A genome-wide approach accounting for body mass index identifies genetic variants influencing fasting glycemic traits and insulin resistance

Abstract

Recent genome-wide association studies have described many loci implicated in type 2 diabetes (T2D) pathophysiology and β-cell dysfunction but have contributed little to the understanding of the genetic basis of insulin resistance. We hypothesized that genes implicated in insulin resistance pathways might be uncovered by accounting for differences in body mass index (BMI) and potential interactions between BMI and genetic variants. We applied a joint meta-analysis approach to test associations with fasting insulin and glucose on a genome-wide scale. We present six previously unknown loci associated with fasting insulin at P < 5 × 10(-8) in combined discovery and follow-up analyses of 52 studies comprising up to 96,496 non-diabetic individuals. Risk variants were associated with higher triglyceride and lower high-density lipoprotein (HDL) cholesterol levels, suggesting a role for these loci in insulin resistance pathways. The discovery of these loci will aid further characterization of the role of insulin resistance in T2D pathophysiology.

Figure 1

Genome-wide plots of the discovery joint meta-analysis (JMA) for fasting insulin (1a top) and fasting glucose (1b top): 17 loci with known associations were observed (red) and 50 loci were taken to the follow-up analysis (light and dark blue). Of these, 12 reached genome-wide significance in the combined discovery and follow-up JMA (dark blue). The P values of these 12 loci from the models fit in the combined discovery and follow-up analyses are presented in Figure 1a (bottom) for fasting insulin and 1b (bottom) for fasting glucose: JMA (red), main effects adjusting for BMI (orange), interaction with continuous BMI (green) and interaction with dichotomous BMI (blue). * G6PC2 JMA P value: 1.7 × 10⁻¹¹³, ** GCK JMA P value: 8.3 ×10⁻⁵⁶, *** MTNR1B JMA P value: 4.38 × 10⁻¹⁰⁵.

Figure 2

Regional plot of the COBLL1-GRB14 genomic locus. The left plot shows the discovery JMA P values in the background and for the SNP taken forward to the follow-up analyses (rs7607980), the P values of the discovery JMA (dark red) and the combined discovery and follow-up JMA (light red), main effects adjusting for BMI (orange), interaction with continuous BMI (green) and interaction with dichotomous BMI (blue). The right plot shows the beta estimates with 95% confidence interval of the increasing allele in the two BMI strata: BMI<28 and BMI≥28 kg/m². Effect = Beta estimates from regression models of ln(FI) adjusting for age, sex and other study-specific covariates.