Causal Associations in Type 2 Diabetes Development

Abstract

Abstract Context Obesity, glucose, insulin resistance [homeostatic model assessment, version 2, for insulin resistance (HOMA2-IR)], and insulin secretion (HOMA2-β) have been associated with type 2 diabetes (T2D) observationally. However, the causal, genetic contribution of each parameter to this risk is largely unknown and important to study because observational data are prone to confounding but genetic, causal data are free of confounding and reverse causation. Objective We examined the causal, genetic contribution of body mass index (BMI), glucose level, C-peptide level, HOMA2-IR, and HOMA2-β to the risk of T2D in 95,540 individuals from the Copenhagen General Population Study and estimated the absolute 10-year risks. Methods Cox regression analysis, instrumental variable analysis, and Poisson regression analysis were performed to estimate the observational hazard ratios, causal, genetic ORs, and absolute 10-year risks of T2D. Results For 1-SD greater level, BMI was associated with an observational 66% (95% CI, 62% to 72%) and causal, genetic 121% (95% CI, 25% to 291%) greater risk of T2D; glucose with an observational 44% (95% CI, 41% to 46%) and causal, genetic 183% (95% CI, 56% to 416%) greater risk of T2D; and HOMA2-IR with an observational 30% (95% CI, 18% to 44%) and causal, genetic 12% (95% CI, 2% to 22%) greater risk of T2D. In contrast, for 1-SD greater level, HOMA2-β was associated with an observational 14% (95% CI, 11% to 16%) and causal, genetic 21% (95% CI, 8% to 32%) lower risk of T2D. The upper tertiles of HOMA2-IR were associated with absolute 10-year diabetes risks of 31% and 37% in obese women and men, age >60 years, and a glucose level of 6.1 to 11.0 mmol/L. Conclusions BMI, glucose level, HOMA2-IR, and HOMA2-β are causally associated with T2D.