Abstract
Standard polygenic risk scores capture overall genetic liability to coronary artery disease (CAD) but do not reveal the biological processes contributing to risk in an individual. Here, we applied data-driven latent factor decomposition to 526 genome-wide significant CAD-associated variants, using their associations with 78 CAD-related phenotypic traits. We identified eight genetic components corresponding to cholesterol-enriched lipoproteins, triglyceride-rich lipoproteins, lipoprotein(a), metabolically unhealthy normal weight, obesity, endothelial dysfunction, coagulation and cardiac remodeling. Process-specific polygenic risk scores (pPRSs) captured biological profiles beyond those reflected by conventional risk factors. In UK Biobank (n = 458,905), the Swedish CArdioPulmonary bioImage Study (n = 24,904) and Mass General Brigham Biobank (n = 53,848), pPRSs were consistently associated with clinical and subclinical CAD and showed distinct biomarker, proteomic and comorbidity profiles. These findings establish a framework for biologically resolved genetic CAD risk and may inform process-specific approaches to precision prevention.