Abstract P548: Epigenome-Wide Association Meta-Analysis of Obesity in African Americans

Abstract

Background: Despite considerable advances in obesity genetics, much remains to be learned regarding the pathway from genes to obesity development. Prior studies have shown that that 40-70% of the variation in body mass index (BMI), a measure of obesity, is due to heritable factors, yet BMI-identified genetic variants only account for a limited portion of the heritability. Research suggests epigenetic variation, which reflects both genetic and environmental influence, may account for some of the missing heritability. However, such studies have largely been conducted in primarily European ancestry populations, despite the disproportionate burden of obesity in African Americans (AAs). Methods: To identify CpG sites associated with BMI in AAs, we conducted an epigenome-wide association study (EWAS) using participants from the Jackson Heart Study (JHS, n=1604) and a subset of AA participants from the Multi-Ethnic Study of Atherosclerosis (MESA, n=179) using the Illumina HumanMethylationEPIC array (~850K CpG sites). Analyses were stratified by study and sex and meta-analyzed using an inverse-variance weighted approach. EWAS was performed using a linear regression model with methylation M-values (n=573,388 CpGs) as the outcome and continuous BMI as the predictor. Covariates in the linear models included age, centered-mean-age-squared, smoking status (never/past/current), 10 genetic ancestry principal components, and Houseman-estimated blood cell proportions. Results: The meta-analysis resulted in 204 CpGs that reached epigenome-wide significance (p < 8.72 x 10 -8 ) after multiple-testing correction, and these include both previously identified and novel BMI-associated sites. 48 of the 204 have been previously reported in EWAS of obesity phenotypes, 11 of which were in studies of AAs. 156 of the methylation sites significantly associated with BMI in this study are novel, and potentially AA-specific. Differentially methylated region (DMR) analysis showed significant association of these CpGs in the promoters of genes previously associated with obesity traits, such as SOCS3, ABCG1, and TGFB1 , although TGFB1 had not previously been associated in AAs. In addition, the DMR analysis revealed novel obesity-associated CpGs in genes, including JAK3, which has previously been shown to have differential gene expression in association with BMI in MESA. Conclusion: We identified novel, potentially AA-specific, methylation sites associated with BMI in the largest African-ancestry EWAS conducted thus far, supporting that continued analysis in diverse populations is crucial for increased understanding of obesity.