Distinct metabolic features of genetic liability to type 2 diabetes and coronary artery disease: a reverse Mendelian randomization study

Type 2 diabetes (T2D) and coronary artery disease (CAD) both have roots in disordered metabolism but the mechanisms through which their associated genetic variants lead to disease onset remain poorly understood. We performed two-sample reverse Mendelian randomization (MR) to estimate effects of genetic liability to T2D and CAD on 249 circulating metabolites from targeted nuclear magnetic resonance spectroscopy in the UK Biobank (N=118,466). We examined the potential for medication use to distort effect estimates by examining effects of disease liability on metformin and statin use and conducting age-stratified metabolite analyses. Using inverse variance weighted models, higher genetic liability to T2D was estimated to increase triglycerides and branched chain amino acids (BCAAs). Estimates for CAD liability suggested an effect on reducing high-density lipoprotein cholesterol (HDL-C) as well as raising levels of very-low-density lipoprotein cholesterol, low-density lipoprotein cholesterol (LDL-C), LDL triglycerides, and apolipoprotein-B. In pleiotropy-robust sensitivity models, T2D liability was still estimated to increase BCAAs, but several effect estimates for higher CAD liability were reversed and estimated to decrease LDL-C and apolipoprotein-B. Higher T2D liability increased the odds of using metformin, whereas higher CAD liability increased the odds of using statins. Estimated effects of CAD liability differed substantially by age tertile for non-HDL-C traits, e.g., pleiotropy-robust models suggested that higher CAD liability lowers LDL-C only at older ages when use of statins is common. Our results support largely distinct metabolic features of genetic liability to T2D and to CAD, and substantial effect modification by medication use in adulthood for CAD in particular.