Epigenome-Wide Association Study of Fasting Measures of Glucose, Insulin, and HOMA-IR in the Genetics of Lipid Lowering Drugs and Diet Network Study

Bertha Hidalgo,M. Ryan Irvin,Jin Sha,Degui Zhi,Stella Aslibekyan,Devin Absher,Hemant K. Tiwari,Edmond K. Kabagambe,Jose M. Ordovas,Donna K. Arnett

DOI: https://doi.org/10.2337/db13-1100

IF: 7.7

2014-01-16

Diabetes

Abstract:Known genetic susceptibility loci for type 2 diabetes (T2D) explain only a small proportion of heritable T2D risk. We hypothesize that DNA methylation patterns may contribute to variation in diabetes-related risk factors, and this epigenetic variation across the genome can contribute to the missing heritability in T2D and related metabolic traits. We conducted an epigenome-wide association study for fasting glucose, insulin, and homeostasis model assessment of insulin resistance (HOMA-IR) among 837 nondiabetic participants in the Genetics of Lipid Lowering Drugs and Diet Network study, divided into discovery (N = 544) and replication (N = 293) stages. Cytosine guanine dinucleotide (CpG) methylation at ∼470,000 CpG sites was assayed in CD4+ T cells using the Illumina Infinium HumanMethylation 450 Beadchip. We fit a mixed model with the methylation status of each CpG as the dependent variable, adjusting for age, sex, study site, and T-cell purity as fixed-effects and family structure as a random-effect. A Bonferroni corrected P value of 1.1 × 10−7 was considered significant in the discovery stage. Significant associations were tested in the replication stage using identical models. Methylation of a CpG site in ABCG1 on chromosome 21 was significantly associated with insulin (P = 1.83 × 10−7) and HOMA-IR (P = 1.60 × 10−9). Another site in the same gene was significant for HOMA-IR and of borderline significance for insulin (P = 1.29 × 10−7 and P = 3.36 × 10−6, respectively). Associations with the top two signals replicated for insulin and HOMA-IR (P = 5.75 × 10−3 and P = 3.35 × 10−2, respectively). Our findings suggest that methylation of a CpG site within ABCG1 is associated with fasting insulin and merits further evaluation as a novel disease risk marker.

endocrinology & metabolism