Development of an Integrated Monogenic and Polygenic Risk Assessment Tool for Coronary Artery Disease and Its Application in a Community-Based Health Care Biobank.

Abstract

Background: Current genetic testing for coronary artery disease (CAD) primarily targets monogenic variants in individuals with severe hypercholesterolemia. Whether supplementing monogenic testing with polygenic risk scores for CAD and Lp(a; lipoprotein[a]) levels [PRS_Lp(a)] improves identification of high-risk individuals in the general population remains understudied.

Methods: A genetic probability for CAD (GenProb_CAD), incorporating monogenic pathogenic variants (PVs), polygenic risk scores for CAD, and PRS_Lp(a) was developed using Cox regression in 226 145 UK Biobank participants, validated in the remaining 226 145 UK Biobank participants, and applied to 20 477 participants of the Genomic Health Initiative, a community-based health care biobank. Predictive performance was evaluated and adjusted for clinical risk factors.

Results: In the UK Biobank development cohort, PVs, polygenic risk scores for CAD and PRS_Lp(a) were each independently associated with CAD. In the UK Biobank validation cohort, GenProb_CAD outperformed PVs and the polygenic risk score in predicting CAD and reclassified risk among PV carriers. GenProb_CAD identified 16% of participants as high risk, 46-fold more than PV carriers (0.35%), with comparable observed CAD prevalence (15.60% versus 15.43%). Nearly 50% of CAD with high-risk GenProb_CAD were premature. GenProb_CAD also independently predicted incident CAD after adjusting for clinical risk factors. Importantly, high-risk individuals defined by GenProb_CAD showed consistently elevated CAD incidence across all low-density lipoprotein cholesterol strata. When UK Biobank-derived coefficients and cutoffs were applied in the Genomic Health Initiative, GenProb_CAD substantially outperformed the PV-only strategy, identifying 1966 high-risk participants (345 developed CAD), far exceeding the 20 PV carriers detected among those with low-density lipoprotein cholesterol ≥190 mg/dL (5 developed CAD). A universal genetic testing strategy using GenProb_CAD would hypothetically identify 20.75% of all incident CAD, 69-fold higher than current monogenic testing (0.3%). Results were generally consistent across ancestry populations, although the sample size of non-European participants was limited.

Conclusions: GenProb_CAD, a novel integrated genetic risk tool combining monogenic PVs, polygenic risk scores for CAD, and PRS_Lp(a), improves identification of individuals at high genetic risk for CAD across diverse populations.