Divergent genetic mediator signatures of adiposity-related metabolic liabilities and coronary artery disease.

Background. Obesity and leanness lie at opposite ends of the weight spectrum, yet both carry coronary risk when metabolically dysfunctional. A 2025 Commission redefined obesity around organ function rather than body mass, but whether that distinction reflects a difference in causal architecture, and whether the two high-risk phenotypes reach disease by distinct routes, is untested. Methods. We conducted two-sample Mendelian randomization of the four combinations of adiposity and metabolic health against coronary artery disease (CARDIoGRAMplusC4D; 60,801 cases), using published adiposity and insulin-resistance instruments. Thirty mediators were interrogated identically in every quadrant by two-step Mendelian randomization with multivariable adjustment, and the pipeline replicated with an independent instrument sharing five of fifty-three variants. Results. Adiposity with preserved metabolic function was protective (odds ratio 0.28, 95% CI 0.19?0.42), whereas adiposity with metabolic dysfunction (1.90, 1.42?2.55) and leanness with metabolic dysfunction (1.86, 1.52?2.27) conferred indistinguishable risk (ratio of odds ratios 1.02; P = 0.91). The unhealthy obese phenotype acted through triglycerides and diabetes; the unhealthy lean through apolipoprotein B, blood pressure and diabetes, sharing no other conduit. High-density lipoprotein cholesterol and triglycerides collapsed despite being the most strongly instrumented exposures, indicating causal redundancy rather than weak-instrument bias. Conclusions. Coronary risk follows the failure of adipose storage capacity, not its magnitude. The two high-risk phenotypes reach the same endpoint by divergent causal architectures that reproduce in full with an independent instrument, and the metabolically unhealthy lean phenotype carries risk equal to clinical obesity while falling outside a framework that requires excess adiposity to be confirmed first.