Polygenic risk scores improve CAD risk prediction in individuals at borderline and intermediate clinical risk
Dariusz Ratman, Placede Tshiaba, Michael G. Levin, Jiayi Sun, Tate Tunstall, Robert J. Maier, Premal Shah, Matthew Rabinowitz, Daniel J. Rader, Akash Kumar, Kate M. Im
Abstract
Polygenic risk scores (PRSs) can improve clinical risk tools for coronary artery disease (CAD). This study assessed a risk model integrating PRS across populations, focusing on individuals with borderline/intermediate clinical risk. We developed ancestry-specific ensemble models combining multi-ancestry PRSs for CAD and type 2 diabetes. The cross-ancestry PRS (caPRS) was integrated with the Pooled Cohort Equations (PCE) to derive the cross-ancestry Integrated Risk Score (caIRS), estimating 10-year CAD risk. The caIRS outperformed the PCE across four cohorts, including UK Biobank and Penn Medicine Biobank, with significant improvements for Hispanic and South Asian individuals. For those at borderline/intermediate PCE risk (5-20%), the caIRS reclassified between 7.0% and 10.7% into the high-risk group, which had higher CAD incidence and hazard ratios ranging from 3.20 to 3.84. The CAD caIRS, combining genetic and clinical factors, enhances high-risk CAD identification across diverse populations, potentially improving treatment guidance.