1709-P: Assessing the Role of 211 “Established” Type 2 Diabetes Variants in American Indians

Abstract

Genome-wide association studies (GWAS) have identified many variants associated with type 2 diabetes (T2D), but there is little information on their effects in high risk populations such as American Indians. We analyzed 211 primary variants, associated with T2D (P<5.0×10-8) in published GWAS, in 7659 American Indian participants in a population study (33.5% with T2D)- updating our previous study of 63 variants. Genotypes came from a custom Axiom array that captures common variation in American Indians; imputation was performed using whole genome sequence data from 266 Pima Indians. Variants were tested for association with T2D in a mixed model accounting for relatedness with adjustment for age, sex and the 1st 5 genetic principal components to account for population stratification. Heterogeneity was tested by comparing odds ratios (ORs) in American Indians with published ORs. A weighted polygenic risk score across all 211 variants was analyzed. Nominally significant (P<0.05) and directionally consistent replication was observed for 21 of the 211 variants. Notable associations were with rs2237895 in KCNQ1 (OR=1.29, P=2.5×10-8) and rs4929965 in INS/IGF2 (OR=1.28, P=9.8×10-8). Significant heterogeneity was seen for 23 variants- 21 had weaker effects in American Indians than in published values. A summary test over all variants showed effects were generally weaker in American Indians (P=4.0×10-11). The polygenic risk score was strongly associated with T2D (OR=1.38 per SD, P=2.8×10-22). The risk score was also associated with reduced insulin secretion measured by a 25 g intravenous glucose tolerance test (by 10% per SD, P=0.005, n=300 all with normal glucose tolerance), but not insulin sensitivity measured by a hyperinsulinemic-euglycemic “clamp” (P=0.78, n=557 all nondiabetic). These findings suggest that established T2D risk variants generally also affect T2D risk in American Indians (although effects tend to be weaker) and that they act predominantly through diminished insulin secretion. Disclosure R.L. Hanson: None. L.E. Wedekind: None. W. Hsueh: None. S. Kobes: None. L.J. Baier: None. C. Bogardus: None. W.C. Knowler: None.