The emergence of networks in human genome epidemiology: challenges and opportunities

Identifying the most important facilitators of open research data sharing and reuse in Epidemiology: A mixed-methods study

To understand how open research data sharing and reuse can be further improved in the field of Epidemiology, this study explores the facilitating role that infrastructural and institutional arrangements play in this research discipline. It addresses two research questions: 1) What influence do infrastructural and institutional arrangements have on open research data sharing and reuse practices in the field of Epidemiology? And 2) how could infrastructural and institutional instruments used in Epidemiology potentially be useful to other research disciplines? First, based on a systematic literature review, a conceptual framework of infrastructural and institutional instruments for open research data facilitation is developed. Second, the conceptual framework is applied in interviews with Epidemiology researchers. The interviews show that two infrastructural and institutional instruments have a very high influence on open research data sharing and reuse practices in the field of Epidemiology, namely (a) access to a powerful search engine that meets open data search needs and (b) support by data stewards and data managers. Third, infrastructural and institutional instruments with a medium, high, or very high influence were discussed in a research workshop involving data stewards and research data officers from different research fields. This workshop suggests that none of the influential instruments identified in the interviews are specific to Epidemiology. Some of our findings thus seem to apply to multiple other disciplines. This study contributes to Science by identifying field-specific facilitators and challenges for open research data in Epidemiology, while at the same time revealing that none of the identified influential infrastructural and institutional instruments were specific to this field. Practically, this implies that open data infrastructure developers, policymakers, and research funding organizations may apply certain infrastructural and institutional arrangements to multiple research disciplines to facilitate and enhance open research data sharing and reuse.

Meta-analysis of genome-wide association studies for litter size in sheep

Litter size and ovulation rate are important reproduction traits in sheep and have important impacts on the profitability of farm animals. To investigate the genetic architecture of litter size, we report the first meta-analysis of genome-wide association studies (GWAS) using 522 ewes and 564,377 SNPs from six sheep breeds. We identified 29 significant associations for litter size which 27 of which have not been reported in individual GWAS for each population. However, we could confirm the role of BMPR1B in prolificacy. Our gene set analysis discovered biological pathways related to cell signaling, communication, and adhesion. Functional clustering and enrichment using protein databases identified epidermal growth factor-like domain affecting litter size. Through analyzing protein-protein interaction data, we could identify hub genes like CASK, PLCB4, RPTOR, GRIA2, and PLCB1 that were enriched in most of the significant pathways. These genes have a role in cell proliferation, cell adhesion, cell growth and survival, and autophagy. Notably, identified SNPs were scattered on several different chromosomes implying different genetic mechanisms underlying variation of prolificacy in each breed. Given the different layers that make up the follicles and the need for communication and transfer of hormones and nutrients through these layers to the oocyte, the significance of pathways related to cell signaling and communication seems logical. Our results provide genetic insights into the litter size variation in different sheep breeds.

Using imputation to provide harmonized longitudinal measures of cognition across AIBL and ADNI

To improve understanding of Alzheimer’s disease, large observational studies are needed to increase power for more nuanced analyses. Combining data across existing observational studies represents one solution. However, the disparity of such datasets makes this a non-trivial task. Here, a machine learning approach was applied to impute longitudinal neuropsychological test scores across two observational studies, namely the Australian Imaging, Biomarkers and Lifestyle Study (AIBL) and the Alzheimer's Disease Neuroimaging Initiative (ADNI) providing an overall harmonised dataset. MissForest, a machine learning algorithm, capitalises on the underlying structure and relationships of data to impute test scores not measured in one study aligning it to the other study. Results demonstrated that simulated missing values from one dataset could be accurately imputed, and that imputation of actual missing data in one dataset showed comparable discrimination (p < 0.001) for clinical classification to measured data in the other dataset. Further, the increased power of the overall harmonised dataset was demonstrated by observing a significant association between CVLT-II test scores (imputed for ADNI) with PET Amyloid-β in MCI APOE-ε4 homozygotes in the imputed data (N = 65) but not for the original AIBL dataset (N = 11). These results suggest that MissForest can provide a practical solution for data harmonization using imputation across studies to improve power for more nuanced analyses.

Calibrating longitudinal cognition in Alzheimer's disease across diverse test batteries and datasets

BACKGROUND

We sought to identify optimal approaches by calibrating longitudinal cognitive performance across studies with different neuropsychological batteries.

METHODS

We examined four approaches to calibrate cognitive performance in nine longitudinal studies of Alzheimer's disease (AD) (n = 10,875): (1) common test, (2) standardize and average available tests, (3) confirmatory factor analysis (CFA) with continuous indicators, and (4) CFA with categorical indicators. To compare precision, we determined the minimum sample sizes needed to detect 25% cognitive decline with 80% power. To compare criterion validity, we correlated cognitive change from each approach with 6-year changes in average cortical thickness and hippocampal volume using available MRI data from the AD Neuroimaging Initiative.

RESULTS

CFA with categorical indicators required the smallest sample size to detect 25% cognitive decline with 80% power (n = 232) compared to common test (n = 277), standardize-and-average (n = 291), and CFA with continuous indicators (n = 315) approaches. Associations with changes in biomarkers changes were the strongest for CFA with categorical indicators.

CONCLUSIONS

CFA with categorical indicators demonstrated greater power to detect change and superior criterion validity compared to other approaches. It has wide applicability to directly compare cognitive performance across studies, making it a good way to obtain operational phenotypes for genetic analyses of cognitive decline among people with AD.

Increasing value and reducing waste in research design, conduct, and analysis

Correctable weaknesses in the design, conduct, and analysis of biomedical and public health research studies can produce misleading results and waste valuable resources. Small effects can be difficult to distinguish from bias introduced by study design and analyses. An absence of detailed written protocols and poor documentation of research is common. Information obtained might not be useful or important, and statistical precision or power is often too low or used in a misleading way. Insufficient consideration might be given to both previous and continuing studies. Arbitrary choice of analyses and an overemphasis on random extremes might affect the reported findings. Several problems relate to the research workforce, including failure to involve experienced statisticians and methodologists, failure to train clinical researchers and laboratory scientists in research methods and design, and the involvement of stakeholders with conflicts of interest. Inadequate emphasis is placed on recording of research decisions and on reproducibility of research. Finally, reward systems incentivise quantity more than quality, and novelty more than reliability. We propose potential solutions for these problems, including improvements in protocols and documentation, consideration of evidence from studies in progress, standardisation of research efforts, optimisation and training of an experienced and non-conflicted scientific workforce, and reconsideration of scientific reward systems.

Data harmonization and federated analysis of population-based studies: the BioSHaRE project

Abstracts

Modern epidemiology and its discontents

The goal of this article is to present a snapshot of an ongoing debate within epidemiology, pitching opposing sides in the struggle to define the path it should follow in the years to come. The debate among epidemiologists in the mid-90s pitted those who defended the idea that epidemiology should necessarily deal with a wide context against those who believed that science and public health are better served by focusing on the individual level. Ian Hacking’s concept of styles of reasoning was used as a theoretical tool. The literature was reviewed using a core set of articles as an entry point, seeking articles that cited them, and then back-tracking the citations of the resulting set in the Scopus database. The main arguments are presented according to levels (ontological, epistemological, axiological and pragmatic), in order to show an even deeper disagreement, in the very conception of science and its relation to social issues and public policy.

The geometric increase in meta-analyses from China in the genomic era

Meta-analyses are increasingly popular. It is unknown whether this popularity is driven by specific countries and specific meta-analyses types. PubMed was used to identify meta-analyses since 1995 (last update 9/1/2012) and catalogue their types and country of origin. We focused more on meta-analyses from China (the current top producer of meta-analyses) versus the USA (top producer until recently). The annual number of meta-analyses from China increased 40-fold between 2003 and 2011 versus 2.4-fold for the USA. The growth of Chinese meta-analyses was driven by genetics (110-fold increase in 2011 versus 2003). The HuGE Navigator identified 612 meta-analyses of genetic association studies published in 2012 from China versus only 109 from the USA. We compared in-depth 50 genetic association meta-analyses from China versus 50 from USA in 2012. Meta-analyses from China almost always used only literature-based data (92%), and focused on one or two genes (94%) and variants (78%) identified with candidate gene approaches (88%), while many USA meta-analyses used genome-wide approaches and raw data. Both groups usually concluded favorably for the presence of genetic associations (80% versus 74%), but nominal significance (P<0.05) typically sufficed in the China group. Meta-analyses from China typically neglected genome-wide data, and often included candidate gene studies published in Chinese-language journals. Overall, there is an impressive rise of meta-analyses from China, particularly on genetic associations. Since most claimed candidate gene associations are likely false-positives, there is an urgent global need to incorporate genome-wide data and state-of-the art statistical inferences to avoid a flood of false-positive genetic meta-analyses.

The power of meta-analysis in genome-wide association studies

Meta-analysis of multiple genome-wide association (GWA) studies has become common practice over the past few years. The main advantage of this technique is the maximization of power to detect subtle genetic effects for common traits. Moreover, one can use meta-analysis to probe and identify heterogeneity in the effect sizes across the combined studies. In this review, we systematically appraise and evaluate the characteristics of GWA meta-analyses with 10,000 or more subjects published up to June 2012. We provide an overview of the current landscape of variants discovered by GWA meta-analyses, and we discuss and assess with extrapolations from empirical data the value of larger meta-analyses for the discovery of additional genetic associations and new biology in the future. Finally, we discuss some emerging logistical and practical issues related to the conduct of meta-analysis of GWA studies.

Meta-analysis methods for genome-wide association studies and beyond

Meta-analysis of genome-wide association studies (GWASs) has become a popular method for discovering genetic risk variants. Here, we overview both widely applied and newer statistical methods for GWAS meta-analysis, including issues of interpretation and assessment of sources of heterogeneity. We also discuss extensions of these meta-analysis methods to complex data. Where possible, we provide guidelines for researchers who are planning to use these methods. Furthermore, we address special issues that may arise for meta-analysis of sequencing data and rare variants. Finally, we discuss challenges and solutions surrounding the goals of making meta-analysis data publicly available and building powerful consortia.

Sequential sentinel SNP Regional Association Plots (SSS-RAP): an approach for testing independence of SNP association signals using meta-analysis data

Genome-Wide Association Studies (GWAS) frequently incorporate meta-analysis within their framework. However, conditional analysis of individual-level data, which is an established approach for fine mapping of causal sites, is often precluded where only group-level summary data are available for analysis. Here, we present a numerical and graphical approach, "sequential sentinel SNP regional association plot" (SSS-RAP), which estimates regression coefficients (beta) with their standard errors using the meta-analysis summary results directly. Under an additive model, typical for genes with small effect, the effect for a sentinel SNP can be transformed to the predicted effect for a possibly dependent SNP through a 2×2 2-SNP haplotypes table. The approach assumes Hardy-Weinberg equilibrium for test SNPs. SSS-RAP is available as a Web-tool (http://apps.biocompute.org.uk/sssrap/sssrap.cgi). To develop and illustrate SSS-RAP we analyzed lipid and ECG traits data from the British Women's Heart and Health Study (BWHHS), evaluated a meta-analysis for ECG trait and presented several simulations. We compared results with existing approaches such as model selection methods and conditional analysis. Generally findings were consistent. SSS-RAP represents a tool for testing independence of SNP association signals using meta-analysis data, and is also a convenient approach based on biological principles for fine mapping in group level summary data.

Scientific inbreeding and same-team replication: type D personality as an example

Replication is essential for validating correct results, sorting out false-positive early discoveries, and improving the accuracy and precision of estimated effects. However, some types of seemingly successful replication may foster a spurious notion of increased credibility, if they are performed by the same team and propagate or extend the same errors made by the original discoveries. Besides same-team replication, replication by other teams may also succumb to inbreeding, if it cannot fiercely maintain its independence. These patterns include obedient replication and obliged replication. I discuss these replication patterns in the context of associations and effects in the psychological sciences, drawing from the criticism of Coyne and de Voogd of the proposed association between type D personality and cardiovascular mortality and other empirical examples.

Knowledge integration in cancer: current landscape and future prospects

Knowledge integration includes knowledge management, synthesis, and translation processes. It aims to maximize the use of collected scientific information and accelerate translation of discoveries into individual and population health benefits. Accumulated evidence in cancer epidemiology constitutes a large share of the 2.7 million articles on cancer in PubMed. We examine the landscape of knowledge integration in cancer epidemiology. Past approaches have mostly used retrospective efforts of knowledge management and traditional systematic reviews and meta-analyses. Systematic searches identify 2,332 meta-analyses, about half of which are on genetics and epigenetics. Meta-analyses represent 1:89-1:1162 of published articles in various cancer subfields. Recently, there are more collaborative meta-analyses with individual-level data, including those with prospective collection of measurements [e.g., genotypes in genome-wide association studies (GWAS)]; this may help increase the reliability of inferences in the field. However, most meta-analyses are still done retrospectively with published information. There is also a flurry of candidate gene meta-analyses with spuriously prevalent "positive" results. Prospective design of large research agendas, registration of datasets, and public availability of data and analyses may improve our ability to identify knowledge gaps, maximize and accelerate translational progress or-at a minimum-recognize dead ends in a more timely fashion.

Melanocortin-1 receptor, skin cancer and phenotypic characteristics (M-SKIP) project: study design and methods for pooling results of genetic epidemiological studies

Background

For complex diseases like cancer, pooled-analysis of individual data represents a powerful tool to investigate the joint contribution of genetic, phenotypic and environmental factors to the development of a disease. Pooled-analysis of epidemiological studies has many advantages over meta-analysis, and preliminary results may be obtained faster and with lower costs than with prospective consortia.

Design and methods

Based on our experience with the study design of the Melanocortin-1 receptor (MC1R) gene, SKin cancer and Phenotypic characteristics (M-SKIP) project, we describe the most important steps in planning and conducting a pooled-analysis of genetic epidemiological studies. We then present the statistical analysis plan that we are going to apply, giving particular attention to methods of analysis recently proposed to account for between-study heterogeneity and to explore the joint contribution of genetic, phenotypic and environmental factors in the development of a disease. Within the M-SKIP project, data on 10,959 skin cancer cases and 14,785 controls from 31 international investigators were checked for quality and recoded for standardization. We first proposed to fit the aggregated data with random-effects logistic regression models. However, for the M-SKIP project, a two-stage analysis will be preferred to overcome the problem regarding the availability of different study covariates. The joint contribution of MC1R variants and phenotypic characteristics to skin cancer development will be studied via logic regression modeling.

Discussion

Methodological guidelines to correctly design and conduct pooled-analyses are needed to facilitate application of such methods, thus providing a better summary of the actual findings on specific fields.

A population approach to precision medicine

The term P4 medicine is used to denote an evolving field of medicine that uses systems biology approaches and information technologies to enhance wellness rather than just treat disease. Its four components include predictive, preventive, personalized, and participatory medicine. In the current paper, it is argued that in order to fulfill the promise of P4 medicine, a "fifth P" must be integrated-the population perspective-into each of the other four components. A population perspective integrates predictive medicine into the ecologic model of health; applies principles of population screening to preventive medicine; uses evidence-based practice to personalize medicine; and grounds participatory medicine on the three core functions of public health: assessment, policy development, and assurance. Population sciences-including epidemiology; behavioral, social, and communication sciences; and health economics, implementation science, and outcomes research-are needed to show the value of P4 medicine. Balanced strategies that implement both population- and individual-level interventions can best maximize health benefits, minimize harm, and avoid unnecessary healthcare costs.

Meta-analysis of genetic association studies under heterogeneity

In multi-cohort genetic association studies or meta-analysis, associations of genetic variants with complex traits across cohorts may be heterogeneous because of genuine genetic diversity or differential biases or errors. To detect the associations of genes with heterogeneous associations across cohorts, new global fixed-effect (FE) and random-effects (RE) meta-analytic methods have been recently proposed. These global methods had improved power over both traditional FE and RE methods under heterogeneity in limited simulation scenarios and data application, but their usefulness in a wide range of practical situations is not clear. We assessed the performance of these methods for both binary and quantitative traits in extensive simulations and applied them to a multi-cohort association study. We found that these new approaches have higher power to detect mostly the very small to small associations of common genetic variants when associations are highly heterogeneous across cohorts. They worked well when both the underlying and assumed genetic models are either multiplicative or dominant. But, they offered no clear advantage for less common variants unless heterogeneity was substantial. In conclusion, these new meta-analytic methods can be used to detect the association of genetic variants with high heterogeneity, which can then be subjected to further exploration, in multi-cohort association studies and meta-analyses.

What should the genome-wide significance threshold be? Empirical replication of borderline genetic associations

BACKGROUND

Robust replication is a sine qua non for the rigorous documentation of proposed associations in the genome-wide association (GWA) setting. Currently, associations of common variants reaching P ≤ 5 × 10(-8) are considered replicated. However, there is some ambiguity about the most suitable threshold for claiming genome-wide significance.

METHODS

We defined as 'borderline' associations those with P > 5 × 10(-8) and P ≤ 1 × 10(-7). The eligible associations were retrieved using the 'Catalog of Published Genome-Wide Association Studies'. For each association we assessed whether it reached P ≤ 5 × 10(-8) with inclusion of additional data from subsequent GWA studies.

RESULTS

Thirty-four eligible genotype-phenotype associations were evaluated with data and clarifications contributed from diverse investigators. Replication data from subsequent GWA studies could be obtained for 26 of them. Of those, 19 associations (73%) reached P ≤ 5 × 10(-8) for the same or a related trait implicating either the exact same allele or one in very high linkage disequilibrium and 17 reached P < 10(-8). If the seven associations that did not reach P ≤ 5 × 10(-8) when additional data were considered are assumed to have been false-positives, the false-discovery rate for borderline associations is estimated to be 27% [95% confidence interval (CI) 12-48%]. For five associations, the current P-value is > 10(-6) [corresponding false-discovery rate 19% (95% CI 7-39%)].

CONCLUSION

A substantial proportion, but not all, of the associations with borderline genome-wide significance represent replicable, possibly genuine associations. Our empirical evaluation suggests a possible relaxation in the current GWS threshold.

Translational epidemiology in psychiatry: linking population to clinical and basic sciences

Translational research generally refers to the application of knowledge generated by advances in basic sciences research translated into new approaches for diagnosis, prevention, and treatment of disease. This direction is called bench-to-bedside. Psychiatry has similarly emphasized the basic sciences as the starting point of translational research. This article introduces the term translational epidemiology for psychiatry research as a bidirectional concept in which the knowledge generated from the bedside or the population can also be translated to the benches of laboratory science. Epidemiologic studies are primarily observational but can generate representative samples, novel designs, and hypotheses that can be translated into more tractable experimental approaches in the clinical and basic sciences. This bedside-to-bench concept has not been explicated in psychiatry, although there are an increasing number of examples in the research literature. This article describes selected epidemiologic designs, providing examples and opportunities for translational research from community surveys and prospective, birth cohort, and family-based designs. Rapid developments in informatics, emphases on large sample collection for genetic and biomarker studies, and interest in personalized medicine--which requires information on relative and absolute risk factors--make this topic timely. The approach described has implications for providing fresh metaphors to communicate complex issues in interdisciplinary collaborations and for training in epidemiology and other sciences in psychiatry.

The meta-analysis of genome-wide association studies

The pressure to publish novel genetic associations has meant that meta-analysis has been applied to genome-wide association studies without the time for a careful consideration of the methods that are used. This review distinguishes between the use of meta-analysis to validate previously reported genetic associations and its use for gene discovery, and advocates viewing gene discovery as an exploratory screen that requires independent replication instead of treating it as the application of hundreds of thousands of statistical tests. The review considers the use of fixed and random effects meta-analyses, the investigation of between-study heterogeneity, adjustment for confounding, assessing the combined evidence and genomic control, and comments on alternative approaches that have been used in the literature.

Phenotype harmonization and cross-study collaboration in GWAS consortia: the GENEVA experience

Genome-wide association study (GWAS) consortia and collaborations formed to detect genetic loci for common phenotypes or investigate gene-environment (G*E) interactions are increasingly common. While these consortia effectively increase sample size, phenotype heterogeneity across studies represents a major obstacle that limits successful identification of these associations. Investigators are faced with the challenge of how to harmonize previously collected phenotype data obtained using different data collection instruments which cover topics in varying degrees of detail and over diverse time frames. This process has not been described in detail. We describe here some of the strategies and pitfalls associated with combining phenotype data from varying studies. Using the Gene Environment Association Studies (GENEVA) multi-site GWAS consortium as an example, this paper provides an illustration to guide GWAS consortia through the process of phenotype harmonization and describes key issues that arise when sharing data across disparate studies. GENEVA is unusual in the diversity of disease endpoints and so the issues it faces as its participating studies share data will be informative for many collaborations. Phenotype harmonization requires identifying common phenotypes, determining the feasibility of cross-study analysis for each, preparing common definitions, and applying appropriate algorithms. Other issues to be considered include genotyping timeframes, coordination of parallel efforts by other collaborative groups, analytic approaches, and imputation of genotype data. GENEVA's harmonization efforts and policy of promoting data sharing and collaboration, not only within GENEVA but also with outside collaborations, can provide important guidance to ongoing and new consortia.

The emergence of translational epidemiology: from scientific discovery to population health impact

Recent emphasis on translational research (TR) is highlighting the role of epidemiology in translating scientific discoveries into population health impact. The authors present applications of epidemiology in TR through 4 phases designated T1-T4, illustrated by examples from human genomics. In T1, epidemiology explores the role of a basic scientific discovery (e.g., a disease risk factor or biomarker) in developing a "candidate application" for use in practice (e.g., a test used to guide interventions). In T2, epidemiology can help to evaluate the efficacy of a candidate application by using observational studies and randomized controlled trials. In T3, epidemiology can help to assess facilitators and barriers for uptake and implementation of candidate applications in practice. In T4, epidemiology can help to assess the impact of using candidate applications on population health outcomes. Epidemiology also has a leading role in knowledge synthesis, especially using quantitative methods (e.g., meta-analysis). To explore the emergence of TR in epidemiology, the authors compared articles published in selected issues of the Journal in 1999 and 2009. The proportion of articles identified as translational doubled from 16% (11/69) in 1999 to 33% (22/66) in 2009 (P = 0.02). Epidemiology is increasingly recognized as an important component of TR. By quantifying and integrating knowledge across disciplines, epidemiology provides crucial methods and tools for TR.

A1166C genetic variation of the angiotensin II type I receptor gene and susceptibility to coronary heart disease: collaborative of 53 studies with 20,435 cases and 23,674 controls

OBJECTIVE

Angiotensin II induces vasoconstriction and vascular smooth muscle growth via stimulation of the angiotensin II type I receptor (AGTR1). Some studies have reported an association between a genetic variant (A1166C) in the 3' un-translated region of AGTR1 and increased risk of coronary heart disease (CHD), but other have yielded apparently conflicting results.

METHODS

Literature-based meta-analyses were performed on 48 papers including 53 studies published before June 2008 in relation to the A1166C polymorphism (NCBI, dbSNP: rs5186) of the AGTR1, involving a total of 20,435 CHD cases and 23,674 controls. We also explored potential sources of heterogeneity and conducted appropriate stratified analyses.

RESULTS

In a combined analysis, the per-allele odds ratio (OR) for CHD of the A1166C polymorphism was 1.11 (95% confidence interval: 1.03-1.19), but there is an indication of publication bias and heterogeneity among the 53 studies. Sample size and study quality were significant sources of heterogeneity among studies of the A1166C polymorphism with possibly overestimates in studies of smaller sample-size and poor-quality. When the analyses were restricted to 11 larger studies (≥500 cases), and to 8 high-quality studies (quality score: ≥11 points), the summary per-allele odds ratios were 0.992 (95% confidence interval, 0.944-1.042) and 0.990 (95% confidence interval, 0.915-1.072), respectively.

CONCLUSIONS

An overall weak association between the A1166C polymorphism and CHD is observed but this is likely to be due to publication bias and heterogeneity between studies. There were no significant associations among the larger sample-size and high-quality studies which are less prone to selective publication and have greater power to detect a true association.

Researching genetic versus nongenetic determinants of disease: a comparison and proposed unification

Research standards deviate in genetic versus nongenetic epidemiology. Besides some immutable differences, such as the correlation pattern between variables, these divergent research standards can converge considerably. Current research designs that dissociate genetic and nongenetic measurements are reaching their limits. Studies are needed that massively measure genotypes, nongenetic exposures, and outcomes concurrently.

Challenges in the use of literature-based meta-analysis to examine gene-environment interactions

Statistical interactions between genes and environmental exposures with respect to disease outcomes may help to identify biologic mechanisms and pathways and inform behavioral interventions. The number of persons required for a single study to have sufficient statistical power to detect such interactions may be considered prohibitively large, making a meta-analysis of published literature an apparently attractive alternative. However, meta-analysis of gene-environment interactions using published literature is challenging, with the conclusions being likely to suffer from bias and lack of generalizability. The authors highlight these challenges and biases using an illustrative example: meta-analysis of interactions between the Pro12Ala variant of the peroxisome proliferator-activated receptor gamma (PPARgamma) gene and various diet and lifestyle factors in the risk of diabetes. The authors conclude that literature-based meta-analysis conducted to examine gene-environment interactions is unlikely to provide a meaningful quantitative conclusion. Alternative strategies are required, including analyses in scientific consortia established to assess main genetic effects, where individual participant data can be shared, allowing both greater power and consistency of analysis methods. However, these consortia are likely to be limited by lack of standardization of the measures of environmental factors. This issue may ultimately only be resolvable by the de novo establishment of large single or multicenter cohorts using comparable methods.

Association of mitochondrial allele 4216C with increased risk for sepsis-related organ dysfunction and shock after burn injury

Impaired mitochondrial activity has been linked to increased risk for clinical complications after injury. Furthermore, variant mitochondrial alleles have been identified and are thought to result in decreased mitochondrial activity. These include a nonsynonymous mitochondrial polymorphism (T4216C) in the nicotinamide adenine dinucleotide dehydrogenase 1 gene (ND1), encoding a key member of complex I within the electron transport chain, which is found almost exclusively among Caucasians. We hypothesized that burn patients carrying ND1 4216C are less able to generate the cellular energy necessary for an effective immune response and are at increased risk for infectious complications. The association between 4216C and outcome after burn injury was evaluated in a cohort of 175 Caucasian patients admitted to the Parkland Hospital with burns covering greater than or equal to 15% of their total body surface area or greater than or equal to 5% full-thickness burns under an institutional review board-approved protocol. To remove confounding unrelated to burn injury, individuals were excluded if they presented with significant non-burn-related trauma (Injury Severity Score > or =16), traumatic or anoxic brain injury, spinal cord injury, were HIV/AIDS positive, had active malignancy, or survived less than 48 h postadmission. Within this cohort of patients, carriage of the 4216C allele was significantly associated by unadjusted analysis with increased risk for sepsis-related organ dysfunction or septic shock (P = 0.011). After adjustment for full-thickness burn size, inhalation injury, age, and sex, carriage of the 4216C allele was associated with complicated sepsis (adjusted odds ratio = 3.7; 95% confidence interval, 1.5-9.1; P = 0.005), relative to carriers of the T allele.

Non-replication of association for six polymorphisms from meta-analysis of genome-wide association studies of Parkinson's disease: large-scale collaborative study

Early genome-wide association (GWA) studies on Parkinson's disease (PD) have not been able to yield conclusive, replicable signals of association, perhaps due to limited sample size. We aimed to investigate whether association signals derived from the meta-analysis of the first two GWA investigations might be replicable in different populations. We examined six single-nucleotide polymorphisms (SNPs) (rs1000291, rs1865997, rs2241743, rs2282048, rs2313982, and rs3018626) that had reached nominal significance with at least two of three different strategies proposed in a previous analysis of the original GWA studies. Investigators from the "Genetic Epidemiology of Parkinson's Disease" (GEOPD) consortium were invited to join in this study. Ten teams contributed replication data from 3,458 PD cases and 3,719 controls. The data from the two previously published GWAs (599 PD cases, 592 controls and 443 sibling pairs) were considered as well. All data were synthesized using both fixed and random effects models. The summary allelic odds ratios were ranging from 0.97 to 1.09 by random effects, when all data were included. The summary estimates of the replication data sets (excluding the original GWA data) were very close to 1.00 (range 0.98-1.09) and none of the effects were nominally statistically significant. The replication data sets had significantly different results than the GWA data. Our data do not support evidence that any of these six SNPs reflect susceptibility markers for PD. Much stronger signals of statistical significance in GWA platforms are needed to have substantial chances of replication. Specifically in PD genetics, this would require much larger GWA studies and perhaps novel analytical techniques.

Thinking big: large-scale collaborative research in observational epidemiology

Efforts to identify risk factors for chronic diseases have tended to involve observational studies characterised by relatively few disease outcomes. In the absence of individual studies of sufficiently large size, synthesis of available evidence from multiple smaller studies can help enhance statistical power and aid appropriate interpretation. While meta-analyses of published findings can help prioritize research hypotheses, they are inherently limited by the scale of the evidence available for review and by vulnerability to potential reporting biases. By contrast, collaborative analyses of individual participant data from a comprehensive set of relevant epidemiological studies can offer several advantages over moderately sized individual studies or meta-analyses of aggregated data. This review describes those advantages with reference to selected examples.

Biomarkers in cancer epidemiology: an integrative approach

There are different reasons for the increase in the use of biomarkers in cancer epidemiology which is as follows: (i) the fact that the identification of new carcinogens, characterized by complex exposure circumstances and weak effects, has become increasingly difficult with traditional epidemiological approaches; (ii) the increasing understanding of mechanisms of carcinogenesis and (iii) technical developments in molecular biology and genetics. While a distinction is made between biomarkers of exposure, intermediate events, disease, outcome and susceptibility, their integration in a unique conceptual model is needed. The use of exposure biomarkers in cancer epidemiology aims at measuring the biologically relevant exposure more validly and precisely. In some instances, there is an obvious improvement in using an exposure biomarker, as in the case of urinary markers of aflatoxin and tobacco-specific nitrosamines. Intermediate (effect) biomarkers measure early--in general non-persistent--biological events that take place in the continuum between exposure and cancer development. These include cellular or tissue toxicity, chromosomal alterations, changes in DNA, RNA and protein expression and alterations in functions relevant to carcinogenesis (e.g. DNA repair, immunological response, etc.). The analysis of acquired TP53 mutations is an example of the potentially important. Biomarkers should be validated and consideration of sources of bias and confounding in molecular epidemiology studies should be no less stringent than in other types of epidemiological studies. The overarching goal is the integration of different types of biomarkers to derive risk and outcome profiles for healthy individuals as well as patients.

Discovery properties of genome-wide association signals from cumulatively combined data sets

Genetic effects for common variants affecting complex disease risk are subtle. Single genome-wide association (GWA) studies are typically underpowered to detect these effects, and combination of several GWA data sets is needed to enhance discovery. The authors investigated the properties of the discovery process in simulated cumulative meta-analyses of GWA study-derived signals allowing for potential genetic model misspecification and between-study heterogeneity. Variants with null effects on average (but also between-data set heterogeneity) could yield false-positive associations with seemingly homogeneous effects. Random effects had higher than appropriate false-positive rates when there were few data sets. The log-additive model had the lowest false-positive rate. Under heterogeneity, random-effects meta-analyses of 2-10 data sets averaging 1,000 cases/1,000 controls each did not increase power, or the meta-analysis was even less powerful than a single study (power desert). Upward bias in effect estimates and underestimation of between-study heterogeneity were common. Fixed-effects calculations avoided power deserts and maximized discovery of association signals at the expense of much higher false-positive rates. Therefore, random- and fixed-effects models are preferable for different purposes (fixed effects for initial screenings, random effects for generalizability applications). These results may have broader implications for the design and interpretation of large-scale multiteam collaborative studies discovering common gene variants.

Genetic variants in association studies--review of strengths and weaknesses in study design and current knowledge of impact on cancer risk

Sequencing of the human genome has recently been completed and mapping of the complete genomic variation is ongoing. During the last decade there has been a huge expansion of studies of genetic variants, both with respect to association studies of disease risk and for studies of genetic factors of prognosis and treatments response, i.e., pharmacogenomics. The use of genetics to predict a patient's risk of disease or treatment response is one step toward an improved personalised prevention and screening modality for the prevention of cancer and treatment selection. The technology and statistical methods for completing whole genome tagging of variants and genome wide association studies has developed rapidly over the last decade. After identifying the genetic loci with the strongest, statistical associations with disease risk, future studies will need to further characterise the genotype-phenotype relationship to provide a biological basis for prevention and treatment decisions according to genetic profile. This review discusses some of the general issues and problems of study design; we also discuss challenges in conducting valid association studies in rare cancers such as paediatric brain tumours, where there is support for genetic susceptibility but difficulties in assembling large sample sizes. The clinical interpretation and implementation of genetic association studies with respect to disease risk and treatment is not yet well defined and remains an important area of future research.

An empirical comparison of meta-analyses of published gene-disease associations versus consortium analyses

PURPOSE

Consortia of investigators currently compile sufficiently large sample sizes to investigate the effects of low-risk susceptibility genetic variants. It is not clear how the results obtained by consortia compare with those derived from meta-analyses of published studies.

METHODS

We performed meta-analyses of published data for 16 genetic polymorphisms investigated by the Breast Cancer Association Consortium, and compared sample sizes, heterogeneity, and effect sizes. PubMed, Web of Science, and Human Genome Epidemiology Network databases were searched for breast cancer case-control association studies.

RESULTS

We found that meta-analyses of published data and consortium analyses were based on substantially different data. Published data by non-consortium teams amounted on average to 26.9% of all available data (range 3.0 -50.0%). Both approaches showed statistically significant decreased breast cancer risks for CASP8 D302H. The meta-analyses of published data demonstrated statistically significant results for five other genes and the consortium analyses for two other genes, but the strength of this evidence, evaluated on the basis of the Venice criteria, was not strong.

CONCLUSIONS

Because both approaches identified the same gene out of 16 candidates, the methods can be complimentary. The expense and complexity of consortium-based studies should be considered vis-à-vis the potential methodological limitations of synthesis of published studies.

Underlying genetic models of inheritance in established type 2 diabetes associations

For most associations of common single nucleotide polymorphisms (SNPs) with common diseases, the genetic model of inheritance is unknown. The authors extended and applied a Bayesian meta-analysis approach to data from 19 studies on 17 replicated associations with type 2 diabetes. For 13 SNPs, the data fitted very well to an additive model of inheritance for the diabetes risk allele; for 4 SNPs, the data were consistent with either an additive model or a dominant model; and for 2 SNPs, the data were consistent with an additive or recessive model. Results were robust to the use of different priors and after exclusion of data for which index SNPs had been examined indirectly through proxy markers. The Bayesian meta-analysis model yielded point estimates for the genetic effects that were very similar to those previously reported based on fixed- or random-effects models, but uncertainty about several of the effects was substantially larger. The authors also examined the extent of between-study heterogeneity in the genetic model and found generally small between-study deviation values for the genetic model parameter. Heterosis could not be excluded for 4 SNPs. Information on the genetic model of robustly replicated association signals derived from genome-wide association studies may be useful for predictive modeling and for designing biologic and functional experiments.

The impact of genotype frequencies on the clinical validity of genomic profiling for predicting common chronic diseases

PURPOSE

Single genetic variants in multifactorial disorders typically have small effects, so major increases in disease risk are expected only from the simultaneous exposure to multiple risk genotypes. We investigated the impact of genotype frequencies on the clinical discriminative accuracy for the simultaneous testing of 40 independent susceptibility genetic variants.

METHODS

In separate simulation scenarios, we varied the genotype frequency from 1% to 50% and the odds ratio for each genetic variant from 1.1 to 2.0. Population size was 1 million and the population disease risk was 10%. Discriminative accuracy was quantified as the area under the receiver-operating characteristic curve. Using an example of genomic profiling for type 2 diabetes, we evaluated the area under the receiver-operating characteristic curve when the odds ratios and genotype frequencies varied between five postulated genetic variants.

RESULTS

When the genotype frequency was 1%, none of the subjects carried more than six of 40 risk genotypes, and when risk genotypes were frequent (> or =30%), all carried at least six. The area under the receiver-operating characteristic curve did not increase above 0.70 when the odds ratios were modest (1.1 or 1.25), but higher genotype frequency increased the area under the receiver-operating characteristic curve from 0.57 to 0.82 and from 0.63 to 0.93 when odds ratios were 1.5 or 2.0. The example of type 2 diabetes showed that the area under the receiver-operating characteristic curve did not change when differences in the odds ratios were ignored.

CONCLUSIONS

Given that the effects of susceptibility genes in complex diseases are small, the feasibility of future genomic profiling for predicting common diseases will depend substantially on the frequencies of the risk genotypes.

Validating, augmenting and refining genome-wide association signals

Studies using genome-wide platforms have yielded an unprecedented number of promising signals of association between genomic variants and human traits. This Review addresses the steps required to validate, augment and refine such signals to identify underlying causal variants for well-defined phenotypes. These steps include: large-scale exact replication across both similar and diverse populations; fine mapping and resequencing; determination of the most informative markers and multiple independent informative loci; incorporation of functional information; and improved phenotype mapping of the implicated genetic effects. Even in cases for which replication proves that an effect exists, confident localization of the causal variant often remains elusive.

Association of mitochondrial allele 4216C with increased risk for complicated sepsis and death after traumatic injury

OBJECTIVES

Numerous studies have linked impaired mitochondrial activity with increased risk for clinical complications after injury. Furthermore, a number of nonsynonymous polymorphisms have been identified within the mitochondrial genome that are believed to impair cellular respiration. These DNA variants include a nonsynonymous polymorphism (T4216C) in the NADH dehydrogenase 1 gene (ND1), which encodes a key member of Complex I of the electron transport chain. We hypothesized that trauma patients who carry the ND1 4216C allele may be less able to generate the cellular energy necessary to mount an effective immune response and are at increased risk for death as well as sepsis complicated by organ dysfunction or shock.

METHODS

We enrolled a cohort of 136 patients admitted to the Parkland Hospital Surgical intensive care unit (ICU) with significant trauma (Injury Severity Score > or = 16), > or =16 years of age, and with a minimum intensive care unit stay of > or =24 hours under a protocol approved by the UTSW and Parkland IRBs. Patients with brain death, spinal cord injury, active malignancy, HIV/AIDS or who survived <48 hours after admission were excluded. Clinical data were collected prospectively and T4216C was genotyped by polymerase chain reaction-restriction fragment length polymorphism.

RESULTS

After multivariate adjustment for mechanism, severity of injury, units of packed red blood cells given in the first 24 hours, age, gender, and race/ethnicity, carriage of the 4216 C-allele was significantly associated with increased risk for sepsis complicated by organ dysfunction or septic shock (adjusted odds ratio [aOR] = 3.68; 95%CI: 1.17-11.52; p = 0.02) as well as death (aOR = 4.56; 95% CI: 1.05-19.79; p = 0.04), relative to carriers of the T-allele.

CONCLUSION

Carriage of the mitochondrial 4216C-allele increases the risk for infectious complications and death after severe trauma.

Genome-wide and candidate gene association study of cigarette smoking behaviors

The contribution of common genetic variation to one or more established smoking behaviors was investigated in a joint analysis of two genome wide association studies (GWAS) performed as part of the Cancer Genetic Markers of Susceptibility (CGEMS) project in 2,329 men from the Prostate, Lung, Colon and Ovarian (PLCO) Trial, and 2,282 women from the Nurses' Health Study (NHS). We analyzed seven measures of smoking behavior, four continuous (cigarettes per day [CPD], age at initiation of smoking, duration of smoking, and pack years), and three binary (ever versus never smoking, < or = 10 versus > 10 cigarettes per day [CPDBI], and current versus former smoking). Association testing for each single nucleotide polymorphism (SNP) was conducted by study and adjusted for age, cohabitation/marital status, education, site, and principal components of population substructure. None of the SNPs achieved genome-wide significance (p<10(-7)) in any combined analysis pooling evidence for association across the two studies; we observed between two and seven SNPs with p<10(-5) for each of the seven measures. In the chr15q25.1 region spanning the nicotinic receptors CHRNA3 and CHRNA5, we identified multiple SNPs associated with CPD (p<10(-3)), including rs1051730, which has been associated with nicotine dependence, smoking intensity and lung cancer risk. In parallel, we selected 11,199 SNPs drawn from 359 a priori candidate genes and performed individual-gene and gene-group analyses. After adjusting for multiple tests conducted within each gene, we identified between two and five genes associated with each measure of smoking behavior. Besides CHRNA3 and CHRNA5, MAOA was associated with CPDBI (gene-level p<5.4x10(-5)), our analysis provides independent replication of the association between the chr15q25.1 region and smoking intensity and data for multiple other loci associated with smoking behavior that merit further follow-up.

Meta-analysis in genome-wide association studies

The advent of genome-wide association studies has allowed considerable progress in the identification and robust replication of common gene variants that confer susceptibility to common diseases and other phenotypes of interest. These genetic effect sizes are almost invariably moderate to small in magnitude and single studies, even if large, are underpowered to detect them with confidence. Meta-analysis of many genome-wide association studies improves the power to detect more associations, and to investigate the consistency or heterogeneity of these associations across diverse datasets and study populations. In this review, we discuss the key methodological issues in the set-up, information gathering and processing, and analysis of meta-analyses of genome-wide association datasets. We illustrate, as an example, the application of meta-analysis methods in the elucidation of common genetic variants associated with Type 2 diabetes. Finally, we discuss the prospects and caveats for future application of meta-analysis methods in the genome-wide setting.

Invited commentary: from genome-wide association studies to gene-environment-wide interaction studies--challenges and opportunities

The recent success of genome-wide association studies in finding susceptibility genes for many common diseases presents tremendous opportunities for epidemiologic studies of environmental risk factors. Analysis of gene-environment interactions, included in only a small fraction of epidemiologic studies until now, will begin to accelerate as investigators integrate analyses of genome-wide variation and environmental factors. Nevertheless, considerable methodological challenges are involved in the design and analysis of gene-environment interaction studies. The authors review these issues in the context of evolving methods for assessing interactions and discuss how the current agnostic approach to interrogating the human genome for genetic risk factors could be extended into a similar approach to gene-environment-wide interaction studies of disease occurrence in human populations.

Calibration of credibility of agnostic genome-wide associations

Genome-wide testing platforms are increasingly used to promote "agnostic" approaches to the discovery of gene variants associated with the risk of many common diseases and quantitative traits. The early track record of genome-wide association (GWA) studies suggests that some proposed associations are replicated quite consistently with large-scale subsequent evidence from multiple studies, others have a more inconsistent replication record, some have failed to be replicated by independent investigators and many more early proposed associations await further replication. An important question is how to calibrate the credibility of these postulated associations. A simple Bayesian method is applied here to achieve such calibration. The variability of the estimated credibility is examined under different assumptions. Empirical examples are drawn from existing GWA studies. It is demonstrated that the credibility of different proposed associations can cover a very wide range. The credibility of specific associations usually remains relatively robust when different plausible assumptions are made (within a reasonable range) for the prior odds of an association being true, or the magnitude of the anticipated effect size for genetic associations. Heterogeneity and bias assumptions can have a more major impact on the credibility estimates and thus they need very careful consideration in each case. Credibility calibration may be used in conjunction with qualitative criteria for the appraisal of the cumulative evidence that take into consideration the amount, consistency, and protection from bias in the data.