Study-specific Covariates Research Articles

A multivariate Polya tree model for meta-analysis with event-time distributions.

We develop a nonparametric Bayesian prior for a family of random probability measures by extending the Polya tree ($\mbox{PT}$) prior to a joint prior for a set of probability measures $G_1,\dots ,G_n$, suitable for meta-analysis with event-time outcomes. In the application to meta-analysis, $G_i$ is the event-time distribution specific to study $i$. The proposed model defines a regression on study-specific covariates by introducing increased correlation for any pair of studies with similar characteristics. The desired multivariate $\mbox{PT}$ model is constructed by introducing a hierarchical prior on the conditional splitting probabilities in the $\mbox{PT}$ construction for each of the $G_i$. The hierarchical prior replaces the independent beta priors for the splitting probability in the PT construction with a Gaussian process prior for corresponding (logit) splitting probabilities across all studies. The Gaussian process is indexed by study-specific covariates, introducing the desired dependence with increased correlation for similar studies. The main feature of the proposed construction is (conditionally) conjugate posterior updating with commonly reported inference summaries for event-time data. The construction is motivated by a meta-analysis over cancer immunotherapy studies.

Validation of a face image assessment technology to study the dynamics of human functional states in the EEG resting-state paradigm.

The article presents the results of a study aimed at finding covariates to account for the activity of implicit cognitive processes in conditions of functional rest of the subjects and during them being presented their own or someone else's face in a joint analysis of EEG experiment data. The proposed approach is based on the analysis of the dynamics of the facial muscles of the subject recorded on video. The pilot study involved 18 healthy volunteers. In the experiment, the subjects were sitting in front of a computer screen and performed the following task: sequentially closed their eyes (three trials of 2 minutes each) and opened them (three trials of the same duration between periods of closed eyes) when the screen was either empty or when it was showing a video recording of their own face or the face of an unfamiliar person of the same gender as the participant. EEG, ECG and a video of the face were recorded for all subjects. In the work a separate subtask of the study was also addressed: validating a technique for assessing the dynamics of the subjects' facial muscle activity using the recorded videos of the "eyes open" trials to obtain covariates that can be included in subsequent processing along with EEG correlates in neurocognitive experiments with a paradigm that does not involve the performance of active cognitive tasks ("resting-state conditions"). It was shown that the subject's gender, stimulus type (screen empty or showing own/other face), trial number are accompanied by differences in facial activity and can be used as study-specific covariates. It was concluded that the analysis of the dynamics of facial activity based on video recording of "eyes open" trials can be used as an additional method in neurocognitive research to study implicit cognitive processes associated with the perception of oneself and other, in the functional rest paradigm.

Abstract 15719: Proteomic Analyses Reveal Mechanistic Links Between Clonal Hematopoiesis of Indeterminate Potential and Coronary Artery Disease

Introduction: Clonal hematopoiesis of indeterminate potential (CHIP) is a phenomenon where hematopoietic stem cells acquire leukemogenic mutations without blood cancer. CHIP is a causal risk factor for coronary artery disease (CAD), yet underlying mechanisms remain unclear. The plasma proteome may provide novel mechanistic insights into the links between CHIP and CAD. Methods: We studied the associations of CHIP with the plasma proteins and CAD risk in four Trans-Omics for Precision Medicine (TOPMed) longitudinal cohorts: ARIC (N=9,084), CHS (N=1,727), JHS (N=1,858), and MESA (N=978). CHIP (variant allele fraction >2%) was identified from whole genome sequences of blood DNA and modeled both as a composite of the most common drivers ( DNMT3A , TET2 , ASXL1 , and JAK2 ) and separately. Levels of ~1,300 serum proteins measured by SomaScan were log-transformed and residualized on study-specific covariates. The association between CHIP and protein levels was estimated across all participants within each study, as well as by sex, and meta-analyzed. For proteins significantly (FDR<0.05) associated with CHIP, pathway analysis and functional validation in Tet2-/- mice were conducted. Results: Across all studies (mean age: 59.8 y; 43.6% male), 720 (5.6%) individuals were identified with CHIP. There were 143 significant CHIP-protein pairs, with carbonic anhydrase 1, lysozyme C, and properdin being the most associated proteins. TET2 had the largest number of, and strongest, associations among examined driver genes. In sex-stratified analysis, follicle-stimulating hormone was inversely associated in females. Pathway analysis implicated the inflammation-related STAT3 and IL-17 signaling pathways. We observed significantly greater expression of Cfp and Lyz (encoding properdin and lysosome C, respectively) in the monocytes of Tet2 -/- female mice compared to wild-type female mice and Tet2 -/- male mice. A number of proteins (including PCSK9) were significantly associated with both CHIP and CAD, suggesting both shared and nonshared mechanisms. Conclusions: CHIP, particularly TET2 , is broadly associated with plasma proteins, as validated by evidence in Tet2-/- mice. Further studies are needed to understand the biological mechanisms linking CHIP to CAD.

Large trans-ethnic meta-analysis identifies AKR1C4 as a novel gene associated with age at menarche.

Does the expansion of genome-wide association studies (GWAS) to a broader range of ancestries improve the ability to identify and generalise variants associated with age at menarche (AAM) in European populations to a wider range of world populations? By including women with diverse and predominantly non-European ancestry in a large-scale meta-analysis of AAM with half of the women being of African ancestry, we identified a new locus associated with AAM in African-ancestry participants, and generalised loci from GWAS of European ancestry individuals. AAM is a highly polygenic puberty trait associated with various diseases later in life. Both AAM and diseases associated with puberty timing vary by race or ethnicity. The majority of GWAS of AAM have been performed in European ancestry women. We analysed a total of 38 546 women who did not have predominantly European ancestry backgrounds: 25 149 women from seven studies from the ReproGen Consortium and 13 397 women from the UK Biobank. In addition, we used an independent sample of 5148 African-ancestry women from the Southern Community Cohort Study (SCCS) for replication. Each AAM GWAS was performed by study and ancestry or ethnic group using linear regression models adjusted for birth year and study-specific covariates. ReproGen and UK Biobank results were meta-analysed using an inverse variance-weighted average method. A trans-ethnic meta-analysis was also carried out to assess heterogeneity due to different ancestry. We observed consistent direction and effect sizes between our meta-analysis and the largest GWAS conducted in European or Asian ancestry women. We validated four AAM loci (1p31, 6q16, 6q22 and 9q31) with common genetic variants at P < 5 × 10-7. We detected one new association (10p15) at P < 5 × 10-8 with a low-frequency genetic variant lying in AKR1C4, which was replicated in an independent sample. This gene belongs to a family of enzymes that regulate the metabolism of steroid hormones and have been implicated in the pathophysiology of uterine diseases. The genetic variant in the new locus is more frequent in African-ancestry participants, and has a very low frequency in Asian or European-ancestry individuals. N/A. Extreme AAM (<9 years or >18 years) were excluded from analysis. Women may not fully recall their AAM as most of the studies were conducted many years later. Further studies in women with diverse and predominantly non-European ancestry are needed to confirm and extend these findings, but the availability of such replication samples is limited. Expanding association studies to a broader range of ancestries or ethnicities may improve the identification of new genetic variants associated with complex diseases or traits and the generalisation of variants from European-ancestry studies to a wider range of world populations. Funding was provided by CHARGE Consortium grant R01HL105756-07: Gene Discovery For CVD and Aging Phenotypes and by the NIH grant U24AG051129 awarded by the National Institute on Aging (NIA). The authors have no conflict of interest to declare.

Rare Variants in the DNA Repair Pathway and the Risk of Colorectal Cancer.

Inherited susceptibility is an important contributor to colorectal cancer risk, and rare variants in key genes or pathways could account in part for the missing proportion of colorectal cancer heritability. We conducted an exome-wide association study including 2,327 cases and 2,966 controls of European ancestry from three large epidemiologic studies. Single variant associations were tested using logistic regression models, adjusting for appropriate study-specific covariates. In addition, we examined the aggregate effects of rare coding variation at the gene and pathway levels using Bayesian model uncertainty techniques. In an exome-wide gene-level analysis, we identified ST6GALNAC2 as the top associated gene based on the Bayesian risk index (BRI) method [summary Bayes factor (BF)BRI = 2604.23]. A rare coding variant in this gene, rs139401613, was the top associated variant (P = 1.01 × 10-6) in an exome-wide single variant analysis. Pathway-level association analyses based on the integrative BRI (iBRI) method found extreme evidence of association with the DNA repair pathway (BFiBRI = 17852.4), specifically with the nonhomologous end joining (BFiBRI = 437.95) and nucleotide excision repair (BFiBRI = 36.96) subpathways. The iBRI method also identified RPA2, PRKDC, ERCC5, and ERCC8 as the top associated DNA repair genes (summary BFiBRI ≥ 10), with rs28988897, rs8178232, rs141369732, and rs201642761 being the most likely associated variants in these genes, respectively. We identified novel variants and genes associated with colorectal cancer risk and provided additional evidence for a role of DNA repair in colorectal cancer tumorigenesis. This study provides new insights into the genetic predisposition to colorectal cancer, which has potential for translation into improved risk prediction.

Evaluating the strength of evidence between long-term NO2 exposure and pediatric asthma incidence for potential inclusion into the Global Burden of Disease Study

Several epidemiological studies have highlighted a relationship between NO2 exposure and increased incidence and prevalence of pediatric asthma. Recently, the Global Burden of Disease (GBD) developed a quantitative evidence scoring approach with the intention to assess the effect and strength of the evidence for exposure-response relationships with consistency across all risk factors. We use this evidence scoring approach to generate a quantitative effect estimate for NO2 exposure and childhood asthma that can be applied for burden of disease assessments. For the purpose of this work, we systematically compiled existing studies on the relationship between NO2 and childhood asthma and extracted relative risks (RR). In addition, we extracted a set of study-specific covariates in order to explain between-study heterogeneity in risk estimates. We then applied the GBD’s newly developed meta-regression tool MR-BRT (MetaRegression - Bayesian, Regularized, Trimmed) to provide a quantitative estimate of effect and strength of evidence based upon unexplained between-study heterogeneity. Preliminary analyses including 28 sources from 11 countries identified within a systematic literature review conducted by Khreis et al. in 2017 were included in the metaregression. Overall, we estimated a RR of 1.10 (95% confidence interval [CI]=0.94-1.30) per 5 ppb increase in annual average NO2 (p-value=0.13) when not accounting for study-specific covariates. A model accounting for all significant study-level covariates (‘outcomes self-reported`, ‘confounding uncontrolled’, ‘selection bias’, and ‘subpopulation’) revealed an effect estimate of 1.28 (95%CI=1.00-1.63) per 5 ppb (p-value=0.024). Findings provide strong evidence for a relationship between NO2 exposure and pediatric asthma when adjusting for between-study heterogeneity. In future steps, we will include 27 additional studies that were published after the 2017 systematic review conducted by Khreis et al. (2017). Furthermore, we will apply an automated covariate selection process in which variables are sub-selected by running a set of log-linear models with a range of Lasso penalty parameters.

GWAS of QRS duration identifies new loci specific to Hispanic/Latino populations.

BackgroundThe electrocardiographically quantified QRS duration measures ventricular depolarization and conduction. QRS prolongation has been associated with poor heart failure prognosis and cardiovascular mortality, including sudden death. While previous genome-wide association studies (GWAS) have identified 32 QRS SNPs across 26 loci among European, African, and Asian-descent populations, the genetics of QRS among Hispanics/Latinos has not been previously explored.MethodsWe performed a GWAS of QRS duration among Hispanic/Latino ancestry populations (n = 15,124) from four studies using 1000 Genomes imputed genotype data (adjusted for age, sex, global ancestry, clinical and study-specific covariates). Study-specific results were combined using fixed-effects, inverse variance-weighted meta-analysis.ResultsWe identified six loci associated with QRS (P<5x10-8), including two novel loci: MYOCD, a nuclear protein expressed in the heart, and SYT1, an integral membrane protein. The top SNP in the MYOCD locus, intronic SNP rs16946539, was found in Hispanics/Latinos with a minor allele frequency (MAF) of 0.04, but is monomorphic in European and African descent populations. The most significant QRS duration association was with intronic SNP rs3922344 (P = 1.19x10-24) in SCN5A/SCN10A. Three other previously identified loci, CDKN1A, VTI1A, and HAND1, also exceeded the GWAS significance threshold among Hispanics/Latinos. A total of 27 of 32 previously identified QRS duration SNPs were shown to generalize in Hispanics/Latinos.ConclusionsOur QRS duration GWAS, the first in Hispanic/Latino populations, identified two new loci, underscoring the utility of extending large scale genomic studies to currently under-examined populations.

Epigenetic Biomarkers Of PTSD: Updates From The EWAS Working Group of The PTSD PGC

Trauma exposure is unfortunately quite common, yet only a minority of individuals develop PTSD following trauma. The Epigenome-Wide Association Study (EWAS) group of the PTSD PGC is working to identify DNA methylation-based epigenetic biomarkers of PTSD that can serve as important indicators of the disorder, and to characterize which of these biomarkers identify individuals who may be most at risk following trauma. Existing, blood-derived DNA methylation microarray data from three civilian and seven military cohorts were analyzed using a common pipeline that tested each CpG site for an association with PTSD, controlling for age, genomic-derived ancestry estimates, leukocyte cell proportions as well as study-specific covariates if applicable (i.e. gender, plate, or bisulfite conversion). A meta-analysis across all cohorts and separately within Civilian and Military cohorts was subsequently conducted to identify robust biomarkers of PTSD. Results from primary analyses of all 1,896 samples revealed that the top four CpG sites fell within the AHRR locus (FDR for all sites <0.0003) and were all associated with reduced DNA methylation in PTSD. However, the associations at all four sites were substantially attenuated (FDR for all sites >0.69) in additional analyses that controlled for smoking. In contrast, the fifth ranked CpG site within RNF6 (FDR=0.007) from the primary analyses became the top-ranked CpG site associated with PTSD in our additional analyses that controlled for smoking (N=1,841), although results did not reach statistical significance (FDR=0.09). The meta-analysis of Civilian cohorts (N = 545) yielded two significant CpG sites within the NRG1 (FDR = 0.023) and HGS (FDR = 0.033) loci. Both loci remained significant after controlling for smoking. A sensitivity analysis of two cohorts found that the association with PTSD was consistent within each cohort after controlling for childhood maltreatment. Longitudinal analyses are ongoing to assess whether these differences pre-date trauma exposure that precipitates PTSD, or whether they associate with PTSD onset.

Genome-wide meta-analysis of 158,000 individuals of European ancestry identifies three loci associated with chronic back pain.

Back pain is the #1 cause of years lived with disability worldwide, yet surprisingly little is known regarding the biology underlying this symptom. We conducted a genome-wide association study (GWAS) meta-analysis of chronic back pain (CBP). Adults of European ancestry were included from 15 cohorts in the Cohorts for Heart and Aging Research in Genomic Epidemiology (CHARGE) consortium, and from the UK Biobank interim data release. CBP cases were defined as those reporting back pain present for ≥3–6 months; non-cases were included as comparisons (“controls”). Each cohort conducted genotyping using commercially available arrays followed by imputation. GWAS used logistic regression models with additive genetic effects, adjusting for age, sex, study-specific covariates, and population substructure. The threshold for genome-wide significance in the fixed-effect inverse-variance weighted meta-analysis was p<5×10−8. Suggestive (p<5×10−7) and genome-wide significant (p<5×10−8) variants were carried forward for replication or further investigation in the remaining UK Biobank participants not included in the discovery sample. The discovery sample comprised 158,025 individuals, including 29,531 CBP cases. A genome-wide significant association was found for the intronic variant rs12310519 in SOX5 (OR 1.08, p = 7.2×10−10). This was subsequently replicated in 283,752 UK Biobank participants not included in the discovery sample, including 50,915 cases (OR 1.06, p = 5.3×10−11), and exceeded genome-wide significance in joint meta-analysis (OR 1.07, p = 4.5×10−19). We found suggestive associations at three other loci in the discovery sample, two of which exceeded genome-wide significance in joint meta-analysis: an intergenic variant, rs7833174, located between CCDC26 and GSDMC (OR 1.05, p = 4.4×10−13), and an intronic variant, rs4384683, in DCC (OR 0.97, p = 2.4×10−10). In this first reported meta-analysis of GWAS for CBP, we identified and replicated a genetic locus associated with CBP (SOX5). We also identified 2 other loci that reached genome-wide significance in a 2-stage joint meta-analysis (CCDC26/GSDMC and DCC).

Non-Parametric Bayesian Multivariate Metaregression: An Application in Environmental Epidemiology

Summary In biomedical research, meta-analysis is a popular tool to combine evidence from multiple studies to investigate an exposure–response association. A two-stage analytical approach is used in meta-analysis for its computational convenience and flexibility. The first stage estimates the association for each study whereas the second stage combines the study-specific estimates correcting for the study-specific error. The second stage often incorporates study-specific covariates (metapredictors) and is called metaregression. One application where the two-stage meta-analysis is useful is an epidemiological study for the health effects of environmental exposure, which often analyses time series data of exposure and health outcome collected from multiple locations. The first stage models location-specific association, which is often represented by multiple parameters as the association is non-linear or delayed, and the second stage conducts a multivariate metaregression with location-specific characteristics as metapredictors. The currently used multivariate metaregression is a form of normal linear regression, which may be limited as it assumes linearity in metapredictors, residual normality and homoscedasticity. In the paper, we propose a flexible multivariate metaregression in a non-parametric Bayesian modelling framework incorporating a residual spatial dependence. The proposed metaregression was evaluated through a simulation study and applied to investigate a temperature–mortality association in the 135 US cities.

Improving the accuracy of likelihood-based inference in meta-analysis and meta-regression

Summary Random-effects models are frequently used to synthesize information from different studies in meta-analysis. While likelihood-based inference is attractive both in terms of limiting properties and of implementation, its application in random-effects meta-analysis may result in misleading conclusions, especially when the number of studies is small to moderate. The current paper shows how methodology that reduces the asymptotic bias of the maximum likelihood estimator of the variance component can also substantially improve inference about the mean effect size. The results are derived for the more general framework of random-effects meta-regression, which allows the mean effect size to vary with study-specific covariates.

Abstract LB-365: Novel susceptibility loci for colorectal cancer: Findings from the colorectal transdisciplinary (CORECT) study OncoArray analysis

Abstract Genome-wide association studies (GWAS) have identified 58 susceptibility alleles across 37 regions associated with the incidence of colorectal cancer (CRC) with P&amp;lt;5E-08, yet much of the disease's familial risk remains to be explained. In its first phase, the Colorectal Transdisciplinary (CORECT) study, in collaboration with the Genetics and Epidemiology of Colorectal Cancer Consortium (GECCO), the Colon Cancer Family Registry (CCFR), the Asian Colorectal Cancer Consortium (ACCC) and the Multiethnic Cohort (MEC), identified 6 of these loci based on 18,299 cases and 19,656 controls. The CORECT consortium recently completed genotyping on approximately 26,000 cases and 18,000 controls using the Illumina Infinium® OncoArray platform with the goals of 1) identifying novel overall and ethnic-specific susceptibility alleles and 2) fine-mapping known CRC risk loci. Newly genotyped samples represent multiple ethnic groups and include individuals primarily of European (75%), Asian (19%) and Hispanic (2.5%) origins. Combining both phases of CORECT, this study constitutes the largest GWAS meta-analysis of CRC to date. Here, we present European-specific results derived from genetically-defined Europeans genotyped on the OncoArray (16,456 cases and 10,442 controls), CORECT Phase I (5,584 cases and 5,329 controls), and CCFR/GECCO (12,715 cases and 14,327 controls), for a total of 34,755 cases and 30,098 controls. OncoArray genotype data were imputed to the 1000 Genomes Phase III reference panel, and summary results from logistic regression adjusting for age, sex, global ancestry, and study-specific covariates were combined in a fixed-effects inverse variance-weighted meta-analysis. Preliminary results indicate at least 10 new low-penetrance risk variants that reach genome-wide significance (P&amp;lt;5E-08) and that are independent of known risk loci. Further, fine-mapping of well-established CRC susceptibility regions is underway. This investigation provides additional insight into the etiology of CRC and informs future risk modeling efforts. Citation Format: Stephanie L. Schmit, Fredrick R. Schumacher, Christopher K. Edlund, Jian Gong, Gad Rennert, Wei Zheng, Loic Le Marchand, Ulrike Peters, Graham Casey, Li Hsu, Stephen B. Gruber, David V. Conti, CORECT, GECCO, CCFR, ACCC, and MEC. Novel susceptibility loci for colorectal cancer: Findings from the colorectal transdisciplinary (CORECT) study OncoArray analysis. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr LB-365.

Long-term follow-up of psychosocial distress after early onset preeclampsia: the Preeclampsia Risk EValuation in FEMales cohort study

Objective: To examine long-term psychosocial distress in women with a history of early onset preeclampsia (PE) compared to a comparison group. Methods: Women with and without a history of early onset PE participating in the ‘Preeclampsia Risk EValuation in FEMales’ (PREVFEM) study were sent questionnaires, on average 14.1 years (SD = 3.2, range 5–23 years) after the index pregnancy. In total 265 (77%) women with PE and 268 (78%) age-matched women without PE returned questionnaires (mean age 43.5, SD =4.6 years). Group differences were examined on indicators of psychosocial distress, depressive symptoms, anxiety, fatigue, loneliness, marital quality, trait optimism and Type D personality, and unadjusted and adjusted for a priori chosen and study-specific covariates. In secondary analyses, the effect of previously detected hypertension was examined, as well as pregnancy-related events within the PE group. Results: Women with a history of PE reported more subsequent depressive symptoms (B = 0.70, 95% CI 0.09–1.32, p = 0.026) and more fatigue (B = 1.12, 95% CI 0.07–2.18, p = 0.037) compared to the non-PE group, but the differences explained less than 1% of the variance. The differences remained after adjustment for age, BMI and education level, and additional adjustment for partner, being unemployed and physical activity. No significant differences were observed for anxiety, loneliness, marital quality, optimism, or Type D personality. These differences were not explained by four-year previously measured elevated blood pressure in the PE group. Having had a stillborn child or early neonatal death during the index pregnancy was associated with higher depressive symptoms, anxiety, fatigue, and loneliness in the PE group, but these factors explained only a small proportion of the variance in these psychosocial distress factors. Conclusion: A history of early PE is associated with slightly higher levels of depressive symptoms and fatigue on average 14 years later, but this is unlikely to be of clinical relevance.

Abstract MP79: Identification of Novel Loci Associated with Anthropometric Traits in African Ancestry Populations

Introduction: African Americans have the highest prevalence of obesity in the U.S. Our prior work on HapMap imputed genome-wide association studies (GWAS) from the African Ancestry Anthropometry Genetics Consortium (AAAGC) revealed seven genome-wide significant loci for body mass index (BMI). Hypothesis: In this study, we extended to use genome-wide imputation to the cosmopolitan 1000 Genomes Phase 1 reference panel to examine individuals of African ancestry for association with BMI (N=53,493), waist-to-hip ratio adjusted for BMI (WHRadjBMI) (N=23,692) and height (N=53,362) in the discovery and replication stages. Methods: In each study, traits were stratified by gender and transformed to normality and adjusted for age, study specific covariates and principal components for single variant association test under an additive model. In the discovery stage, meta-analyses using fixed-effects inverse variance weighted method were performed to combine association results in all and sex-stratified samples. Variants with P&lt;1E-4 were selected for subsequent replications in both individuals of African ancestry and European ancestry, the latter results were obtained by imputation of summary statistics from the GIANT consortium (N=322,154 for BMI, 210,062 for WHRadjBMI, and 253,252 for height). Results: For BMI, we observed genome-wide significant associations (P&lt;5E-8) at seven established loci (near SEC16B, TMEM18, GNPDA2, GALNT10, KLHL32, FTO and MC4R) in meta-analysis of all African ancestry individuals from the discovery and replication stages. Sex-stratified meta-analysis revealed two novel loci near IRX4-IRX2 in women, and near MLC1 in men, respectively. Meta-analysis of all individuals from African and European ancestry revealed an additional novel locus near ARAP1. For WHRadjBMI, we observed genome-wide significant associations at one established locus (ADAMTS9) in all African ancestry individuals, and three novel loci (near SSX2IP, RREB1 and PDE3B) in women. For height, we observed 29 established loci and three novel loci (near NCOA2, P4HA1 and TGFB3) in all African ancestry individuals, and three novel loci (near CRB1, MIR4268 and LINC00704) in women. Among 12 lead variants at novel loci associated with anthropometry traits, four variants are low frequencies in our African ancestry populations (0.005&lt;MAF &lt;0.05). On the contrary, all but one of the lead variants at established loci were common. Conclusions: We identified 12 novel loci associated with anthropometry traits in sex-combined and sex-stratified analyses of African ancestry populations and additional European populations. Our findings support that imputation to higher density reference panels such as 1000 Genomes improves the power to detect associations at low frequency variants, which is particularly useful for African ancestry populations with a low degree of linkage disequilibrium.

Bayesian meta‐analysis of test accuracy in the absence of a perfect reference test applied to bone scintigraphy for the diagnosis of complex regional pain syndrome

There is conflicting evidence about the accuracy of bone scintigraphy (BS) for the diagnosis of complex regional pain syndrome 1 (CRPS 1). In a meta-analysis of diagnostic studies, the evaluation of test accuracy is impeded by the use of different imperfect reference tests. The aim of our study is to summarize sensitivity and specificity of BS for CRPS 1 and to identify factors to explain heterogeneity. We use a hierarchical Bayesian approach to model test accuracy and threshold, and we present different models accounting for the imperfect nature of the reference tests, and assuming conditional dependence between BS and the reference test results. Further, we include disease duration as explanatory variable in the model. The models are compared using summary ROC curves and the deviance information criterion (DIC). Our results show that those models which account for different imperfect reference tests with conditional dependence and inclusion of the covariate are the ones with the smallest DIC. The sensitivity of BS was 0.87 (95% credible interval 0.73-0.97) and the overall specificity was 0.87 (0.73-0.95) in the model with the smallest DIC, in which missing values of the covariate are imputed within the Bayesian framework. The estimated effect of duration of symptoms on the threshold parameter was 0.17 (-0.25 to 0.57). We demonstrate that the Bayesian models presented in this paper are useful to address typical problems occurring in meta-analysis of diagnostic studies, including conditional dependence between index test and reference test, as well as missing values in the study-specific covariates.

Merging multiple longitudinal studies with study-specific missing covariates: A joint estimating function approach.

Merging multiple datasets collected from studies with identical or similar scientific objectives is often undertaken in practice to increase statistical power. This article concerns the development of an effective statistical method that enables to merge multiple longitudinal datasets subject to various heterogeneous characteristics, such as different follow-up schedules and study-specific missing covariates (e.g., covariates observed in some studies but missing in other studies). The presence of study-specific missing covariates presents great statistical methodology challenge in data merging and analysis. We propose a joint estimating function approach to addressing this challenge, in which a novel nonparametric estimating function constructed via splines-based sieve approximation is utilized to bridge estimating equations from studies with missing covariates to those with fully observed covariates. Under mild regularity conditions, we show that the proposed estimator is consistent and asymptotically normal. We evaluate finite-sample performances of the proposed method through simulation studies. In comparison to the conventional multiple imputation approach, our method exhibits smaller estimation bias. We provide an illustrative data analysis using longitudinal cohorts collected in Mexico City to assess the effect of lead exposures on children's somatic growth.

Abstract P157: Does Physical Activity Modify the Association of 15 Well-established Obesity Loci with BMI: The ARIC Study

This study investigates gene-by-physical activity interaction with 15 well-replicated body mass index (BMI) loci in the Atherosclerosis Risk in Communities (ARIC) Study, a prospective, multi-center, bi-racial population-based cohort of adults (N=15,792 at baseline (1987-1989), aged 45-64). Our analyses were restricted to subjects of self-reported European descent with complete genotype and phenotype data (n=8059, 53% female, mean age 54.3 (SD 5.7)). BMI was computed from measured height and weight. Physical activity (PA) was measured using a modified validated Baecke questionnaire, resulting in an ordinal score ranging from 1.00-10.00 for sport and leisure-time activity. Physical activity was dichotomized at the sex-specific median for high/low PA, with high PA as the referent. BMI was regressed on single nucleotide polymorphism (SNP), PA level, SNP*PA level, and study-specific covariates assuming an additive genetic model. To account for multiple testing, interaction p-values less than 0.006 (0.10/15) were considered statistically significant. Mean BMI was 27.0 (SD 4.8) kg/m2, and sports and leisure time physical activity was 5.00 (SD 1.13, range 1.75-9.25) units. We replicated the main effects of SNP on BMI for FTO (rs9939609, p=8.52E-11) and SEC16B/RASAL2 (rs10913469, p=0.001), and NEGR1 was borderline significant (rs2815752, p=0.006). MAF was the only SNP that displayed a directionally inconsistent effect estimate with earlier findings (Beta: -0.02, SE -0.16, 0.13). Four of the interactions had a p-value &lt;0.1, including SEC16B, PTER, NEGR1, and SH2B1. The largest magnitude effect estimates among the nominally significant interactions were observed for SNPs rs10913469 (locus SEC16B/RASAL2, frequency: 0.2024) and rs10508503 (locus PTER, frequency: 0.0871). For rs10913469, each additional copy of the risk allele resulted in an increased BMI of 0.58 kg/m2 for those in the low PA category, but an increase of only 0.07 kg/m2 for those in the high PA category (p-interaction=0.007). For rs10508503, each additional copy of the risk allele resulted in an increased BMI of 0.51 kg/m2 for those in the low PA category, and a decreased BMI of 0.14 kg/m2 for those in the high PA category (p-interaction=0.013). Mean BMIs for those in the low physical activity category increase with additional copies of the C (risk) allele (mean BMI for TT=25.6, TC=27.2, CC=27.6 kg/m2, p=0.04), while mean BMIs of those with high PA do not differ between genotypes (mean BMI for TT=26.7, TC=26.8, CC=26.6 kg/m2, p=0.51). While we found a significant main effect of the SEC16B/RASAL2 SNP on BMI (p=0.001), for the PTER SNP we did not (p=0.229). These novel findings demonstrate the importance of context specific genetic effects and contribute to the growing literature highlighting the possible importance of physical activity in the attenuation of underlying genetic risk.

Generalised interval estimation in the random effects meta regression model

The explanation of heterogeneity when combining different studies is an important issue in meta analysis. Besides including a heterogeneity parameter in the analysis, it is also important to understand the possible causes of heterogeneity. A possibility is to incorporate study-specific covariates in the model that account for between-trial variability. This leads to the random effects meta regression model. Commonly used methods for constructing confidence intervals for the regression coefficients are examined and two new methods based on generalised inference principles are proposed. The different methods are compared by an extensive simulation study with respect to coverage probability and average length.

Common genetic variants in the 9p21 region and their associations with multiple tumours

Background:The chromosome 9p21.3 region has been implicated in the pathogenesis of multiple cancers.Methods:We systematically examined up to 203 tagging SNPs of 22 genes on 9p21.3 (19.9–32.8 Mb) in eight case–control studies: thyroid cancer, endometrial cancer (EC), renal cell carcinoma, colorectal cancer (CRC), colorectal adenoma (CA), oesophageal squamous cell carcinoma (ESCC), gastric cardia adenocarcinoma and osteosarcoma (OS). We used logistic regression to perform single SNP analyses for each study separately, adjusting for study-specific covariates. We combined SNP results across studies by fixed-effect meta-analyses and a newly developed subset-based statistical approach (ASSET). Gene-based P-values were obtained by the minP method using the Adaptive Rank Truncated Product program. We adjusted for multiple comparisons by Bonferroni correction.Results:Rs3731239 in cyclin-dependent kinase inhibitors 2A (CDKN2A) was significantly associated with ESCC (P=7 × 10−6). The CDKN2A-ESCC association was further supported by gene-based analyses (Pgene=0.0001). In the meta-analyses by ASSET, four SNPs (rs3731239 in CDKN2A, rs615552 and rs573687 in CDKN2B and rs564398 in CDKN2BAS) showed significant associations with ESCC and EC (P<2.46 × 10−4). One SNP in MTAP (methylthioadenosine phosphorylase) (rs7023329) that was previously associated with melanoma and nevi in multiple genome-wide association studies was associated with CRC, CA and OS by ASSET (P=0.007).Conclusion:Our data indicate that genetic variants in CDKN2A, and possibly nearby genes, may be associated with ESCC and several other tumours, further highlighting the importance of 9p21.3 genetic variants in carcinogenesis.

A Meta‐Analysis of the Effects of Common Management Actions on the Nest Success of North American Birds

Management strategies for the recovery of declining bird populations often must be made without sufficient data to predict the outcome of proposed actions or sufficient time and resources necessary to collect these data. We quantitatively reviewed studies of bird management in Canada and the United States to evaluate the relative efficacy of 4 common management interventions and to determine variables associated with their success. We compared how livestock exclusion, prescribed burning, removal of predators, and removal of cowbirds (Molothrus ater) affect bird nest success and used meta-regression to evaluate the influence of species and study-specific covariates on management outcomes. On average, all 4 management interventions increased nest success. When common species and threatened, endangered, or declining species (as defined by long-term trend data from the North American Breeding Bird Survey) were analyzed together, predator removal was the most effective management option. The difference in mean nest success between treatment and control plots in predator-removal experiments was more than twice that of either livestock exclusion or prescribed burning. However, when we considered management outcomes from only threatened, endangered, or declining species, livestock exclusions resulted in the greatest mean increase in nest success, more than twice that of the 3 other treatments. Our meta-regression results indicated that between-species variation accounted for approximately 86%, 40%, 35%, and 7% of the overall variation in the results of livestock-exclusion, prescribed-burn, predator-removal, and cowbird-removal studies, respectively. However, the covariates we tested explained significant variation only in outcomes among prescribed-burn studies. The difference in nest success between burned and unburned plots displayed a significant, positive trend in association with time since fire and was significantly larger in grasslands than in woodlands. Our results highlight the importance of comparative studies on management effects in developing efficient and effective conservation strategies.