Abstract 052: Genome-wide Methylation Study of Body Mass Index (BMI) in African American Adults: Preliminary Data from the ARIC Study

Abstract

Background and Objective: DNA methylation patterns are influenced by environmental factors, alter gene expression, and can point to genomic regions affected by behavioral and lifestyle factors, such as obesity. However, analytic strategies for high dimensional methylation data are still evolving. Here we present preliminary analyses examining cross-sectional associations between DNA methylation level and BMI in African-American adults enrolled in the Atherosclerosis Risk in Communities (ARIC) study. Methods: BMI was measured at the same study visit as the DNA sample used for analysis. We used the Illumina Infinium HumanMethylation450 BeadChip to measure average methylation levels (beta values) in bisulfite-converted peripheral blood DNA obtained from participants from the Jackson, MS and Forsyth County, NC field centers. After excluding outlier samples and CpG sites using quality control filters, a model adjusting BMI (continuous) for batch (plate) and a model additionally adjusting for a small set of potential confounders (age, sex, center, smoking) were tested, with average beta value as the dependent variable. Robust standard errors were used for statistical testing and Bonferroni-corrected p value for significance was p<1 x 10 -7 . Results: A total of 2,873 individuals and 473,788 CpG sites entered the analysis. In the minimally-adjusted analysis, there were over 300 regions distributed across all 22 autosomes with at least one significant association between BMI and CpG methylation. Adjusting for confounders sharply reduced the evidence for association to a total of approximately 20 regions on 11 autosomes. Conclusion: Use of the 450K methylation BeadChip in large epidemiologic cohorts has potential to help identify genes that are transcriptionally altered by common behavioral factors such as obesity, leading to improved understanding of related diseases. However, whereas genetic associations with disease are generally unconfounded by demographic and other covariate factors, methylation associations can be strongly affected by selection of covariates as well as other analytic choices.