Dissecting the Genetic Link Between Insulin Resistance and Coronary Heart Disease: A Multi-Trait GWAS and Proteomic-Metabolomic Mediation Analysis

Abstract

Insulin resistance (IR) is a hallmark feature of Type 2 Diabetes Mellitus and a well-established risk factor for coronary heart disease (CHD), yet the pathways through which IR contributes to CHD remain incompletely understood. To comprehensively describe the genetic architecture of IR, we collected three direct IR indicators--- fasting insulin (N = 151,013), the Modified Stumvoll insulin sensitivity index (N = 53,657), and pro-insulin (N = 45,861), and conducted joint analysis on them using Linkage Disequilibrium (LD) Score Regression and factor analysis. A single latent IR factor was generated which captures 60.5% of the shared genetic variance. Based on the IR factor, a multivariate GWAS was conducted within Genomic Structural Equation Modelling and then we built a polygenic risk score (PRS) of IR for the UK Biobank European cohort (N = 407,767; 55,729 CHD cases and 352,038 controls). Then we performed the association study to quantify the relationship between genetically predicted IR and CHD. Causal mediation analyses were also performed through 2,923 plasma proteins (Olink) and 168 metabolites (Nightingale NMR), followed by serial-mediation models (IR -> protein -> metabolite -> CHD) over all prescreened pairs. Single-mediator screening nominated 938 proteins and 159 metabolites; this yielded 3,669 significant "protein -> metabolite -> CHD" serial-mediation pathways, from which 33 broadly-acting core mediating proteins were prioritized. These acted predominantly through the large and very large high-density lipoprotein (HDL) particle subclasses: 30 of the 33 core proteins and all 14 HDL-subclass metabolites formed 306 pathways, of which 303 (99.0%) amplify the CHD risk. These results map a proteomic and metabolomic axis linking the genetic component of IR to CHD and highlight HDL lipoprotein-subclass biology as a candidate therapeutic and biomarker space.