Multi-ethnic genome-wide association study identifies novel locus for type 2 diabetes susceptibility

Abstract

Genome-wide association studies (GWAS) have traditionally been undertaken in homogeneous populations from the same ancestry group. However, with the increasing availability of GWAS in large-scale multi-ethnic cohorts, we have evaluated a framework for detecting association of genetic variants with complex traits, allowing for population structure, and developed a powerful test of heterogeneity in allelic effects between ancestry groups. We have applied the methodology to identify and characterise loci associated with susceptibility to type 2 diabetes (T2D) using GWAS data from the Resource for Genetic Epidemiology on Adult Health and Aging, a large multi-ethnic population-based cohort, created for investigating the genetic and environmental basis of age-related diseases. We identified a novel locus for T2D susceptibility at genome-wide significance (P<5 × 10(-8)) that maps to TOMM40-APOE, a region previously implicated in lipid metabolism and Alzheimer's disease. We have also confirmed previous reports that single-nucleotide polymorphisms at the TCF7L2 locus demonstrate the greatest extent of heterogeneity in allelic effects between ethnic groups, with the lowest risk observed in populations of East Asian ancestry.

Figure 1

Power to detect association (at genome-wide significance, P<5 × 10⁻⁸) as a function of the allelic effect size for: (i) the logistic regression model after adjustment for 10 AGV as covariates; and (ii) fixed-effects meta-analysis of summary statistics across populations via inverse-variance weighting of effect sizes. The four panels correspond to alternative models of heterogeneity of allelic effects between ancestry groups (defined in Supplementary Table S1). Within each panel, results are presented for extreme, moderate and no population structure (defined in Supplementary Table S2).

Figure 2

Power to detect heterogeneity in allelic effects between ancestry groups (at genome-wide significance, P<5 × 10⁻⁸), as a function of allelic effect size for: (i) the logistic regression model by including interactions of the variant with the first two AGV (and with adjustment for the first 10 as covariates); and (ii) Cochran's Q statistic from fixed-effects meta-analysis of summary statistics across populations via inverse-variance weighting of effect sizes. The four panels correspond to alternative models of heterogeneity of allelic effects between ancestry groups (defined in Supplementary Table S1). Within each panel, results are presented for extreme, moderate and no population structure (defined in Supplementary Table S2).