Fig. 4. External validation of GPSMult and benchmarking against published polygenic scores for CAD across multiple ancestries in Million Veteran Program and Genes & Health studies.
The OR/SD with 95% CI for prevalent CAD risk was assessed for each polygenic score in a logistic regression model adjusted for age, sex, genotyping array and the first ten principal components of ancestry in the same group of individuals per cohort: n = 33,096 independent African ancestry individuals in the Million Veteran Program; n = 124,467 independent European ancestry individuals in the Million Veteran Program; n = 16,433 independent Hispanic ancestry individuals in the Million Veteran Program; n = 16,874 independent South Asian ancestry individuals in the Genes & Health Study, using high-performing published scores from the Polygenic Score Catalog (GPS2018 (ref. 9), metaGRS8, metaPRSCAD67, AnnoPredCAD68, PRSCSCHD69 and PRS2022 (ref. 27), as well as GPSMult28. Results for these and additional CAD polygenic scores published in the Polygenic Score Catalog are available in Supplementary Tables 6 and 7.