Multi-omics Risk Prediction of Coronary Artery Disease Using Polygenic Scores and Metabolomics

Background: Coronary artery disease (CAD) remains a major global health burden and diagnostic challenge. Improved tools for estimating disease risk and refining patient stratification are of substantial clinical value. Genomics and metabolomics provide complementary insights into genetic predisposition and downstream metabolic processes that may enhance disease prediction in this patient group.<br/><br/>Methods: This study included 4062 patients with symptoms suggestive of CAD referred for diagnostic testing by coronary computed tomography angiography (CCTA). Genetic profiles based on common genetic variants were used to calculate polygenic scores, while nuclear magnetic resonance (NMR) metabolomics quantified circulating metabolites, with extensive quality control applied to reduce technical variation. Presence of CAD was defined as &gt;50% diameter reduction in a major vessel on CCTA. Risk prediction models integrating polygenic scores, metabolomic features, and clinical risk factors were constructed using penalized regression (glmnet) with repeated fivefold cross-validation.<br/><br/>Preliminary results and significance: This study aims to identify genetic and metabolomic biomarkers of CAD and integrate them into predictive models. These models will be benchmarked against established clinical risk tools to assess the added value of genetic and metabolomic data for CAD risk assessment. Demonstrating incremental predictive performance could support more precise diagnosis, reduce unnecessary testing, and lower healthcare costs.<br/>