different gray-level intensities. Structural features capture the architectural composition of the parenchyma by characterizing the tissue complexity, the directionality of flow-like structures in the breast, and intensity variations between central and neighboring pixels. Finally, multi-resolution/spectral features use spatial frequency transforms, such as Fourier, wavelet/Gabor, and the Power spectrum, to characterize intrinsic periodic texture structures that repeat over multiple scales.
## Towards a new breast cancer risk assessment paradigm based on mammographic texture descriptors
The proposed methodologies have been applied primarily to digitized film-screen mammograms and more recently on full-field digital mammograms. Texture descriptors have been evaluated in a few prospective and a larger number of retrospective case-control studies, where their discriminatory capacity in breast cancer prediction was typically assessed in terms of the area under the ROC curve (AUC) measuring their ability to distinguish between cancer cases and controls [fig_ref] Table 3: Breast cancer prediction capacity of automated characterization of the parenchymal patterns [/fig_ref]. The potential of mammographic texture in breast cancer risk assessment has also been investigated in studies with BRCA1/2 mutation carriers, where the AUC was evaluated in terms of the performance of the texture features in predicting a woman's risk of carrying this high-risk genetic mutation. Although hereditary breast cancers account for 5-10 % of incident breast cancers, women who inherit a mutated form of the BRCA1/2 gene have up to 87 % risk of developing breast cancer by the age of 70 years [bib_ref] Risks of cancer in BRCA1-mutation carriers, Ford [/bib_ref]. As such, and considering that mammographic PD has not been associated with BRCA1/2 mutation status [bib_ref] Mammographic density and breast cancer risk in BRCA1 and BRCA2 mutation carriers, Mitchell [/bib_ref] [bib_ref] Mammographic density does not differ between unaffected BRCA1/2 mutation carriers and women..., Gierach [/bib_ref] [bib_ref] Comparative analysis of image-based phenotypes of mammographic density and parenchymal patterns in..., Li [/bib_ref] , the ability of texture to identify potential BRCA1/2 carriers could have important value in risk stratification.
Associations of parenchymal texture with breast cancer in case-control studies
Byng et al. [bib_ref] Automated analysis of mammographic densities and breast carcinoma risk, Byng [/bib_ref] were the first to evaluate automatically calculated parenchymal texture descriptors directly as independent risk factors for breast cancer. The authors reported on data from a prospective case-control study using 354 incident cases diagnosed with histologically verified invasive breast carcinoma at least 1 year after their entry in the Canadian National Breast Screening Study, and 354 age-matched controls with at least 7 years of negative follow-up. Two grey-level intensity histogram texture features were estimated in screen-film mammograms; specifically, skewness averaged over individual 6.2 × 6.2 mm 2 patches in the breast and the fractal dimension estimated by considering the entire breast as a single ROI. For both features, the results showed moderate relative risk (RR) after adjustments for the effects of other risk factors, i.e., age at menarche, menopausal status, age at first time pregnancy, number of live births, family history of breast carcinoma, height, and weight (RR = 3.35 and RR = 3.35 for skewness and fractal dimension, respectively), while no additional contribution to risk was found in models that incorporated breast density measures. Similar conclusions were reported by Torres-Mejia et al. [bib_ref] Mammographic features and subsequent risk of breast cancer: a comparison of qualitative..., Torres-Mejia [/bib_ref] who estimated the same texture features and lacunarity, a measure of the degree of structural variation in image intensities within the breast, from prospectively collected data of 111 breast cancer cases and 3100 controls. The promising results of these early studies were followed by retrospective studies using more complex parenchymal texture descriptors [bib_ref] A novel and automatic mammographic texture resemblance marker is an independent risk..., Nielsen [/bib_ref] [bib_ref] Mammographic texture resemblance generalizes as an independent risk factor for breast cancer, Nielsen [/bib_ref] [bib_ref] Association of computerized mammographic parenchymal pattern measure with breast cancer risk: a..., Wei [/bib_ref] [bib_ref] An anatomically oriented breast coordinate system for mammogram analysis. Med Imaging, Brandt [/bib_ref] [bib_ref] Breast cancer risk analysis based on a novel segmentation framework for digital..., Chen [/bib_ref] [bib_ref] Comparison of mammographic parenchymal patterns of normal subjects and breast cancer patients, Wu [/bib_ref]. Wei et al. [bib_ref] Association of computerized mammographic parenchymal pattern measure with breast cancer risk: a..., Wei [/bib_ref] investigated the associations of breast cancer risk with run-length features, using two different implementations of run-length statistics: namely, the conventional approach for calculating the runs of pixels in one direction and an extension for the two-dimensional space [bib_ref] Comparison of mammographic parenchymal patterns of normal subjects and breast cancer patients, Wu [/bib_ref]. The authors found that the run-length measures calculated in the retroareolar region of the breast could serve as an additional risk factor that could not be explained by established breast cancer risk factors (i.e., age, BMI, family history of breast cancer, and number of previous biopsies) and breast density. A mammographic texture resemblance (MTR) marker based on multi-scale Gaussian features was proposed by Nielsen et al. [bib_ref] A novel and automatic mammographic texture resemblance marker is an independent risk..., Nielsen [/bib_ref]. This marker demonstrated high casecontrol discriminatory performance (AUC = 0.60-0.63) in two independent cohorts within the Dutch screening program [bib_ref] A novel and automatic mammographic texture resemblance marker is an independent risk..., Nielsen [/bib_ref] and the Mayo Mammography Health Study [bib_ref] A novel and automatic mammographic texture resemblance marker is an independent risk..., Nielsen [/bib_ref] [bib_ref] Mammographic texture resemblance generalizes as an independent risk factor for breast cancer, Nielsen [/bib_ref] , while performance was optimized by an aggregate marker combining MTR with density measures (AUC = 0.66). Gaussian derivative features at multiple scales were examined in a cross-sectional study with MLO-view film mammograms of 245 cancer cases and 245 controls from the Nijmegen risk-assessment study [bib_ref] An anatomically oriented breast coordinate system for mammogram analysis. Med Imaging, Brandt [/bib_ref]. In this work, derivative features were extracted using an anatomically oriented breast coordinate system and, compared to breast PD, demonstrated enhanced breast cancer prediction ability (AUC = 0.63 versus 0.56). Finally, a preliminary study on the dual-tree complex wavelet transform showed that wavelet features alone may have value in risk assessment [bib_ref] Breast cancer risk analysis based on a novel segmentation framework for digital..., Chen [/bib_ref].
In an attempt to identify highly discriminative texture descriptors from multiple feature groups and develop optimal combinations that maximize the case-control classification performance, research groups have also explored comprehensive sets of multi-parametric features reflecting various aspects of mammographic texture [bib_ref] Parenchymal texture analysis in digital mammography: a fully-automated pipeline for breast cancer..., Zheng [/bib_ref] [bib_ref] Using multiscale texture and density features for near-term breast cancer risk analysis, Sun [/bib_ref] [bib_ref] Characterizing mammographic images by using generic texture features, Häberle [/bib_ref] [bib_ref] Texture features from mammographic images and risk of breast cancer, Manduca [/bib_ref] [bib_ref] A new approach to develop computer-aided detection schemes of digital mammograms, Tan [/bib_ref] [bib_ref] Assessment of a four-view mammographic image feature based fusion model to predict..., Tan [/bib_ref]. Following an evaluation of more than 1000 co-occurrence, run-length, Laws, wavelet, and Fourier features in prior film mammograms of 246 cases and 522 controls, Manduca et al. [bib_ref] Texture features from mammographic images and risk of breast cancer, Manduca [/bib_ref] identified individual features which, when estimated at a coarse scale of a single ROI covering the entire breast, provided strong prediction for future breast cancer (odds ratio per 1 SD = 1.36-1.50, AUC = 0.61-0.62). In another retrospective study with 864 cancer cases and 418 controls, a three-step variable selection process [bib_ref] Characterizing mammographic images by using generic texture features, Häberle [/bib_ref]. When fed to multivariable logistic regression models adjusted for established breast cancer risk factors, these features demonstrated an AUC of 0.79 and an odds ratio of 2.88, while the additional inclusion of breast PD did not lead to any further performance improvement. Promising results from rich feature sets were also recently reported for digital mammograms. In a study with CC-view digital mammograms of 141 cases and 199 controls, a total number of 765 features were computed from ROIs defined at multiple density scales [bib_ref] Using multiscale texture and density features for near-term breast cancer risk analysis, Sun [/bib_ref]. From these features, an optimal set of 12 features was selected and yielded an AUC of 0.73 in separating the two study subgroups using a support vector machine classifier. Zheng et al. [bib_ref] Parenchymal texture analysis in digital mammography: a fully-automated pipeline for breast cancer..., Zheng [/bib_ref] retrospectively analyzed MLO-view digital mammograms of 106 cases and 318 controls, where 30 candidate features were extracted from multiple adjacent ROIs covering the entire parenchyma. The authors showed a collective discriminatory capacity of AUC = 0.85, with the fractal dimension, run-length, co-occurrence, and graylevel histogram features being more frequently selected than local binary and edge-enhancing index features in classification models. Furthermore, preliminary comparisons of the parenchymal patterns of estrogen-receptor positive (ER+) and negative (ER-) cancer cases measured with the same methodology [bib_ref] Parenchymal texture analysis in digital mammography: a fully-automated pipeline for breast cancer..., Zheng [/bib_ref] showed that subtypespecific breast cancer risk assessment based on mammographic textures may also be feasible [bib_ref] Breast density and parenchymal texture measures as potential risk factors for estrogen-receptor..., Keller [/bib_ref]. Finally, to assess the combined discriminatory ability of texture analysis in CC and MLO views, Tan et al. [bib_ref] Assessment of a four-view mammographic image feature based fusion model to predict..., Tan [/bib_ref] designed an artificial neural network model to fuse the features extracted from the two views. Following an evaluation of 79 features calculated from a single ROI, corresponding to either the entire breast or the dense tissue areas of the breast, on 430 cases and 440 controls, the highest performance of the proposed fusion model (AUC = 0.73) was obtained for the run-length features of the dense tissue. The authors also demonstrated a classification performance of similar magnitude (AUC = 0.71) for the same fusion model when applied on texture features of the entire breast for a larger dataset of 821 cancer cases and 1084 controls [bib_ref] A new approach to develop computer-aided detection schemes of digital mammograms, Tan [/bib_ref].
## Assessing the risk of carrying a high-risk gene mutation
The potential of mammographic texture in breast cancer risk assessment has also been demonstrated in studies with BRCA1/2 carriers, where texture features from a single 25.6 × 25.6 mm 2 retro-areolar ROI in CC mammographic views were shown to predict a woman's risk of carrying this high-risk genetic mutation. The first study addressing this topic extracted a comprehensive feature set of grey-level intensity statistics, co-occurrence features, and multi-scale texture measures based on Fourier transform analysis [bib_ref] Computerized analysis of mammographic parenchymal patterns for breast cancer risk assessment: feature..., Huo [/bib_ref]. In film mammograms of 30 BRCA1/2 carriers and 142 low-risk women, most features demonstrated high individual discriminatory capacity (AUC > 0.68), while the collective performance of the features that were deemed significant in multivariable models raised AUC values of 0.91 and 0.92 in the entire database and in an age-matched subgroup, respectively [bib_ref] Computerized analysis of digitized mammograms of BRCA1 and BRCA2 gene mutation carriers, Huo [/bib_ref]. Using the same image dataset, the authors also showed a promising individual classification performance for structural measures such as edge frequency (AUC = 0.78) [bib_ref] Computerized texture analysis of mammographic parenchymal patterns of digitized mammograms, Li [/bib_ref] , for different implementations of the fractal dimension (AUC = 0.74-0.93) [bib_ref] Computerized texture analysis of mammographic parenchymal patterns of digitized mammograms, Li [/bib_ref] [bib_ref] Fractal analysis of mammographic parenchymal patterns in breast cancer risk assessment, Li [/bib_ref] , and for power law spectral analysis (AUC = 0.90) [bib_ref] Power spectral analysis of mammographic parenchymal patterns for breast cancer risk assessment, Li [/bib_ref].
These results were recently replicated and validated in datasets with digital mammograms [bib_ref] Comparative analysis of image-based phenotypes of mammographic density and parenchymal patterns in..., Li [/bib_ref] [bib_ref] Computerized analysis of mammographic parenchymal patterns on a large clinical dataset of..., Li [/bib_ref] and larger numbers of high-risk women [bib_ref] Comparative analysis of image-based phenotypes of mammographic density and parenchymal patterns in..., Li [/bib_ref] [bib_ref] Computerized analysis of mammographic parenchymal patterns on a large clinical dataset of..., Li [/bib_ref] [bib_ref] Relationships between computer-extracted mammographic texture pattern features and BRCA1/2 mutation status: a..., Gierach [/bib_ref]. A similar design of texture analysis in the retroareolar breast region combined with a Bayesian Artificial Neural Network (BANN) for the classification task was applied to 1) film mammograms of 137 mutation carriers and 100 low-risk women [bib_ref] Relationships between computer-extracted mammographic texture pattern features and BRCA1/2 mutation status: a..., Gierach [/bib_ref] , and 2) digital mammograms of 53 mutation carriers, 75 women with unilateral cancer, and 328 low-risk women [bib_ref] Comparative analysis of image-based phenotypes of mammographic density and parenchymal patterns in..., Li [/bib_ref] [bib_ref] Computerized analysis of mammographic parenchymal patterns on a large clinical dataset of..., Li [/bib_ref]. The first analysis conferred a two-fold increase in the odds of predicting BRCA1/2 mutation status, and an AUC of 0.68 for texture features alone and 0.72 for the features plus breast PD [bib_ref] Relationships between computer-extracted mammographic texture pattern features and BRCA1/2 mutation status: a..., Gierach [/bib_ref]. In the second analysis, AUC values of 0.82 and 0.73 were obtained between mutation carriers and lowrisk women, and between unilateral cancer and low-risk women, respectively [bib_ref] Computerized analysis of mammographic parenchymal patterns on a large clinical dataset of..., Li [/bib_ref] ; these evaluation results were also retained in age-matched subgroup analysis (0.81 and 0.70, respectively) [bib_ref] Computerized analysis of mammographic parenchymal patterns on a large clinical dataset of..., Li [/bib_ref] without any significant improvement from the inclusion of breast PD (0.81 and 0.68, respectively) [bib_ref] Comparative analysis of image-based phenotypes of mammographic density and parenchymal patterns in..., Li [/bib_ref].
## Beyond established risk factors in breast cancer risk assessment
A comparison of the evaluation results published to date [fig_ref] Table 3: Breast cancer prediction capacity of automated characterization of the parenchymal patterns [/fig_ref] , focusing primarily on cross-validated experiments, suggests that more comprehensive sets of multi-parametric texture features [bib_ref] Parenchymal texture analysis in digital mammography: a fully-automated pipeline for breast cancer..., Zheng [/bib_ref] [bib_ref] Using multiscale texture and density features for near-term breast cancer risk analysis, Sun [/bib_ref] [bib_ref] Computerized analysis of mammographic parenchymal patterns for assessing breast cancer risk: effect..., Li [/bib_ref] [bib_ref] Characterizing mammographic images by using generic texture features, Häberle [/bib_ref] [bib_ref] Comparative analysis of image-based phenotypes of mammographic density and parenchymal patterns in..., Li [/bib_ref] [bib_ref] A new approach to develop computer-aided detection schemes of digital mammograms, Tan [/bib_ref] [bib_ref] Computerized analysis of mammographic parenchymal patterns on a large clinical dataset of..., Li [/bib_ref] may be more effective in predicting breast cancer than a single feature group. However, the literature lacks extensive comparative studies on the same datasets and generalized conclusions should, therefore, be limited. While the implementation of texture analysis, including both the location and size of ROIs [bib_ref] Parenchymal texture analysis in digital mammography: a fully-automated pipeline for breast cancer..., Zheng [/bib_ref] [bib_ref] Kontos D, editors. A fullyautomated software pipeline for integrating breast density and..., Zheng [/bib_ref] [bib_ref] Computerized analysis of mammographic parenchymal patterns for assessing breast cancer risk: effect..., Li [/bib_ref] and the specific texture measures, appears to have an effect on texture classification performance, all studies have consistently shown the highly promising, independent role of automated texture analysis in breast cancer risk assessment. Specifically, parenchymal texture descriptors have demonstrated a strong cross-validated ability in predicting both risk for breast cancer (0.58 ≤ AUC ≤ 0.85) and BRCA1/2 mutation status (0.53 ≤ AUC ≤ 0.93). Moreover, texture performance has been shown to be either comparable or significantly higher than the performance of breast PD (0.51 ≤ AUC ≤ 0.62 and 0.53 ≤ AUC ≤ 0.59, respectively), as reported in studies where texture and density measures were comparatively evaluated on the same datasets [bib_ref] Parenchymal texture analysis in digital mammography: a fully-automated pipeline for breast cancer..., Zheng [/bib_ref] [bib_ref] Characterizing mammographic images by using generic texture features, Häberle [/bib_ref] [bib_ref] A novel and automatic mammographic texture resemblance marker is an independent risk..., Nielsen [/bib_ref] [bib_ref] Texture features from mammographic images and risk of breast cancer, Manduca [/bib_ref] [bib_ref] Comparative analysis of image-based phenotypes of mammographic density and parenchymal patterns in..., Li [/bib_ref] [bib_ref] Association of computerized mammographic parenchymal pattern measure with breast cancer risk: a..., Wei [/bib_ref] [bib_ref] An anatomically oriented breast coordinate system for mammogram analysis. Med Imaging, Brandt [/bib_ref] [bib_ref] Breast cancer risk analysis based on a novel segmentation framework for digital..., Chen [/bib_ref] [bib_ref] Relationships between computer-extracted mammographic texture pattern features and BRCA1/2 mutation status: a..., Gierach [/bib_ref].
In addition, a number of related findings suggest that texture analysis is able to provide complementary information about a woman's risk of developing breast cancer which cannot be captured by breast PD and other established risk factors. Texture descriptors have been weakly or moderately correlated with breast PD [bib_ref] Computerized analysis of digitized mammograms of BRCA1 and BRCA2 gene mutation carriers, Huo [/bib_ref] [bib_ref] A novel and automatic mammographic texture resemblance marker is an independent risk..., Nielsen [/bib_ref] [bib_ref] Mammographic texture resemblance generalizes as an independent risk factor for breast cancer, Nielsen [/bib_ref] [bib_ref] Comparative analysis of image-based phenotypes of mammographic density and parenchymal patterns in..., Li [/bib_ref] [bib_ref] Mammographic features and subsequent risk of breast cancer: a comparison of qualitative..., Torres-Mejia [/bib_ref] [bib_ref] Association of computerized mammographic parenchymal pattern measure with breast cancer risk: a..., Wei [/bib_ref] [bib_ref] Relationships between computer-extracted mammographic texture pattern features and BRCA1/2 mutation status: a..., Gierach [/bib_ref] [bib_ref] Parenchymal texture analysis in digital breast tomosynthesis for breast cancer risk estimation:..., Kontos [/bib_ref] [bib_ref] Digital breast tomosynthesis parenchymal texture analysis: comparison with digital mammography and implications..., Kontos [/bib_ref] , and weakly correlated with risk factors as reflected in the Gail and Claus risk scores [bib_ref] Computerized analysis of mammographic parenchymal patterns for breast cancer risk assessment: feature..., Huo [/bib_ref] [bib_ref] Parenchymal texture analysis in digital breast tomosynthesis for breast cancer risk estimation:..., Kontos [/bib_ref]. In addition, texture descriptors deemed as strong predictors of breast cancer retained significance when breast PD, age, BMI, family history of breast cancer, parity, age at first term pregnancy, number of previous breast biopsies, menopause, and hormonal use, all shown to be associated with breast cancer risk, were simultaneously considered in classification models [fig_ref] Table 3: Breast cancer prediction capacity of automated characterization of the parenchymal patterns [/fig_ref]. Finally, with age-matched datasets or model adjustments for age, most studies evaluating the capacity of parenchymal texture features in risk assessment have ruled out possible confounding due to differences in age, a major breast cancer risk factor, thereby showing a strong potential for computerized texture descriptors in augmenting breast cancer risk assessment.
## Future directions
Moving forward, experiments evaluating the relative performance of different implementations of texture analysis, using the same evaluation methodology (i.e., dataset and classification model), are necessary to develop more robust and reproducible quantitative mammographic phenotypes of breast cancer risk. Future studies to test the incremental value added by computerized textural measures in predicting breast cancer will require: (a) the design of large age-matched datasets; (b) the selection of an effective classification model, where different previously used models [fig_ref] Table 3: Breast cancer prediction capacity of automated characterization of the parenchymal patterns [/fig_ref] could be comparatively examined; (c) model adjustments to rule out possible confounding due to differences in major risk factors; and (d) validation of the classification performance in independent datasets.
In an attempt to add an anatomical meaning in texture analysis which may also give additional discrimination power to feature classification, increasing attention is currently given to the incorporation of breast anatomy in texture analysis pipelines. Brandt et al. [bib_ref] An anatomically oriented breast coordinate system for mammogram analysis. Med Imaging, Brandt [/bib_ref] first introduced an anatomically oriented breast coordinate system which allows for anatomical correspondences across mammograms of the same woman or different women. In preliminary analyses using the proposed coordinate system, the authors have demonstrated that anatomydriven Gaussian derivative features are able to (a) effectively separate cancer cases and controls [bib_ref] An anatomically oriented breast coordinate system for mammogram analysis. Med Imaging, Brandt [/bib_ref] , (b) quantify the effect of hormone replacement therapy as a change in the breast parenchymal patterns [bib_ref] Anisotropic diffusion tensor applied to temporal mammograms: an application to breast cancer..., Karemore [/bib_ref] , and (c) demonstrate specific regions of the breast parenchyma where breast cancer risk is mainly expressed [bib_ref] A method to determine the mammographic regions that show early changes due..., Karemore [/bib_ref]. More recently, Gastounioti et al. [bib_ref] Towards a breast-anatomy-weighted parenchymal texture signature for breast cancer risk assessment, Gastounioti [/bib_ref] [bib_ref] Associations of dense and fatty breast-tissue heterogeneity with breast cancer risk: Preliminary..., Gastounioti [/bib_ref] showed that the discriminatory capacity of texture descriptors is further enhanced by an anatomy-driven polar grid for anatomical breast sampling and a breast-anatomy-weighted texture signature which considers the spatial position and the underlying tissue composition of individual ROIs to summarize the parenchymal texture properties of the breast.
Another emerging technology is deep learning [bib_ref] Deep learning, Lecun [/bib_ref] , which may prove a valuable addition in texture analysis for breast cancer risk assessment [bib_ref] Unsupervised deep learning applied to breast density segmentation and mammographic risk scoring, Kallenberg [/bib_ref] [bib_ref] Breast tissue segmentation and mammographic risk scoring using deep learning, Petersen [/bib_ref] [bib_ref] An initial investigation on developing a new method to predict short-term breast..., Qiu [/bib_ref]. Deep learning involves automated learning, from raw image data, of hierarchical representations useful for pattern detection and classification, in a supervised mode via neural networks with multiple hidden layers or in an unsupervised mode via autoencoders. The few available studies which have applied deep learning in the particular field show a promising role in risk scoring (AUC = 0.61-0.65) [bib_ref] Unsupervised deep learning applied to breast density segmentation and mammographic risk scoring, Kallenberg [/bib_ref] [bib_ref] An initial investigation on developing a new method to predict short-term breast..., Qiu [/bib_ref]. Further, preliminary comparisons against two previously presented methodologies with handcrafted texture features [bib_ref] Characterizing mammographic images by using generic texture features, Häberle [/bib_ref] [bib_ref] Mammographic texture resemblance generalizes as an independent risk factor for breast cancer, Nielsen [/bib_ref] suggest that it may be better to "let the data speak" instead of modeling prior assumptions [bib_ref] Unsupervised deep learning applied to breast density segmentation and mammographic risk scoring, Kallenberg [/bib_ref]. Additional experimentation with deep learning, as well as future comparisons with the state-of-the-art texture analysis techniques, is warranted to better explore the potential of this novel technology.
Digital breast tomosynthesis (DBT), an emerging x-ray technology [bib_ref] Breast cancer screening using tomosynthesis in combination with digital mammography, Friedewald [/bib_ref] in which quasi three-dimensional (3D) images are reconstructed from a limited number of lowdose x-ray source projections [bib_ref] A review of breast tomosynthesis. Part II. Image reconstruction, processing and analysis,..., Sechopoulos [/bib_ref] [bib_ref] A review of breast tomosynthesis. Part I. The image acquisition process, Sechopoulos [/bib_ref] , is increasingly being implemented clinically due to improvements in sensitivity and specificity compared to imaging with digital mammography alone [bib_ref] Digital breast tomosynthesis: a brave new world of mammography screening, Houssami [/bib_ref]. By imaging the breast in 3D, DBT alleviates the effect of tissue superimposition, offering superior tissue visualization, which in turn may allow for better characterization of the breast parenchyma compared to two-dimensional mammography [bib_ref] Parenchymal texture analysis in digital breast tomosynthesis for breast cancer risk estimation:..., Kontos [/bib_ref] [bib_ref] Digital breast tomosynthesis parenchymal texture analysis: comparison with digital mammography and implications..., Kontos [/bib_ref]. The extension of the parenchymal texture analysis descriptors for volumetric texture analysis in DBT is, therefore, an important future challenge towards developing superior texture features which can optimize image-driven breast cancer risk assessment.
Another challenging future step which would establish the predictive value of texture analysis is the validation of parenchymal texture measures in prospectively collected data. Large-scale studies involving multiple screening centers, imaging machines, and image acquisition settings are also of major importance towards validating their predictive capacity and robustness to heterogeneous image data [bib_ref] Parenchymal texture analysis in digital mammography: robust texture feature identification and equivalence..., Keller [/bib_ref]. Furthermore, the literature lacks large-scale longitudinal studies monitoring longitudinal changes in automated parenchymal texture descriptors over successive mammograms, which could elucidate the mechanisms of breast cancer development [bib_ref] Raised mammographic density: causative mechanisms and biological consequences, Sherratt [/bib_ref] and the causal relations between the texture risk scoring and breast cancer [bib_ref] Mammographic tissue, breast cancer risk, serial image analysis, and digital mammography. Part..., Heine [/bib_ref] [bib_ref] Mammographic tissue, breast cancer risk, serial image analysis, and digital mammography: Part..., Heine [/bib_ref]. Finally, crucial questions to be addressed in such rich datasets are the causes of inter-woman variation in mammographic parenchymal patterns [bib_ref] Pilot study demonstrating potential association between breast cancer image-based risk phenotypes and..., Li [/bib_ref] [bib_ref] Mammary gland architecture as a determining factor in the susceptibility of the..., Russo [/bib_ref] and in the relation of texture risk markers to the subsequent location and grading of tumors, disease mortality, and treatment effects [bib_ref] Risk factors and tumor characteristics of interval cancers by mammographic density, Holm [/bib_ref] [bib_ref] Early stage triple-negative breast cancer: imaging and clinical-pathologic factors associated with recurrence, Bae [/bib_ref].
The valuable risk markers provided by parenchymal texture analysis could also leverage the relatively new, yet promising, paradigms of radiomics [bib_ref] Radiomics: extracting more information from medical images using advanced feature analysis, Lambin [/bib_ref] and radiogenomics [bib_ref] Behind the numbers: decoding molecular phenotypes with radiogenomics-guiding principles and technical considerations, Kuo [/bib_ref] for breast cancer, aiming to convert breast images into comprehensive measurable data and to delve into the interaction between these data and genetic variants. These novel approaches may pave the way to revealing correlations with the genomic diversity present in breast cancer, understanding how biological processes are reflected in quantitative breast imaging phenotypes, and defining novel clinical biomarkers or biological surrogates [bib_ref] Radiogenomic analysis of breast cancer using MRI: a preliminary study to define..., Yamamoto [/bib_ref] [bib_ref] Radiogenomic analysis of breast cancer: luminal B molecular subtype is associated with..., Mazurowski [/bib_ref] [bib_ref] Quantitative breast MRI radiomics for cancer risk assessment and the monitoring of..., Mendel [/bib_ref] [bib_ref] Prediction of clinical phenotypes in invasive breast carcinomas from the integration of..., Guo [/bib_ref] , thus improving personalized breast cancer screening, monitoring, and treatment selection.
# Conclusions
Automated breast parenchymal texture analysis has the potential to elucidate imaging phenotypes of breast cancer risk, which is valuable in accelerating the translation of individualized risk stratification into routine breast cancer screening and prevention strategies. Future work addressing technical challenges in this field and large prospective studies are expected to further enhance and establish the predictive value of parenchymal texture measures for inclusion in breast cancer risk assessment models in clinical practice. Authors' contributions AG performed the literature research and drafted the manuscript. EFC participated in the design of the review and drafting of the manuscript. DK coordinated the review, participated in the design of the review, and drafting of the manuscript. All authors have read and approved the final manuscript and agree to be accountable for all aspects of the work.
[fig] Figure 1: Regions of interest (ROIs) used in texture analysis. a single ROIs selected in the retro-areolar breast area, b the entire breast and the largest rectangular box inscribed within the breast, studied as single ROIs, c multiple ROIs at multiple scales of density, and d multiple ROIs defined by a lattice covering the entire breast [/fig]
[fig] Figure 2: Characterization of parenchymal patterns using computerized texture analysis. Examples of feature maps showing the distribution of texture values in the breast, generated by the application of the lattice-based strategy of Zheng et al. [51] to an MLO-view full-field digital mammogram. (a) Grey-level histogram, (b) Co-occurrence, (c) Run-length, (d) Structural, and (e) Multi-resolution [/fig]
[fig] Abbreviations 3D: Three-dimensional; AUC: Area under the ROC curve; CC: Cranio-caudal; DBT: Digital breast tomosynthesis; ER-: Estrogen receptor negative; ER+: Estrogen receptor positive; GLCM: Grey-level co-occurrence matrix; MLO: Medio-lateral oblique; MTR: Mammographic texture resemblance; PD: Percent density; ROI: Region of interest; RR: Relative risk Funding The authors wish to acknowledge support by the National Cancer Institute at the National Institutes of Health via a research grant award (5R01CA161749-04), the Population-based Research Optimizing Screening through Personalized Regimens (PROSPR) Network (U54CA163313), and a Resource-Related Research Project-Cooperative Agreement (1U24CA189523). [/fig]
[table] Table 1: Key studies in automated parenchymal texture analysis for breast cancer risk assessment TheTable describes the image data used in each study, including type of mammograms and dataset size, as well as methodological details for the computerized texture analysis, the technique of breast sampling, and algorithm implementation of texture features IMPRS International Max Planck Research School for Optics and Imaging, IPOFG Instituto Português de Oncologia Francisco Gentil, LSHTM London School of Hygiene and Tropical Medicine, Moffitt Moffitt Cancer Center and Research Institute, NCI-NIH National Cancer Institute, National Institutes of Health, RadboudUMC Radboud University Nijmegen Medical Centre, TTUHS Texas Tech University Health Sciences, UCL University College London, UCLA University of California at Los Angeles, UNAM Universidad Nacional Autónoma de México, USUHS Uniformed Services University of the Health Sciences, WRNMC Walter Reed National Military Medical Center [/table]
[table] Table 2: Parenchymal texture descriptors for breast cancer risk assessment; texture descriptors which have been examined in association with breast cancer risk, classified to five feature groups [/table]
[table] Table 3: Breast cancer prediction capacity of automated characterization of the parenchymal patterns [/table]
|
Variation in KCNQ1 is associated with therapeutic response to sulphonylureas
Background:We aimed to analyse quantitative effects of treatment with sulphonylurea in addition to metformin on parameters of glycemic control in relation to KCNQ1 genotypes, and to identify factors predictive for the response to sulphonylurea treatment.Material/Methods: Effect of 6-month sulphonylurea therapy in addition to metformin on glycemic control according to KCNQ1 genotypes was evaluated in 87 patients with type 2 diabetes who failed to achieve glycemic control on metformin monotherapy. KCNQ1 rs163184 (T>G) polymorphism was determined by real-time PCR with melting analysis of unlabeled probe.Results:The reduction in fasting plasma glucose (ΔFPG) after 6-month sulphonylurea therapy significantly differed among 3 KCNQ1 genotype groups (ANOVA, p=0.017). In a recessive genetic model, carriers of the T-allele (TT+TG) achieved significantly lower FPG levels in comparison with patients with the GG genotype (6.95±0.13 vs. 7.50±0.21 mmol/L, p=0.033). Consequently, ΔFPG was significantly higher in the TT+TG group compared to the GG group (1.58±0.13 vs. 1.04±0.18 mmol/L, p=0.016). In multiple linear regression analysis KCNQ1 genotype (p=0.016) and baseline FPG (p<0.001) were the only significant independent predictors of ΔFPG (R 2 =0.48).Conclusions:Our results suggest that the magnitude of FPG reduction after 6-month sulphonylurea treatment in addition to metformin in patients with type 2 diabetes is related to the variation in KCNQ1. The FPG response to sulphonylureas was significantly lower in carriers of the risk GG genotype. key words: pharmacogenetics - sulphonylureas - KCNQ1 - glycemic control - type 2 diabetes Full-text PDF:
# Background
Type 2 diabetes is a disease with significant genetic predisposition. In recent years almost 40 genes associated with type 2 diabetes have been identified by genome-wide association studies [bib_ref] Genomics, type 2 diabetes, and obesity, Mccarthy [/bib_ref]. Among them, 2 independent genomewide studies in East Asian populations reported that several polymorphisms of potassium voltage-gated channel KQTlike subfamily member 1 (KCNQ1) gene were consistently associated with type 2 diabetes [bib_ref] SNPs in KCNQ1 are associated with susceptibility to type 2 diabetes in..., Unoki [/bib_ref] [bib_ref] Variants in KCNQ1 are associated with susceptibility to type 2 diabetes mellitus, Yasuda [/bib_ref]. This association was subsequently replicated in several population cohorts of Asian and European ancestry [bib_ref] Common variants in KCNQ1 are associated with type 2 diabetes and impaired..., Qi [/bib_ref] [bib_ref] Variants in KCNQ1, AP3S1, MAN2A1, and ALDH7A1 and the risk of type..., Zhou [/bib_ref] [bib_ref] Variations in KCNQ1 are associated with type 2 diabetes and beta cell..., Hu [/bib_ref] [bib_ref] A variant in the KCNQ1 gene predicts future type 2 diabetes and..., Jonsson [/bib_ref].
KCNQ1 gene encodes the pore-forming subunit of a voltage-gated K + channel (KvLQT1) that plays a key role in the repolarization of the cardiac action potential, as well as water and salt transport in epithelial tissues [bib_ref] K(V)LQT1 and IsK (mink) proteins associate to form the I(Ks) cardiac potassium..., Barhanin [/bib_ref] [bib_ref] Differential expression of KvLQT1 and its regulator IsK in mouse epithelia, Demolombe [/bib_ref]. Both human and animal studies suggest that mutations in KCNQ1 can result in the K + channel dysfunction and cause the hereditary long QT syndrome and familial atrial fibrilation [bib_ref] Positional cloning of a novel potassium channel gene: KVLQT1 mutations cause cardiac..., Wang [/bib_ref] [bib_ref] KCNQ1 gain-of-function mutation in familial atrial fibrillation, Chen [/bib_ref]. KCNQ1 is also expressed in pancreatic islets and insulin-secreting cell lines [bib_ref] SNPs in KCNQ1 are associated with susceptibility to type 2 diabetes in..., Unoki [/bib_ref] [bib_ref] Variants in KCNQ1 are associated with susceptibility to type 2 diabetes mellitus, Yasuda [/bib_ref] [bib_ref] Effects of I (Ks) channel inhibitors in insulin-secreting INS-1 cells, Ullrich [/bib_ref]. Several studies indicated that various KCNQ1 polymorphisms were related either to impaired insulin secretion [bib_ref] Variations in KCNQ1 are associated with type 2 diabetes and beta cell..., Hu [/bib_ref] [bib_ref] A variant in the KCNQ1 gene predicts future type 2 diabetes and..., Jonsson [/bib_ref] [bib_ref] Genetic variation in KCNQ1 associates with fasting glucose and b-cell function. A..., Tan [/bib_ref] [bib_ref] The type 2 diabetes associated minor allele of rs2237895 KCNQ1 associates with..., Holmkvist [/bib_ref] [bib_ref] Association of type 2 diabetes candidate polymorphism in KCNQ1 with incretin and..., Müssig [/bib_ref] or to impaired incretin secretion [bib_ref] Association of type 2 diabetes candidate polymorphism in KCNQ1 with incretin and..., Müssig [/bib_ref].
The recommended initial therapeutic interventions in type 2 diabetes include lifestyle changes and pharmacotherapy with metformin [bib_ref] Management of hyperglycaemia in type 2 diabetes: a consensus algorithm for initiation..., Nathan [/bib_ref]. In patients with metformin monotherapy failure, sulphonylureas are frequently used as a secondline treatment. Sulphonylureas act as insulin secretagogues through the stimulation of insulin secretion via the sulfonylurea receptor 1 in pancreatic b-cells [bib_ref] Sulphonylurea action revisited: the post cloning era, Gribble [/bib_ref]. Considerable interindividual variation in the hypoglycaemic response to sulphonylureas likely reflects variations in the b-cell secretory reserve, and may relate to variations in genes involved in regulating b-cell function [bib_ref] Association of sulfonylurea receptor 1 genotype with therapeutic response to gliclazide in..., Zhang [/bib_ref] [bib_ref] Ser1369Ala variant in sulfonylurea receptor gene ABCC8 is associated with antidiabetic efficacy..., Feng [/bib_ref] [bib_ref] Variation in TCF7L2 influences therapeutic response to sulfonylureas, Pearson [/bib_ref] [bib_ref] Effect of sulphonylurea treatment on glycaemic control is related to TCF7L2 genotype..., Schroner [/bib_ref].
Since genetic variation in KCNQ1 is associated with fasting glucose and b-cell function [bib_ref] Genetic variation in KCNQ1 associates with fasting glucose and b-cell function. A..., Tan [/bib_ref] , we hypothesised that the magnitude of sulphonylurea treatment effect might be related to the KCNQ1 genotype. Therefore, the aim of the present pharmacogenetic pilot study was to analyse quantitative effects of treatment with sulphonylurea in addition to metformin on parameters of glycaemic control with respect to KCNQ1 genotypes in patients with type 2 diabetes.
# Material and methods
## Patients
Patients with type 2 diabetes diagnosed according to the American Diabetes Association criteriarecruited from 3 out-patient clinics participated in the study, which was conducted in a university hospital setting. Patients were eligible for the study if they were on previous metformin monotherapy for at least 6 months, and failed to maintain HbA 1c <7.0% on maximal tolerated doses of metformin at 2 consecutive visits within a 3-month period. Inclusion criteria were HbA 1c of 7.0-11.0%, fasting glycaemia of 6-15 mmol/L, age 35-70 years, and BMI 20-35 kg/m 2 . Patients with malignancies, hypothyroidism, chronic renal failure, severe liver disease, systemic inflammatory disease, and receiving corticosteroid treatment were excluded. The study was approved by the L. Pasteur University Hospital Review Board, and all subjects gave written consent to participate in the study.
Anthropometric data and diabetes duration were recorded at the baseline visit. Blood samples were taken for biochemical measurements of FPG, HbA 1c , lipid levels and genotyping. Sulphonylurea treatment was started with gliclazide, glimepiride, glipizide or glibenclamide. Sulphonylurea dose could have been adjusted after 3 months based on blood glucose self-monitoring results. Measurements of body weight, FPG, HbA 1c and serum lipids were repeated after 6 months following initiation of sulphonylurea therapy.
## Biochemical analyses
In all patients, peripheral venous blood samples were collected between 7-8 a.m. following an overnight 12-hour fast. Glucose was measured by glucose oxidase method, and cholesterol, triglycerides, low-density lipoprotein (LDL) and high-density lipoprotein (HDL) cholesterol were measured by routine enzymatic methods (Pliva-Lachema, Czech Republic) on a Beckman autoanalyser. HbA 1c was measured using an immunoturbidimetric method (Roche Diagnostica, France).
## Genotyping of kcnq1 rs163184 (t>g)
Genomic DNA was extracted using a Wizard Genomic DNA purification kit (Promega Corp., Wisconsin, USA). PCR was performed in 10 µl of reaction volume on a LightScanner 32 instrument (Idaho Technology Inc., Salt Lake City, USA) at asymmetric primer ratio (1:10). Master mix was composed of 1x LCGreen Plus+ PCR conditions were the following: initial denaturation at 95°C for 5 min, 60 cycles at 95°C for 10 s, 56°C for 15 s and 72°C for 15 s. Amplification was performed at the thermal transition rate of 10°C/s for all steps, and was immediately followed by melting analysis with a denaturation at 95°C for 30 s and renaturation at 45°C for 1 minute. Data were acquired over a 45-90°C range at the thermal transition rate of 0.1°C/s. Genotypes were identified by the melting temperatures indicated by peaks on the derivate plots after normalization using LightScanner 32 software 1.0.0.23 (Idaho Technology Inc.). The probe was designed to exactly match the allele T. The T allele homozygotes had a derivative melting peak at 60°C and G allele homozygous samples had a melting peak at 53°C, whereas heterozygotes showed both peaks.
## Statistical analyses
Statistical analyses were performed using SPSS 17.0 for Windows software (SPSS Inc., Chicago, IL, USA). Relative frequencies of sexes were compared using the c 2 -test. Continuous variables are presented as mean ± standard error of mean (SE). For comparison of means, either unpaired Student's t-test or analysis of variance (ANOVA) was used. General linear models and multiple linear regression analyses were used to account for baseline differences and other confounding factors for the 2 primary outcomes -change in FPG (ΔFPG) and HbA 1c (ΔHbA 1c ) after treatment. All models were adjusted for age, sex, BMI, and either for baseline FPG or baseline HbA 1c values.
# Results
The effect of 6-month treatment with sulphonylureas in the whole study group is shown in . Mean HbA 1c level decreased by 1.0%, mean FPG by 1.5 mmol/L, and mean triglyceride levels by 0.3 mmol/L (p<0.001, p<0.001, p=0.036, respectively). In contrast, no significant differences were recorded in body weight, BMI, total cholesterol, LDL cholesterol and HDL cholesterol after sulphonylurea treatment.
The distribution of KCNQ1 rs163184 genotypes followed Hardy-Weinberg equilibrium (p=0.39). Twenty subjects were homozygous for the T-allele (TT genotype), 39 were carriers of 1 risk G-allele (TG genotype), and 28 were homozygotes for the G-allele (GG genotype).
Baseline clinical and biochemical characteristics, as well as the indices of glycaemic control after sulphonylurea treatment of 3 genotype groups are displayed in [fig_ref] Table 2: Clinical and biochemical characteristics of the subjects according to KCNQ1 rs163184 genotypes [/fig_ref]. No significant differences were observed in sex representation, age, weight, BMI, diabetes duration, cholesterol, LDL-cholesterol, HDL-cholesterol, triglycerides and either baseline FPG or HbA 1c levels among the genotype groups.
The reduction in FPG (ΔFPG) after 6-month sulphonylurea therapy significantly differed among the 3 KCNQ1 genotype groups (ANOVA, p=0.017). The post-hoc analysis revealed greater ΔFPG in the TG compared to the GG genotype group (p=0.015) [fig_ref] Table 2: Clinical and biochemical characteristics of the subjects according to KCNQ1 rs163184 genotypes [/fig_ref].
In further analyses, a recessive genetic model was tested in which carriers of the T-allele were pooled (TT+TG) and compared with patients with the GG genotype . After sulphonylurea therapy, patients in the TT+TG group achieved significantly lower FPG levels in comparison with patients with the GG genotype (6.95±0.13 vs. 7.50±0.21 mmol/L, p=0.033). Consequently, ΔFPG was significantly higher in the TT+TG group compared to the GG group (1.58±0.13 vs. 1.04±0.18 mmol/L, p=0.016) .
In multiple linear regression analysis with ΔFPG as a dependent variable and KCNQ1 genotype, age, sex, BMI, and baseline FPG as independent variables, KCNQ1 genotype (p=0.016) and baseline FPG (p<0.001) were the only significant independent predictors of ΔFPG (R 2 =0.48).
# Discussion
In the present study we have shown that the homozygous carriers of the risk G-allele of the KCNQ1 rs163184 gene polymorphism responded to treatment with sulphonylurea by significantly lower reduction in fasting glycaemia. In multivariate analysis, KCNQ1 genotype and baseline FPG levels were the only independent significant predictors of FPG reduction after sulphonylurea treatment.
The reports on the potential relationships between the effect of sulphonylurea therapy and genes variants associated with type 2 diabetes are scarce. Sulphonylurea receptor 1 is encoded by ATP-binding cassette transporter sub-family C member 8 (ABCC8) gene. A recent study in 115 Chinese patients revealed that carriers of the risk Ala-allele in the nonsynonymous ABCC8 Ser1369Ala polymorphism were more sensitive to gliclazide treatment with respect to reduction of HbA 1c level in comparison with Ser/Ser homozygotes [bib_ref] Association of sulfonylurea receptor 1 genotype with therapeutic response to gliclazide in..., Zhang [/bib_ref]. This finding was confirmed in a larger study of 1268 Chinese subjects with type 2 diabetes, in which 8-week treatment with gliclazide showed higher therapeutic efficacy on fasting glycemia, but not on HbA 1c levels, observed in patients with Ala/Ala genotype when compared with patients with Ser/Ser genotype [bib_ref] Ser1369Ala variant in sulfonylurea receptor gene ABCC8 is associated with antidiabetic efficacy..., Feng [/bib_ref]. factor 7-like 2 (TCF7L2) polymorphisms and the response to sulphonylurea therapy in type 2 diabetic patients. In that study, 901 Scottish patients with type 2 diabetes treated with sulphonylurea were analysed. The probability for early sulphonylurea treatment failure was almost doubled in subjects with the risk TT genotypes of both rs12255372 and rs7901346 polymorphisms compared to those homozygous for the reference non-risk genotypes [bib_ref] Variation in TCF7L2 influences therapeutic response to sulfonylureas, Pearson [/bib_ref]. Furthermore, our recent study demonstrated that carriers of the risk T-allele of TCF7L2 rs7903146 polymorphism responded to treatment with sulphonylurea by significantly smaller reductions in HbA 1c and FPG levels in comparison with patients with the CC genotype [bib_ref] Effect of sulphonylurea treatment on glycaemic control is related to TCF7L2 genotype..., Schroner [/bib_ref].
In the present study we observed a significant association between the KCNQ1 rs163184 gene polymorphism and the efficacy of sulphonylurea treatment in type 2 diabetic patients who failed to achieve adequate glycaemic control on monotherapy with metformin -patients carrying 2 risk G-alleles of KCNQ1 (GG genotype) had significantly smaller reductions in FPG levels after 6-month therapy than those with at least 1 non-risk T-allele (TT+TG genotypes). The differences in mean FPG (0.54 mmol/l) observed between different genotypes of KCNQ1 rs163184 polymorphism after sulphonylurea treatment in the present report were similar to those observed for TCF7L2 polymorphism [bib_ref] Effect of sulphonylurea treatment on glycaemic control is related to TCF7L2 genotype..., Schroner [/bib_ref]. To our best knowledge the present findings are the first to provide evidence of differences in the FPG reductions between various KCNQ1 genotype groups. Nevertheless, similar to Feng et al who analysed the treatment effect with respect to ABCC8 genotypes [bib_ref] Ser1369Ala variant in sulfonylurea receptor gene ABCC8 is associated with antidiabetic efficacy..., Feng [/bib_ref] , we did not observe significant effect of the KCNQ1 rs163184 gene polymorphism on ΔHbA 1c following sulphonylurea treatment.
Besides the KCNQ1 genotype, baseline FPG also independently predicted the FPG reduction after sulphonylurea therapy. Importantly, the effect of KCNQ1 gene polymorphism on the reduction in FPG was independent of the baseline FPG levels in the present study.
The mechanism underlying the effects of KCNQ1 polymorphism on the therapeutic effect of sulphonylureas is poorly understood. In a physiological study in INS-1 cell cultures, the blockade of the KvLQT1 channel with KCNQ1 protein inhibitor 293B stimulated insulin secretion in the presence of sulphonylurea drug tolbutamide [bib_ref] Effects of I (Ks) channel inhibitors in insulin-secreting INS-1 cells, Ullrich [/bib_ref] , suggesting that KvLQT1 channels might play a role in fine-tuning of insulin secretion during sulphonylurea treatment.
There are limitations in the current pilot study, such as the small size of the study group. However, the approximately 35% difference in ΔFPG after 6-month therapy between the carriers of risk GG genotype and the carriers of nonrisk T-allele suggests that the sample size of this study, although limited, is enough to gain understanding on the role of KCNQ1 gene polymorphisms in the pharmacologic response to sulphonylureas. Furthermore, although the duration of the present study was longer than the previous ones [bib_ref] Association of sulfonylurea receptor 1 genotype with therapeutic response to gliclazide in..., Zhang [/bib_ref] [bib_ref] Ser1369Ala variant in sulfonylurea receptor gene ABCC8 is associated with antidiabetic efficacy..., Feng [/bib_ref] , our results reflect only the effect of sulphonylurea therapy during the first 6 months after initiation of this treatment; therefore, our results cannot be automatically extrapolated beyond this period. Nevertheless, in the GoDARTS study, higher probability of sulphonylurea failure was observed up to 12 months following initiation of therapy in patients with the risk TCF7L2 genotypes [bib_ref] Variation in TCF7L2 influences therapeutic response to sulfonylureas, Pearson [/bib_ref]. Therefore, the KCNQ1 genotype-related differences in glucose lowering response to sulphonylureas observed in the present study might be preserved beyond 6 months. Further studies are needed to address this question. Importantly, investigation of a homogenous group of patients with type 2 diabetes, in whom sulphonylurea was started in a typical clinical situation after metformin monotherapy failure with mean HbA 1c level of approximately 8% represents a strength of the present study.
# Conclusions
The present study showed for the first time that the variation in KCNQ1 gene is related to therapeutic response to sulphonylurea treatment in addition to previous metformin monotherapy. Large-scale pharmacogenetic studies are needed to examine combined effect of risk alleles of several genes and their eventual interactions on response to sulphonylureas. Such observations might lead to the development of genotype-based personalized strategies for the treatment of type 2 diabetes in the future.
references:
[table] Table 2: Clinical and biochemical characteristics of the subjects according to KCNQ1 rs163184 genotypes.Data are expressed as mean ±SE. * p refers to χ 2 -test, otherwise p refers to ANOVA. ∆FPG and ∆HbA1c means were adjusted in general linear models for gender, age, BMI and baseline FPG or HbA1c values, respectively. ** p=0.015 for comparison TG vs. GG (Bonferroni test). BMI -body mass index; FPG -fasting plasma glucose; LDL -low-density lipoprotein; HDL -high-density lipoprotein. [/table]
|
Enhancement, ethics and society: towards an empirical research agenda for the medical humanities and social sciences
For some time now, bioethicists have paid close attention to issues associated with 'enhancement'; specifically, the appropriate use and regulation of substances and artefacts understood by some to improve the functioning of human bodies beyond that associated with 'normal' function. Medical humanities scholars (aside from philosophers and lawyers) and social scientists have not been frequent participants in debates around enhancement, but could shine a bright light on the range of dilemmas and opportunities techniques of enhancement are purported to introduce. In this paper, we argue that empirical research into the notion and practice of enhancement is necessary and timely. Such work could fruitfully engage with-and further develop -existing conceptual repertoires within the medical humanities and social sciences in ways that would afford benefit to scholars in those disciplines. We maintain that empirical engagements could also provide important resources to bioethicists seeking to regulate new enhancements in ways that are sensitive to societal context and cultural difference. To this end, we outline an empirical agenda for the medical humanities and social sciences around enhancement, emphasising especially how science and technology studies could bring benefits to-and be benefitted by-research in this area. We also use the example of ( pharmaceutical) cognitive enhancement to show how empirical studies of actual and likely enhancement practices can nuance resonant bioethical debates.
# Introduction
Biomedical and technological developments, regarded as having the capacity to improve the body's appearance or function, are often termed 'enhancements'. Over the last 15 years or more, bioethical and policy discourses have questioned the appropriateness of enhancements for individuals and societies. Discussions have turned on the problems of differentiating enhancement from therapy, the hubris of changing nature and the potential for exacerbating social inequalities. Such questioning has resulted in sometimes acrimonious debate playing out in journals, conferences and in the meeting rooms of learned societies and policy advisory committees-as well as in more public arenas. These widely circulating questions and concerns around enhancement are, no doubt, important matters for consideration, but we believe they are not the only issues of import. More needs to be understood about the social, political and historical contexts in which 'enhancements' arise, as well as the reception and uptake of specific drugs or devices. [bib_ref] Cognitive enhancement? Exploring modafinil use in social context, Coveney [/bib_ref] [bib_ref] The sociology of cognitive enhancement: medicalisation and beyond, Coveney [/bib_ref] [bib_ref] Pharmaceutical cognitive enhancement: interrogating the ethics, addressing the issues, Martin [/bib_ref] Further, the potential of new technologies to reconfigure health, medicine, social relations and subjectivities is also relevant to bioethical debate. Yet, if and when this complex issue appears within bioethics discourse, it is often theorised in isolation from other traditions that have taken the relationships between materials, bodies and societies to be key foci.
In this article, we suggest that the empirical medical humanities and social sciences can contribute meaningfully to societal discourse around enhancement in two key ways. First, and most importantly, by attending to issues less commonly considered in current debates. Second, by providing methodologically robust studies of actual practices involving technologies and chemicals regarded as enhancing (which might intersect directly withand in some cases challenge-bioethical deliberation). [bib_ref] Critical bioethics: beyond the social science critique of applied ethics, Hedgecoe [/bib_ref] We take these matters in turn, with the first and substantive focus of this paper examining how research foci within the medical humanities and social sciences might (re)direct analytical attention to the tools, bodies and institutions that populate debates around enhancement. Then, we move to a more specific account of 'cognitive enhancement' using pharmaceuticals meant for attention disorders. These are employed as a case study of empirical engagement with a matter of bioethical concern, in order to consider the benefits to bioethics that such analyses might afford. Debates around drugs taken to have enhancing properties to the human brain are wide-ranging; such pharmaceuticals attract broad attention due to the wider social and legal contexts of their use and the potential implications for personal responsibility and liability that come with this. An exploration of extant studies is indicative of new empirical research trajectories that might be considered, and the salience of such scholarship for bioethical deliberation.
## Exploring new terrain
We outline some research questions and directions that might contribute to research on enhancement that intersects with existing bioethics scholarship. As a consequence of our own inclination towards the field of science and technology studies (STS), we start with specifying questions that direct the gaze of scholars to the objects around which Open Access Scan to access more free content i We place the word 'enhancement' within quotes in order to signal our reticence to ontologise particular tools and substances as intrinsically 'enhancing' (or, indeed, 'therapeutic'). Our reasons for this will become clearer as the analysis unfolds.
bioethical debate circulates (eg, pharmaceuticals), before considering the sociotechnical networks that enable, legitimise and sustain their production and consumption. Like others, we feel that the substantial body of interdisciplinary STS scholarship on the social shaping of technology is of immediate relevance and import to our considerations. Such work seeks to open up the 'black box' of technology and consider the ways in which a range of social and economic factors construct the eventual form and feel of an object. In the case of enhancement, we might ask: what ideas about society, of the body, and of individual hopes and fears are imagined and/or channelled within processes of innovation? What values-clinical, scientific, economic, societal and so on-are 'inscribed' in particular substances and technologies deemed to be enhancing?How are the practices of potential users configured in the process?Scrutiny of the circulation and transformation of different kinds of value and values within the innovation processes producing technological enhancements will likely represent an interesting case for scholars who have attended to these issues in more straightforwardly biomedical contexts (such as in drug trials, trial recruitment and marketing).Disaggregating the different values and practices of valuation constitutive of the innovation and consumption of materials purchased (in formal, grey or black markets) for their enhancing properties also serves to trouble the more essentialist claims of some bioethical commentators who present these as unproblematic social and individual goods. Regulating biomedical lives As a range of scholars have shown, economic value relates closely to how any new tool, technology or drug is apprehended by law, constituted through legal processes and regulated on the market.Questions of law and regulation, then, are central to innovation processes, both during development and after the finished products are released. Recent work at the interface of regulatory studies and STS has illustrated how particular modes of governance are part of the societal context that can be built into technologies, helping to shape their design and mode(s) of delivery. 16-18 Sheila Jasanoff, for example, uses cross-cultural case study methods to demonstrate how political cultures affect the way societies assess evidence, evaluate risk and engage with publics around innovations in science and technology.Such perspectives could contribute considerably to understanding the differences regarding how policy issues around enhancements are framed, as well as what-and whose-values become embedded in the institutions governing their use.
Social histories of technologies are also often omitted from studies of 'enhancements', but can be key to understanding the interpretation of certain technologies. For example, the burgeoning science of optogenetics uses optics and gene modification techniques to control the activity of specific cells. ii In addition to the promise of modulating cell function in neurological disorders, scientists suggest they can alter addictions, depression and mood disorders and other behavioural and psychosocial conditions. Previous controversies over gene therapy and psychosurgery will surely colour how such emerging techniques will be viewed from both regulatory and ethical perspectives. Additionally, popular culture, including science fiction themes of mind control, have been known to affect legal frameworks in which such brain interventions would be viewed for therapeutic purposes, much less enhanced function. There has already been considerable scholarly interest in the interactions of law and neuroscience, particularly in light of questions of altered autonomy, identity, personal responsibility and liability with the use of brain stimulation techniques. [bib_ref] I am who I am": on the perceived threats to personal identity..., Baylis [/bib_ref] [bib_ref] Did my brain implant make me do it? Questions raised by DBS..., Klaming [/bib_ref] Accordingly, it will be fruitful to examine how legal processes shape the production and deployment of tools and substances that are understood to enhance human bodies and cognition. This is in terms of how actual regulation directs innovation, and if and how imaginaries of future governance inform the development and marketing of objects framed as enhancements. Court cases where these figure would likewise be important to examine, in terms of how object ontologies (eg, purported nature of particular drugs and devices, and hence, the 'appropriate' uses to which they are put) and subjective properties (eg, authenticity, personhood, etc.) are co-produced. It is also important to attend to the ways that such innovations are situated within so-called knowledge-intensive societies, with their emphasis on performance, productivity, competition and connectivity. Markets are likely to help shape legal and regulatory environments as much as the technologies themselves.
Documenting how 'enhancements' are dealt with in clinical practice especially-including who is empowered to control access to them, and why-will be a key empirical task for those in the medical humanities and social sciences. Not enough is known about how physicians represent to patients (and themselves consider) the benefits, risks and harms of potentially enhancing products and treatments, nor how 'legitimate' use is configured. [bib_ref] Dr would you give me a pill to help me?" A national..., Hotze [/bib_ref] Further to reflecting on the philosophical or policy implications of how products might be classified as therapeutic or enhancing, scholars might ask: what comes to count in how health professionals (not solely medical practitioners) decide whether the use of one particular drug or technology counts as 'therapy' or as 'enhancement'? For instance, do physicians consider the moral or legal implications of prescribing attention deficit-hyperactivity disorder (ADHD) drugs for offlabel purposes, including enhancing cognition? To what extent are they aware of how patients may be using such prescribed drugs, including selling them to others? [bib_ref] Everyday drug diversions: a qualitative study of the illicit exchange and non-medical..., Vrecko [/bib_ref] The literature on ethical decision-making in clinical practice will afford benefit to empirical analysis of such questions, which will in turn enrich scholarship around ethics-in-practice. Ethnographic and conversation analytic work might usefully examine the physician-patient dance involved in representing a subjective state as a legitimate target of optimisation, not least to cast further light on the cultural logics shaping discourses of legitimacy today. [bib_ref] Emily's scars: surgical shapings, technoluxe, and bioethics, Frank [/bib_ref] In cases where physicians may be reluctant to prescribe 'enhancements' to individuals who ask for them, where do individuals who wish to consume these then turn? Research exploring such issues may have much to offer sociolegal studies in terms of understanding how grey markets are formed and negotiated by a range of individuals. Further, interviews and other methodological strategies might be employed to examine how users construct the benefits, risks and harms of substances and tools used to enhance human bodies (in particular, where access is achieved beyond the clinic, such as via the internet). Users may employ these notions in ways that are quite distinct from those featuring within biomedical and bioethical discourses, and the regulatory regimes that are in part constituted through these (as we will show below-drawing on a range of work-for the case of cognitive enhancement). [bib_ref] Pharmaceutical cognitive enhancement: interrogating the ethics, addressing the issues, Martin [/bib_ref] For example, comparative work between prescription and non-prescription 'enhancements' (such as the use of nicotine ii Specifically, light-responsive DNA segments are inserted into cells and when externally activated by a light source, can be switched on or off to observe effects in living subjects, or modulate cell function. replacement patches as stimulants) regarding publics' constructions of risk would be both timely and have import for scholarly interrogations of the production of risk, harm and safety in other (non-)medical domains. Since public engagement plays an important role in in shaping policy approaches to new and emerging technologies, consideration will be required of how this is leveraged as a regulatory device to either promote or curtail the use of supposed enhancements. [bib_ref] Regulatory or regulating publics? The European Union's regulation of emerging health technologies..., Flear [/bib_ref] Relatedly, more critical work might document the establishment of new professional networks (both formal and informal) of scientists and ethicists (such as the International Neuroethics Society), and the ways in which calls for regulation and ethical oversight can function as a means of rendering particular sociotechnical pathways legitimate. [bib_ref] Field of dreams: a social history of neuroethics, Conrad [/bib_ref] The legitimating function of bioethics has been analysed for 'therapies' rather than 'enhancements' (eg, using the case of pharmacogenetics), but any impact of (inter)national ethics bodies-if it exists-on the consumption of products that can be used to enhance bodily or cognitive function has not been formally examined. [bib_ref] The drugs don't work: expectations and the shaping of pharmacogenetics, Hedgecoe [/bib_ref] Further, we might also explore the degree to which the act of calling out issues of bioethical or regulatory concern can be employed as a vehicle for establishing organisational authority and legitimacy. In particular, under the Bush administration in the USA, the President's Council on Bioethics garnered considerable attention through bioethical agenda-setting and what some regard as overcautionary reports, including on enhancement.Both historical and ethnographic research could reveal how and to what extent ethical claims-making and organisational legitimacy relate to each other. Such scholarship could engage productively with feminist bioethics; this has cast a critical eye upon issues of authority and legitimacy, and renders problematic the universal subject that is often assumed by official ethics bodies. Rhetorical analyses of formal reports, documents and position papers deployed by such organisations will also illuminate embedded political rationalities (E Barr. The President's Council on Bioethics' 'Happy Souls': discipline, citizenship and true selves, unpublished manuscript, 2014).
## Diverse bodies
Above we noted some of the diverse kinds of empirical attention that could be paid to the development, production and marketing of technologies and pharmaceuticals characterised in various spheres as enhancements. Cutting across this must also be an analytic sensitivity to the ways through which the intertwined issues of gender, ethnicity, sexuality, sociodemographic status and other markers of identity are implicated in practices of innovation and consumption.How are particular ideas of gender, for instance, embedded within enhancements, and how are new drugs and technologies gendered within media and marketing portrayals (and, indeed, within academic writing)? How do they impact on experienced and ascribed gender? Viagra is perhaps the case per excellence here, and empirical, STS-inflected research on this has cast bright light on its gendered marketing and uptake-but also on sites of resistance, and communities of users (eg, women) who have appropriated this drug and re-inscribed its embedded scripts so as to situate it within new regimes of personal life. [bib_ref] Potency in all the right places: Viagra as a technology of the..., Mamo [/bib_ref] Likewise, painkillers have been used in Indonesia to enhance libido, and hormonal therapies have been documented as being repurposed by individuals self-identifying as men who seek to feminise their bodies. [bib_ref] Chemical sexualities: the use of pharmaceutical and cosmetic products by youth in..., Hardon [/bib_ref] Research in this vein reveals the complex social lives of 'enhancements' that exceed anticipatory imaginaries of their uses and actions, and that also enrich medical humanities and social scientific understandings of the relationships between biomedicine and wider society.A concern with gender-as well as ethnicity and other somatic markers of identity-connects with themes of embodiment more generally. Debates in this area might stimulate a consideration of any changes to how bodies 'dys-appear' (ie, come to subjective attention through pain, disability or impairment), disappear and reappear in societal practices and discourses associated with the drugs and technologies that some deem to enhance human traits and states.What aspects of 'normal' bodies could come to be regarded as sites of intervention in contexts where diverse forms of customisation are possible? The use of human growth hormone to simultaneously 'treat' and 'enhance' height is one example where the differences between 'therapy' and 'enhancement' emerge as complex, contingent and readily reworkable. With a focus on the 'treatment' of short boys, such distinctions are also highly gendered. [bib_ref] Growth hormone, enhancement and the pharmaceuticalisation of short stature, Morrison [/bib_ref] Further, the human brain seems often to only become salient in personal discourse through its dys-appearance due to injury or cognitive impairment-yet, in an era of proliferating enhancements, it may increasingly be viewed as a plastic organ operable upon through techniques of 'objective self-fashioning' via pharmaceuticals or electroceuticals. [bib_ref] Bonding brains to machines: ethical implications of electroceuticals for the human brain, Clausen [/bib_ref] [bib_ref] The changing brain: neuroscience and the enduring import of everyday experience, Pickersgill [/bib_ref] [bib_ref] The plastic brain: neoliberalism and the neuronal self, Pitts-Taylor [/bib_ref] At the same time, 'dys-appearing' forms of (especially) neurological difference are increasingly subject to projects of de-stigmatisation (eg, in the case of autism). [bib_ref] Wired up differently': autism, adolescence and the politics of neurological identities, Ortega [/bib_ref] In this light, understandings of techniques that operate upon human difference (be those drugs, devices or psychological interventions) can shift from being conceptualised as part of a therapyenhancement continuum to being questioned over whether they are unwanted (and perhaps even coercive) tools of normalisation. The use of enhancements therefore represents the opening up of a new set of cases for empirical examination both in terms of how 'differences' between humans come to be recognised as such, and of how much difference populations are prepared to accept. Historians, anthropologists, geographers and others are particularly well placed to parse out how the uses of enhancement technologies might recast what bodily properties are taken as valued in particular contexts, by who and to what ends, and how the practices and consequences of valuation interact with actual engagements with enhancing technologies.A reinvigoration of research around cosmetic surgery is one starting point for the interrogation of these questions, especially including non-Western spaces and places where empirical ethnographic research might render problematic any assumptions that emerge regarding such practices based on their contemporary uses in Europe and North America.In Brazil, for example, the experimental use of testosterone to enhance vigour and verve means that there is no ready place to position this along any kind of linear 'therapy'/'enhancement' continuum. [bib_ref] Medical borderlands: engineering the body with plastic surgery and hormonal therapies in..., Edmonds [/bib_ref] Rather, as Alex Edmonds puts it, 'health and aesthetics become entangled'. [bib_ref] Can medicine be aesthetic? Disentangling beauty and health in elective surgeries, Edmonds [/bib_ref] Cross-cultural empirical research could further illustrate (and perhaps contribute to the destabilisation of ) Anglo-American societal norms that assume what counts as 'therapy' and what comprises 'enhancement' (simultaneously decentring 'the West' within bioethical analyses in this area).
## Constructing communities and contexts
Thinking about changing cross-cultural experiences and identities also directs our attention to the formation of new social groups orientated around apparent enhancements. Debates around the uses to which objects deemed enhancing might be put have helped to stimulate a range of neologisms (such as 'bioconservative' and 'transhumanist') that will be of interest to researchers keen to document, explore and understand how these become an idiom through which communities coalesce and expand. There are identity politics at play that are important to attend to: different communities can and could respond to discourses and practices centring on drugs and devices associated with human enhancement in diverse ways that speak to broader (and diverging) conceptions of humanity, sociality and public good.Disability studies scholars have begun to unpick some of the wider moral and political dimensions and implications of the imaginaries fuelling the speculation and advocacy of self-identifying 'transhumanists' (individuals who propose widespread adoption of a range of radical enhancements), for example. Their concerns regarding new forms of discrimination are suggestive of the need for empirical work with normative bite. [bib_ref] Why NBIC? Why human performance enhancement?, Wolbring [/bib_ref] Further, philosopher Trijsje-Marie Franssen has examined how the myth of Prometheus features within and configures debates around human enhancement and transhumanism (T-M Franssen. Prometheus through the ages: from ancient trickster to future human. University of Exeter: Unpublished doctoral thesis 2014. Available at: https://ore.exeter.ac.uk/repository/ bitstream/handle/10871/15889/FranssenT.pdf?sequence=1). Scholarship of this kind provides insight into how images and imaginaries come to be instantiated within the identities of novel communities, and the discussions about medical technologies that these in turn propel.
A further point of departure for empirical work around the broad issue of 'enhancement' might be to interrogate how ethical debates and social practices (including but not limited to processes of innovation) mutually shape one another. [bib_ref] The co-production of science, ethics, and emotion, Pickersgill [/bib_ref] The co-productionist tradition of STS-which foregrounds the dynamism between materiality and sociality-could provide rich conceptual resources for consideration of the reciprocal constitution of objects, moral discourse and social practices.Co-productionist studies of enhancement might also attend to how any routine use of enhancements could change their cultural contexts. How, for instance, does the promotion and use of particular technologies-such as drugs to enhance cognition -respond to issues of social concern (eg, educational attainment)? In turn: how, and to what extent, does the embedding of enhancements within societies recast the meanings of the terms by which societal issues are debated? Engagement with such a question necessarily entails an historical, rhetorical and ethnographic analysis of what 'individual attainment' is taken to be within specified cultures, and what cultural actors judge to be legitimate means of performing accomplishments. Accordingly, research that takes the practices of and politics around enhancement as a starting point could produce textured and nuanced scholarship that contributes to conceptual and empirical literatures far beyond the case study under primary examination.
Finally, a co-productionist perspective also enjoins us to engage with the emergence of the debate around 'therapy' and 'enhancement' itself. As indicated above, we argue that these concepts come to find meaning through particular configurations of social conventions, epistemic norms and biomedical tools. Intrinsically normative characterisations, their position in relation to each other is ambiguous and ever-changing, since the terrains of treatment and of enhancement have boundaries that shift according to mutations in the same sociotechnical practices that constitute them. The flexibility of these spaces underscores what we regard as a central problem for medical humanities and social science scholars: how, why and where do particular technologies come to count as 'enhancing'? Since the categories of 'therapy' and 'enhancement' are emergent through social praxis, any definitional settlements are of course essentially contingent, yet historical, geographical and anthropological research tracing processes of stabilisation is both timely and important.
## Empirical analyses of pharmaceutical cognitive enhancement
Throughout our analysis, we have sought to take care over how we use the term 'enhancement', indicating what empirical research agendas might be propelled through an acknowledgement that there are a number of individuals and groups who actively use this word and ascribe it to actual or imagined objects around which social practices are orientated-as well as many more people who (seek to) 'enhance' their bodies and brains in the absence of this idiom. In addition to being of import for the wider medical humanities and social sciences, we believe (and in so doing take cues from scholars such as Adam Hedgecoe) that empirical attention to the discourses and objects associated with enhancement might also better ground bioethical claims and appraisals in this area. [bib_ref] Critical bioethics: beyond the social science critique of applied ethics, Hedgecoe [/bib_ref] Given that qualitative and quantitative methods have become more common in bioethics, we suggest that these methods can help to illuminate how ethically significant issues related to enhancement play out in everyday life. [bib_ref] Appropriate methodologies for empirical bioethics: it's all relative, Ives [/bib_ref] Practices of enhancing cognition, for example, have been central to much of the bioethical discussion that focuses explicitly on enhancement, especially the use of pharmaceuticals that promote concentration and alertness (for instance, brand names Adderall and Ritalin). Bioethicists of a more libertarian bent argue that such drugs should be more widely and freely available to those who want them, and that potentially citizens have a duty to consume them. 14 Arguments like these can be seen as situated within a broader discourse of what social scientists have referred to as 'biological citizenship'-the implicit or explicit connection of citizenship to somatic states or processes. [bib_ref] Biological citizenship, Rose [/bib_ref] Work is being produced that is seeking to quantify the degree to which drugs like Adderall are being used (eg, by university students). [bib_ref] Neuroenhancement among German university students: motives, expectations, and relationship with psychoactive lifestyle..., Eickenhorst [/bib_ref] [bib_ref] To dope or not to dope: neuroenhancement with prescription drugs and drugs..., Maier [/bib_ref] [bib_ref] Robust resilience and substantial interest: a survey of pharmacological cognitive enhancement among..., Singh [/bib_ref] Interview studies are concerning themselves with the texture of this use and the meanings that are ascribed to it. Notably, for instance, a recent paper by Singh et al 57 underscored that many UK students were not familiar with cognitive enhancement. Such scholarship is needed in order to ensure that bioethical analysis is grounded in what seems to actually be happening in schools, workplaces and in the homeand what feasibly, based on empirical research on current practice, could start to happen. 7 This is in contrast to a common bioethical focus on more speculative matters, and by commentators who may be geographically, professionally and personally distant from the figures that populate their imaginaries.
One useful resource is Catherine Coveney's sociological study of a range of individuals (including university students, health professionals and other shift workers) asked to envision the use of a drug promoting wakefulness (specifically, modafinil; brand name: Provigil). Her findings showed that potential uses of this enhancement included ensuring safety at work (eg, ameliorating the dangers posed by sleepy surgeons), but also as a study aid. However, perceptions both of legitimate use and the legitimacy of the drug per se were tightly interwoven with its legality, and how readily available it was (eg, from local pharmacists, without a prescription, or from physicians only). [bib_ref] Perceptions of assisted cognitive and sport performance enhancement among university students in..., Vargo [/bib_ref] Accordingly, what Coveney's research shows is that public attitudes towards enhancement drugs are linked to broader feelings towards pharmaceuticals more generally-and, specifically, the ways in which (to borrow historian Keith Wailoo's term) the 'identities' of drugs are constituted through regulatory regimes. This insight has salience for bioethical debate that might take the meanings of pharmaceuticals to be solely a function of personal mores, as opposed to being powerfully shaped by the social frameworks in which they and their consumers are embedded. [bib_ref] Pharmaceuticals and society: power, promises and prospects, Gabe [/bib_ref] [bib_ref] The pharmaceuticalisaton of society? A framework for analysis, Williams [/bib_ref] [bib_ref] Medicalisation or customisation? Sleep, enterprise and enhancement in the 24/7 society, Williams [/bib_ref] Other work by social psychologist Ilina Singh directs attention to the actual users of pharmaceuticals. One major theme around cognitive enhancement is 'authenticity', and bioethical scholarship focusing on this has until recently operated rather tangentially from empirical research. Questions discussed within the bioethics literature include what it might mean to live authentically, whether regular pharmaceutical use necessarily results in an 'inauthentic' life and if medication can in itself enhance authenticity. Conclusions may seek to direct the use of enhancements, including both restrictions and expansions of access to these drugs and technologies.A major contribution from Singh is her qualitative work with children diagnosed with ADHD, in which discussions took place around the relationship between psychopharmaceutical consumption and understandings of authenticity. For this work, Singh interviewed more than 150 children between 9 and 15 in both the USA and UK. As she shows, drugs are not viewed as necessarily challenging or compromising characteristics associated in the bioethics literature with authenticity, and in some cases are figured by children as supporting, for instance, moral agency. [bib_ref] Not robots: children's perspectives on authenticity, moral agency and stimulant drug treatments, Singh [/bib_ref] This is suggestive of the degree to which the use of psychopharmaceuticals that act on cognition by consumers who do not have a psychiatric diagnosis may also refrain from seeing the use of Adderall or Ritalin as compromising their autonomy. There are implications here for bioethical debates around whether individuals should imbibe these substances in order to preserve their authenticity. Empirical work on authenticity potentially is of interest to philosophers concerned more directly with this concept per se.
Perspectives from users of 'enhancements' have also been the focus of sociologist Scott Vrecko's research. His participants were university students at a US institution who used Adderall and Ritalin as study aids. Vrecko asked young men and women about their experiences of consuming such drugs, and elicited striking narratives of their practices. In particular, he revealed the diverse kinds of illicit exchanges through which prescription drugs come to be used for non-prescribed purposes, as well as the highly emotional impact of drug effects, whereby his respondents detailed how it felt to be 'on' psychopharmaceuticals that were understood by them to improve their work. Students discussed feelings of channelled interest and enjoyment (ie, a tight focus on, and pleasure in, the studies they were immediately employed in and not other things happening around them), 'drivenness', and feeling mentally and/or physically 'up'. [bib_ref] Just how cognitive is "cognitive enhancement"? On the significance of emotions in..., Vrecko [/bib_ref] Such findings resonate with the more recent research of Margit Anne Petersen and colleagues, who likewise have underscored the role of prescription stimulants in sharpening students' focus on their work, as well as on the subjective experience of work per se (and also on the moral ambivalences that characterise drug use in this context). These experiences challenge the more 'unemotional' (as Vrecko puts it), sometimes strongly rationalist, accounts of decision-making and use that have found a home within much bioethical deliberation on cognitive enhancement. Hence, Vrecko's and Petersen et al's findings are suggestive of the need to bring in the perspectives of actual users into the discursive spheres of bioethical and policy deliberation.
Work such as that summarised above thus complicates bioethical claims-making by underscoring the ways in which multiple meanings adhere to substances. As Coveney, Singh, Vrecko, Petersen and others remind us, drugs are never just their chemical constituents. Rather, they may take on a variety of 'identities' (and have different effects) depending on the context within which they are (or realistically might be) used. Many standard analyses of risk and benefits of pharmaceutical means of enhancing cognition take for granted what a drug 'really' is; that is, they are concerned solely with its active ingredients, mode of biological action and physiological effects. In so doing, commentaries and appraisals can elide the diverse and shifting societal understandings of the nature of the substance in question. Besides holding science and society as separate domains, in these cases recommendations for policy and practice can be problematic as they might lack social salience. If a culturally impoverished understanding of a drug is mobilised in debate, the complexities of deciding whether or not this substance can, should or must be used are obfuscated, and so the legitimacy and utility of any conclusions reached are lessened. Our point is one that will be familiar to scholars concerned with the social lives of pharmaceuticals: that understandings of drugs and their effects are mediated through their use, the context of this, and who drugs are being used by and for what reasons. This illustrates the necessity of the empirical studies of enhancement already being undertaken. Further, it suggests the need for more in-depth quantitative, qualitative and interpretive research (informed by relevant theoretical and conceptual literatures from the medical humanities and social sciences) that can attend more precisely to the contextual issues we believe are so necessary to integrate into bioethical deliberation, and which have begun to be charted by the authors discussed above.
# Conclusion
Our aim in this analysis has been twofold. First, and most importantly, in common with scholars like Coveney, Morrison, Singh and Vrecko, we want to more firmly fix the attention of the empirical medical humanities and social sciences to the objects, bodies, subjectivities and discourses that populate debates around the enhancement of human bodies and minds. Our hope is that this can be achieved in ways that are not dismissive of traditional normative analysis or the debates around autonomy and responsibility that are so prevalent within bioethics, but which nevertheless uncover new territory for empirical exploration and theorisation. Indeed, perspectives from the wider medical humanities and social sciences can, we believe, also be brought to question the very terms through which bioethicists, policymakers, entrepreneurs and others are currently framing and advocating their positions and movements (eg, as enhancement 'bioconservatives' or proponents). Examining substances and devices that some regard as having the capacity to enhance human bodies and cognition can inform our understandings of the interactions between biomedicine and society more broadly, and hence, represent important case studies for scholars in cultural studies, history, law and STS, to name just a few disciplines.
Second, in mapping the terrain that the empirical medical humanities and social sciences might chart (and, indeed, have begun to explore), we want to make a renewed call for the importance of drawing the expertise of these scholars into the spheres of policy-orientated bioethical decision-making (and potentially disrupt them). We recognise the quality and rigour of many contributions from philosophers and others to understandings of enhancements in society; nevertheless, we suggest that a still more nuanced understanding of these issues, incorporating (or at least, drawing on) the expertise of other humanities scholars and of social scientists, might further enrich bioethical discussions. We believe that the appraisal and analysis of any new sociotechnical practices associated with enhancing human capabilities must be constituted through sensitivity to the larger historical, cultural and political context within which they find form-and likewise should be sensitive to how technologies of enhancement might reconfigure those contexts. [bib_ref] Enhancement technologies and the body, Hogle [/bib_ref] Key here is research that engages a range of publics with decision-making around enhancements. In so doing, there is an increased likelihood that recommendations can be reached which resonate with the lived experiences of policymakers, scientists, clinicians, patients and a range of other users. Accordingly, workable solutions to the governance of contested drugs and devices associated with enhancement might more readily be achieved.
Twitter Follow Martyn Pickersgill at @PickersgillM
[table] 3: 10 29 71 72 [/table]
|
HIV Testing Implementation in Two Urban Cities: Practice, Policy, and Perceived Barriers
Background: Although funding has supported the scale up of routine, opt-out HIV testing in the US, variance in implementation mechanisms and barriers in high-burden jurisdictions remains unknown.Methods: We conducted a survey of health care organizations in Washington, DC and Houston/Harris County to determine number of HIV tests completed in 2011, policy and practices associated with HIV testing, funding mechanisms, and reported barriers to testing in each jurisdiction and to compare results between jurisdictions.Results: In 2012, 43 Houston and 35 DC HIV-testing organizations participated in the survey. Participants represented 85% of Department of Health-supported testers in DC and 90% of Department of Health-supported testers in Houston. The median number of tests per organization was 568 in DC and 1045 in Houston. Approximately 50% of organizations in both DC and Houston exclusively used opt-in consent and most conducted both pre-and post-test counseling with HIV testing (80% of organizations in DC, 70% in Houston). While the most frequent source of funding in DC was the Department of Health, Houston organizations primarily billed the patient or third-party payers. Barriers to testing most often reported were lack of funding, followed by patient discomfort/refusal with more barriers reported in DC.Conclusions: Given unique policies, resources and programmatic contexts, DC and Houston have taken different approaches to support routine testing. Many organizations in both cities reported opt-in consent approaches and pre-test counseling, suggesting 2006 national HIV testing recommendations are not being followed consistently. Addressing the barriers to testing identified in each jurisdiction may improve expansion of testing.
# Introduction
HIV testing is an important step of the HIV care continuum and a critical component of prevention programs throughout the United States (US). Nationally, an estimated 18% of infected persons remain undiagnosed.Awareness of HIV infection decreases risk-taking behaviors, [bib_ref] Meta-analysis of high-risk sexual behavior in persons aware and unaware they are..., Marks [/bib_ref] with earlier initiation of care resulting in decreased HIV-associated morbidity and mortality. [bib_ref] Declining morbidity and mortality among patients with advanced human immunodeficiency virus infection...., Palella [/bib_ref] [bib_ref] The late diagnosis and consequent short-term mortality of HIV-infected heterosexuals, Chadborn [/bib_ref] In 2006, the US Centers for Disease Control and Prevention (CDC) released revised recommendations for routine, opt-out HIV screening in healthcare settings. Specifically, the revised recommendations include screening patients for HIV using an opt-out approach, the elimination of a separate written consent for HIV testing, and optional pre-testing prevention counseling. [bib_ref] Revised recommendations for HIV testing of adults, adolescents, and pregnant women in..., Branson [/bib_ref] CDC funding has since supported the scale up of routine, optout screening nationwide,yet implementation varies widely by jurisdiction as each determines how to expand testing given diverse contexts. Initiatives to scale up testing have included social marketing campaigns in Miami, FL and Oakland, CA, door-todoor testing outreach in San Diego, CA and Philadelphia, PA, [bib_ref] Geography Should Not Be Destiny: Focusing HIV/AIDS Implementation Research and Programs on..., Nunn [/bib_ref] and policy change to mandate the offer of HIV testing in primary care settings in New York State. [bib_ref] Legislated Human Immunodeficiency Virus Testing in New York State Emergency Departments: Reported..., Egan [/bib_ref] Some models, such as that of the Bronx Knows HIV Testing Initiative, have demonstrated success through combination approaches in partnership with the local community and clinical providers. [bib_ref] Geography Should Not Be Destiny: Focusing HIV/AIDS Implementation Research and Programs on..., Nunn [/bib_ref] [bib_ref] Assessing the Impact of a Community-Wide HIV Testing Scale-Up Initiative in a..., Myers [/bib_ref] This article reports on some of the similarities and differences in HIV testing observed in two jurisdictions highly impacted by HIV, Washington DC and Houston, TX. The jurisdictions are two of the twelve urban areas that represent 44% of the nation's AIDS cases.DC has a generalized epidemic with 14,465 living HIV/AIDS cases and a prevalence of 2.7%, and Houston has 19,943 living HIV/AIDS cases and a concentrated epidemic with a prevalence of 0.6%. Each jurisdiction's local epidemic and the associated policies, programmatic priorities, and resources that influence testing implementation are reported in [fig_ref] Table 1: HIV Prevalence, Testing Policies and Programs by City [/fig_ref].
Policy in both jurisdictions has been relatively permissive of the expansion of routine HIV testing. Statutes relevant to the 2006 recommendations in DC and Texas were either neutral or consistent with all routine, opt-out recommendations. [bib_ref] Consistency of state statutes with the Centers for Disease Control and Prevention..., Mahajan [/bib_ref] Specifically, in DC, there are no regulations requiring written consent or pre/post-test counseling. The Insurance Coverage for HIV Testing in Emergency Departments (ED) Amendment Act, which was passed in 2008, attempts to facilitate third party coverage of testing in DC EDs. Under Texas Health and Safety Code, informed consent for HIV testing may be verbal as long as test explanation and consent are documented in a patient's medical record. Additionally, an opt-out approach is required by statute for pregnant women, where HIV screening is required at the first visit and during the third trimester. No statute requires pre-test prevention counseling and post-test counseling is only required upon delivery of a positive test result.Since 2006, the DC Department of Health (DOH) has supported targeted testing and routine, opt-out HIV testing implementation as a standard of care in health care settings. [bib_ref] Implementing a novel citywide rapid HIV testing campaign in, Castel [/bib_ref] The DOH began this initiative by providing free rapid test kits and funding to testing programs at community-based organizations (CBOs), clinics, DC Department of Corrections, hospitals, and EDs. Between 2007 and 2008, DC expanded testing to additional hospitals, primary medical settings, managed care organizations, and CBOs.The ''Ask for the Test/Offer the Test'' initiative launched in 2009 to educate providers on how to establish routine HIV testing protocols that emphasize the use of standard blood panels. The DOH also engaged the DC Department of Health Care Finance on promoting standard HIV testing as part of the DC Medicaid and Alliance programs. In Both DC and Houston health departments have provided support for the provision and funding of HIV testing in community and health care settings, including implementation of social marketing campaigns. HIV testing and promotion of HIV testing was further scaled up with the influx of HIV prevention activities funded by the CDC's Enhanced Comprehensive HIV Prevention Planning (ECHPP) Project in both DC and Houston. ECHPP was a 3-year demonstration project designed to improve program planning and implementation in support of the National HIV/AIDS Strategy.As part of ECHPP efforts, partnerships were formed between academic institutions and public health departments to improve program planning of HIV prevention activities in both DC and Houston. These partnerships ensured that local research capacity was improved and policy and programming stakeholders were invested in the study's outcomes, both of which are enabling factors for success in operational research. [bib_ref] Operational research in low-income countries: what, why, and how?, Zachariah [/bib_ref] While operational research has been utilized in low-income countries, [bib_ref] Operational research in low-income countries: what, why, and how?, Zachariah [/bib_ref] [bib_ref] Is operational research delivering the goods? The journey to success in low-income..., Zachariah [/bib_ref] it has been suggested that US jurisdictions could greatly benefit from applying an operational research approach to identify barriers and inform policy and decision-making in HIV prevention programs. [bib_ref] Operational research to improve HIV prevention in the United States, Herbst [/bib_ref] Therefore, in response to ECHPP's goal of improving implementation, we used descriptive operational research to gain knowledge of HIV testing implementation and organizationallevel testing barriers in Washington, DC and Houston, Texas. Although implementation and barriers have been assessed among EDs, health centers, or clinics, [bib_ref] Barriers and facilitators to enhancing HIV testing in publicly funded primary care..., Myers [/bib_ref] [bib_ref] Barriers and facilitators to routine HIV testing: perceptions from Massachusetts Community Health..., Johnson [/bib_ref] [bib_ref] Less encouraging lessons from the front lines: barriers to implementation of an..., Mumma [/bib_ref] assessment across different types of organizations and jurisdictions has been limited. This study examined implementation mechanisms, testing volume, and organizational barriers for two cities with a high HIV/AIDS burden and provides a comprehensive jurisdictional-level perspective on HIV testing.
# Methods
## Survey instrument
Surveys were designed to gather organizational-level data on sites conducting HIV testing in each jurisdiction in 2011. Selected survey questions were adapted from the National Association of Community Health Centers survey on HIV testing, [bib_ref] National survey of health center HIV testing, prevention, care and treatment practices, Charlebois [/bib_ref] a method used previously in similar research. [bib_ref] Barriers to routine HIV testing among Massachusetts community health center personnel, Mimiaga [/bib_ref] In both DC and Houston, organizations were questioned regarding their policies, funding, implementation practices, testing volume, and barriers to testing for calendar year 2011. Testing volume in 2011 was assessed by each organization providing counts by test type. Review of administrative and/or clinical data was the primary methodology used for gathering this data (80.0% of organizations in DC and 81.4% of organizations in Houston), followed by estimates from organizational staff members (17.1% DC and 11.6% Houston). Barriers to HIV testing were assessed by asking respondents to indicate the extent to which they agreed with possible barriers using a 5-point scale ranging from no barrier (1) to major barrier [bib_ref] Revised recommendations for HIV testing of adults, adolescents, and pregnant women in..., Branson [/bib_ref]. Possible barriers included attitudinal conflicts, staff capacity, and available resources and funding.
## Sampling frame
Sampling frames were designed to best fit the model of HIV prevention activities unique to each jurisdiction. In DC, the sampling strategy was developed through a partnership between DC DOH and researchers at the George Washington University as part of ECHPP. In 2011, DC DOH's HIV/AIDS, Hepatitis, STD, and TB Administration (HAHSTA)'s prevention model primarily supported HIV testing through distribution of HIV rapid test kits at no cost to approximately forty organizations. With stakeholder and DOH input, it was determined that these organizations conducted the vast majority of testing within the city. Therefore, organizations were identified for survey inclusion if their HIV testing services were HAHSTA-supported. These organizations included CBOs, clinics, and hospitals.
In Houston, the sampling frame and sampling tool were designed jointly by staff and researchers from HDHHS, Houston Area HIV Services Ryan White Planning Council's Office of Support, and Baylor-UTHouston Center for AIDS Research, also as part of ECHPP. The HDHHS funded two distinct HIV testing services at 10 organizations in 2011: 1) targeted HIV testing by CBOs that included prevention counseling, and 2) routine, opt-out HIV testing in EDs and county clinics. HIV testing in Houston was supported by a diverse mix of both public and private support, thus sampled organizations included both those that were and were not financially supported by the HDHHS. Utilizing a methodology described previously, [bib_ref] Assessing HIV testing and linkage to care activities and providing academic support..., Giordano [/bib_ref] an exhaustive list of known HIV testing organizations was created from stakeholder input. The list included CBOs, substance abuse treatment centers, homeless shelters, hospitals, clinics and universities. This list was then prioritized to focus on obtaining information from HDHHS HIV prevention contractors, major public and private hospitals and clinics, and known HIV testing organizations.
## Survey administration
In DC, surveys were emailed to HIV testing coordinators at each organization. Confidential surveys were administered online using Research Electronic Data Capture (REDCap). If no response was received within one month of invitation, paper surveys were mailed in addition to follow up phone calls. Participants were compensated for survey completion.
In Houston, each organization was emailed a formal invitation letter, followed by phone calls if there was no response via email. Initial contact was with HIV program directors, nurse managers, and/or lab directors. Respondents were selected based on their authority to make HIV testing program decisions and/or their knowledge of HIV testing activities within the organization. When possible, surveys were completed by an in-person interview, otherwise a telephone interview was conducted. Participants were not compensated for survey completion.
# Data analysis
In DC, data collection and storage were via REDCap. Data analysis was conducted using SAS 9.2 (SAS Institute, Cary, NC). In Houston, all data were entered and managed in Excel. Data analysis was conducted using SAS 9.3 (SAS Institute, Cary, NC).
Summary data from both jurisdictions were compiled via Excel and analyzed collaboratively by researchers from both jurisdictions.
## Consent
This study was approved by the Institutional Review Boards of The George Washington University, the District of Columbia Department of Health (Washington, DC) and the Baylor College of Medicine (Houston). In Houston, the Institutional Review Board approved verbal consent, with completion of the survey as documentation of consent. Consent was also documented in a secure database by the interviewer. Written consent would have been the only record linking the participant and survey; therefore, the potential harm resulting from a breach of confidentiality would have increased since no other personal identifiers were gathered. In DC, the Institutional Review Board approved the use of written consent. The survey instructions asked participants to mark ''yes'' or ''no'' to indicate their consent to the survey prior to answering any further questions. In both jurisdictions, the survey was voluntary and deemed no more than minimal risk to participants.
# Results
## Respondents & organizational characteristics
In DC, 41 organizations were contacted for study inclusion. Of these, 35 (85.4%) organizations participated in the survey, representing 103 facilities that conducted HIV testing. Many of the DC organizations were community-based organizations/ community service organizations (40.0%) [fig_ref] Table 2: HIV Testing Policy, Funding, and Implementation Practices [/fig_ref].
In Houston, 84 organizations were contacted for study inclusion. Of these, 55 (65.5%) organizations participated in the survey. Forty-three participating organizations, representing 114 facilities in the Houston/Harris County area, conducted HIV testing. This analysis focused only on the 43 organizations conducting HIV testing of which over half (51.2%) were outpatient clinics or university health centers [fig_ref] Table 2: HIV Testing Policy, Funding, and Implementation Practices [/fig_ref].
## Policy and practices, funding, and testing volume
The majority of organizations reported that a specific individual coordinated HIV testing efforts for the organization. DC organizations more frequently reported a coordinator than did Houston (82.9% and 62.8%, respectively) [fig_ref] Table 2: HIV Testing Policy, Funding, and Implementation Practices [/fig_ref]. Similarly, 82.9% of surveyed organizations in DC reported having a written policy for HIV testing, while only 46.5% of surveyed organizations in Houston reported having a policy for their organization.
The two jurisdictions reported similar approaches to provider education about HIV testing. The top three sources of provider education for DC were continuing education (74.3%), DOH or clinic training (65.7%), and peer-to-peer best practices (57.1%). Houston organizations most frequently identified DOH or clinic training (58.1%), manuals/guidelines/literature (48.8%), and continuing education (46.5%).
A similar proportion of organizations exclusively used an opt-in consent approach in DC and Houston (51.4% and 48.8% respectively). However, a substantial number of organizations in Houston used a combination of opt-in and opt-out approaches (44.2%). Further analysis revealed that 11.6% (n = 5) of organizations in Houston using a combination approach only used optout consent during pregnancy. While only three (7.0%) organizations in Houston reported using only the opt-out approach, 48.6% of the surveyed organizations in DC exclusively used an opt-out approach.
Despite CDC recommendations that removed the need for pretest counseling, most organizations in both jurisdictions reported conducting both pre-and post-test counseling with HIV testing (80.0% in DC and 69.8% in Houston). Very few organizations reported conducting only post-test counseling (11.4% in DC and 16.3% in Houston).
For all HIV testing conducted by surveyed organizations, the most frequent funding sources for DC were the DOH (74.3%), research monies (34.3%), and federal funding other than that from CDC and DOH (33.3%). In contrast, Houston organizations most frequently identified the payer as the patient or the patient's insurance (65.1%), Medicaid (53.5%), Medicare (32.6%), and other sources (32.6%) such as the organization's general revenue or funds received as donations. Funding specifically for routine, opt-out HIV testing also varied between the two cities. The most frequent funding source in DC was the DOH (68.6%), while Houston organizations primarily paid for HIV testing by billing the patient or patient's insurance (54.5%) or Medicaid (50.0%).
The median total number of HIV tests performed per organization in 2011 was higher in Houston than in DC (1045 and 568 respectively). While the total number of rapid tests performed in DC was higher than in Houston (78,765 in DC vs. 40,910 in Houston), the median number of rapid tests performed per organization was lower in DC than in Houston (593 vs. 1478). Similarly, although the median number of venipuncture tests per organization was higher in DC (1200 in DC vs. 585 in Houston), the total number of venipuncture tests was higher in Houston than in DC (169,635 in Houston vs. 4,558 in DC).
## Barriers to hiv testing
On a scale of 1 (no barrier) to 5 (major barrier), the highest reported barriers to HIV testing in both DC and Houston were patient discomfort/refusal (median score = 2) and lack of funding for testing (median score = 3). When categorizing all items dichotomously (no barrier vs. any barrier), DC organizations most frequently selected ''HIV not a problem for client population'' (68.6% of organizations), ''lack of funding'' (60.0%), and ''patient discomfort or refusal for testing'' (59.4%) as barriers [fig_ref] Figure 1: Attitudinal barriers to HIV testing by City [/fig_ref]. Houston organizations most frequently selected ''lack of funding'' (59.5% of organizations), ''patient discomfort or refusal for testing'' (51.2%), and ''staff knowledge, skill, experience'' (41.9%).
In comparison to Houston organizations, DC organizations more frequently reported every area as a barrier except for staff knowledge, skill, experience (41.9% Houston vs. 32.4% DC) and provider/staff resistance (20.9% vs. 20.6%) [fig_ref] Figure 1: Attitudinal barriers to HIV testing by City [/fig_ref]. However, 11 Houston organizations (25.6%) scored ''other'' barriers to testing at 3 or higher, including managing data and/or paperwork associated with testing and funding for testing (n = 4) and challenges in contacting patients for results delivery and/or care referrals (n = 3). The largest variances observed between the two jurisdictions were ''HIV is not a problem for the client population'' (20.9% Houston vs. 68.6% DC), ''limited staff time to provide testing'' (30.2% vs. 52.9%), and ''limited staff size to provide testing'' (37.2% vs. 58.8%).
We examined whether barriers to testing implementation differed by organization type (CBOs, hospitals, and clinics; data not shown). Lack of funding was reported frequently across all organizational types. Clinics were particularly challenged by cost or reimbursement for testing and patient refusal, while CBOs reported staff size as a barrier. Among hospitals, the highest ranked barrier to testing in both jurisdictions was the perception that HIV was not a problem in their client population. In DC, barriers to testing were most frequently reported among hospitals in comparison to other site types. The most frequent barriers among DC hospitals included the perception that HIV is not a problem in the client population (83.3% of organizations), staff Community service organizations include homeless services, substance abuse recovery centers, life skills programs, housing assistance programs, faith-based organizations, and other community-oriented service organizations. b Five organizations participating in the Houston survey only use opt-out consent during pregnancy as required by Texas law but do not use opt-out consent otherwise. Therefore, the bulk of consent used for these organizations would be opt-in. Results if these organizations are re-classified into the ''opt-in approach only'' category: 26 (60.5%) organizations use opt-in approach only, 3 (7.0%) organizations use opt-out approach only, and 14 (32.6%) organizations use a combination of opt-in and optout consent. time (75.0%), staff size (75.0%), and lack of funding (66.7%). In contrast, in Houston, hospitals reported the fewest barriers, while clinics reported the most. The most frequent barriers among Houston clinics included lack of funding (68.2%), refusal to get tested (63.6%), cost/reimbursement (50.0%), and staff knowledge/ skill/experience (50.0%).
# Discussion
Utilizing a framework of operational research for HIV prevention, [bib_ref] Operational research to improve HIV prevention in the United States, Herbst [/bib_ref] this study identified differences and similarities in the approaches two high prevalence cities have taken to implement routine HIV testing. While both jurisdictions reported a high level of provider education on HIV testing, many organizations in both cities reported opt-in consent approaches and pre-test counseling, suggesting the 2006 CDC recommendations are not being followed consistently. The cause of inconsistency merits further research since statutes do not pose a substantial barrier in either jurisdiction. [bib_ref] Consistency of state statutes with the Centers for Disease Control and Prevention..., Mahajan [/bib_ref] Previous implementation research has shown that this inconsistency may be due to 1) a lack of awareness or misunderstanding of the recommendations, [bib_ref] Barriers and facilitators to routine HIV testing: perceptions from Massachusetts Community Health..., Johnson [/bib_ref] [bib_ref] Internal Medicine and Emergency Medicine Physicians Lack Accurate Knowledge of Current CDC..., Mohajer [/bib_ref] [bib_ref] Are VA primary care providers aware of HIV testing recommendations for Veterans?..., Arya [/bib_ref] disagreement with the recommendations, [bib_ref] Patient and clinician ethical perspectives on the 2006 Centers for Disease Control..., Merchant [/bib_ref] and/ or 3) organizational barriers that impede application of the recommendations. [bib_ref] Barriers and facilitators to routine HIV testing: perceptions from Massachusetts Community Health..., Johnson [/bib_ref] [bib_ref] Less encouraging lessons from the front lines: barriers to implementation of an..., Mumma [/bib_ref] [bib_ref] Barriers to routine HIV testing among Massachusetts community health center personnel, Mimiaga [/bib_ref] [bib_ref] Providers' perceptions of the factors influencing the implementation of the New York..., Schnall [/bib_ref] [bib_ref] Development of a measure of barriers to HIV testing among individuals at..., Awad [/bib_ref] Both cities may benefit from further research into why organizations continue to use both preand post-test counseling and do not use opt-out testing.
The two jurisdictions have taken different approaches to support and fund the expansion of routine testing. DC DOH's approach has involved considerable funding for the distribution of free rapid test kits to local implementation sites including clinics, hospitals and CBOs. This is in contrast to the HDHHHS' approach which has encouraged performing venipuncture tests as part of routine visits and funding large volumes of HIV testing in hospitals and clinics with rapid results of standard testing. [bib_ref] Using nonrapid HIV technology for routine, opt-out HIV screening in a high-volume..., Hoxhaj [/bib_ref] This divergence of approaches is reflected in the greater total number of rapid tests being utilized in DC and much greater total number of venipuncture tests being performed in Houston.
Funding sources for testing also varied between the two jurisdictions which could have implications for the long term sustainability and scalability of these testing programs. Houston testing organizations reported much more third party billing and reimbursement from Medicaid, Medicare, and private insurance as a primary means to support their testing programs, compared to DC sites which relied much more heavily on direct support from DC DOH. Because of this, Houston organizations may be better positioned to more quickly adapt and benefit from expanded coverage through the Affordable Care Act (ACA), especially now that routine testing is rated a recommended service for coverage by the US Public Health Services Task Force (USPHSTF). However, this strong reliance on third party reimbursement does leave these organizations more vulnerable to reimbursement challenges and inadequate reimbursement rates that may not cover the total cost of testing provision. The monitoring of HIV testing post-ACA and USPHSTF implementation will allow for further exploration of the impact of these policies on testing programs.
Overall, few barriers to testing implementation were reported by organizations in both jurisdictions however, among those reported, lack of funding and organizational capacity to conduct testing were key barriers. Capacity barriers, such as staff size and time were moderate barriers for the surveyed organizations and may have been related to lack of funding to support these efforts at the organizational level, particularly among organizations in DC. Although multiple methods of routine HIV testing have been promoted in DC, there is more reliance on rapid testing, which can be more time and labor-intensive than venipuncture testing. Time-related barriers and competing priorities have been extensively reported in literature, [bib_ref] Barriers and facilitators to enhancing HIV testing in publicly funded primary care..., Myers [/bib_ref] [bib_ref] Barriers and facilitators to routine HIV testing: perceptions from Massachusetts Community Health..., Johnson [/bib_ref] [bib_ref] Less encouraging lessons from the front lines: barriers to implementation of an..., Mumma [/bib_ref] [bib_ref] Barriers to routine HIV testing among Massachusetts community health center personnel, Mimiaga [/bib_ref] [bib_ref] Internal Medicine and Emergency Medicine Physicians Lack Accurate Knowledge of Current CDC..., Mohajer [/bib_ref] [bib_ref] Providers' perceptions of the factors influencing the implementation of the New York..., Schnall [/bib_ref] [bib_ref] A provider participatory implementation model for HIV testing in an ED, Chen [/bib_ref] [bib_ref] General internists' beliefs, behaviors, and perceived barriers to routine HIV screening in..., Korthuis [/bib_ref] but research has found that providers cite consent and counseling requirements as time-intensive even when statutes have removed these barriers in a jurisdiction. [bib_ref] Barriers and facilitators to enhancing HIV testing in publicly funded primary care..., Myers [/bib_ref] Interestingly, despite the generalized epidemic and high HIV prevalence in DC, a commonly reported barrier included the perception that HIV was not a problem among the patient population. This perception may depend on the age of the population being served as a previous survey conducted by DC DOH found that almost 40% of providers did not perceive HIV to be a problem among those 50 years of age and older . Barriers elicited in this study suggest that DC may benefit from campaigns that emphasize the necessity of HIV testing, while Houston may benefit from increasing testing knowledge and skill among providers. Patient discomfort or refusal for testing was a barrier in both cities and has been reported by providers elsewhere [bib_ref] Barriers and facilitators to enhancing HIV testing in publicly funded primary care..., Myers [/bib_ref] [bib_ref] Barriers and facilitators to routine HIV testing: perceptions from Massachusetts Community Health..., Johnson [/bib_ref] [bib_ref] General internists' beliefs, behaviors, and perceived barriers to routine HIV screening in..., Korthuis [/bib_ref] with clients often refusing due to low-perceived risk for HIV, recent testing elsewhere, [bib_ref] Patient perceptions and acceptance of routine emergency department HIV testing, Brown [/bib_ref] [bib_ref] Risk, reasons for refusal, and impact of counseling on consent among ED..., Ubhayakar [/bib_ref] fear of loss, fatalism, confidentiality concerns, and structural barriers. [bib_ref] Development of a measure of barriers to HIV testing among individuals at..., Awad [/bib_ref] Future qualitative research in both DC and Houston, focusing first on the commonly reported barriers presented here, may best elicit additional barriers and facilitators of sustained HIV testing.
This study determined current implementation practices and perceived barriers to HIV testing, a crucial step to design interventions and guide policy to improve HIV testing in cities with a high burden of HIV, such as DC and Houston. In 2010, New York State enacted a statute requiring primary care settings to offer HIV testing to all patients between the ages of 13 to 64. Since this law was enacted, one study found that only 65% of emergency departments implemented HIV testing as required by the law. [bib_ref] Legislated Human Immunodeficiency Virus Testing in New York State Emergency Departments: Reported..., Egan [/bib_ref] While changes to policy guidance at the jurisdictional level may increase testing, [bib_ref] Effects of written informed consent requirements on HIV testing rates: evidence from..., Wing [/bib_ref] statute alone is not the solution. [bib_ref] Legislated Human Immunodeficiency Virus Testing in New York State Emergency Departments: Reported..., Egan [/bib_ref] [bib_ref] Providers' perceptions of the factors influencing the implementation of the New York..., Schnall [/bib_ref] Further increases in testing uptake may be realized with policy change and implementation at the organizational level, [bib_ref] Simplifying consent for HIV testing is associated with an increase in HIV..., Zetola [/bib_ref] [bib_ref] Greater HIV testing after Veterans Health Administration policy change: the experience from..., Nayak [/bib_ref] especially in jurisdictions with low uptake such as some organizations in Houston. Organizational leadership to drive local policy formulation, revision, and implementation is critical. Other facilitators include organizational buy-in from both providers and administration, [bib_ref] Barriers and facilitators to routine HIV testing: perceptions from Massachusetts Community Health..., Johnson [/bib_ref] [bib_ref] Less encouraging lessons from the front lines: barriers to implementation of an..., Mumma [/bib_ref] written procedures for HIV testing, [bib_ref] Barriers to routine HIV testing among Massachusetts community health center personnel, Mimiaga [/bib_ref] provider trainings, [bib_ref] Barriers and facilitators to routine HIV testing: perceptions from Massachusetts Community Health..., Johnson [/bib_ref] and dedicated staff and funding. [bib_ref] Providers' perceptions of the factors influencing the implementation of the New York..., Schnall [/bib_ref] There are a number of limitations to this study. The survey was a convenience sample that could be subject to non-response bias. The self-reported responses were limited to the organizational knowledge, perceptions and/or experiences of the person completing the survey, thus the responses may not accurately reflect the practices or opinions of all providers or staff members. The primary limitation of the parts of this study that compared DC and Houston was differences in survey administration between the two cities. Non-response was potentially influenced in Houston (65.5% vs. 85.4% in DC) by the lack of compensation for survey completion. The differing routes of survey administration (inperson in Houston and online in DC) may have also led to variances in social desirability or comprehension of the survey questions. In addition, social desirability may have biased the results, especially in reporting of testing barriers. Furthermore, our ability to statistically analyze barrier data was limited as many of the barriers received low rankings. Finally, generalizability of these findings may be limited, particularly in jurisdictions that have statutes in conflict with the 2006 CDC recommendations. Results presented herein may be useful to such jurisdictions after their statutes transition.
Despite both DC and Houston having high HIV prevalence, each jurisdiction has tailored their HIV testing strategies to meet particular funding options, resources, and statutes. The range of approaches was appropriate given the distinct context of the epidemic in each jurisdiction. Reported implementation practices, such as types of tests used and the funding mechanisms most often used to pay for HIV testing, highlight this variation. Regardless of different implementation approaches, our results suggest many organizations in at least two high burden jurisdictions have not aligned with all aspects of national testing recommendations. Research has shown that neither awareness of CDC guidelines nor jurisdictional policy change alone fully scales up testing to the levels recommended in the 2006 guidelines. [bib_ref] Legislated Human Immunodeficiency Virus Testing in New York State Emergency Departments: Reported..., Egan [/bib_ref] [bib_ref] Internal Medicine and Emergency Medicine Physicians Lack Accurate Knowledge of Current CDC..., Mohajer [/bib_ref] [bib_ref] General internists' beliefs, behaviors, and perceived barriers to routine HIV screening in..., Korthuis [/bib_ref] It is likely that a multi-faceted approach is needed that includes awareness campaigns, policy change, and reduction of jurisdictional and organizational-specific barriers. Few multi-faceted approaches have been rigorously evaluated to determine if they can widely span multiple jurisdictions with diverse epidemics, policy, and resources. Further implementation research and dissemination of implementation research findings is critical to the expansion of routine HIV testing and early diagnosis of HIV infection.
[fig] c: Respondents ould hek more than one response. d Other responses in DC inlude general revenue (n = 1). Other responses in Houston inlude donations (n = 12) and general revenue (n = 2). e Other responses in DC inlude general revenue (n = 1). Other responses in Houston inlude donations (n = 8) and general revenue (n = 2). One Houston organization seleted both donations and general revenue for a total of n = 9 organizations seleting ''other''. f One Houston organization estimated 1-2 tests were ompleted. The average, 1.5 tests, was reorded as the response. doi:10.1371/journal.pone.0110010.t002 [/fig]
[fig] Figure 1: Attitudinal barriers to HIV testing by City. doi:10.1371/journal.pone.0110010.g001 [/fig]
[fig] Figure 2: Resources and funding barriers to HIV testing by City. doi:10.1371/journal.pone.0110010.g002 [/fig]
[fig] Figure 3: Staff capacity barriers to HIV testing by City. doi:10.1371/journal.pone.0110010.g003 [/fig]
[table] Table 1: HIV Prevalence, Testing Policies and Programs by City.Mandatory testing of all offenders upon entry and release within Texas Department of Criminal Justice Texas law allows for mandatory testing in county and municipal jails Texas law allows for mandatory testing in county and municipal jails ETI) started in 2008 when the Houston Department of Health and Human Services (HDHHS) began supporting routine, opt-out HIV testing in the EDs of two large hospitals and two community health centers. The Initiative grew to encompass four large EDs by 2009. With funding from the Texas Department of State Health Services (DSHS), routine testing was further expanded in 2010 to five additional EDs and twelve additional community health centers. Targeted testing was also funded by the HDHHS at CBOs for outreach to high risk populations and geographic areas. In 2011, the HDHHS funded over 107,000 tests in the Houston area. [/table]
[table] Table 2: HIV Testing Policy, Funding, and Implementation Practices. [/table]
|
Recapitulation of selective nuclear import and export with a perfectly repeated 12mer GLFG peptide
In this manuscript, Ng et al. aim at engineering an artificial barrier in vitro to mimic what happened in the native nuclear pore by using extracted GLFG12mer peptides and then testing the selectivity of the GLFG-assembled barrier in nuclear transport. In more detail, three are three major steps included in the draft: 1) in vitro mimic of the NPC's selectivity barrier by using GLFG domains; 2) test importinand exportin-mediated cargo transport through this artificial barrier; and 3) use DAPI/ Hoechst dyes to block nuclear transport after UV-photo-induction. The major contribution of this study is mainly from the first two measurements, however, the similar experiments and conclusions have been published from the corresponding author's lab 15 years ago (Frey et al. Science, 314:815, 2006). Also, the manuscript was poorly written by having many inaccurate and confusing statements, which made it hard to read. Although substantial data was presented, more importantly, both the significance and innovations of this manuscript are very limited. Thus, this reviewer doesn't support to publish it in Nature Communications.The manuscript was not well organized by including numerous errors. Only some of the major and minor issues are listed below: 1) Given its relatively small size, NTF2 (30 kDa) is likely to freely diffuse into the Nucleus when not bound to RanGAP. This makes the assumption that this transport factors interaction with the FG motifs is evidence of a selectively permeable FG region imperfect. Given that the key thesis of this paper is to reproduce the selectivity barrier, it is problematic that they have chosen a nuclear transport factor smaller than the diffusion barrier.2) In the background, the author references irregular inter-FG spacers. Then in the results section, design their experimental FG region to move hydrophobic residues from within the inter-FG spacers to the FG motifs. The paper would benefit from a brief explanation in the background explaining why (if at all) FG spacers are important. Authors state they adjusted inter-FG spacers to be an identical length. What length? This is important, as they moved the hydrophobic amino acids from the spacers to the FG regions, and the distance between the regions will provide insight into the efficacy of the system being a result of the FG/FG-like motifs or if it is a byproduct of a concentration of hydrophobic amino acids. Are the spacer regions intrinsically disordered?3) As for the experiments on NTF2, there are many more issues to be taken care of as follows: A) It should be specified in the text that the NTF2 is tagged with Alexa488; B) The FG regions are derived from algae (Tetrahymena thermophila) and the NTF2 is derived from Rat. It is unclear how similar the engineered FG regions are to the native FG regions within Rats. C) Where on the NTF2 is the Alexa488 binding? Are they using an antibody labeling strategy or is the NTF2 covalently bound to the Alexa488? D) This is critical to know asfig 1 and 2depict key evidence for their central thesis, specifically, that their FG aggregates selectively interact with NTF2, but not mCherry. Depending upon where the Alexa488 is associated with NTF2, the FG interaction region could be occluded. This would then imply that the interaction is driven by Alexa488 (a highly charged and prolific binder) and not NTF2. E) Transport times are compared to Imp-B1 import, are dynamics of NTF2 dynamics known in any capacity? The comparison of import dynamics needs to recognize how imperfect of a comparison this is.4) This sentence "The selective phase of NPCs is just one extreme in a range of similar biological condensates that govern developmental programs, RNA-metabolism, RNP assembly, stress, and disease conditions30-37." was listed as a separated paragraph? Is this a typo or real? 5) Many statements are inaccurate and confusing throughout the manuscript. For example, in the
abstract it says that "NPCs contain numerous distinct FG domains, each comprising variable repeats.", but soon after in the section of introduction, the authors stated that "Vertebrate NPCs contain ≈10 different FG domains". It's hard to guess which one that the authors want readers to follow.
6) It might be too arbitrary and inaccurate by stating that "the barrier can be described as a condensed phase assembled from cohesive FG repeat domains." because there are still disputes about the assembly of the NPC's selectivity barrier and also many experiments argued against the hypothesis of "a condensed phase" in the NPC.
7) It is stated that Serine is used to replace Proline in the proline free FG sequence. However, the logic for why they chose to remove proline is not readily apparent to me.
8) The results highlight that the contents of the inter-FG spacers do have an impact upon function of the FG region. The background section should include a statement summing up current thinking on the function of the inter-FG spacers. 9) Authors state that the hydrophobic side chain on the surface of GFP enhances FG interactions, this is to be expected. That said, why is there an explanation why the relatively hydrophobic mCherry does not interact with the FG phases? (source for mCherry being much more hydrophobic than DsRed: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3394910/) Reviewer #2: Remarks to the Author: Ng et al create a synthetic nuclear transport permeability barrier with minimalistic composition. In four intermediates steps they created a synthetic protein that contains 52 identical GLFG_12mers, all of which phase separate in a similar way as compared to wt protein. Sequence variation affected amyloid forming propensity. The authors further identify a contribution of the GLEBS domain to phase separation. Various experiments proof that the engineered FG-phase recapitulates the cargo/NTR entry and release behavior of 'non-engineered' FG-Nups.
A very laudable innovation is the demonstration of cargo complex release upon addition of RanGTP-RFP. Also the Ran-GTP facilitated entry of an export complex is a very nice experiment! Those data finally show that in vitro FG-phases also recapitulate the most essential steps of active nuclear transport.
Otherwise, different fluorescent dyes are tested for entry into the phase. The authors demonstrate that dye attachment to proteins entering the permeability barrier affects their partitioning coefficient and that NPCs can be inactivated by illumination, both of which are very neat experiments and important for the field.
The paper is very interesting. Why there are so many different types of FG repeats is a major outstanding question. The demonstration that this variation, although conserved, is not required to recapitulate active nuclear transport in vitro, comprises an important step forward. As always, the paper is technically well done and taken together, a strong candidate for publication in Nat Comm after some revisions.
## Some comments:
Having that said, the weakness of the paper is that it is solely based on an in vitro approach. Engineering an in vivo NPC that would only contain a single engineered FG-Nup is challenging. Although seemingly feasible in yeast if done gradually, it would be lots of work and thus beyond the scope of the present study. However, one really wishes to know if a yeast strain with all the same FG repeats would be viable or alternatively, gradually develop fitness defects once native FG-Nups are changed one by one. You could at least refer to in the discussion. I understand that this is not the exact same idea. It is still interesting in this context that many repeats can be removed in vivo and cells are still viable.
Style: As pointed out above, this reviewer appreciates the middle part of the manuscript, but it does interrupt the flow of the paper. I do not have a good suggestion where to place it elsewhere but you may consider trimming it down -or integrate it better.
Page 13: Why did the authors use an RFP fusion of Ran? This should perfectly work with native RanQ69LDeltaC-GTP, correct? It would be important to know if release becomes faster, consistent with the idea that non-tagged Ran can diffuse into the phase -or not.
Introduction on page 4: While the 3rd paragraph emphasizes the remarkable sequence conservation of Nup98FG, the 4th paragraph states that 'substantial progress has been hampered by the sequence diversity within given Nup98FG domains'. For readability, you may want to include an additional sentence explaining the concept of comparing sequences of FG repeats within the same protein to contrast previous paragraph better (most people think in terms of comparing sequences across proteins).
Related to that, if the different properties of individual repeats are conserved across different species but a simple repeat recapitulates nuclear transport, what could be the selection pressure?
Some of the authors have previously shown that MacNup98 is bit special in the sense that it contains many glycines and forms a very tight hydrogel already at comparably low intra-particle FG domain concentrations (eLife paper from 2013). Why is it a well-justified choice for this study that hopefully allows for generic conclusions? The statement that it is well-characterized is understandable. However, it makes the reader wonder about the relevance, because the source species is not exactly a main stream model organism. A bit of additional introduction may be helpful.
The authors state that they expect Hoechst dyes to become useful for staining FG-phases in vitro, but they also show that even a single dye attached to GFP has a quite massive effect. This does not come across very sound and maybe explained better. What type of experiments could be done despite the fact that the dye massively changes the given phase?
This reviewer does not like the term 'authentic NPC', because it inversely suggests that phase separated FG-bodies are 'non-authentic NPCs'. Why can't it just be an 'NPC'?
Reviewer #3: Remarks to the Author: The underlying mechanisms governing the selective barrier imposed by the nuclear pore complex (NPC) remains to be fully understood. This paper takes an innovative protein-engineering approach to define a "simple" sequence that can recapitulate the fundamental selective sorting properties of the NPC in vitro. By beginning with a Nup98 FG-domain from Tetrahymena, the authors sequentially minimize the sequence to a near perfect repeating GLFG peptide. A critical control is a similar peptide lacking proline residues. Using these reagents, they demonstrate that they can form cohesive networks in vitro that phase-separate into condensates that can recapitulate both active import and export while excluding an inert mCherry protein. Along the way, they also interrogate how these condensates interact with a series of fluorescent probes. The major advance here is that they demonstrate that DAPI can crosslink the FG-network, which could provide a useful tool for the field. In general, the data are of high quality and are rigorous. The manuscript is written well although it is a bit disjointed due to the "detour" exploring interactions with probes. It should make a valuable contribution to Nature Communications as it will provide valuable concepts and tools for multiple fields.
Major Point 1: The paper is written with considerable tunnel vision as to how the NPC acts as a selective barrier as it is solely focused on the importance of cohesive interactions. It would be nice if the authors at least considered that there are non-cohesive FG-networks in the NPC that are known to be important functionally. No additional data is required, but a discussion of the disconnect, for example, between the ability of the GLFG-domain to exclude mCherry in vitro and the observation that mCherry can freely pass through the NPC in vivo would seem warranted, particularly in the context of work defining the NPC barrier in vivo as being "soft" (e.g. Timney, JCB).
Major Point 2: Hydrogel versus liquid? A description of the condensates in terms of their physical characteristics would be helpful. Although it is acknowledged that there is a continuum of physical states between liquids and gels, there is also the concept that cohesive interactions can transition or "age" between liquids and more solid states, the latter being less functional/pathological. The photobleaching in may, for example, be indicating that the GLEBS domain drives a gel-like state. Again, I don't think any additional experimentation is needed, but some discussion of this in the context of the phase-separation literature should be considered.
## Major changes to the manuscript:
- We have expanded the first part of the introduction to make clear that NPCs display great transport selectivity already towards GFP-sized mobile species. This addresses comments by reviewers 1 and 3.
- We have expanded the Introduction to cover also non-cohesive FG domains. This addresses comments by reviewers 1, 2, and 3.
- We have included three additional ± NTR controls to document the transport selectivity of our perfectly repeated FG phase. The effects are quite striking, covering factors of 200 to 3500. This addresses a concern of reviewer 1.
- We expanded the Perspectives by a paragraph on non-cohesive FG domains and their possible cooperation with the Nup98 phase system.
## Answers to the reviewers' comments
(for clarity, we repeat their points in blue in front of each of our replies)
Reviewer #1 (Remarks to the Author):
In this manuscript, Ng et al. aim at engineering an artificial barrier in vitro to mimic what happened in the native nuclear pore by using extracted GLFG12mer peptides and then testing the selectivity of the GLFG-assembled barrier in nuclear transport. In more detail, three are three major steps included in the draft: 1) in vitro mimic of the NPC's selectivity barrier by using GLFG domains; 2) test importin-and exportin-mediated cargo transport through this artificial barrier; and 3) use DAPI/ Hoechst dyes to block nuclear transport after UV-photoinduction. The major contribution of this study is mainly from the first two measurements, however, the similar experiments and conclusions have been published from the corresponding author's lab 15 years ago . This is an incorrect account of our 2006 Science paper, where we demonstrated for the first time cohesive FG domain interactions, FG hydrogel formation and provided evidence for inter FG cohesion being required for viability. However, the paper did not contain a single piece of data addressing transport selectivity and did not contain any experiment to simplify the sequence of the FG domain while keeping it cohesive, nor any attempt to show that sequence heterogeneity of the native FG domain is not important.
Also, the manuscript was poorly written by having many inaccurate and confusing statements, which made it hard to read. Although substantial data was presented, more importantly, both the significance and innovations of this manuscript are very limited. Thus, this reviewer doesn't support to publish it in Nature Communications.
The manuscript was not well organized by including numerous errors. Only some of the major and minor issues are listed below: 1) Given its relatively small size, NTF2 (30 kDa) is likely to freely diffuse into the Nucleus when not bound to RanGAP. This makes the assumption that this transport factors interaction with the FG motifs is evidence of a selectively permeable FG region imperfect. Given that the key thesis of this paper is to reproduce the selectivity barrier, it is problematic that they have chosen a nuclear transport factor smaller than the diffusion barrier.
First, there is no reason to believe that RanGAP binds NTF2. RanGAP triggers RanGTP to hydrolyze its nucleotide, while NTF2 mediates the nuclear import of RanGDP.
Second, it is an error to assume that 30kDa proteins freely diffuse through NPCs. Instead, nuclear pores are highly selective already at this size level (see: Frey et al., 2018 and references therein). We have now expanded the first part of the introduction to explain this point.
NTF2 (29 kDa) passes NPCs ~700 times faster than mCherry (26.5kDa) and even 6000 times faster than the most surface-inert GFP variant (26.7kDa). These differences in NPC-passage rates are remarkably well represented by differences in FG phase-partition coefficients.
There is a straightforward explanation of why NPCs need to be selective towards 20-30kDasized proteins, which relates to Ran (25kDa). NPCs need to suppress the dissipation of the nucleocytoplasmic RanGTP-gradient, thus restricting the efflux of RanGTP and ensure that only NTR-bound RanGTP exits cell nuclei (otherwise, the coupling between GTP-hydrolysis with transport would break down). This concept is supported by the fact that RanGDP subsequently employs NTF2 for efficient retrieval into nuclei. Thus, the comparison of 30kDa mobile species is highly relevant for NPC function and was consciously chosen.
The reviewer is also incorrect in implying that we had restricted our analysis to small mobile species. He/ she missed that we also studied the tetrameric GFP NTR _3B7C (110ka), Importin β-IBB GFP complexes (130kDa), Xpo1-export complexes (170kDa), as well as importin β·IBB-GFP-MBP-M9·transportin complexes (270kDa) (in Figs. 6, 8 and 9).
2) In the background, the author references irregular inter-FG spacers. Then in the results section, design their experimental FG region to move hydrophobic residues from within the inter-FG spacers to the FG motifs. The paper would benefit from a brief explanation in the background explaining why (if at all) FG spacers are important.
The spacers are foremost required for adjusting the overall FG-density within the FG domains. Spacer-less (GLFG)n polymers will form a very compact phase that is impossible to cross by any NTR.
Different FG domains feature different types of spacers, which in turn can have a profound impact on the phase behavior. Highly charged spacers, for example, confer high water solubility and counteract cohesive interactions. We have expanded on that in the Introduction, Results
and Discussion, as well as in the Perspectives.
Authors state they adjusted inter-FG spacers to be an identical length. What length?
Repeat units have been adjusted to a length of 12 residues. If we define "GLFG" as the FG motif, this leaves eight residues for the spacers. A narrower definition of "FG" as the FG motif leaves ten residues as a spacer. This should be clear from (apparently missed by this reviewer) and the text.
This is important, as they moved the hydrophobic amino acids from the spacers to the FG regions, and the distance between the regions will provide insight into the efficacy of the system being a result of the FG/FG-like motifs or if it is a byproduct of a concentration of hydrophobic amino acids. Are the spacer regions intrinsically disordered?
The entire domain is intrinsically disordered, and so are the spacers.
3) As for the experiments on NTF2, there are many more issues to be taken care of as follows:
A) It should be specified in the text that the NTF2 is tagged with Alexa488;
The fact that NTF2 has been labelled has been mentioned four times in the figure legends as well as in the Methods and the supplements.
B) The FG regions are derived from algae (Tetrahymena thermophila) and the NTF2 is derived from Rat. It is unclear how similar the engineered FG regions are to the native FG regions within Rats.
Tetrahymena is a ciliate and, as such, evolutionary very distant from algae. Rat FG domains are no subject of this study, one reason being that their analysis is complicated by O-GlcNAc modifications that modulate inter-FG cohesion.
We previously surveyed Nup98 FG phases from a very wide range of eukaryotic species (mammals, lancelets, insects, nematodes, fungi, plants, amoebas, ciliates, and excavates) and found not only rather similar phase properties but also rather similar partition coefficients for both yeast and mammalian NTF2. A species mismatch between NTR and FG phase is therefore unlikely to be a serious issue. C) Where on the NTF2 is the Alexa488 binding? Are they using an antibody labeling strategy or is the NTF2 covalently bound to the Alexa488?
We have used maleimide chemistry for labeling as previously described and with a labeling density of 0.8 dye molecules per NTF2 dimer. The dimer contains six cysteines, and we assume random labeling of those.
An antibody labeling would not have been practical as 150kDa-sized immunoglobulins cannot enter the FG phase.
D) This is critical to know as depict key evidence for their central thesis, specifically, that their FG aggregates selectively interact with NTF2, but not mCherry. Depending upon where the Alexa488 is associated with NTF2, the FG interaction region could be occluded. This would then imply that the interaction is driven by Alexa488 (a highly charged and prolific binder) and not NTF2.
This is highly unlikely for several reasons:
First, the effect of a single Alexa488 moiety is far too weak to explain the NTF2 partition coefficient of 2000. When labeling efGFP_8Q-Cys with Alexa488 maleimide, we see no more than a 3-fold increase in partition coefficient (see the right).
The NTF2 structure is consistent with the assumption that all six cysteines are part of FG-binding sites so that the modification might cause some loss as well as some gain in FG-philicity, the result being perhaps a rather neutral effect.
Second, the new demonstrates that also unlabelled NTF2 interacts strongly with the Perf.GLFG12mer phase and increases the RanGDP partition coefficient 170-fold.
Third, we did not rely blindly just on labeling with unshielded fluorophores. Instead, we also used GFPs as probes, whose fluorophores are fully buried within the beta-barrel of the protein.
This includes surface-engineered GFPs that behave like NTRs (compare e.g. mCherry with GFP NTR -3B7C in as well as NES and IBB-GFP fusions that show a striking NTRmediated influx into the FG phase .
Please also note that negative charges actually counteract a partitioning into a Nup98 FG phase. Please see Frey et al., 2018 and Discussion therein. See also Supp. , which compares Cy3 and Cy5 with their more negatively charged Sulfo-Cy3 and Sulfo-Cy5 variants.
E) Transport times are compared to Imp-B1 import, are dynamics of NTF2 dynamics known in any capacity?
Dwell times of importin β-cargo complexes (~10 msec) and NTF2 (~4 msec) at NPCs have been reported previously by other groups .
## The comparison of import dynamics needs to recognize how imperfect of a comparison this is.
It is unclear to us which comparison and what imperfection the reviewer is referring to.
4) This sentence "The selective phase of NPCs is just one extreme in a range of similar biological condensates that govern developmental programs, RNA-metabolism, RNP assembly, stress, and disease conditions30-37." was listed as a separated paragraph? Is this a typo or real?
This is "real", but is has now been moved to the Perspectives section.
## Figure for reviewers' information.
Perf.GLFG phase incubated with: efGFP8Q-Cys efGFP8Q-Cys-Alexa488
Part. coeff.: 0.45 1.4 5) Many statements are inaccurate and confusing throughout the manuscript. For example, in the abstract it says that "NPCs contain numerous distinct FG domains, each comprising variable repeats.", but soon after in the section of introduction, the authors stated that "Vertebrate NPCs contain ≈10 different FG domains". It's hard to guess which one that the authors want readers to follow.
"Numerous" was perhaps not the most appropriate term. We changed it to "several". Thank you for pointing out this inaccurate phrase. It is unclear why this reviewer used the term "many" but identified just one.
6) It might be too arbitrary and inaccurate by stating that "the barrier can be described as a condensed phase assembled from cohesive FG repeat domains." because there are still disputes about the assembly of the NPC's selectivity barrier and also many experiments argued against the hypothesis of "a condensed phase" in the NPC.
It would have helped to detail those many experiments that argue against the hypothesis of a condensed FG phase forming the NPC permeability barrier. This would have allowed for a factbased discussion. We wish to add that the barrier is defined by function and that this definition by no means implies that all FG domains are necessarily assembled as a condensed phase. There are also non-cohesive domains or sub-domains.
7) It is stated that Serine is used to replace Proline in the proline free FG sequence. However, the logic for why they chose to remove proline is not readily apparent to me.
The logic stems from matching amino acid compositions in the perfectly repeated domains as closely as possible to the starting wildtype sequence. The closest match is the Perf.GLFG12mer, which, however, overrepresents proline (one occurrence instead of 0.58 per repeat). The next best match to the authentic composition was, therefore, to replace proline with another residue. This rationale has been described in the text, and after re-reading this, we still feel that this was clearly presented.
8) The results highlight that the contents of the inter-FG spacers do have an impact upon function of the FG region. The background section should include a statement summing up current thinking on the function of the inter-FG spacers.
We added a summary about the inter-FG spacers to the introduction as well as two paragraphs to the Results and Discussion and the Perspectives. 9) Authors state that the hydrophobic side chain on the surface of GFP enhances FG interactions, this is to be expected. That said, why is there an explanation why the relatively hydrophobic mCherry does not interact with the FG phases? (source for mCherry being much more hydrophobic than DsRed: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3394910/)
Only solvent-accessible residues should matter for FG phase entry/ exclusion. The quoted paper studied something different, namely the environments of the fully buried fluorophore of the two red fluorescent protein variants. Surface-wise, mCherry is very hydrophilic.
Ng et al create a synthetic nuclear transport permeability barrier with minimalistic composition. In four intermediates steps they created a synthetic protein that contains 52 identical GLFG_12mers, all of which phase separate in a similar way as compared to wt protein.
Sequence variation affected amyloid forming propensity. The authors further identify a contribution of the GLEBS domain to phase separation. Various experiments proof that the engineered FG-phase recapitulates the cargo/NTR entry and release behavior of 'nonengineered' FG-Nups.
A very laudable innovation is the demonstration of cargo complex release upon addition of RanGTP-RFP. Also the Ran-GTP facilitated entry of an export complex is a very nice experiment! Those data finally show that in vitro FG-phases also recapitulate the most essential steps of active nuclear transport.
Otherwise, different fluorescent dyes are tested for entry into the phase. The authors demonstrate that dye attachment to proteins entering the permeability barrier affects their partitioning coefficient and that NPCs can be inactivated by illumination, both of which are very neat experiments and important for the field.
The paper is very interesting. Why there are so many different types of FG repeats is a major outstanding question. The demonstration that this variation, although conserved, is not required to re-capitulate active nuclear transport in vitro, comprises an important step forward. As always, the paper is technically well done and taken together, a strong candidate for publication in Nat Comm after some revisions.
Thank you very much for this enthusiastic summary! Some comments:
Having that said, the weakness of the paper is that it is solely based on an in vitro approach.
Engineering an in vivo NPC that would only contain a single engineered FG-Nup is challenging. Although seemingly feasible in yeast if done gradually, it would be lots of work and thus beyond the scope of the present study. However, one really wishes to know if a yeast strain with all the same FG repeats would be viable or alternatively, gradually develop fitness defects once native FG-Nups are changed one by one. You could at least refer to Strawn et al, NCB 2004 in the discussion. I understand that this is not the exact same idea. It is still interesting in this context that many repeats can be removed in vivo and cells are still viable.
We included the Strawn reference as suggested. The genetic experiments would actually address two questions, one being how much FG mass is required? And the other, which mix of FG domain types is optimal for fitness?
Indeed, yeast contains different sets of FG domains, very cohesive and non-charged ones (e.g., from Nup100, Nup116, Nup49, Nup54), but also domains with highly charged and rather noncohesive spacers (Nsp1, Nup159, Nup2, Nup1). We see a function not only for the cohesive but also for the non-cohesive ones. The latter might pose additional filter layers or fine-tune the tightness of the cohesive layers. Too much cohesive FG mass (as a result of FG domain exchange experiments) might be deleterious and lead to NPCs that are too restrictive for supporting high-capacity nuclear transport. More complex experimental systems are needed to address how the different FG domains cooperate with each other. Considering this, we added the following paragraph to the Perspectives:
Nup98 FG phases function in NPCs that contain a considerable mass of less cohesive or even non-cohesive FG domains. These additional domains are typically localized at the cytoplasmic or nuclear periphery of NPCs and probably represent platforms for rapid disassembly of NTR·cargo complexes. They might also form additional filter zones to fine-tune the selectivity of NPCs. Cohesive FG-FG interactions compete with NTR-binding to FG motifs. Non-cohesive FG domains might therefore be more efficient in NTR-capture, thus functioning as "collectors" for incoming NTR-cargo complexes and increasing the capacity of NPCs for active transport. Indeed, it will be very exciting to reconstitute the cooperation of Nup98 FG domains with other FG domains and test those assumptions. A layer-wise FG phase assembly system might be the first step in this direction.
Style: As pointed out above, this reviewer appreciates the middle part of the manuscript, but it does interrupt the flow of the paper. I do not have a good suggestion where to place it elsewhere but you may consider trimming it down -or integrate it better.
We agree and have tried to better integrate the part.
Page 13: Why did the authors use an RFP fusion of Ran? This should perfectly work with native RanQ69LDeltaC-GTP, correct? It would be important to know if release becomes faster, consistent with the idea that non-tagged Ran can diffuse into the phase -or not.
We used the Ran-fusion to counter the formal argument that cargo-dissociation from the import happens already inside the FG phase and not concomitantly with an exit from the barrier. Using a non-fused Ran, efflux becomes faster by a factor of two (now in Supp. , consistent with the idea of non-fused Ran diffusing more rapidly and being better able to access the FG phase's interior. It should be noted, though, that fluorophore-labeled RanGTP hardly enters the FGphase (see -even though the attached fluorophore itself favors the FG-phase. Free, unlabeled RanGTP should therefore show an even more prominent exclusion from the FG phase.
Introduction on page 4: While the 3rd paragraph emphasizes the remarkable sequence conservation of Nup98FG, the 4th paragraph states that 'substantial progress has been hampered by the sequence diversity within given Nup98FG domains'. For readability, you may want to include an additional sentence explaining the concept of comparing sequences of FG repeats within the same protein to contrast previous paragraph better (most people think in terms of comparing sequences across proteins).
Agreed. We changed the sentence to read: "However, substantial progress has been hampered by any given Nup98 FG repeat domain being irregular along its sequence, with variable FG motifs, inter-FG distances, and inter-FG spacers.".
Related to that, if the different properties of individual repeats are conserved across different species but a simple repeat recapitulates nuclear transport, what could be the selection pressure?
Excellent question. We would assume that the bulk phase properties impose just one constraint (keeping, e.g., the overall FG density, "best" FG motifs, and amino acid composition constant). Another constraint might be how to fill the central NPC channel evenly; this could explain why the FG density within a given domain increases with distance from their anchor points. Otherwise, we think that additional molecular interactions must account for the remarkable conservation, in particular of the Nup98 FG domain amongst vertebrates. These could relate to special requirements when transporting very large cargoes, linear motifs targeted by posttranslational modifications or being required for NPC assembly.
Some of the authors have previously shown that MacNup98 is bit special in the sense that it contains many glycines and forms a very tight hydrogel already at comparably low intra-particle FG domain concentrations (eLife paper from 2013). Why is it a well-justified choice for this study that hopefully allows for generic conclusions? The statement that it is well-characterized is understandable. However, it makes the reader wonder about the relevance, because the source species is not exactly a main stream model organism. A bit of additional introduction may be helpful.
Well, our choice was pragmatic and also driven by trying to avoid the expected experimental difficulties with fungal and animal Nup98 FG domains. Animal Nup98 FG domains (from mammals, frogs, insects, nematodes) carry O-GlcNAc-modifications, which tune the cohesiveness of the domain. Lack of this modification causes a very restrictive phase, while over-modification impedes phase separation . This additional dimension complicates the analysis considerably.
The glycosylation density is adjusted in cells through antagonistic activities (OGT and deglycosylating enzymes, with NTRs probably masking some of the modification sites). This is very hard to recapitulate with the purified enzymes.
Furthermore, any change in sequence might also alter glycosylation sites. For sure, having one sugar per repeat (in a perfectly repeated sequence) would imply over-glycosylation and result in an FG domain of low cohesiveness.
The yeast Nup98-like FG domains from Nup100 and Nup116 are rather NQ-rich and prone to a prion-and amyloid-like behavior. We expect any regularization of their sequences will exaggerate their amyloid propensity. Another complication is that N and Q are encoded by just two codons each, meaning that perfectly repeated versions will be encoded by rather repetitive DNA sequences that are hard to synthesize and maintain as stable plasmids.
The FG domain from the Tetrahymena thermophila macronuclear Nup98A shows none of these complications and is by all means very well behaved. We have added a few words (in the first paragraph of Results and Discussion) to explain some of these considerations.
The authors state that they expect Hoechst dyes to become useful for staining FG-phases in vitro, but they also show that even a single dye attached to GFP has a quite massive effect. This does not come across very sound and maybe explained better. What type of experiments could be done despite the fact that the dye massively changes the given phase?
A covalently linked fluorophore indeed increases the partition coefficient of GFP (and probably of other molecules) in an FG phase. The Hoechst dyes are not covalently coupled. They are just added at a low concentration to the mixture. Indeed, we found that they hardly change the FG phase behavior and the partition coefficient of other mobile species (see , as long as they are not irradiated by UV with less than 400nm. We clarified the text accordingly.
This reviewer does not like the term 'authentic NPC', because it inversely suggests that phase separated FG-bodies are 'non-authentic NPCs'. Why can't it just be an 'NPC'?
Agreed. We now call them simply 'NPCs'.
Reviewer #3 (Remarks to the Author):
The underlying mechanisms governing the selective barrier imposed by the nuclear pore complex (NPC) remains to be fully understood. This paper takes an innovative proteinengineering approach to define a "simple" sequence that can recapitulate the fundamental selective sorting properties of the NPC in vitro. By beginning with a Nup98 FG-domain from Tetrahymena, the authors sequentially minimize the sequence to a near perfect repeating GLFG peptide. A critical control is a similar peptide lacking proline residues. Using these reagents, they demonstrate that they can form cohesive networks in vitro that phase-separate into condensates that can recapitulate both active import and export while excluding an inert mCherry protein. Along the way, they also interrogate how these condensates interact with a series of fluorescent probes. The major advance here is that they demonstrate that DAPI can crosslink the FG-network, which could provide a useful tool for the field. In general, the data are of high quality and are rigorous. The manuscript is written well although it is a bit disjointed due to the "detour" exploring interactions with probes. It should make a valuable contribution to Nature Communications as it will provide valuable concepts and tools for multiple fields.
Thank you very much! Major Point 1: The paper is written with considerable tunnel vision as to how the NPC acts as a selective barrier as it is solely focused on the importance of cohesive interactions. It would be nice if the authors at least considered that there are non-cohesive FG-networks in the NPC that are known to be important functionally. No additional data is required, but a discussion of the disconnect, for example, between the ability of the GLFG-domain to exclude mCherry in vitro and the observation that mCherry can freely pass through the NPC in vivo would seem warranted, particularly in the context of work defining the NPC barrier in vivo as being "soft" (e.g. Timney, JCB).
We added a section to discuss non-cohesive interactions as suggested. We wish to add, though, that mCherry passes NPCs 700 times more slowly than NTF2 (Frey et al., 2018; see also . We wouldn't call this a free passage. In fact, the retention of 25-30 kDa-sized proteins is functionally important because it allows to build up a steep nucleocytoplasmic RanGTPgradient. We have re-written the first part of the introduction to clarify these points.
Major Point 2: Hydrogel versus liquid? A description of the condensates in terms of their physical characteristics would be helpful.
We already estimated diffusion coefficients of the FG domains within the phases and have expanded on this.
Although it is acknowledged that there is a continuum of physical states between liquids and gels, there is also the concept that cohesive interactions can transition or "age" between liquids and more solid states, the latter being less functional/pathological. The photobleaching in may, for example, be indicating that the GLEBS domain drives a gel-like state. Again, I don't think any additional experimentation is needed, but some discussion of this in the context of the phase-separation literature should be considered.
We have observed that the Nup116 FG phase accumulates Thioflavin-T positive structures over the course of ≥10 hours (Supp. , which can be interpreted as aging. Apart from this, we haven't seen any aging effects in terms of FG phase selectivity. Given that NPCs are long-lived structures, there was probably selective pressure against age-related deterioration of barrier function. We do fully agree that this is a very important topic, in particular for defining the position of FG phases within the universe of biological condensates. However, we feel that we cannot adequately cover this topic within a few sentences and therefore wish to postpone the discussion for a comprehensive review (also given that the present manuscript is already very long).
Reviewer #2 (Remarks to the Author):
Reviewer #1: Remarks to the Author: The revised manuscript has improved in terms of presentation and clarity, but my major concerns remain about the novelty of the method and the conclusions being drawn.As stated in my previous comments, both in vitro mimic of the NPC's selectivity barrier by using FG domains to form hydrogels and then tests of importins and exportins interacting with this hydrogel have been fully studied in their previous papers(Science , 2006, 314:815;Cell, 2007, 130:512). Since the author denied it, this reviewer has to include figure 4 of that Science paper and also the abstract of 2007 Cell paper in the attachment, in which clearly the interactions between NTRs (importins & exportins) and hydrogel have been fully studied and the conclusion that FG-hydrogel can mimic the NPC's permeability barrier has already been drawn. Thus, compared to these studies published more than ten years ago, small increments presented in the current manuscript would unlikely bring anything new or impactful into the field of NPC and nucleocytoplasmic transport.Reviewer #2: Remarks to the Author: The authors have carefully addressed my comments. I strongly recommend this manuscript for publication in Nat Comm.Reviewer #3: Remarks to the Author: I was supportive of this paper after the first submission. I appreciate that the authors took the time to address my few comments. |
Glomerulonephritis in Youth With Dystrophic Epidermolysis Bullosa
# Introduction
E pidermolysis bullosa (EB) is a rare group of genetic conditions involving mutations in genes such as COL7A1 that encode structural proteins required to maintain skin integrity. These mutations lead to skin fragility and subsequent nonhealing erosions and scarring. Because these mutations are also partially expressed in other epithelial and mesenchymal tissues, patients with EB may have secondary consequences, such as malnutrition and extracutaneous manifestations involving the renal system. Those with junctional EB generalized severe and recessive dystrophic EB generalized severe are particularly susceptible to renal involvement and may progress to renal failure. [bib_ref] Genitourinary complications of inherited epidermolysis bullosa: experience of the National Epidermolysis Bullosa..., Fine [/bib_ref] For example, the National Epidermolysis Bullosa Registry reported renal failure was the attributed cause of death in 3.6% of adults with recessively dystrophic EB generalized severe (RDEB) with a mean age at death of 24 years, and a cumulative risk of death from renal failure of 12.3% at age 35 years. [bib_ref] Genitourinary complications of inherited epidermolysis bullosa: experience of the National Epidermolysis Bullosa..., Fine [/bib_ref] Renal involvement in both the adult and pediatric EB population has also been described in case reports owing to a variety of causes, predominantly postinfectious glomerulonephritis (GN), immunoglobulin (Ig) A nephropathy, chronic interstitial nephritis, secondary amyloidosis, and congenital nephrotic syndrome. [bib_ref] Nephro-urological complications of epidermolysis bullosa in paediatric patients, Chan [/bib_ref] [bib_ref] Fatal systemic amyloidosis (AA type) in two sisters with dystrophic epidermolysis bullosa, Bourke [/bib_ref] [bib_ref] Renal involvement in epidermolysis bullosa patients: a case series study, Cavagnaro [/bib_ref] [bib_ref] Congenital focal segmental glomerulosclerosis associated with beta4 integrin mutation and epidermolysis bullosa, Kambham [/bib_ref] [bib_ref] Nephrotic syndrome and aberrant expression of laminin isoforms in glomerular basement membranes..., Hata [/bib_ref] However, much remains unknown about the pathophysiology of renal disease in patients with EB. This case series describes 5 pediatric patients treated at our institution for dystrophic EB in whom GN developed and who underwent renal biopsy. We describe their unique clinical presentation and histopathologic findings.
## Case description patient 1
Patient 1 was a Caucasian male with a tentative diagnosis of dominant dystrophic EB [fig_ref] Table 1: Summary of renal histopathology, treatment, and clinical outcome of patients with dystrophic... [/fig_ref]. He had many signs of RDEB, so it is possible that he had a cryptic second COL7A1 mutation that was not detected. He also had a history of chronic cutaneous Staphylococcus infections and a baseline C-reactive protein (CRP) level of approximately 5 to 7 mg/dl. At age 18 he had coffee-colored urine and an elevated serum creatinine (SCr) level of 1.5 mg/dl. He was referred to a pediatric nephrologist, who obtained a renal profile with similar results: a urinalysis with >300 mg/dl of protein and large blood; urine microscopy with red blood cells too numerous to count per high-power field; normal complement levels (C3 level 148 mg/dl and C4 level 27 mg/dl); normal anti-DNAse B level; normal antistreptolysin O titer; negative antinuclear antibody; negative perinuclear antineutrophil cytoplasmic antibody; and negative cytoplasmic antineutrophil cytoplasmic antibody. A urine protein-tourine creatinine (UPC) ratio was not available for review. Renal ultrasonographic images showed normal results. Because of the persistence of hematuria and proteinuria, he underwent a renal biopsy 1 month after presentation. Histopathologic study showed 21 glomeruli, all with mesangial hypercellularity on light microscopy and globular mesangial staining for C3 (3þ) and IgA (2þ) on immunofluorescence (IF). Staining for C1q was negative. Electron microscopy showed thickened glomerular basement membrane owing to multifocal subendothelial electron-dense immune complex deposits. A diagnosis of IgA nephropathy was made, and the patient was treated with long-term steroid and lisinopril therapy. Initially there was improvement in the SCr level (0.86 mg/dl), hematuria (31 red blood cells per high-power field on microscopy), and proteinuria (30 mg/dl), but he experienced a relapse of gross hematuria, nephrotic-range proteinuria, and renal dysfunction (SCr, 1.9 mg/dl) while being weaned from steroids. Repeat renal biopsy at age 21 years showed IgA dominant (3þ to 4þ) immune complex proliferative GN with crescents plus strong (3þ) C3 deposition. In addition, electron microscopy showed hump-like subepithelial deposits, suggesting a possible component of infection-associated GN superimposed on chronic IgA nephropathy. Because of concern that the chronic cutaneous Staphylococcus infections represented a possible antigenic stimulus for his GN, he was treated with cephalexin and steroid therapy was resumed. Although he did not complete the antibiotic course, there was resolution of gross hematuria and some improvement in the SCr level (to 1.5 mg/dl), so he was again weaned from steroids. Three months after steroid discontinuation, he was admitted for acute kidney injury with an SCr level of 3.8 mg/dl in the setting of lower extremity cellulitis. He received antibiotics and lisinopril was held because of his acute kidney injury. His primary nephrologist believed he had postinfectious GN that would improve after resolution of the infection. A repeat renal profile obtained 3 months later (at age 22) showed no improvement in the SCr level (3.3 mg/dL), so the patient came to our institution for a second opinion. He was noted to have nephrotic-range proteinuria (UPC ratio, 9.2 mg/mg) and microscopic hematuria. The IgA level was mildly elevated (396 mg/ dl), and both C3 and C4 levels were normal. He underwent a third renal biopsy, which showed signs of chronic GN with crescents, segmental sclerosis, and mesangial proliferation. There was no definitive evidence of immune complex deposition in the 1 glomerulus available for IF, and electron microscopy showed mesangial, paramesangial, and subepithelial hump-like deposits. It was ultimately considered he had IgA nephropathy based on review of his previous biopsy findings. He received intravenous pulse and oral steroids in addition to mycophenolate, with some improvement in SCr (to 2.5 mg/dl) and reduction in proteinuria (UPC ratio 4.0 mg/mg). As of the date of writing, the goal was to continue mycophenolate therapy and with gradual weaning from steroids.
## Patient 2
Patient 2 was a Caucasian male with RDEB and chronic Staphylococcus, Streptococcus, and Pseudomonas cutaneous infections [fig_ref] Table 1: Summary of renal histopathology, treatment, and clinical outcome of patients with dystrophic... [/fig_ref]. His baseline CRP level was approximately 10 to 12 mg/dl. At age 16 gross hematuria and proteinuria developed, so was referred to a pediatric nephrologist, who clinically diagnosed IgA nephropathy and started enalapril therapy. Several months later gross hematuria, worse proteinuria (UPC ratio, 5.4 mg/mg), and an increase in the SCr level from baseline 0.3 mg/dl to 0.8 mg/dl developed. The cystatin C level at the time was 1.3 mg/ (cystatin Cestimated glomerular filtration rate [GFR] 65 ml/min per 1.73 m 2 ). The C3 level was slightly elevated (171 mg/dl). The patient underwent a renal biopsy, with pathologic findings showing up to 21 glomeruli with increased proliferation and occasional neutrophils within capillary loops, as well as interstitial inflammation composed primarily of plasma cells and occasional neutrophils [fig_ref] Figure 1: Renal histopathologic findings from patient 2 [/fig_ref]. IF showed strong C3 mesangial deposition and no immunoglobulin or C1q deposition. Electron microscopy showed mesangial matrix deposits. He was diagnosed with GN with dominant C3 and prominent interstitial inflammation, possibly infectious in nature. Because of concern that his renal disease may have an infectious etiology, he was empirically treated with cefdinir followed by prophylactic cephalexin. After antibiotic treatment the gross hematuria resolved, although microscopic hematuria persisted (20-29 red blood cells per highpower field). The proteinuria markedly improved, with an UPC ratio nadir of 0.58 mg/mg 6 months after antibiotic treatment. His renal function also markedly improved with an SCr level of 0.23 mg/dl (Schwartz formula-calculated GFR, 273 ml/min per 1.73 m 2 ) and a cystatin C level of 0.869 mg/L (cystatin C estimated GFR level, 111 ml/min per 1.73 m 2 ) 6 months after antibiotic therapy. At the time of writing, his renal function remained normal. [fig_ref] Table 1: Summary of renal histopathology, treatment, and clinical outcome of patients with dystrophic... [/fig_ref]. Her baseline CRP level was approximately 10 mg/dl. Coffee-colored urine developed at age 14. A renal profile was obtained and showed a rise in SCr from 0.3 to 0.68 mg/dl. Urinalysis showed more than 50 red blood cells per high-power field, 30 mg/dL protein, and moderate leukocyte esterase. She was presumed to have a urinary tract infection and began a regimen of sulfamethoxazoletrimethoprim. Further evaluation showed an antistreptolysin O titer elevated at 2669 unit/ml and a mildly low C3 level (56 mg/dl). A nephrology consultation was obtained and postinfectious GN was suspected. Gross hematuria continued and she had nephrotic-range proteinuria (UPC ratio, 3.2 mg/mg), so she underwent a renal biopsy 1 month after initial presentation. Histopathologic findings showed 18 glomeruli with mesangial proliferation and extensive tubulointerstitial inflammation, primarily composed of neutrophils, plasma cells, and lymphocytes [fig_ref] Figure 2: Renal histopathologic findings from patient 4 [/fig_ref]. IF showed strong (3þ) mesangial staining for C3 and no immunoglobulin, C1q, or C4 deposition [fig_ref] Figure 3: Renal histopathologic findings from patient 4 [/fig_ref]. Electron microscopy showed a mild increase in the mesangial matrix with scattered paramesangial and mesangial deposits. At the time of biopsy, urinalysis showed nitrite, moderate leukocyte esterase, and 2þ bacteria. A diagnosis of GN with dominant C3 was made. Because of concern that there was an infectious etiology given the tubulointerstitial inflammation, she was treated with cefdinir followed by prophylactic amoxicillin. After initiation of antibiotics, she had subsequent resolution of gross hematuria and proteinuria within 12 weeks. At the time of writing, she was doing well from a renal standpoint, maintaining her baseline SCr, and had a normal urinalysis.
## Patient 5
Patient 5 was a Caucasian female adolescent with RDEB and chronic Staphylococcus and Streptococcus infections [fig_ref] Table 1: Summary of renal histopathology, treatment, and clinical outcome of patients with dystrophic... [/fig_ref]. Her baseline CRP level was approximately 12 to 15 mg/dl. At age 13 microscopic hematuria and proteinuria (UPC ratio, 1.0 mg/mg) developed, so she was referred to a pediatric nephrologist. Further evaluation included a normal renal profile, normal C3 and C4 levels, and a negative antinuclear antibody level. Coke-colored urine developed at age 14 in addition to worsening proteinuria (UPC ratio, 7.3 mg/mg) and a rise in SCr from 0.4 to 0.96 mg/dl. Renal biopsy showed 15 glomeruli with increased mesangial proliferation, 2 small crescentic lesions, and interstitial inflammation. IF showed C3 mesangial deposition and no immunoglobulin, C1q, or C4 deposition. Electron microscopy showed a mild increase in mesangial matrix with scattered paramesangial and mesangial deposits [fig_ref] Figure 4: Renal histopathologic findings from patient 5 [/fig_ref]. A diagnosis of GN with dominant C3 was made, and the patient was treated with alisinopril, but her renal function worsened with a rise in SCr to 1.6 mg/dl. Additional evaluation showed elevated antistreptolysin O titer and elevated streptozyme, so clindamycin therapy was started out of concern for a Streptococcus cutaneous infection. Subsequently the SCr level and gross hematuria improved and the UPC ratio was mildly decreased to 4.0 mg/mg, but a relapse occurred after switching to rifampin after development of diarrhea during clindamycin use. A second renal biopsy at age 15 showed acute proliferative and exudative GN with cellular and fibrosing crescents, interstitial inflammation, and fibrosis with focal tubular atrophy. IF again showed strong mesangial C3 deposition. A transition was made from rifampin therapy to amoxicillin prophylaxis. At age 16 she was considered to have achieved partial remission of chronic GN because of improvement in SCr to 0.87 mg/dl and improvement in microscopic hematuria. However, at age 17 she was hospitalized for decompensated septic shock in the setting of a diarrheal illness and purulent drainage from cutaneous lesions; vasopressor therapy was started, and she received broad-spectrum antibiotics, including vancomycin and cefepime. On admission she had significantly worse renal function with hyperkalemia and proteinuria (SCr, 4.9 mg/dL; serum potassium, 8.6 mg/dL; UPC ratio, 8.0 mg/mg). Her hyperkalemia was initially medically managed. She ultimately declined dialysis, opting for palliative measures instead. Her renal function continued to decline (SCr, 7.62 mg/dl), and she died 2 weeks after admission as the result of complications of renal failure.
# Discussion
We describe the unique renal histopathologic findings in 5 pediatric patients with dystrophic EB and GN.
Histologic findings in all patients showed C3 immunofluorescent staining, mesangial proliferation, and mesangial electron dense deposits. Based on the 2013 consensus report, 7 4 of our 5 patients who underwent biopsy met the criteria for GN with dominant C3, which to the best of our knowledge has not been reported to date in children with EB. Our patients had histologic variability, rendering a single etiology difficult to determine. Infection may cause GN with dominant C3 staining, more specifically termed infection-associated GN (IAGN). All of our patients had either chronic Staphylococcus or Streptococcus cutaneous infections, or both, which are known causes of IAGN. In additional, most of our patients had renal interstitial inflammation, a finding suggestive of IAGN. However, other immunofluorescent staining such as IgG and C1q was notably absent in all our cases, which is less consistent with IAGN. Although "masked" Ig deposits may be detectable by IF on paraffin-embedded tissue, this staining was unavailable in our department at the time of our patients' renal biopsies. Regardless, the lack of Ig deposition was not the only inconsistency with IAGN. A predominance of intracapillary neutrophils was not found in all patients, which is also atypical of IAGN. In light of these discrepant findings, we propound that C3 GN due to primary complement dysregulation was a possible contributing etiology in patients 2 through 5. Supportive of this hypothesis is the preponderance of mesangial deposits and lack of codeposition of immune reactants in 4 of the patients. Patients with EB in whom C3 GN develops may have an underlying genetic abnormality, as has been described in other patients with C3 GN, including mutations in CFH, complement factor H-related genes, and the presence of C3 nephritic factor. [bib_ref] Identification of a mutation in complement factor H-related protein 5 in patients..., Gale [/bib_ref] [bib_ref] Acquired and genetic complement abnormalities play a critical role in dense deposit..., Servais [/bib_ref] Unfortunately, we did not perform extensive complement-related serologic and genetic testing in our patients, and further research is needed to investigate this possibility. Overall, our patients demonstrated histologic variability and we cannot exclude that both IAGN and C3 GN were involved to varying degrees. Renal disease in patients with EB may represent a spectrum of multifactorial causes, including infectious and immunologic components.
Interestingly, another case series of pediatric EB patients with renal dysfunction described 1 patient with renal pathologic findings similar to those of our patients: mesangial expansion and moderate C3 deposition without Ig deposition in the setting of concurrent Staphylococcus and Streptococcus cutaneous infections. [bib_ref] Nephro-urological complications of epidermolysis bullosa in paediatric patients, Chan [/bib_ref] That patient's renal dysfunction progressed to end-stage renal disease, for which he received dialysis but died a few years later from complications of fluid overload and cardiac failure. [bib_ref] Nephro-urological complications of epidermolysis bullosa in paediatric patients, Chan [/bib_ref] The authors similarly opined that his histopathologic findings could be consistent with IAGN or membranoproliferative GN, which includes C3 GN in current classification systems.
Our patient 1 was initially diagnosed with IgA nephropathy and was the only patient who did not meet criteria for GN with dominant C3. Ours is not the first report of a patient with dystrophic EB whose biopsy findings were suggestive of IgA nephropathy; a similar patient case was published in 2008. S1 However, our patient differs in that he additionally had strong C3 IF staining and both subepithelial and subendothelial deposits, findings atypical of IgA nephropathy. His renal disease was likely due to Staphylococcus IAGN, a disease entity in which varying degrees of both IgA and C3 deposition may be present. S2 Alternatively, it is possible our patient's renal disease was due to primary IgA nephropathy with associated complement dysregulation. Studies have shown an association between complement activation and IgA nephropathy activity, S3 with C3 and C4d mesangial deposition an independent risk factor for progression of IgA disease. S4 Therefore, it is possible that underlying complement dysregulation contributed to the aggressive form of IgA nephropathy witnessed in our patient.
Infections, either as the primary cause or as a trigger for complement dysregulation, were suspected to play a . Teaching points Glomerulonephritis may be an unrecognized cause of renal disease in children with epidermolysis bullosa.
Preceding cutaneous bacterial infections may play a role in the development of glomerulonephritis with dominant C3 in children with epidermolysis bullosa. Therefore, antibiotics targeting cutaneous bacterial infections should be considered as a potential-first line therapy in these patients.
Cystatin C is likely a more accurate marker of renal function in malnourished patients with epidermolysis bullosa compared with serum creatinine.
## Nephrology rounds
E Hughley et al.: Glomerulonephritis in EB Youth role in the onset of GN in our patients. Three patients therefore received a trial of antibiotic therapies targeting pathogenic bacteria previously cultured from their cutaneous tissues. In patients 2 and 4, markers of renal dysfunction improved after antibiotic therapy without use of immunosuppression. In patient 5, markers of renal dysfunction transiently improved after antibiotic therapy and then acutely worsened in the setting of septic shock. Although the correlation of antibiotic use with renal improvement does not prove causation, we believe the antibiotics contributed at least in part to improvement in our patients' conditions. To our knowledge, there have been no reports of antibiotic therapy as a potential curative treatment for EB patients with GN. Immunosuppressive therapies, such as mycophenolate, steroids, and rituximab, have historically been used to treat C3-dominant GN with some success. S5,S6 These therapies also remain an option to treat C3-dominant GN in patients with EB. It has been our practice to consider this as second-line therapy if signs of renal dysfunction do not first resolve with a course of antibiotics, or if the renal biopsy pathologic findings show severe glomerular proliferative disease.
A final point of interest is the discrepancy between SCrbased (Schwartz formula-calculated) and cystatin C-based estimations of renal function in patients 2 and 3. For both patients, Schwartz formula-calculated renal function was markedly higher, likely because both patients were malnourished and had low muscle mass as secondary consequences of EB. Overall, these findings suggest cystatin C is a more accurate marker of renal function in EB patients with malnutrition, although additional research is needed to substantiate this hypothesis.
In summary, we describe 5 patients with dystrophic EB in whom GN developed, all of whom had C3 deposition on renal biopsy pathologic findings and 2 of whom had sustained improvement in renal dysfunction after antibiotic treatment targeting chronic cutaneous infections. We postulate that ongoing cutaneous bacterial infections play a role in the development of GN with C3 deposition in patients with EB and propose that antibiotics targeting these infections be considered as a potential first-line therapy for children with EB and GN . We additionally hypothesize cystatin C is a more accurate marker of renal function in EB patients with malnutrition compared with serum creatinine.
[fig] Figure 1: Renal histopathologic findings from patient 2. Glomerulus shows increased mesangial cellularity and matrix, and inflammatory cells in capillary loops (arrows; periodic acid-Schiff stain, original magnification Â400). [/fig]
[fig] Figure 3: Renal histopathologic findings from patient 4. Strong (3þ) C3 staining shows a granular mesangial pattern (immunofluorescence staining, original magnification Â200). [/fig]
[fig] Figure 4: Renal histopathologic findings from patient 5. Electron microscopy shows a mild increase in the mesangial matrix with scattered mesangial and paramesangial deposits (arrow). [/fig]
[fig] Figure 2: Renal histopathologic findings from patient 4. Prominent mixed interstitial inflammatory infiltrate is present, with lymphocytes, plasma cells, rare eosinophils, and neutrophils (hematoxylin-eosin stain, original magnification 200). [/fig]
[table] Table 1: Summary of renal histopathology, treatment, and clinical outcome of patients with dystrophic EB Death from renal failure at age 18 in the setting of preceding sepsis ACEI, angiotensin converting enzyme inhibitor; DDEB, dominant dystrophic epidermolysis bullosa; EB, epidermolysis bullosa; GN, glomerulonephritis; IgA, immunoglobulin A; RDEB-GS, recessive dystrophic epidermolysis bullosa-generalized severe [/table]
|
Global Origin of Mycobacterium tuberculosis in the Midlands, UK
DNA fi ngerprinting data for 4,207 Mycobacterium tuberculosis isolates were combined with data from a computer program (Origins). Largest population groups were from England (n = 1,031) and India (n = 912), and most prevalent strains were the Euro-American (45%) and East African-Indian (34%) lineages. Combining geographic and molecular data can enhance cluster investigation.
K nowledge and understanding of transmission dynamics of Mycobacterium tuberculosis have been improved by development of rapid molecular techniques that are being more extensively applied [bib_ref] Evaluation of a two-step approach for large-scale, prospective genotyping of Mycobacterium tuberculosis..., Cowan [/bib_ref] [bib_ref] High-resolution minisatellite-based typing as a portable approach to global analysis of Mycobacterium..., Mazars [/bib_ref]. Globally, application of molecular techniques has identifi ed major M. tuberculosis lineages associated with geographic origin [bib_ref] Restricted structural gene polymorphism in the Mycobacterium tuberculosis complex indicates evolutionarily recent..., Sreevatsan [/bib_ref] [bib_ref] Strain identifi cation of Mycobacterium tuberculosis by DNA fi ngerprinting: recommendations for..., Van Embden [/bib_ref] [bib_ref] Variable host-pathogen compatibility in Mycobacterium tuberculosis, Gagneux [/bib_ref]. Previous studies on transmission dynamics of M. tuberculosis have usually analyzed patient-declared population groups to identify associations [bib_ref] Variable host-pathogen compatibility in Mycobacterium tuberculosis, Gagneux [/bib_ref].
We describe a novel software (Origins; Experian, Nottingham, UK) that assigns cultural, ethnic, and linguistic (CEL) groups on the basis of given and family names. Records from 12 countries containing 1,600,000 family and 600,000 given names were analyzed to construct >200 origin types based on CEL factors associated with given and family names. This approach is applicable worldwide and is more accurate and has better coverage than other software. The fi rst use of Origins in healthcare was identifi cation of how a European CEL group came to emergency departments in the United Kingdom [bib_ref] Use of the emergency department by Polish migrant workers, Leaman [/bib_ref].
The aim of this study was to combine mycobacterial fi ngerprinting data and patient origin as assigned by Origins to relate the occurrence of major global M. tuberculosis lineages in populations originating from around the world. Combining data obtained from universal typing and associated cultural and social links identifi ed by Origins provides the potential for a deeper understanding of the causes for distribution of prevalent strains in specifi c population groups.
## The study
Nonduplicate initial M. tuberculosis complex isolates (n = 4,207) were referred from the Midlands region of the United Kingdom (population 9.5 million) to our center during January 2004-December 2007. These isolates were incubated, identifi ed, and analyzed by mycobacterial interspersed repetitive units containing variable numbers of tandem repeats (MIRU-VNTR) typing [bib_ref] Cluster of human tuberculosis caused by Mycobacterium bovis: evidence for person-to-person transmission..., Evans [/bib_ref]. MIRU-VNTR typing analyzes the number of repetitive DNA sequences at multiple independent genetic loci. These data were compared with those in an online database (MIRU-VNTRplus), which was developed by Allix-Beguec et al. [bib_ref] Evaluation and strategy for use of MIRU-VNTRplus, a multifunctional database for online..., Allix-Beguec [/bib_ref]. This database was used to assign M. tuberculosis strains to 1 of 6 lineages: East African-Indian, East Asian, Euro-American, Indo-Oceanic, West African-1, or West African-2.
The given and family names of 4,207 patients were entered into Origins to obtain a CEL group for each, which was then assigned a continent on the basis of the United Nations Standard Country and Area Codes Classifi cation Scheme. Origins can assign a CEL group when the given and family names are present in a dataset.
Within the study population are predominant CEL groups that originate from each continent: 1,031 (25%) from England in Europe, 912 (22%) from India in Asia, and 130 (3%) from Somalia in Africa Using the 15 MIRU-VNTR loci, we matched 4,117 (98%) of 4,207 typed strains to strains in the MIRU-VN-TRplus database. The 90 strains that did not match with 1 of the 6 major global lineages were M. bovis (24 strains) or could not be defi nitively assigned (66 strains) to 1 of the 6 global lineages. Continental and regional origins of patients as assigned by Origins and global lineage were then combined to identify the distribution of global M. tuberculosis lineages within each population .
The Euro-American lineage was the most prevalent lineage in our study. It contained 1,894 (45%) strains and was present in each continental human population group. The Euro-American strain was the most prevalent lineage in patients originating from Africa (125), the Americas (11), and Europe (1,072) and was the second most prevalent lineage in patients originating from Asia (663). The most prevalent M. tuberculosis lineage in patients originating from Asia was the East African-Indian lineage .
Combining geographic data assigned by Origins and DNA fi ngerprinting data could affect public health efforts to control tuberculosis because this approach can identify strains in CEL groups in which specifi c global M. tuberculosis lineages are not present. The MIRU-VNTR profi le 424352332515333 (East Asian lineage) was identifi ed in 23 patients from the Midlands. Of these 23 patients, 20 resided within a 5-mile radius of each other. Within this geographically restricted cluster, 12 (60%) of these patients were assigned to the Europe CEL group and 8 patients to part of the Asia CEL group. The fi rst strain was identifi ed in 2004, and subsequent strains were identifi ed in each year of this study.
The MIRU-VNTR profi le 422352542517333 was identifi ed in 102 patients during 2004-2007. This profi le was matched with the East African-Indian lineage; 98 (96%) patients originated from Asia and 4 (4%) from Europe. This strain was identifi ed in various locations in the Midlands within an ≈40-mile radius that included all patients.
# Conclusions
We studied >4,000 M. tuberculosis isolates typed in the United Kingdom. Our study demonstrated that the combination of molecular and population group data provided by novel software can provide information about the molecular epidemiology of M. tuberculosis. The 2 example MIRU-VNTR profi les show that molecular and social data identifi ed an East Asian strain in an unsuspected CEL group (Europe) and limited transmission of an East African-Indian strain between CEL groups. Geographic restriction of the 424352332515333 East Asian strain in the European CEL group identifi ed possible recent transmission within this population group. The 422352542517333 East African-Indian strain infected a large number of patients (102) and showed wide geographic spread with limited transmission into the European CEL group (4/102 patients). This fi nding indicates that this strain is widely distributed in southern Asia and has not been transmitted between CEL groups. Its wide distribution in the United Kingdom refl ects areas of residence for this CEL group.
Data from our study support previous fi ndings and extend the dataset for Europe. Our results also include a large number of strains from southern Asia, which were underrepresented in other studies [bib_ref] Major Mycobacterium tuberculosis lineages associate with patient country of origin, Reed [/bib_ref].
Origins identifi ed CEL groups within a country (e.g., Kashmir in Pakistan or northern India) and divided Great Britain and Ireland into 4 CEL groups . This enhanced differentiation could be useful in future population-based studies because migration patterns may be localized to specifi c areas within countries and common social networks could be identifi ed. CEL groups can be assigned to any dataset in which the patient's name is known. Traditional epidemiologic identifi cation of ethnic groups requires a questionnaire, but if patient names are not in a dataset, then CEL groups cannot be assigned. Origins showed some discrepancies because the black Caribbean CEL group usually has British names and will be assigned as a British CEL group. However, the utility of Origins is maximized when it is applied to diverse populations.
Many countries now routinely type M. tuberculosis isolates by using MIRU-VNTR typing. This analysis identifi es clusters of strain types across place and time. By using Origin software for identifi cation of CEL groups, public health offi cials can identify and investigate possible cultural links for transmission of M. tuberculosis.
[fig] Table 2 *: Distribution of Mycobacterium tuberculosis isolates according to lineage and continent of patient origin on the basis of CEL group, the Midlands, UK* No. (%) isolates in each M. CEL, cultural, ethnic, and linguistic. Unknown indicates that the continent was identified but without a specific region. The United Kingdom (Great Britain and Northern Ireland) is located in northern Europe and India, Pakistan, and Bangladesh are located in southern Asia. A total of 90 (2%) of 4,207 strains were not assigned to 1 of the 6 major lineages. [/fig]
|
Effects of 5-Aminolevulinic Acid as a Supplement on Animal Performance, Iron Status, and Immune Response in Farm Animals: A Review
Simple Summary: 5-aminolevulinic acid is an amino acid that promotes the formation of heme-an essential constituent of hemoglobin. It has been recently used as a novel feed supplement to enhance the productivity of farm animals, but the current understanding of its effects on livestock is not clear. We systematically evaluated the literature for the effects of 5-aminolevulinic acid supplementation on animal performance, iron status, and immune response in farm animals. Extensive search of PubMed and Web of Science resulted into 16 eligible controlled trials. Findings revealed that iron status and immunity were most responsive to 5-aminolevulinic acid. Other parameters displayed hardly any tangible effect. Studies were highly heterogeneous (regarding species, dose, treatment duration, use of other supplements), which may limit the conclusion. Standard procedures and outcome measures are needed to confirm the benefits of 5-aminolevulinic acid. Attention should also be paid to any adverse effects.Abstract: Efforts directed toward enhancing animals' productivity are focused on evaluating the effects of non-traditional feed additives that are safer than antibiotics, which have been banned because of their health hazards. Many studies used an amino acid that contributes to heme biosynthesis, known as 5-aminolevulinic acid (5-ALA), to promote the productivity of farm animals. However, these studies demonstrate inconsistent results. In order to develop a clear understanding of the effects of 5-ALA in farm animals, we comprehensively searched PubMed and Web of Science for studies evaluating 5-ALA effects on the performance, iron status, and immune response of different farm animals. The search retrieved 1369 publications, out of which 16 trials were relevant. The 5-ALA-relevant data and methodological attributes of these trials were extracted/evaluated by two independent researchers, based on a set of defined criteria. Samples were comprised of pigs, chickens, and dairy cows. The 5-ALA doses ranged from 2 mg to 1 g/kg of feed, and treatment duration ranged from 10 to 142 days. Overall, 5-ALA improved iron status in most studies and increased white blood cells count in 3 out of 10 studies, in addition to improving animals' cell-mediated immune response following immune stimulation with lipopolysaccharide. Inconsistent findings were reported for growth performance and egg production; however, a combination of 10 mg/kg of 5-ALA with 500 mg/kg of vitamin C promoted the highest egg production. In addition, 5-ALA improved milk protein concentration. In conclusion, 5-ALA can enhance farm animals' iron status and immune response; however, the heterogeneity of the reviewed studies limits the generalizability of the findings. Standard procedures and outcome measures are needed to confirm the benefits of 5-ALA. Attention should also be paid to any adverse effects.
# Introduction
The ban set on the use of antibiotics that were mainly used to decrease energy loss and improve animals' productivity forced those concerned with animal breeding to search for safer and better alternatives [bib_ref] Use of antibiotics as feed additives: A burning question, Chattopadhyay [/bib_ref]. Researchers are obliged to explore novel animal production systems that provide high-quality products at low cost, such as the use of molecular nutritional techniques [bib_ref] Molecular nutrition: Interaction of nutrients, gene regulations and performances, Sato [/bib_ref]. Some of the prevailing approaches to enhance animals' productivity are based on diet enrichment with amino acids aimed to improve milk protein [bib_ref] Responses to amino acid imbalances and deficiencies in lactating dairy cows, Weekes [/bib_ref] or the supplementation of essential minerals, such as iron [bib_ref] Effects of mineral content of bovine drinking water: Does iron content affect..., Mann [/bib_ref].
Unfortunately, direct addition of iron to feed may interfere with the digestion and absorption of other nutrients, such as the formation of insoluble complexes of P and Fe in the digestive tract (e.g., FePO 4 ), which in turn leads to deficiency of both elements. As a result, several undesirable consequences may ensue, such as oxidative stress, bacterial infection, diarrhea, and reduced weight gain. Thus, fortifying feed with iron may not be an effective enrichment method, especially because iron toxicity (detected from a darker rumen color) has been reported with high-Fe doses [bib_ref] Effects of mineral content of bovine drinking water: Does iron content affect..., Mann [/bib_ref]. In addition, it has been indicated that animals prefer drinking water without added Fe compared with water containing Fe sources with different anionic moieties [bib_ref] Preference and drinking behavior of lactating dairy cows offered water with different..., Genther [/bib_ref].
Evolving research on feed additives involves new non-traditional ones, such as 5-aminolevulinic acid (5-ALA), a product of condensing succinyl-CoA and glycine through the catalytic activity of 5-ALA synthase [bib_ref] 5-Aminolevulinic acid regulates the inflammatory response and alloimmune reaction, Fujino [/bib_ref]. After a chain of chemical interactions, 5-ALA changes into protoporphyrin IX-a heme precursor. Then, with the help of ferrochelatase, the porphyrin ring of protoporphyrin IX acquires an iron atom to finally produce heme, which is an essential constituent of hemoglobin [bib_ref] Heme deficiency interferes with the Ras-mitogen-activated protein kinase signaling pathway and expression..., Zhu [/bib_ref].
Based on this understanding, it has been theorized that introducing 5-ALA as a dietary supplement to livestock can affect the synthesis of heme and positively influence the iron or hemoglobin status of animals [bib_ref] Effect of δ-aminolevulinic acid on growth performance, nutrient digestibility, blood parameters and..., Chen [/bib_ref]. 5-ALA has a wide range of applications in plants. On the one hand, it is used as an herbicide, insecticide, and growth-promoting factor. On the other hand, it is also used to foster plants' tolerance to salt and cold temperatures [bib_ref] Biosynthesis, biotechnological production and applications of 5-aminolevulinic acid, Sasaki [/bib_ref]. Furthermore, the literature documents an array of 5-ALA uses in the medical field, especially for tumor diagnosis and cancer treatment [bib_ref] Preliminary study of photodynamic diagnosis using 5-aminolevulinic acid in gastric and colorectal..., Nakamura [/bib_ref]. Similarly, 5-ALA has been introduced to farm animals as a dietary supplement to improve blood iron status [bib_ref] Use of δ-aminolevulinic acid in swine diet: Effect on growth performance, behavioral..., Mateo [/bib_ref] [bib_ref] Effects of delta-aminolevulinic acid and vitamin C supplementation on feed intake, backfat,..., Wang [/bib_ref] [bib_ref] Effects of dietary delta-aminolevulinic acid and vitamin C on growth performance, immune..., Wang [/bib_ref] , enhance immunity, increase resistance to disease [bib_ref] Effect of δ-aminolevulinic acid on growth performance, nutrient digestibility, blood parameters and..., Chen [/bib_ref] [bib_ref] Dietary supplementation with 5-aminolevulinic acid modulates growth performance and inflammatory responses in..., Sato [/bib_ref] [bib_ref] Effects of dietary supplementation with delta-aminolevulinic acid on growth performance, hematological status,..., Wang [/bib_ref] , improve growth [bib_ref] δ-Aminolevulinic acid, and lactulose supplements in weaned piglets diet: Effects on performance,..., Hossain [/bib_ref] , and enhance egg production and quality [bib_ref] Evaluation of δ-aminolevulinic acid on serum iron status, blood characteristics, egg performance..., Chen [/bib_ref] [bib_ref] Effects of δ-aminolevulinic acid and vitamin C supplementation on iron status, production..., Wang [/bib_ref] [bib_ref] Evaluation of dietary supplementation of delta-aminolevulinic acid and chito-oligosaccharide on production performance,..., Yan [/bib_ref] , as well as milk composition [bib_ref] Effects of delta-aminolevulinic acid and vitamin C supplementation on feed intake, backfat,..., Wang [/bib_ref] [bib_ref] Effects of 5-aminolevulinic acid supplementation on milk production, iron status, and immune..., Hendawy [/bib_ref] [bib_ref] Dietary supplementation of delta-aminolevulinic acid to lactating sows improves growth performance and..., Lee [/bib_ref]. However, available studies report mixed findings. To our knowledge, the benefits of 5-ALA supplement in farm animals have not been systematically reviewed. Though an existing narrative review theoretically enumerates the mechanism and benefits of 5-ALA [bib_ref] δ-aminolevulinic acid (ALA) as a potential feed additive in pig: A review, Cho [/bib_ref] , no sharp cutting decisions about the effectiveness of 5-ALA could be drawn from that review. To bridge the gap, the current study aims to systematically address the beneficial effects of 5-ALA for iron status, immune response, and animal performance.
# Materials and methods
This review was conducted according to the preferred reporting items for systematic reviews and meta-analyses (PRISMA) statement guidelines [bib_ref] Academia and Clinic Annals of Internal Medicine Preferred Reporting Items for Systematic..., Moher [/bib_ref].
## Literature search
A five-stage, comprehensive search was conducted to identify eligible studies. In the first stage, two databases, namely PubMed and Web of Science, were searched to obtain all relevant 5-ALA studies that were published before March 2020. The used search strategy involved a combination of keywords: (5-aminolevulinic acid or δ-aminolevulinic acid or delta-aminolevulinic acid or aminolevulinic acid or Animals 2020, 10, 1352 3 of 15 aminolevulinate) and (cow or cattle or calves or heifer or buffalo or bull or steer or sheep or ewe or lamb or ram or goat or kid or duck or goose or geese or poultry or hen or broiler or chick or chicken or quail or pigeon or turkey or ostrich or rabbit or sow or swine or pig). The search was not restricted by language, date, or study type.
In the second stage, the total hits from both databases were pooled and duplicates were removed. The third stage involved screening of the retrieved articles by reading the article titles and abstracts. In the fourth stage, the full-length individual manuscripts were screened, and papers not satisfying the inclusion criteria (the next section) were excluded. In the fifth stage, to obtain additional data, we conducted a backward manual search of the reference lists of the selected articles.
## Inclusion and exclusion criteria
To be eligible for inclusion, studies should have met the following inclusion criteria: (1) used 5-ALA as a feed supplement, either alone or in combination with other additives; (2) incorporated samples of any farm animal species; and (3) measured the outcome criteria of iron status, immune response, and animal performance. We excluded articles that were not published in English, conference proceedings, editorials, commentaries, and book chapters/book reviews.
## Data extraction
A special tabular form was designed for data collection. Two independent researchers screened all the selected studies to extract data related to the species, treatment and control groups, doses of the supplemented 5-ALA, combinations with other additives, duration of the trial, outcome measures, and the main findings.
## Quality criteria
To evaluate the methodological quality of the reviewed articles, we developed our quality criteria by referring to the Cochrane tool for reporting randomized trials, and also by adapting some relevant criteria from available systematic reviews in the field of animal production [bib_ref] Effect of ractopamine on lipid metabolism in vivo-A systematic review, Ferreira [/bib_ref] [bib_ref] In ovo feeding of carbohydrates for broilers-A systematic review, Retes [/bib_ref]. According to this method, studies are assessed for risk of different biases on a three-point rating scale: 2 = low risk (criteria properly performed/described), 1 = unclear risk (criteria not described), and 0 = high risk (criteria not met). The following criteria were considered for quality assessment:
(i) Selection bias: trials that used a randomization procedure were ranked as low-risk; when the method of randomization was not clearly described, they were ranked as unclear risk; however, when the study was non-randomized, it was ranked as high-risk.
(ii) Allocation concealment: trials that described procedures to conceal allocation were ranked as low-risk, while when allocation was not concealed trials were ranked as high-risk; when the procedure used for concealment was not reported, it was ranked as unclear risk.
(iii) Performance bias: when treatment was conducted by experimenters who were blinded to the study aim/hypothesis, the study was ranked as low-risk; when experimenters were not blinded to the study aim/hypothesis, the study was rated as high-risk; however, when blinding of experimenters was not reported, it was ranked as unclear risk.
(iv) Detection bias: when laboratory analyses were evaluated by examiners who were blinded to the study aim/hypothesis, studies were rated as low-risk; when examiners were not blinded to the study aim/hypothesis, the studies were rated as high-risk; however, when blinding was not clearly reported, these studies were rated unclear risk.
(v) Reporting bias: when all outcomes described in the methods section were reported in the results (and reports were consistent between the abstract, results, and discussion) the studies were rated as low-risk; studies with missing outcomes were rated unclear risk; whereas studies with inconsistent reports between the abstract, results, and discussion were rated as high-risk.
(vi) Sample size: the number of animals per group was determined according to the type of species-pigs and cows (up to 30 was rated as high-risk; more than 30 was rated as low-risk), and birds (up to 50 was rated as high-risk; more than 50 was rated as low-risk).
(vii) Breed or genetic line: when breed or genetic line was detailed, trials were rated as low-risk, but when not described, trials were rated as high-risk.
# Results
## Literature search
Of all the obtained publications (1369), 1347 studies were excluded after the initial screening of titles and abstracts: 250 papers were duplicated, 52 papers were not in English, 45 papers were review papers and meta-analyses, and 1000 articles were irrelevant. We examined the full text of 22 articles, and 6 papers were further excluded, as they were repeated reports from the same samples. In total, 16 studies were included in this review [fig_ref] Figure 1: Summary of the search strategy based on the preferred reporting items for... [/fig_ref]. The selected articles were published between 2006 and 2019.
Animals 2020, 10, x 4 of 15 rated as low-risk; studies with missing outcomes were rated unclear risk; whereas studies with inconsistent reports between the abstract, results, and discussion were rated as high-risk.
(vi) Sample size: the number of animals per group was determined according to the type of species-pigs and cows (up to 30 was rated as high-risk; more than 30 was rated as low-risk), and birds (up to 50 was rated as high-risk; more than 50 was rated as low-risk).
(vii) Breed or genetic line: when breed or genetic line was detailed, trials were rated as low-risk, but when not described, trials were rated as high-risk.
# Results
## Literature search
Of all the obtained publications (1369), 1347 studies were excluded after the initial screening of titles and abstracts: 250 papers were duplicated, 52 papers were not in English, 45 papers were review papers and meta-analyses, and 1000 articles were irrelevant. We examined the full text of 22 articles, and 6 papers were further excluded, as they were repeated reports from the same samples. In total, 16 studies were included in this review [fig_ref] Figure 1: Summary of the search strategy based on the preferred reporting items for... [/fig_ref]. The selected articles were published between 2006 and 2019.
## Description of included studies
Sixteen controlled trials were included in this review. The duration of 5-ALA supplementation ranged from 10 days to 69 days, with the exception of one long-term study in which 5-ALA was supplemented to sows during gestation and lactation periods (including the weaning-to-estrus intervals) [bib_ref] Effects of iron injection at birth on neonatal iron status in young..., Wang [/bib_ref]. The dose of 5-ALA supplemented in these studies varied from 2 mg to 1 g 5-ALA/kg of feed. Seven studies included experimental groups that were fed 5-ALA combined with vitamin C [bib_ref] Effects of delta-aminolevulinic acid and vitamin C supplementation on feed intake, backfat,..., Wang [/bib_ref] [bib_ref] Effects of dietary delta-aminolevulinic acid and vitamin C on growth performance, immune..., Wang [/bib_ref] [bib_ref] Effects of δ-aminolevulinic acid and vitamin C supplementation on iron status, production..., Wang [/bib_ref] , antibiotics [bib_ref] Effects of dietary supplementation with delta-aminolevulinic acid on growth performance, hematological status,..., Wang [/bib_ref] , chito-oligosccharide [bib_ref] Evaluation of dietary supplementation of delta-aminolevulinic acid and chitooligosaccharide on growth performance,..., Yan [/bib_ref] , oriental medicinal plants [bib_ref] The effects of replacement of antibiotics with by-products of oriental medicinal plants..., Kang [/bib_ref] , and iron injection [bib_ref] Effects of iron injection at birth on neonatal iron status in young..., Wang [/bib_ref].
Out of 16 studies, nine used pigs' samples (three studies had sow subjects, while six studies were comprised of weanling or finishing pigs). The remaining seven studies included chicken samples-three studies involved broiler chickens and three studies included laying hens. One study included dairy cows.
All studies measured at least one indicator of growth performance, such as average daily gain (ADG), average daily feed intake, feed efficiency (G/F), feed conversion ratio (F/G), body weight (BW), BW at birth and weaning, born-alive litter size, and nutrient digestibility, such as dry matter, nitrogen, and energy digestibility. Moreover, seven studies assessed production performance. Four studies examined milk composition, e.g., milk fat, milk protein, milk casein, milk glucose, milk lactose, total solid, and Fe concentration [bib_ref] Effects of delta-aminolevulinic acid and vitamin C supplementation on feed intake, backfat,..., Wang [/bib_ref] [bib_ref] Effects of 5-aminolevulinic acid supplementation on milk production, iron status, and immune..., Hendawy [/bib_ref] [bib_ref] Dietary supplementation of delta-aminolevulinic acid to lactating sows improves growth performance and..., Lee [/bib_ref] [bib_ref] Effects of iron injection at birth on neonatal iron status in young..., Wang [/bib_ref]. Egg production and quality were also assessed in three studies: egg weight, shell breaking strength, shell thickness, Haugh unit, yolk color unit, yolk index, shells color, albumin height, and Fe concentration in yolk [bib_ref] Evaluation of δ-aminolevulinic acid on serum iron status, blood characteristics, egg performance..., Chen [/bib_ref] [bib_ref] Effects of δ-aminolevulinic acid and vitamin C supplementation on iron status, production..., Wang [/bib_ref] [bib_ref] Evaluation of dietary supplementation of delta-aminolevulinic acid and chito-oligosaccharide on production performance,..., Yan [/bib_ref].
The effect of 5-ALA supplementation on iron status was studied in 14 studies that used five indices: hemoglobin, hematocrit, Fe, total iron binding capacity (TIBC), and red blood cells (RBC). In two studies, iron was measured in plasma, and in the rest of the studies it was measured in serum. Herein, iron status refers to serum iron levels, unless otherwise specified.
In this review, we considered indices reflecting different levels of defense mechanisms of the immune system: white blood cell (WBC) count, differential count, and cell-mediated immunity, which involves soluble proteins and bioactive small molecules released by the activated cells, such as cytokines, as well as the membrane-bound receptors and cytoplasmic proteins that bind to molecular patterns expressed on the surfaces of invading pathogens. Hence, the immune response in the evaluated studies was addressed through an estimation of WBC count and lymphocyte count, as well as measurement of tumor necrosis factor (TNF)-α; insulin-like growth factor (IGF)-1; haptoglobin; plasma cortisol [bib_ref] Effect of δ-aminolevulinic acid on growth performance, nutrient digestibility, blood parameters and..., Chen [/bib_ref] ; cluster of differentiation antigen positive cells 2, 4, and 8 (CD2+, CD4+, CD8+, respectively); the ratio of CD4+ to CD8+ cells; B-cells; major histocompatibility complex classes I and II [bib_ref] Effects of dietary supplementation with delta-aminolevulinic acid on growth performance, hematological status,..., Wang [/bib_ref] ; rates of phagocytosis in blood mononuclear cells (MNC); mitogen concanavalin A and phytohemagglutinin-induced proliferation of blood MNC [bib_ref] Dietary supplementation with 5-aminolevulinic acid modulates growth performance and inflammatory responses in..., Sato [/bib_ref] [bib_ref] Effects of 5-aminolevulinic acid supplementation on milk production, iron status, and immune..., Hendawy [/bib_ref] ; levels of cluster of differentiation 3 (CD3) mRNA in the spleen; plasma thiobarbituric acid reactive substances (TBARS); plasma ceruloplasmin concentration; expression of interferon-γ; inducible nitric oxide synthase (iNOS); interleukin (IL)-6 and TNF-like ligand 1A mRNA; levels of IL-2; toll-like receptor 2, 4, and 7 mRNA in the spleen during Escherichia coli lipopolysaccharide (LPS) stimulation; ceruloplasmin oxidase [bib_ref] Dietary supplementation with 5-aminolevulinic acid modulates growth performance and inflammatory responses in..., Sato [/bib_ref] ; differential count of WBC [bib_ref] Use of δ-aminolevulinic acid in swine diet: Effect on growth performance, behavioral..., Mateo [/bib_ref] ; immunoglobin G (IgG) [bib_ref] Effect of δ-aminolevulinic acid on growth performance, nutrient digestibility, blood parameters and..., Chen [/bib_ref] [bib_ref] Effects of delta-aminolevulinic acid and vitamin C supplementation on feed intake, backfat,..., Wang [/bib_ref] [bib_ref] Evaluation of dietary supplementation of delta-aminolevulinic acid and chitooligosaccharide on growth performance,..., Yan [/bib_ref] ; and the weight of immune organs, such as spleen, bursa of Fabricius, liver, and the thymus [bib_ref] Effects of dietary delta-aminolevulinic acid and vitamin C on growth performance, immune..., Wang [/bib_ref] [bib_ref] Dietary supplementation with 5-aminolevulinic acid modulates growth performance and inflammatory responses in..., Sato [/bib_ref] [bib_ref] Utilization of δ-aminolevulinic acid for livestock: Blood characteristics and immune organ weight..., Chen [/bib_ref]. Readers are encouraged to refer to Supplementary Table S1 for further details.
The quality of the selected articles was evaluated according to the aforementioned criteria. The maximum achieved quality score was 11 points [bib_ref] Effects of dietary delta-aminolevulinic acid and vitamin C on growth performance, immune..., Wang [/bib_ref] [bib_ref] Effects of dietary supplementation with delta-aminolevulinic acid on growth performance, hematological status,..., Wang [/bib_ref] [bib_ref] δ-Aminolevulinic acid, and lactulose supplements in weaned piglets diet: Effects on performance,..., Hossain [/bib_ref] [bib_ref] Evaluation of δ-aminolevulinic acid on serum iron status, blood characteristics, egg performance..., Chen [/bib_ref] [bib_ref] Evaluation of dietary supplementation of delta-aminolevulinic acid and chito-oligosaccharide on production performance,..., Yan [/bib_ref] [bib_ref] Utilization of δ-aminolevulinic acid for livestock: Blood characteristics and immune organ weight..., Chen [/bib_ref] , while the minimum score was eight points [bib_ref] Effects of delta-aminolevulinic acid and vitamin C supplementation on feed intake, backfat,..., Wang [/bib_ref] [bib_ref] The effects of replacement of antibiotics with by-products of oriental medicinal plants..., Kang [/bib_ref]. [fig_ref] Figure 2: Quality assessment of each included studies [/fig_ref] show that except for one study [bib_ref] Use of δ-aminolevulinic acid in swine diet: Effect on growth performance, behavioral..., Mateo [/bib_ref] , all included studies used randomization procedures; thus, they were low-risk for selection bias. Not a single article indicated or described the use of any procedure of allocation concealment, blinding of treatment performers, (in farms) or blinding of analyses performers (in laboratories). All papers described the breed/animal genetic line. Reporting bias was noticed in 12.5% of the papers [bib_ref] Effects of delta-aminolevulinic acid and vitamin C supplementation on feed intake, backfat,..., Wang [/bib_ref] [bib_ref] The effects of replacement of antibiotics with by-products of oriental medicinal plants..., Kang [/bib_ref]. Less than half the studies (44%) used more than 30 pigs/cows or 50 birds [bib_ref] Use of δ-aminolevulinic acid in swine diet: Effect on growth performance, behavioral..., Mateo [/bib_ref] [bib_ref] Effects of dietary delta-aminolevulinic acid and vitamin C on growth performance, immune..., Wang [/bib_ref] [bib_ref] Effects of dietary supplementation with delta-aminolevulinic acid on growth performance, hematological status,..., Wang [/bib_ref] [bib_ref] δ-Aminolevulinic acid, and lactulose supplements in weaned piglets diet: Effects on performance,..., Hossain [/bib_ref] [bib_ref] Evaluation of δ-aminolevulinic acid on serum iron status, blood characteristics, egg performance..., Chen [/bib_ref] [bib_ref] Evaluation of dietary supplementation of delta-aminolevulinic acid and chito-oligosaccharide on production performance,..., Yan [/bib_ref] [bib_ref] Utilization of δ-aminolevulinic acid for livestock: Blood characteristics and immune organ weight..., Chen [/bib_ref]. It is worth noting that two studies, comprising four experiments, were rated as high-risk for sampling bias, because animal numbers in some groups were small, though some groups had adequate numbers [bib_ref] Dietary supplementation with 5-aminolevulinic acid modulates growth performance and inflammatory responses in..., Sato [/bib_ref] [bib_ref] Effects of iron injection at birth on neonatal iron status in young..., Wang [/bib_ref]. studies, comprising four experiments, were rated as high-risk for sampling bias, because animal numbers in some groups were small, though some groups had adequate numbers [bib_ref] Dietary supplementation with 5-aminolevulinic acid modulates growth performance and inflammatory responses in..., Sato [/bib_ref] [bib_ref] Effects of iron injection at birth on neonatal iron status in young..., Wang [/bib_ref]. studies, comprising four experiments, were rated as high-risk for sampling bias, because animal numbers in some groups were small, though some groups had adequate numbers [bib_ref] Dietary supplementation with 5-aminolevulinic acid modulates growth performance and inflammatory responses in..., Sato [/bib_ref] [bib_ref] Effects of iron injection at birth on neonatal iron status in young..., Wang [/bib_ref]. [fig_ref] Table 1: Effects of 5-aminolevulinic acid [/fig_ref] summarizes effects of 5-ALA in different farm animals. Overall, no significant effects of 5-ALA supplementation on growth performance were observed, except in two studies that reported improved ADG and G/F [bib_ref] Effects of delta-aminolevulinic acid and vitamin C supplementation on feed intake, backfat,..., Wang [/bib_ref] [bib_ref] δ-Aminolevulinic acid, and lactulose supplements in weaned piglets diet: Effects on performance,..., Hossain [/bib_ref]. Significant increases were reported for BW [bib_ref] Effects of delta-aminolevulinic acid and vitamin C supplementation on feed intake, backfat,..., Wang [/bib_ref] [bib_ref] Dietary supplementation with 5-aminolevulinic acid modulates growth performance and inflammatory responses in..., Sato [/bib_ref] [bib_ref] δ-Aminolevulinic acid, and lactulose supplements in weaned piglets diet: Effects on performance,..., Hossain [/bib_ref] [bib_ref] Effects of iron injection at birth on neonatal iron status in young..., Wang [/bib_ref] , born-alive litter size, and BW at birth and weaning [bib_ref] Effects of iron injection at birth on neonatal iron status in young..., Wang [/bib_ref]. Regarding nutrient digestibility, only two studies reported significant positive effects of supplementation of 5-ALA in different dietary levels for dry matter [bib_ref] Effect of δ-aminolevulinic acid on growth performance, nutrient digestibility, blood parameters and..., Chen [/bib_ref] [bib_ref] δ-Aminolevulinic acid, and lactulose supplements in weaned piglets diet: Effects on performance,..., Hossain [/bib_ref] and nitrogen digestibility at the end of treatment [bib_ref] Effect of δ-aminolevulinic acid on growth performance, nutrient digestibility, blood parameters and..., Chen [/bib_ref]. Lactating dairy cows ND + + 0 [bib_ref] Effects of 5-aminolevulinic acid supplementation on milk production, iron status, and immune..., Hendawy [/bib_ref] +: Positive effect; 0: no effect; ND: not determined; : growth performance indices were measured in progenies.
## Effects of 5-ala on animal performance
Concerning production performance, 5-ALA supplementation improved milk protein [bib_ref] Effects of delta-aminolevulinic acid and vitamin C supplementation on feed intake, backfat,..., Wang [/bib_ref] [bib_ref] Effects of 5-aminolevulinic acid supplementation on milk production, iron status, and immune..., Hendawy [/bib_ref] [bib_ref] Dietary supplementation of delta-aminolevulinic acid to lactating sows improves growth performance and..., Lee [/bib_ref] , milk fat [bib_ref] Effects of delta-aminolevulinic acid and vitamin C supplementation on feed intake, backfat,..., Wang [/bib_ref] , milk casein [bib_ref] Effects of 5-aminolevulinic acid supplementation on milk production, iron status, and immune..., Hendawy [/bib_ref] , and Fe concentration in milk [bib_ref] Effects of iron injection at birth on neonatal iron status in young..., Wang [/bib_ref] , while milk glucose, milk lactose, and total solids were unaffected. Similarly, 5-ALA supplementation improved egg production [bib_ref] Effects of δ-aminolevulinic acid and vitamin C supplementation on iron status, production..., Wang [/bib_ref] [bib_ref] Evaluation of dietary supplementation of delta-aminolevulinic acid and chito-oligosaccharide on production performance,..., Yan [/bib_ref] and quality, such as egg weight [bib_ref] Evaluation of dietary supplementation of delta-aminolevulinic acid and chito-oligosaccharide on production performance,..., Yan [/bib_ref] , Haugh unit [bib_ref] Evaluation of δ-aminolevulinic acid on serum iron status, blood characteristics, egg performance..., Chen [/bib_ref] [bib_ref] Effects of δ-aminolevulinic acid and vitamin C supplementation on iron status, production..., Wang [/bib_ref] [bib_ref] Evaluation of dietary supplementation of delta-aminolevulinic acid and chito-oligosaccharide on production performance,..., Yan [/bib_ref] , yolk color unit [bib_ref] Effects of δ-aminolevulinic acid and vitamin C supplementation on iron status, production..., Wang [/bib_ref] [bib_ref] Evaluation of dietary supplementation of delta-aminolevulinic acid and chito-oligosaccharide on production performance,..., Yan [/bib_ref] , eggshell color, Fe concentration in yolk [bib_ref] Effects of δ-aminolevulinic acid and vitamin C supplementation on iron status, production..., Wang [/bib_ref] , and egg yolk index [bib_ref] Evaluation of δ-aminolevulinic acid on serum iron status, blood characteristics, egg performance..., Chen [/bib_ref]. However, there were no changes in some of the aforementioned studies in egg production, egg weight, eggshell breaking strength, eggshell thickness, yolk color unit, and albumin height.
## Effect of 5-ala on iron status
Out of the 13 studies that measured hemoglobin concentration, 5-ALA improved hemoglobin in 8 out of 15 results (e.g., in sows and their piglets). As for hematocrit concentration, three out of seven results revealed some 5-ALA related improvements. Regarding iron concentration, 5-ALA improved plasma and serum Fe concentration in 11 out of 15 results. In relation to TIBC concentration, only 3 of the 12 results witnessed an increase of TIBC concentration because of 5-ALA supplementation. However, RBC count was improved in 8 out of 12 results as a result of the use of 5-ALA.
## Effect of 5-ala on immune response
Supplementation of 5-ALA was associated with improved IGF-1 (2 h) and WBCs (12 h) post-challenge by LPS [bib_ref] Effect of δ-aminolevulinic acid on growth performance, nutrient digestibility, blood parameters and..., Chen [/bib_ref]. Also, 5-ALA increased levels of CD2+, CD8+, B-cells, major histocompatibility complex classes I and II [bib_ref] Effects of dietary supplementation with delta-aminolevulinic acid on growth performance, hematological status,..., Wang [/bib_ref] , the rate of phagocytosis and mitogen-induced proliferation of peripheral [bib_ref] Dietary supplementation with 5-aminolevulinic acid modulates growth performance and inflammatory responses in..., Sato [/bib_ref] [bib_ref] Effects of 5-aminolevulinic acid supplementation on milk production, iron status, and immune..., Hendawy [/bib_ref] , CD3 mRNA in the spleen, and plasma TBARS [bib_ref] Dietary supplementation with 5-aminolevulinic acid modulates growth performance and inflammatory responses in..., Sato [/bib_ref]. On the other hand, dietary use of 5-ALA reduced plasma cortisol and TNF-α concentration at 2 h post-challenge by LPS [bib_ref] Effect of δ-aminolevulinic acid on growth performance, nutrient digestibility, blood parameters and..., Chen [/bib_ref]. Similarly, another trial revealed noticeable decreases of IL-2 levels, plasma ceruloplasmin concentration (24 h), expression of interferon-γ, iNOS, IL-6, and TNF-like ligand 1A mRNA (3 h) after LPS injection [bib_ref] Dietary supplementation with 5-aminolevulinic acid modulates growth performance and inflammatory responses in..., Sato [/bib_ref]. However, findings were mixed, as some studies showed no changes in lymphocyte and haptoglobin post-challenge by LPS [bib_ref] Effect of δ-aminolevulinic acid on growth performance, nutrient digestibility, blood parameters and..., Chen [/bib_ref] ; CD4+ and the ratio of CD4+ to CD8+ [bib_ref] Effects of dietary supplementation with delta-aminolevulinic acid on growth performance, hematological status,..., Wang [/bib_ref] ; interferon-γ mRNA expression; and levels of CD3, IL-2, and toll-like receptors 2, 4, and 7 mRNA in the spleen during LPS stimulation [bib_ref] Dietary supplementation with 5-aminolevulinic acid modulates growth performance and inflammatory responses in..., Sato [/bib_ref]. There was no change in the differential count of WBCs in one study [bib_ref] Use of δ-aminolevulinic acid in swine diet: Effect on growth performance, behavioral..., Mateo [/bib_ref]. While the number of WBCs and the percentage of lymphocytes were examined in 10 studies, only three studies reported an increase of WBCs [bib_ref] Effect of δ-aminolevulinic acid on growth performance, nutrient digestibility, blood parameters and..., Chen [/bib_ref] [bib_ref] Evaluation of dietary supplementation of delta-aminolevulinic acid and chito-oligosaccharide on production performance,..., Yan [/bib_ref] [bib_ref] Effects of 5-aminolevulinic acid supplementation on milk production, iron status, and immune..., Hendawy [/bib_ref] and percentage of lymphocytes [bib_ref] Effects of δ-aminolevulinic acid and vitamin C supplementation on iron status, production..., Wang [/bib_ref] [bib_ref] Evaluation of dietary supplementation of delta-aminolevulinic acid and chito-oligosaccharide on production performance,..., Yan [/bib_ref] as a return for 5-ALA supplementation. Meanwhile, IgG was measured in three studies, and it was increased in two of them [bib_ref] Effects of delta-aminolevulinic acid and vitamin C supplementation on feed intake, backfat,..., Wang [/bib_ref] [bib_ref] Evaluation of dietary supplementation of delta-aminolevulinic acid and chitooligosaccharide on growth performance,..., Yan [/bib_ref]. Captured in a sole study, the plasma ceruloplasmin oxidase activity of piglets decreased due to 5-ALA supplementation [bib_ref] Effects of iron injection at birth on neonatal iron status in young..., Wang [/bib_ref]. Regarding the relative weight of the immune organs, mixed findings were indicated. In one study, 5-ALA increased the weight of both the spleen and the bursa of Fabricius [bib_ref] Utilization of δ-aminolevulinic acid for livestock: Blood characteristics and immune organ weight..., Chen [/bib_ref]. On the contrary, two other studies indicated no change in the weight of the spleen, the liver, the bursa of Fabricius, or the thymus gland [bib_ref] Effects of dietary delta-aminolevulinic acid and vitamin C on growth performance, immune..., Wang [/bib_ref] [bib_ref] Dietary supplementation with 5-aminolevulinic acid modulates growth performance and inflammatory responses in..., Sato [/bib_ref] (see [fig_ref] Table 1: Effects of 5-aminolevulinic acid [/fig_ref] for further details).
## Effects of 5-ala compared with other alternative treatments
In three studies using vitamin C as an alternative treatment, 5-ALA alone had superior effects to vitamin C alone and vitamin C combined with 5-ALA on milk composition (in sows), serum Fe levels, hemoglobin, and RBC count, both in poultry and porcine species . Meanwhile, vitamin C on its own had no effect on any indicators of Fe status, except for hematocrit (in one study) [bib_ref] Effects of dietary delta-aminolevulinic acid and vitamin C on growth performance, immune..., Wang [/bib_ref]. Interestingly, piglets of sows receiving a combination of vitamin C with 5-ALA exhibited significant increases in BW at weaning and ADG compared with other groups [bib_ref] Effects of delta-aminolevulinic acid and vitamin C supplementation on feed intake, backfat,..., Wang [/bib_ref]. Nonetheless, a combination of 5-ALA and vitamin C had a better effect on lymphocytes than each treatment on its own in poultry [bib_ref] Effects of δ-aminolevulinic acid and vitamin C supplementation on iron status, production..., Wang [/bib_ref]. In laying hens, solo 5-ALA treatment significantly improved albumin height, yolk color unit, eggshell color, and Haugh unit, while vitamin C alone improved only the last three parameters. Strangely, combinations of 5-ALA and vitamin C demonstrated no effect on any of these egg qualities, though eggshell thickness and iron concentration in yolk significantly increased [bib_ref] Effects of δ-aminolevulinic acid and vitamin C supplementation on iron status, production..., Wang [/bib_ref].
Two studies compared the effects of antibiotics (apramycin) [bib_ref] Effects of dietary supplementation with delta-aminolevulinic acid on growth performance, hematological status,..., Wang [/bib_ref] and chito-oligosccharides [bib_ref] Evaluation of dietary supplementation of delta-aminolevulinic acid and chitooligosaccharide on growth performance,..., Yan [/bib_ref] with that of 5-ALA on iron status in weanling pigs. Neither antibiotics nor chito-oligosccharides could express any positive effect except for lymphocyte count (only chito-oligosccharides). On the other hand, solo 5-ALA increased Fe levels, TIBC, RBC [bib_ref] Evaluation of dietary supplementation of delta-aminolevulinic acid and chitooligosaccharide on growth performance,..., Yan [/bib_ref] , hemoglobin, and hematocrit [bib_ref] Effects of dietary supplementation with delta-aminolevulinic acid on growth performance, hematological status,..., Wang [/bib_ref]. In the meantime, treatments involving 5-ALA in combination with antibiotics or chito-oligosccharides could not produce any significant effect. 5-ALA with or without antibiotics was associated with an increase of the levels of CD2+, CD8+, B-cells, and major histocompatibility complex classes I and II [bib_ref] Effects of dietary supplementation with delta-aminolevulinic acid on growth performance, hematological status,..., Wang [/bib_ref]. In a single study, inorganic iron (FeSO 4 ) was used as a positive control, and authors compared the effects of 5-ALA and iron injection (iron dextran) on iron status and growth performance in young pigs. Both 5-ALA and iron injection significantly increased Fe, hematocrit, hemoglobin, and BW at weaning. Animals receiving a combination of 5-ALA and iron injection have more positive effects than animals receiving only a single treatment [bib_ref] Effects of iron injection at birth on neonatal iron status in young..., Wang [/bib_ref]. Further details are shown in .
## Mechanism of action of 5-ala
Heme formation process initiated by 5-ALA starts in the cytoplasm, where 5-ALA sequentially generates porphobilinogen, hydroxymethylbilane, uroporphyrinogen III, and eventually coproporphyrinogen III. Coproporphyrinogen III translocates into the mitochondria, where it gets metabolized to protoporphyrinogen IX and protoporphyrin IX, into which iron is inserted through a ferrochelatase-catalyzed reaction, resulting in heme synthesis [bib_ref] 5-Aminolevulinic acid regulates the inflammatory response and alloimmune reaction, Fujino [/bib_ref]. It seems that the action of 5-ALA is adaptive, i.e., it varies in response to changes in the internal microenvironment of the body. In this respect, 5-ALA enhanced the levels of hemoglobin, RBCs, and serum Fe concentration in young pigs [bib_ref] Effect of δ-aminolevulinic acid on growth performance, nutrient digestibility, blood parameters and..., Chen [/bib_ref] [bib_ref] Effects of delta-aminolevulinic acid and vitamin C supplementation on feed intake, backfat,..., Wang [/bib_ref] [bib_ref] δ-Aminolevulinic acid, and lactulose supplements in weaned piglets diet: Effects on performance,..., Hossain [/bib_ref] , which may suggest that the effect of 5-ALA on heme generation occurs under conditions of iron deficiency. This is because young pigs commonly exhibit iron deficiency for a number of reasons: low hepatic iron stores in newborn pigs, low iron concentrations in the milk of the sows, and rapid increase in RBC volume and body mass [bib_ref] Bioavailability of iron from amino acid complex in weanling pigs, Yu [/bib_ref]. In support of this view, research shows that heme has a feedback effect on 5-ALA synthase, which gets downregulated in conditions involving increased heme concentrations, resulting in suppression of heme synthesis [bib_ref] Utilization of δ-aminolevulinic acid for livestock: Blood characteristics and immune organ weight..., Chen [/bib_ref] [bib_ref] Tissue-specific regulation of iron metabolism and heme synthesis: Distinct control mechanisms in..., Ponka [/bib_ref].
From another perspective, levels of endogenous protoporphyrin induced by systemic administration of 5-ALA vary in different tissues of the body-protoporphyrin was not detected in muscle, but relatively higher levels (18%) were detected in the liver, a main detoxifying organ, and much higher levels were detected in tumor tissues. As a result, protoporphyrin suppressed energy phosphate metabolism, causing tumor adenosine triphosphate levels to drop to near zero, leading to high necrosis and the inhibition of tumor growth. In that study, the activities of certain selected enzymes involved in heme biosynthesis had no association with porphyrin concentrations in different tissues [bib_ref] Effectiveness of delta-aminolevulinic acid-induced protoporphyrin as a photosensitizer for photodynamic therapy in..., Hua [/bib_ref]. In the current review, 5-ALA had no effect on heme synthesis in some studies (e.g., involving dairy cows); however, it boosted immune functioning [bib_ref] Effects of 5-aminolevulinic acid supplementation on milk production, iron status, and immune..., Hendawy [/bib_ref]. Evidence associates inflammation and oxidative stress that foster senescence with levels of iron overload-related mitochondrial ferritin. As an adaptive response, the master antioxidant molecule, nuclear factor erythroid 2 (NRF2, gets activated in order to upregulate levels of phase II antioxidants, such as heme oxygenase-1 (HO-1). Phase II antioxidants get activated in pathological conditions as a defense against insults induced by cellular toxins. Therefore, phase II antioxidants trigger an array of cellular activities that aim at restoring cellular structure and function [bib_ref] Hippocampal lipocalin 2 is associated with neuroinflammation and iron-related oxidative stress in..., Jin [/bib_ref]. In this respect, the above-reviewed studies signify an immunomodulatory effect of 5-ALA. It seems that the biological activities of 5-ALA, such as being an antioxidant, anti-inflammatory, and immunomodulator underlie the reported effects [fig_ref] Figure 4: The possible mechanism of action of 5-aminolevulinic acid [/fig_ref]. Herein, we explore the literature for the molecular actions that promote the positive effects of 5-ALA.
A wealth of studies reports a potent antioxidant effect of 5-ALA [bib_ref] 5-Aminolevulinic acid regulates the inflammatory response and alloimmune reaction, Fujino [/bib_ref] [bib_ref] 5-Aminolevulinic acid combined with ferrous iron enhances the expression of heme oxygenase-1, Nishio [/bib_ref]. In details, 5-ALA promotes the production of internal antioxidants, such as HO-1 [bib_ref] 5-Aminolevulinic acid regulates the inflammatory response and alloimmune reaction, Fujino [/bib_ref] [bib_ref] 5-Aminolevulinic acid combined with ferrous iron enhances the expression of heme oxygenase-1, Nishio [/bib_ref]. HO-1 promotes the production of other antioxidant molecules, such as biliverdin, ferrous iron, and carbon monoxide. The action underlying this activity is promoting the degradation of heme [bib_ref] 5-Aminolevulinic acid combined with ferrous iron enhances the expression of heme oxygenase-1, Nishio [/bib_ref] [bib_ref] 5-Aminolevulinic acid regulates the immune response in LPS-stimulated RAW 264.7 macrophages, Sugiyama [/bib_ref].
Similar to the findings reported by Sugiyama et al. [bib_ref] 5-Aminolevulinic acid regulates the immune response in LPS-stimulated RAW 264.7 macrophages, Sugiyama [/bib_ref] , we detected a slight increase of TNF-α in peripheral blood MNC (obtained from untreated dairy cows) treated with 5-ALA in the absence of LPS, while co-treatment with LPS resulted in a significant reduction in the mRNA expression of TNF-α (unpublished data). The literature documents that mild levels of free radicals, such as reactive oxygen species (ROS) and reactive nitrogen species (RNS), promote optimal biological functioning. For example, ROS/RNS support the production of antioxidants through the activation of NRF2, and they accelerate the degradation of tumor cells and enhance autophagy and intracellular defenses against pathogens [bib_ref] IFN-γ + LPS induction of iNOS is modulated by ERK, JNK/SAPK, and..., Chan [/bib_ref] [bib_ref] 4-Hydroperoxy-2-decenoic acid ethyl ester protects against 6-hydroxydopamine-induced cell death via activation of..., Inoue [/bib_ref]. It is well-known that LPS treatment elevates ROS and RNS production [bib_ref] IFN-γ + LPS induction of iNOS is modulated by ERK, JNK/SAPK, and..., Chan [/bib_ref]. Meanwhile, the production of HO-1 increases following treatment with a range of stress stimuli [bib_ref] 5-Aminolevulinic acid regulates the inflammatory response and alloimmune reaction, Fujino [/bib_ref]. In addition to their antioxidant activities, HO-1 and NRF2 also suppress inflammation through inhibition of the key inflammatory pathway: nuclear factor kappa light chain enhancer of activated B cells (NF-kB) [bib_ref] 4-Hydroperoxy-2-decenoic acid ethyl ester protects against 6-hydroxydopamine-induced cell death via activation of..., Inoue [/bib_ref] [bib_ref] Bee honey protects astrocytes against oxidative stress: A preliminary in vitro investigation, Ali [/bib_ref] [bib_ref] Royal Jelly and Its Components Promote Healthy Aging and Longevity: From Animal..., Kunugi [/bib_ref].
Therefore, the anti-inflammatory effects of 5-ALA may be a mere translation of its antioxidant effects. In other words, pretreatment with 5-ALA would induce a mild degree of oxidative stress, resulting in cellular pre-conditioning to ROS and RNS. Subsequent treatment with LPS would result in the activation of cellular signaling necessary for cell protection against the high emission of free radicals and inflammatory mediators in pre-conditioned cells, compared with untreated cells [bib_ref] 4-Hydroperoxy-2-decenoic acid ethyl ester protects against 6-hydroxydopamine-induced cell death via activation of..., Inoue [/bib_ref]. The cytoprotective role demonstrated by HO-1 is documented in several studies. In this regard, HO-1-deficient cells display limited stress resistance [bib_ref] Reduced stress defense in heme oxygenase 1-deficient cells, Poss [/bib_ref] , and people expressing deficiency in genes involved in the production of HO-1 demonstrate higher production of inflammatory markers [bib_ref] Oxidative stress causes enhanced endothelial cell injury in human heme oxygenase-1 deficiency, Yachie [/bib_ref]. Likewise, an in vivo investigation involving treatment of HO-1-deficient rodents with endotoxins is associated with the body flooding with cytokines, and necrosis in key organs involved in immune protection, such as the liver and the spleen [bib_ref] Heme oxygenase-1 modulates early inflammatory responses: Evidence from the heme oxygenase-1-deficient mouse, Matthias [/bib_ref]. Finally, the 5-ALA mechanism of actions seem to be adaptive to internal needs of body environment, promoting heme synthesis under conditions of deficiency and promoting proper cellular housekeeping via enhancement of the activity of internal antioxidants. However, more studies are needed to examine the plausibility of these pathways.
# Discussion
To our knowledge, this is the first attempt to systematically investigate the benefits of 5-ALA in farm animals. To determine its potential effects on animal performance, iron status, and immune response, we summarized data from 16 studies that used 5-ALA as a supplement in various farm animals. Except for milk protein, animal performance indices revealed few significant benefits of 5-ALA for growth performance, or for egg production and quality. Meanwhile, mixed findings were reported, with some significant effects observed for indices of iron status and immune response. However, studies indicating significance exhibited some heterogeneity with regard to 5-ALA doses; duration of supplementation; and animal species/breed, sex, and age. Generalizability of the findings is limited because of the heterogeneity of the techniques and outcome measures; in addition, samples consisted only of pigs, chickens, and dairy cows.
Despite the fact that an existing review of 5-ALA as a potential feed additive had been conducted by Cho and Kim, that review is narrative and theory-based-in a few instances, it addressed examples of the effects of 5-ALA from few, subjectively selected studies [bib_ref] δ-aminolevulinic acid (ALA) as a potential feed additive in pig: A review, Cho [/bib_ref]. We located another narrative review of the role of 5-ALA in the regulation of iron metabolism in animals, but it is in Chinese [bib_ref] Advance in the application of 5-aminolevulinic acid in the regulation of animal..., Longtao [/bib_ref]. Therefore, our attempt to systematically examine the benefits of 5-ALA in farm animals uniquely expands the existing knowledge about the effects of 5-ALA in farm animals.
# Discussion
To our knowledge, this is the first attempt to systematically investigate the benefits of 5-ALA in farm animals. To determine its potential effects on animal performance, iron status, and immune response, we summarized data from 16 studies that used 5-ALA as a supplement in various farm animals. Except for milk protein, animal performance indices revealed few significant benefits of 5-ALA for growth performance, or for egg production and quality. Meanwhile, mixed findings were reported, with some significant effects observed for indices of iron status and immune response. However, studies indicating significance exhibited some heterogeneity with regard to 5-ALA doses; duration of supplementation; and animal species/breed, sex, and age. Generalizability of the findings is limited because of the heterogeneity of the techniques and outcome measures; in addition, samples consisted only of pigs, chickens, and dairy cows.
Despite the fact that an existing review of 5-ALA as a potential feed additive had been conducted by Cho and Kim, that review is narrative and theory-based-in a few instances, it addressed examples of the effects of 5-ALA from few, subjectively selected studies [bib_ref] δ-aminolevulinic acid (ALA) as a potential feed additive in pig: A review, Cho [/bib_ref]. We located another narrative review of the role of 5-ALA in the regulation of iron metabolism in animals, but it is in Chinese [bib_ref] Advance in the application of 5-aminolevulinic acid in the regulation of animal..., Longtao [/bib_ref]. Therefore, our attempt to systematically examine the benefits of 5-ALA in farm animals uniquely expands the existing knowledge about the effects of 5-ALA in farm animals.
In theory, 5-ALA does not directly influence the animal bodies; it has a role in heme synthesis, through which heme or hemoglobin content improves [bib_ref] δ-aminolevulinic acid (ALA) as a potential feed additive in pig: A review, Cho [/bib_ref]. Some effects of 5-ALA on oxidative stress have been indicated. In one study, plasma ceruloplasmin oxidase activity-which increases as a result of the deficiency of antioxidant enzymes during iron deficiency-decreased in response to a dose of 90 mg/kg of 5-ALA, both with and without iron injection [bib_ref] Effects of iron injection at birth on neonatal iron status in young..., Wang [/bib_ref]. On the other hand, another study revealed an adverse effect of 5-ALA over long-term use, where plasma TBARS, a marker of lipid peroxidation and oxidative stress, increased in broiler chickens treated by 5-ALA compared with the control groups and chickens treated by the same 5-ALA dose for 10 days [bib_ref] Dietary supplementation with 5-aminolevulinic acid modulates growth performance and inflammatory responses in..., Sato [/bib_ref]. However, TBARS concentration in the loin meat of swine treated by a combination of 5-ALA and oriental medicinal plants was significantly lower than in the control group [bib_ref] The effects of replacement of antibiotics with by-products of oriental medicinal plants..., Kang [/bib_ref].
No overall improvement was noticed in most growth indices. However, the highest significant effect on growth performance was noticed in few studies that used high doses (500 mg and 1 g 5-ALA/kg of feed) of 5-ALA [bib_ref] δ-Aminolevulinic acid, and lactulose supplements in weaned piglets diet: Effects on performance,..., Hossain [/bib_ref] and the long-term use of 10 mg 5-ALA/kg of feed [bib_ref] Dietary supplementation with 5-aminolevulinic acid modulates growth performance and inflammatory responses in..., Sato [/bib_ref]. Furthermore, some significant changes were noted at certain stages of development-e.g., ADG significantly improved during the first week of life in piglets of sows fed on 5-ALA [bib_ref] Effects of iron injection at birth on neonatal iron status in young..., Wang [/bib_ref] , while two studies showed an increase in ADG on day 21 [bib_ref] Effect of δ-aminolevulinic acid on growth performance, nutrient digestibility, blood parameters and..., Chen [/bib_ref] and from 2 to 5 weeks of treatment [bib_ref] Evaluation of dietary supplementation of delta-aminolevulinic acid and chitooligosaccharide on growth performance,..., Yan [/bib_ref].
Pertaining to the remaining outcome indicators, data from the reviewed studies did not follow a certain pattern. For example, milk protein and fat improved with supplementation of 10 mg/kg of 5-ALA for 28 days [bib_ref] Effects of delta-aminolevulinic acid and vitamin C supplementation on feed intake, backfat,..., Wang [/bib_ref] , whereas milk Fe improved with supplementation of 90 mg/kg of 5-ALA to sows during the gestation and lactation periods [bib_ref] Effects of iron injection at birth on neonatal iron status in young..., Wang [/bib_ref]. Interestingly, the use of trivial doses of 5-ALA (2 and 4 mg 5-ALA/kg) resulted in improvement of egg production [bib_ref] Evaluation of dietary supplementation of delta-aminolevulinic acid and chito-oligosaccharide on production performance,..., Yan [/bib_ref]. Nonetheless, administration of 10 mg/kg of 5-ALA combined with 500 mg/kg of vitamin C gave the highest egg production and improved egg qualities [bib_ref] Effects of δ-aminolevulinic acid and vitamin C supplementation on iron status, production..., Wang [/bib_ref]. Blood-related outcomes, especially Fe, seemed to be generally improved in both standalone and combined different doses of 5-ALA for different durations. As hypothesized, 5-ALA as feed additive turned out to be more effective than iron supplement for enhancing indices of Fe status. In fact, Fe as a feed additive has failed to maintain indices of Fe status in pigs [bib_ref] Effects of dietary iron supplementation on growth performance, hematological status, and whole-body..., Rincker [/bib_ref]. Similarly, standalone administration of 5-ALA produced immunity improvement similar to the combinations, or even had better effects. Accordingly, it is fair to say that if the use 5-ALA is intended to improve animals' iron status and immunity, then for cost-effectiveness it should be used solo.
It is noteworthy that in studies that measured WBC counts and lymphocyte counts (10 studies), there was a considerable increase in only three studies. It is well-known that cellular immunity is activated during times of infection [bib_ref] Overview of the immune response, Chaplin [/bib_ref] ; hence, the reported lack of effect of 5-ALA on WBC count could be because involved animals were infection-free. Notably, indices of the immune response other than cellular immunity (soluble proteins and bioactive molecules) were estimated in a few studies, and similar to the cellular immune response, few overall improvements were reported. On the other hand, in two studies LPS was used to mimic the effect of pathogen invasion, and the pro-inflammation markers noticeably improved in response to LPS stimulation only in the 5-ALA groups compared with the control groups: in one study, the level of IGF-1 increased, whereas TNF-α and cortisol decreased 2 h after LPS injection [bib_ref] Effect of δ-aminolevulinic acid on growth performance, nutrient digestibility, blood parameters and..., Chen [/bib_ref] , and plasma ceruloplasmin, interferon-γ, iNOS, IL-6, and TNF-like ligand 1A decreased 3 h after LPS injection [bib_ref] Dietary supplementation with 5-aminolevulinic acid modulates growth performance and inflammatory responses in..., Sato [/bib_ref]. This finding is supported by results from an existing review that reported that 5-ALA regulates the inflammatory response in several conditions in humans [bib_ref] 5-Aminolevulinic acid regulates the inflammatory response and alloimmune reaction, Fujino [/bib_ref].
Most improvements in hemoglobin, Fe, and RBCs were noticed in young pigs directly supplemented with 5-ALA [bib_ref] Effect of δ-aminolevulinic acid on growth performance, nutrient digestibility, blood parameters and..., Chen [/bib_ref] [bib_ref] Use of δ-aminolevulinic acid in swine diet: Effect on growth performance, behavioral..., Mateo [/bib_ref] [bib_ref] Effects of dietary supplementation with delta-aminolevulinic acid on growth performance, hematological status,..., Wang [/bib_ref] [bib_ref] δ-Aminolevulinic acid, and lactulose supplements in weaned piglets diet: Effects on performance,..., Hossain [/bib_ref] [bib_ref] Evaluation of dietary supplementation of delta-aminolevulinic acid and chitooligosaccharide on growth performance,..., Yan [/bib_ref] , or those nursing sows supplemented with 5-ALA as a result of increased milk content of iron [bib_ref] Effects of delta-aminolevulinic acid and vitamin C supplementation on feed intake, backfat,..., Wang [/bib_ref] [bib_ref] Effects of iron injection at birth on neonatal iron status in young..., Wang [/bib_ref]. In addition, 5-ALA increased milk protein both in sows and dairy cows [bib_ref] Effects of delta-aminolevulinic acid and vitamin C supplementation on feed intake, backfat,..., Wang [/bib_ref] [bib_ref] Effects of 5-aminolevulinic acid supplementation on milk production, iron status, and immune..., Hendawy [/bib_ref] [bib_ref] Dietary supplementation of delta-aminolevulinic acid to lactating sows improves growth performance and..., Lee [/bib_ref] , which indicates that 5-ALA-supplemented animals can be a source of naturally iron-fortified milk. Furthermore, 5-ALA increased egg yolk iron concentration, Haugh unit, eggshell color, egg yolk index, and egg yolk color unit [bib_ref] Evaluation of δ-aminolevulinic acid on serum iron status, blood characteristics, egg performance..., Chen [/bib_ref] [bib_ref] Effects of δ-aminolevulinic acid and vitamin C supplementation on iron status, production..., Wang [/bib_ref] [bib_ref] Evaluation of dietary supplementation of delta-aminolevulinic acid and chito-oligosaccharide on production performance,..., Yan [/bib_ref]. Accordingly, 5-ALA may be most beneficial in young pigs, lactating animals, and laying hens. Additionally, 5-ALA expressed positive effects on the immune response, both under normal conditions [bib_ref] Effects of dietary supplementation with delta-aminolevulinic acid on growth performance, hematological status,..., Wang [/bib_ref] [bib_ref] Effects of 5-aminolevulinic acid supplementation on milk production, iron status, and immune..., Hendawy [/bib_ref] and under infection-like conditions involving challenge with LPS [bib_ref] Effect of δ-aminolevulinic acid on growth performance, nutrient digestibility, blood parameters and..., Chen [/bib_ref] [bib_ref] Dietary supplementation with 5-aminolevulinic acid modulates growth performance and inflammatory responses in..., Sato [/bib_ref]. Therefore, it might be recommended to supplement animals at risk for infection (e.g., around parturition) with 5-ALA to attain prophylactic effects.
In short, results from studies involving the use of high doses as well as long-term use of low doses of 5-ALA seem to be mostly positive. Nevertheless, from the available data we cannot clearly recommend a dose, combination, or duration of feeding. Peculiarly, the use of combinations yielded mixed results; for instance, use of 10 mg/kg of 5-ALA with 500 mg/kg of vitamin C for 28 days in mother pigs immediately postpartum significantly improved piglet BW and ADG [bib_ref] Effects of delta-aminolevulinic acid and vitamin C supplementation on feed intake, backfat,..., Wang [/bib_ref]. In contrast, the same combination had no effect on the growth performance of broilers [bib_ref] Effects of dietary delta-aminolevulinic acid and vitamin C on growth performance, immune..., Wang [/bib_ref]. We can conclude that some species might benefit better than others from 5-ALA combinations. This may be related to major differences in stomach structure and metabolism between different animal species.
Our findings should be considered within the boundaries of the heterogeneity, as well as the validity threats of the included studies. Although most studies used randomization, no studies mentioned that the person who conducted the randomization was blinded to the treatment, which implies unclear or high risk of selection bias, due to lack of allocation concealment. Also, not in a single study were experimenters who conducted animal treatment either in farms or laboratories blinded to treatment assignments, which can give rise to performance and detection biases. We also noted some instances of reporting errors, where values in tables were not consistent with the ones indicating the same outcomes in the text and abstract. In brief, within the context of the detected methodological flaws and the obvious variations in procedures, the spotted significant effects cannot be fully attributed to use of 5-ALA, because variations in animals' characteristics and the environment can affect animals' response to treatment. Moreover, the noted non-significance in most outcome indicators is somewhat alarming, because it puts the cost-effectiveness of 5-ALA in question. Future cost-effectiveness studies that use standardized procedures and unified outcome measures are needed to clarify the effects of 5-ALA in farm animals.
This review has some strengths, given that PubMed and Web of Science databases were extensively searched through a comprehensive list of key words. It also has some limitations. Three non-English studies were excluded, and the available studies exclusively were comprised from samples of pigs, chickens, and dairy cows, which limit the generalization of the findings. In addition, some outcomes were addressed in a limited number of studies-e.g., most indices of the cell-mediated immune response and cortisol hormone were measured only in single studies [bib_ref] Effect of δ-aminolevulinic acid on growth performance, nutrient digestibility, blood parameters and..., Chen [/bib_ref] [bib_ref] Dietary supplementation with 5-aminolevulinic acid modulates growth performance and inflammatory responses in..., Sato [/bib_ref] [bib_ref] Effects of dietary supplementation with delta-aminolevulinic acid on growth performance, hematological status,..., Wang [/bib_ref]. Noticeably, the majority of studies reported in this review were produced by a single group of researchers. In addition, the enormous heterogeneity between studies with regard to doses and outcomes measured, as well as the poorly reported findings, limited our potential for meta-analysis.
# Conclusions
Use of 5-ALA as a dietary supplement had a significant beneficial effect on iron status and immune response. However, growth performance witnessed almost no change in response to 5-ALA. Nevertheless, 5-ALA would produce the most beneficial effects in young pigs by correcting anemia, in lactating animals by increase milk protein and iron contents, and in laying hens by improving egg production and egg quality. These effects may be attained in a short time (within 2 or 3 weeks of the commencement of treatment). According to cited studies, the minimal effective doses of 5-ALA are 2 mg/kg feed in laying hens; 10 mg/kg feed in broilers, weanling pigs, and dairy cows; and 500-1000 mg/kg feed in sows.
Although only a single study noted oxidative stress with long-term use, this is alarming, and future studies should address this outcome. Generally speaking, reviewed studies were limited in number and quality; outcomes were numerous, which made it difficult to group and report tangible outcomes. Standard procedures and established evaluation measures are essential for future studies, in order to fully reveal the true effects of 5-ALA supplementation.
Supplementary Materials: The following are available online at http://www.mdpi.com/2076-2615/10/8/1352/s1, [fig_ref] Table 1: Effects of 5-aminolevulinic acid [/fig_ref] : Summary of the general data of papers evaluating the effect of 5-ALA supplementation on farm animals. : Evaluation of the quality of the papers according to the selection criteria. : Effects of 5-aminolevulininc acid compared with other alternative treatments, and 5-aminolevulininc acid combination with other treatments.
## Author
[fig] Figure 1: Summary of the search strategy based on the preferred reporting items for systematic reviews and meta-analyses (PRISMA) statement guidelines. [/fig]
[fig] Figure 2: Quality assessment of each included studies. [/fig]
[fig] Figure 3: Overall quality assessment summary. [/fig]
[fig] Figure 4: The possible mechanism of action of 5-aminolevulinic acid. Abbreviations: ROS, reactive oxygen species; RNS, reactive nitrogen species; NRF2, nuclear factor erythroid 2; HO-1, heme oxygenase-1; NF-kB, nuclear factor kappa light chain enhancer of activated B cells. [/fig]
[table] Table 1: Effects of 5-aminolevulinic acid (5-ALA) supplementation on different farm animals. [/table]
|
Model‐based meta‐analysis of changes in circulatory system physiology in patients with chronic heart failure
This is an open access article under the terms of the Creat ive Commo ns Attri butio n-NonCo mmerc ial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.AbstractTo characterize and compare various medicines for chronic heart failure (CHF), changes in circulatory physiological parameter during pharmacotherapy were investigated by a model-based meta-analysis (MBMA) of circulatory physiology. The clinical data from 61 studies mostly in patients with heart failure with reduced ejection fraction (HFrEF), reporting changes in heart rate, blood pressure, or ventricular volumes after treatment with carvedilol, metoprolol, bisoprolol, bucindolol, enalapril, aliskiren, or felodipine, were analyzed. Seven cardiac and vasculature function indices were estimated without invasive measurements using models based on appropriate assumptions, and their correlations with the mortality were assessed. Estimated myocardial oxygen consumption, a cardiac load index, correlated excellently with the mortality at 3, 6, and 12 months after treatment initiation, and it explained differences in mortality across the different medications. The analysis based on the present models were reasonably consistent with the hypothesis that the treatment of HFrEF with various medications is due to effectively reducing the cardiac load. Assessment of circulatory physiological parameters by using MBMA would be insightful for quantitative understanding of CHF treatment.Study HighlightsWHAT IS THE CURRENT KNOWLEDGE ON THE TOPIC?In clinical studies of chronic heart failure (CHF), important physiological indices, such as cardiac load, could not be assessed without invasive measurements.WHAT QUESTION DID THIS STUDY ADDRESS?Cardiac and vasculature indices, including cardiac load, were estimated from clinical studies with various medications in CHF by using a new type of model-based metaanalysis (MBMA).WHAT DOES THIS STUDY ADD TO OUR KNOWLEDGE?The reductions on a cardiac load index, myocardial oxygen consumption estimated by MBMA, reasonably correlated with the decrease in odds ratio of mortality of patients with heart failure with reduced ejection fraction.|TAKAOKA eT Al.
# Introduction
In chronic heart failure (CHF), the blood supply to the body is insufficient, thus feed-back mechanisms, such as the sympathetic nervous and renin-angiotensin systems, are often overactive. [bib_ref] ACC/AHA/HFSA Focused Update of the 2013 ACCF/AHA Guideline for the Management of..., Yancy [/bib_ref] This further overloads the heart and causes it to become exhausted. There may be an increased risk of an events occurring if the heart is continuously under heavy loads compared to its capacity; reduction of the cardiac loads is considered necessary for recovery from CHF. CHF is classified principally in heart failure with reduced ejection fraction (HFrEF) and heart failure with preserved ejection fraction (HFpEF). In the former, EF is less than 40%, but in the latter, it is 40 to 50% or more. In reality, however, CHF includes even more diverse groups of patients, each of whom requires an appropriate treatment.
Current treatment guidelines (2017 American College of Cardiology Foundation/American Heart Association [ACCF/AHA] Guideline 1 and European Society of Cardiology [ESC] Guidelines 2 ) primarily recommend the use of β-blocker together [bib_ref] Comparison of carvedilol and metoprolol on clinical outcomes in patients with chronic..., Poole-Wilson [/bib_ref] [bib_ref] Effect of metoprolol CR/XL in chronic heart failure: metoprolol CR/XL randomised intervention..., Merit-Hf Study [/bib_ref] with angiotensin-converting enzyme (ACE) inhibitor [bib_ref] Effect of enalapril on survival in patients with reduced left ventricular ejection..., Investigators [/bib_ref] or angiotensin II receptor blocker (ARB) or angiotensin receptor neprilysin inhibitor (ARNI) for patients with HFrEF. In addition, diuretics are used widely for those who have peripheral edema (e.g., swelling of legs). So-called guideline-directed management and therapy (GDMT) β-blockers include carvedilol, bisoprolol, and sustained-release metoprolol. Bisoprolol and metoprolol selectively block β 1 receptor, and carvedilol blocks α 1 , β 1 , and β 2 receptors. On the other hand, bucindolol, which blocks β receptors nonselectively, and short-acting metoprolol tartrate are not recommended because of insufficient efficacy. Reduction in mortality was observed in patients with sinus rhythm by treatments with β-blocker, but not with atrial fibrillation in an umbrella review and metaanalytic assessment. [bib_ref] Beta-blocker efficacy across different cardiovascular indications: an umbrella review and metaanalytic assessment, Ziff [/bib_ref] Calcium channel blocker (CCB) and direct renin inhibitor (DRI) were not effective for CHF treatment but the reason is not fully understood. In patients with HFpEF, no pharmacotherapy has improved the mortality whereas diuretics show some beneficial effects.
To compare these therapeutic agents and to understand the pathology of CHF, evaluations of clinical changes in the circulatory system's physiological parameters, including cardiac function indices (cardiac output , EF, endsystolic elastance [E es ], or maximum elastance), cardiac load indices (myocardial oxygen ] consumption or myocardial volume oxygen, and effective arterial elastance [E a ]), and vasculature indices (total peripheral vascular resistance and total arterial compliance [TAC]) would be important. [bib_ref] Selective heart rate reduction with ivabradine unloads the left ventricle in heart..., Reil [/bib_ref] CO and EF have been routinely used for the diagnosis of CHF. MVO 2 is proportional to the energy consumption by the heart and should be a good marker of cardiac load theoretically. However, some of these indices, such as E es , E a , and MVO 2 have not been evaluated in most clinical trials since the measurements are invasive. Instead, heart rate (HR) and B-type natriuretic peptide are evaluated often in clinical trials because they correlate with the prognosis for some medications, [bib_ref] Meta-analysis: beta-blocker dose, heart rate reduction, and death in patients with heart..., Mcalister [/bib_ref] [bib_ref] Clinical outcome endpoints in heart failure trials: a European Society of Cardiology..., Zannad [/bib_ref] although they cannot explain therapeutic effects for all the medications. On the other hand, in the field of circulatory physiology, various parameters described above have been related mathematically to HR, blood pressure, and ventricular volume. [bib_ref] Physiological insights of recent clinical diagnostic and therapeutic technologies for cardiovascular diseases, Shigemi [/bib_ref] Therefore, it may be possible to estimate them based on carefully constructed models.
Current efficacious medicines, β-blockers, and ACE inhibitors exert an initial effect on the circulatory system, which causes direct changes in HR and blood pressure, and, in addition, they may have secondary therapeutic effects to the heart. In the sympathetic nervous system, β-blockers improve β-receptor density and suppress G protein uncoupling, which are decreased by long-term catecholamine stimulation. [bib_ref] Improvement of postreceptor events by metoprolol treatment in patients with chronic heart..., Böhm [/bib_ref] [bib_ref] Reduced beta-adrenergic receptor activation decreases G-protein expression and beta-adrenergic receptor kinase activity..., Ping [/bib_ref] In addition, β-blockers improve heart contraction and reduced circulating levels of vasoconstrictor substances, such as norepinephrine, renin, and endothelin. It has been considered that these changes may be associated with delays in the progression of myocardial injury. 14,15 ACE inhibitors act on the reninangiotensin-aldosterone (RAA) and kallikrein-kinin systems and reduce blood pressure. In the myocardium, ACE inhibitors may also delay cardiac remodeling by reducing oxidative stress. [bib_ref] Inhibitory effects of antioxidants on neonatal rat cardiac myocyte hypertrophy induced by..., Nakamura [/bib_ref] However, the clinical significance of these secondary effects of β-blockers and ACE inhibitors are unknown.
In this study, to characterize and compare various medicines for CHF, we collected noninvasive observations of HR, blood pressure, and left ventricular volume during CHF treatment from clinical trials of β-blockers, ACE inhibitors, DRI, and CCB. The goal of this study is to apply mathematical models of circulatory system physiology to these data in order to understand the progression of treatment of CHF quantitatively, and in particular to find a good indicator of mortality odds ratio. To our knowledge, this is the first report HOW MIGHT THIS CHANGE DRUG DISCOVERY, DEVELOPMENT, AND/ OR THERAPEUTICS? Assessment of circulatory physiological parameters by using MBMA would be beneficial for quantitative understanding of CHF treatment.
## | 1083
## Mbma of circulatory system physiology in chf
of mechanism-oriented model-based meta-analysis (MBMA) of circulatory system physiology.
# Methods
## Selection of drugs
β-blockers (carvedilol, metoprolol, bisoprolol, and bucindolol), ACE inhibitor (enalapril), DRI (aliskiren), or CCB (felodipine) were selected because odds ratio of mortality compared to the control group (β-blockers) or placebo group (other drugs) were obtained.
## Data collection of odds ratios of mortality
The inclusion and exclusion criteria of the trials used in this study were listed in ,b, respectively. The number of trials and subjects for each drug are described in and detailed background characteristics and the reference of each trial are described in . For β-blockers, the results of network meta-analysis reported by Chatterjee et al. were used. [bib_ref] Benefits of β blockers in patients with heart failure and reduced ejection..., Chatterjee [/bib_ref] For enalapril, aliskiren, and felodipine, odds ratio of mortality was extracted from the literature.
## Data collection of clinical parameters
The inclusion and exclusion criteria of the trials, the number of trials and subjects for each drug, and detailed background characteristics and the reference of each study were similarly listed in ,d, Tables S1 and S3, respectively. The means and SDs or standard errors of systolic arterial blood pressure (SAP), mean blood pressure (MAP), HR, end-systolic volume (ESV) and end-diastolic volume (EDV) at baseline and 3, 6, and 12 months after treatment initiation were extracted from the literature. It should be noted that, considering the inclusion criteria for both mortality and clinical parameter collection, the target population of this study was mostly patients with HFrEF, but a relatively small number of borderline patients were also included. In addition, as stated in the footnote of , all the searches were conducted in 2014, so trials of recently developed drugs, such as ARNI, were not included.
## Correction of between-trial heterogeneity and meta-analysis
To correct the between-trial heterogeneity, multiple regression analyses were performed on baseline values and 3-, 6-, and 12-month changes (only for β-blockers) in each Controlled randomized clinical trial in patients of heart failure with reduced ejection fraction, and reported mortality.
2
The average baseline EF was 45% or less.
3
The average NYHA class was 1.8-3.2.
(b) Exclusion criteria of studies for collection of mortality b
## Id criteria
1 Non-randomized clinical trial.
2 Number of patients was <100.
## 3
A follow-up period of less than three months. [bib_ref] Effect of metoprolol CR/XL in chronic heart failure: metoprolol CR/XL randomised intervention..., Merit-Hf Study [/bib_ref] The average age of the drug was more than 70. 2
The average baseline EF was 45% or less.
3
The average NYHA class was 1.8-3.2.
## 4
The average baseline HR was 70-90 bpm.
## 5
The average baseline systolic blood pressure was 100-140 mm Hg. of the observed parameters (SAP, MAP, HR, ESV, and EDV). Potential covariates included age, New York Heart Association (NYHA) class, proportions of men and patients with ischemic heart failure, treatment duration, baselines of each parameter, and β-blocker used. The stepwise forward selection based on Akaike Information Criterion was performed to select the statistically significant covariates, and multicollinearity between variable factors was evaluated using variance inflation factor (VIF). Based on the results of multiple regression analyses, the baselines and changes for each parameter were corrected to be the values when all identified covariates were the mean of the studies for odds ratios of mortality in each drug class.
## (d) exclusion criteria of studies for collection of clinical parameters
## Id criteria
The mean values of the corrected parameters and their standard errors at 3, 6, and 12 months after treatment initiation were calculated with a random-effects model (DerSimonian Laird method, R 3.5.2). 18
## Calculation of parameters of circulatory system physiology
If the MAP was not described but SAP and diastolic arterial pressure (DAP) were described, MAP (mm Hg) was conventionally estimated by Equation 1.
If the ESV or EDV was not described but end-systolic diameter (ESD) or end-diastolic diameter (EDD; cm) was described, the ESV or EDV (ml) was estimated by Equations 2 and 3 (Teichholz method).
The left ventricular volume was expressed in ml/m 2 assuming that body surface area is 1.7 m 2 for a body weight of 60 kg and height of 170 cm.
Stroke volume (SV; ml) to EDV, CO (L/min), and EF (%) were calculated from the following Equations 4 to 6.
Based on the cardiac mathematical models, cardiac function indices, cardiac load indices, and vasculature indices were calculated from the mean values of observed parameters . E es is the maximum elastance at the end of systole, and is the slope of the end systolic pressure-volume relationship (ESPVR; . E es is considered to be an index of the cardiac contraction force. Sunagawa et al. and reported that E es can be calculated simply as a ratio of end-systolic pressure E a is the total amount of resistive and pulsatile afterload derived from the ventricular pressure-volume relationship, [bib_ref] Left ventricular interaction with arterial load studied in isolated canine ventricle, Sunagawa [/bib_ref] and is calculated by the ratio of P es to SV. 21 E a (mm Hg/ml) was estimated from Equation 8.
[formula] (1) MAP = DAP + SAP − DAP 3 (2) ESV = 7.0 × ESD 3 2.4 + ESD (3) EDV = 7.0 × EDD 3 2.4 + EDD (4) SV = EDV − ESV (5) CO = SV × HR (6) EF = SV EDV F I G U R E 1 [/formula]
The pressure-volume area (PVA) links hemodynamic factors to MVO 2 and corresponds to the total automatic energy in one beat [bib_ref] Hemodynamic and mechanical determinants of myocardial O 2 consumption in normal human..., Vanoverschelde [/bib_ref] [bib_ref] Total mechanical energy of a ventricle model and cardiac oxygen consumption, Suga [/bib_ref]. This is based on the physics concept that the product of pressure and volume change is equivalent to energy. In the ventricular pressure-volume relationship, PVA is the sum of the area surrounded by the trajectory of the left ventricle pressure and the left ventricle volume while the heart beats once (extra work) and the area surrounded by ESPVR and the pressure-volume trajectory (potential energy). PVA correlates with myocardial oxygen consumption. [bib_ref] Selective heart rate reduction with ivabradine unloads the left ventricle in heart..., Reil [/bib_ref] MVO 2 (J/min/m 2 ) was estimated using Equation [bib_ref] Clinical outcome endpoints in heart failure trials: a European Society of Cardiology..., Zannad [/bib_ref] , which was obtained from the correlation between PVA and MVO 2 in table 2 of the report by Takaoka et al. [bib_ref] Comparison of hemodynamic determinants for myocardial oxygen consumption under different contractile states..., Takaoka [/bib_ref] A detailed explanation of the derivation of this equation was described in the Supplement. TPR (dynes·s/cm 5 ) was estimated by Equation 11 approximating venous pressure to 0 where 1 mm Hg/L/min =79.98 dynes·sec/cm where CVP represents the central venous pressure, which was considered negligible compared to MAP. TAC (mL/mm Hg) is an index of blood vessel compliance and was estimated by the ratio of the difference between SAP and DAP to SV. [bib_ref] Total arterial compliance estimated by stroke volume-to-aortic pulse pressure ratio in humans, Chemla [/bib_ref] Standard errors were calculated based on the propagation rule of error. All data analyses were conducted using R version 3.5.2.
# Results
## Data characteristics
Based on the selection criteria, 61 eligible studies were identified in which the circulatory parameters (SAP, MAP, HR, ESV, or EDV) were evaluated. Detailed background characteristics of each study are described in Tables S2 and S3. Significant covariates on the baseline and the treatment effects of observed circulatory parameters for β-blockers are shown in Tables S4 and S5. The average EF values were considerably less than 40% for all the trials. shows the integrated mean changes in observed circulatory parameters after treatment. Both SAP and MAP were decreased by the ACE inhibitor, DRI, and CCB. Regarding β-blockers, SAP and MAP were decreased by carvedilol and bisoprolol, and transiently decreased by metoprolol, but not by bucindolol. HR was clearly decreased by β-blockers (carvedilol, metoprolol, bisoprolol, and bucindolol) whereas no change was observed with the other drugs. Decreases in HR were consistent for all β-blockers and for the entire treatment period of 12 months. ESV and EDV were decreased most efficiently by β-blockers,
## Treatment effects on blood pressure, heart rate, and left ventricle volume
[formula] (7) E es = P es ESV ≈ MAP ESV (8) E a = P es SV ≈ MAP SV = TPR × HR (9) PVA = 1.33333 ⋅ 10 − 4 ⋅ MAP ⋅ ESV 2 + (SAP + MAP) ⋅ (EDV − ESV) 2(10)MVO 2 = 2.42 ⋅ PVA + 0.4 ⋅ E es − 2.4 + 0.017 ⋅ HR (11) TPR = (MAP − CVP) CO × 79.98 (12) TAC = SV (SAP − DAP) (13) TAC = (EDV − ESV) 3 2 × (SAP − MAP) [/formula]
followed by the ACE inhibitor. ESV and EDV were not changed by DRI and CCB.
## Treatment effects on estimated cardiac function indices
Estimated CO was decreased by carvedilol, metoprolol and bisoprolol within 6 months, in line with the general class effect of β-blockers. [bib_ref] Beta-blockade in heart failure, Metra [/bib_ref] On the other hand, estimated CO was not changed by bucindolol, ACE inhibitor, DRI, and CCB at 3 months. Estimated EF was continuously improved by β-blockers and reached a 30% increase at 12 months. ACE inhibitor and CCB improved estimated EF and reached a 10% and 15% increase at 12 months, respectively. DRI exerted almost no change compared to the control group. Estimated E es was improved by the β-blockers and the ACE inhibitor, but no difference was observed in DRI and CCB.
## Treatment effects on estimated cardiac load indices
Regarding β-blockers, estimated MVO 2 was decreased by more than 20% by carvedilol, metoprolol, and bisoprolol, and 13% by bucindolol at 3 months. Decreases in estimated MVO 2 by β-blockers were stable for 12 months. Estimated MVO 2 was decreased 17% by ACE inhibitors at 3 months. With DRI and CCB, almost no change was observed compared with the control group. Estimated E a showed a tendency to decrease with β-blockers and CCB, but was not changed by ACE inhibitors and DRI.
## F i g u r e 2
Estimated time-course changes of circulatory physiological parameter after treatment initiation. Data represent mean change ratio and relative standard error of the change. By multiple regression analysis, the variable factors for the baseline and the change ratio (in β-blocker group) were identified and corrected. The corrected results of 61 trials were integrated by meta-analysis based on the random effects model (DerSimonian Laird method). The mean change ratio of function indices (CO, EF, E es ), cardiac load indices (MVO 2 , E a ), and vasculature indices (TPR, TAC) were estimated by hemodynamic mathematical model. The standard error was calculated based on the error propagation. CO, cardiac output; E a , effective arterial elastance; EDV, end-diastolic volume; E es , end-systolic elastance; EF, ejection fraction; ESPVR, end systolic pressurevolume relationship; ESV, end-systolic volume; EW, extra work (energy required to pump blood); HR, heart rate; MAP, mean blood pressure; MVO 2 , myocardial oxygen consumption; PE, potential energy (energy required for the basic metabolism and contraction of the heart); P es , endsystolic pressure; PVA, pressure-volume area (consumption energy consumed per heartbeat); SAP, systolic arterial blood pressure; TAC, total arterial compliance; TPR, total peripheral vascular resistance
## Treatment effects on estimated vasculature indices
Estimated TPR was markedly decreased by CCB, which is consistent with the reported class effect. [bib_ref] Pathophysiological background for the use of calcium antagonists, Aoki [/bib_ref] On the other hand, β-blockers were prone to increase estimated TPR. No difference was observed with ACE inhibitor and DRI. Estimated TAC was increased by carvedilol, with a change ratio of 28% at 12 months. No clear change was observed with other drugs compared to the control group. . Correlations for all the parameters at 6 and 12 months are shown in .
## Relationships between odds ratio of mortality and cardiac indices
# Discussion
## Assumptions of this study
In this study, parameters of the circulatory system physiology were estimated based on information generally available from clinical trials, avoiding necessity of invasive measurements.
## F i g u r e 3
Correlations between change of circulatory physiological parameter and odds ratio of mortality after 3 months from treatment initiation. The horizontal axis represents mean change percent and relative standard error from the baseline, and the vertical axis represents mean and standard error of odds ratio. The correlation between mean change ratios of clinical parameters (SAP, MAP, HR, ESV, and EDV) and estimated indices (CO, EF, E es , MVO 2 , E a , TPR, and TAC) integrated by meta-analysis from 61 trials, and the mean odds ratios of mortality obtained from 21 large-scale clinical trials was represented. CO, cardiac output; E a , effective arterial elastance; EDV, end-diastolic volume; E es , end-systolic elastance; EF, ejection fraction; ESPVR, end systolic pressure-volume relationship; ESV, end-systolic volume; EW, extra work (energy required to pump blood); HR, heart rate; MAP, mean blood pressure; MVO 2 , myocardial oxygen consumption; PE, potential energy (energy required for the basic metabolism and contraction of the heart); P es , end-systolic pressure; PVA, pressure-volume area (consumption energy consumed per heartbeat); SAP, systolic arterial blood pressure; TAC, total arterial compliance; TPR, total peripheral vascular resistance For this purpose, several assumptions were made including P es ≈ MAP, V 0 ≈ 0, and the left ventricular maximum pressure ≈ SAP. In addition, we adopted the observed correlation for the relationship between MVO 2 and PVA in one particular study. All of these are based on some theories or circumstantial evidences 11,19-25 but have not been fully validated yet. Errors may occur between the true and approximate values. However, because the above assumptions were made for all drugs in the same way, the relative changes (e.g., EF improves with beta-blockers, ACE, and CCBs, in that order) are considered reliable. Another important assumption for calculating MVO 2 is that all the energy is being used for the beating of the heart. It assumes that the energy used is transferred from the heart to the bloodstream and surrounding tissues. However, if the heart is using energy for its own recovery, metabolism, or secretion, there will be errors in the estimation. In order to obtain reliable results from the MBMA analysis, the backgrounds of the trials should be homologous, as required in the meta-analysis. To correct for heterogeneity between trials, multiple regression analysis was performed for clinical parameters using various covariates as described in the Method section. Moreover, because CHF is heterogenous, it may be necessary to account for differences in diverse types of systolic heart failure, including idiopathic, hypertensive, and valvular origin. However, these were not fully included in the analysis because of incomplete description in some manuscripts. These covariates, as well as other covariates not included in this analysis may have affected the accuracy of MBMA if there were notable differences between trials.
## Effects of each medication on prognosis
An interesting finding of this research is that the reductions on MVO 2 estimated by MBMA from many studies in patients with HFrEF correlated reasonably with the decrease in odds ratio of mortality. The β-blockers reduce HR and ventricular volume, and ACE inhibitors reduce ventricular volume and pressure. Considering estimation method of MVO 2 , it could be an excellent cardiac load marker, summing up pulsatile load, volume load, and pressure load. The hypothesis that reduction in cardiac load is important for the treatment of CHF is widely known, the present analysis is valuable in that it showed that the results of many studies with various medications were consistent with the hypothesis.
Changes in estimated MVO 2 were in line with differences in the effectiveness between β-blockers included in this study. These differences were considered to ascribe to the differences in the changes in HR and blood pressure between β-blockers. Consistent with previous findings, most βblockers markedly reduced HR, however, bucindolol only led to a weak reduction. Blood pressure was clearly decreased by carvedilol and bisoprolol, whereas the effect of the other β-blockers was transient or insignificant. Some differences between β-blockers can be explained by the inhibitory action of carvedilol via α-receptor and by the high selectivity to the β 1 receptor of bisoprolol. [bib_ref] Pharmacodynamic profile of the selective beta 1-adrenoceptor antagonist bisoprolol, Harting [/bib_ref] However, the reasons why bucindolol showed reduced effects on HR and blood pressure are unknown.
The transient decrease in estimated CO following carvedilol treatment leads to reduction of cardiac load, F I G U R E 4 Determination coefficients between circulatory physiological parameters and odds ratio of mortality after 3, 6, and 12 months from treatment initiation. Determination coefficients (R 2 ) between clinical parameters (SAP, MAP, HR, ESV, EDV) and estimated indices (CO, EF, E es , MVO 2 , E a , TPR, TAC), and odds ratio of mortality at 3, 6, and 12 months were represented. Blue, orange and red bars represent R 2 at 3, 6, and 12 months, respectively. CO, cardiac output; E a , effective arterial elastance; EDV, end-diastolic volume; E es , end-systolic elastance; EF, ejection fraction; ESPVR, end systolic pressure-volume relationship; ESV, end-systolic volume; EW, extra work (energy required to pump blood); HR, heart rate; MAP, mean blood pressure; MVO 2 , myocardial oxygen consumption; PE, potential energy (energy required for the basic metabolism and contraction of the heart); P es , end-systolic pressure; PVA, pressure-volume area (consumption energy consumed per heartbeat); SAP, systolic arterial blood pressure; TAC, total arterial compliance; TPR, total peripheral vascular resistance however, it may simultaneously worsen the quality of life of patients. The decrease in estimated CO was less evident in the other β-blockers. On the other hand, improvements in estimated E es consistently continued until 6 months with all β-blockers, suggesting that this is a more timeconsuming process of the recovery of cardiac function compared with immediate decreases in HR and MVO 2 . The improvement in estimated E es with metoprolol was better than that with carvedilol at 12 months. On the other hand, E a at 6 and 12 months is lower for carvedilol than for metoprolol, resulting in little difference between two drugs in E es /E a which represents the efficiency of left ventricular contraction that pumps blood through the arteries. [bib_ref] Physiological insights of recent clinical diagnostic and therapeutic technologies for cardiovascular diseases, Shigemi [/bib_ref] For carvedilol, changes in MAP, CO, EF, E a , and TPR were continuous through 12 months.
Blood pressure was effectively reduced by ACE inhibitors in this analysis, as shown in previous studies. [bib_ref] ESC/ESH Guidelines for the management of arterial hypertension, Williams [/bib_ref] Estimated MVO 2 was reduced by ACE inhibitors but to a weaker degree than β-blockers. Increases in estimated E es and EF with treatment were also less noticeable with ACE inhibitors. Although further research is necessary to confirm the relationship between cardiac sympathetic nerve activity and changes in estimated E es and EF, a difference may exist in the progression of reverse remodeling during treatments with β-blockers and ACE inhibitors.
Calcium channel blocker acts as a vasodilator, as indicated by the reduced estimated TPR and blood pressure. However, it did not decrease the cardiac load itself. The effects of CCB or DRI were evaluated in addition to the standard treatment, mostly with an ACE inhibitor, in the clinical trials analyzed in this study. Thus, analyzing the effects of CCB or DRI monotherapy on blood pressure or estimated MVO 2 is not possible in this study. In the ESC Guidelines, 2 CCB and DRI are not recommended as routine treatment for patients with heart failure with reduced ejection fraction as there is insufficient evidence on clinical outcome.
Following recommendations in the 2017 ACCF/AHA Guideline, 1 β-blockers and ACE inhibitors are the most frequently used medicines for the pharmacological treatment of patients with HFrEF at ACCF/AHA stage C (i.e., NYHA II and III). Among β-blockers, carvedilol, bisoprolol, and sustained-release metoprolol are considered to be effective in reducing the risk of death, whereas bucindolol is considered to lack stable effectiveness across different populations. In the present study, the odds ratios of mortality were consistent with the recommendations and evidence described in the current treatment guidelines. This is reasonable because the clinical trial for odds ratio used in this study was partly the same as the studies on which the recommended therapy guidelines are based on. [bib_ref] Comparison of carvedilol and metoprolol on clinical outcomes in patients with chronic..., Poole-Wilson [/bib_ref] [bib_ref] Effect of metoprolol CR/XL in chronic heart failure: metoprolol CR/XL randomised intervention..., Merit-Hf Study [/bib_ref] [bib_ref] Effect of enalapril on survival in patients with reduced left ventricular ejection..., Investigators [/bib_ref] [bib_ref] A trial of the beta-blocker bucindolol in patients with advanced chronic heart..., Beta-Blocker ; Eichhorn [/bib_ref] [bib_ref] Effect of aliskiren on postdischarge mortality and heart failure readmissions among patients..., Gheorghiade [/bib_ref] [bib_ref] Effect of the calcium antagonist felodipine as supplementary vasodilator therapy in patients..., Cohn [/bib_ref] Contrary to the present analysis, Kaye et al. reported that MVO 2 was not significantly changed by carvedilol treatment within 3 months in patients with CHF. 32 MVO 2 was measured directly by an invasive method in this study, unlike our analysis. Because ESV and EDV were not described in the report by Kaye et al. we calculated them from cardiac output and left ventricular ejection fraction (LVEF) described, and estimated MVO 2 by our method for comparison. MVO 2 was reduced by 25% which was consistent with the 27% reduction obtained in our MBMA analysis for carvedilol. There were some ambiguities in this analysis. The estimated ESV and EDV were somewhat larger than usual, and the blood pressure was not decreased in Kaye's study, which was slightly deviated from the general observations in carvedilol treatments. Even so, if this analysis is correct, there is a discrepancy between the oxygen consumption calculated from the work by the beating and the oxygen consumption actually measured from the oxygen concentration in the blood. Given the conservation of energy, the heart may be consuming energy for some untraceable activities. Although further studies are obviously warranted to clarify this issue, MBMA seems to have provided a new perspective for understanding cardiac load and CHF treatment.
In this study, changes in circulatory system physiology as well as changes in blood pressure, HR, and left ventricular volume during CHF treatments were reasonably illustrated by analyses based on the numerous observations in clinical studies. Results of the present analysis can be useful to understand treatment of HFrEF, because of overall consistency. The interest in MBMA has increasing, mainly in the drug development field because it enables an objective and efficient evaluation of new drug candidates. The present study indicates that MBMA is also useful to compare various treatments and to understand their mechanisms of action.
# Limitations
There are several limitations to this study. First, many parameters were estimated based on the theory of circulatory system physiology and therefore, validation with more direct (but invasive) methods would be needed. In particular, more comparisons with MVO 2 by assessing O 2 and CO 2 concentrations in the blood near the heart would be needed. In addition, assessments of E es and E a are important. Second, it is difficult to estimate class effects for ACE inhibitor, CCB, and DRI because the number of analyzed drugs for these groups was insufficient. Furthermore, HF being a complex clinical syndrome with large phenotypic heterogeneity, the results of this study would be potentially inappropriate for some types of patients. Last, it should be noted that all the analysis is retrospective. In the future, prospective evaluations of the circulatory system physiology are necessary to fully validate the present study. |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.