Development and validation of a trans‑ancestry polygenic risk score for coronary artery disease in the chinese population

Trans-ancestry polygenic risk scores (PRSs) have been proposed to address the limited transferability of European-specific PRSs to East Asian populations. However, their ability to improve risk prediction for coronary artery disease (CAD) in the Chinese population remains unclear. We aimed to develop and validate a trans-ancestry CAD PRS in the Chinese population by integrating European and East Asian genome-wide association studies (GWAS). Using summary statistics from large-scale CAD GWAS meta-analyses of East Asian and European populations, we constructed trans-ancestry PRSs via PRS-CSx in two configurations: PRS-meta, derived exclusively from GWAS results and linkage disequilibrium reference panels; and PRS-combine, optimized in a Chinese training set comprising 2,800 CAD cases and 2,055 controls. For comparison, we generated East Asian-specific PRSs using LDpred and pruning-and-thresholding methods, and additionally evaluated previously published CAD PRSs. The top-performing PRSs for each method in the training set were then externally validated in 72,782 individuals with genetic data from the China Kadoorie Biobank (CKB), a large-scale population-based cohort in China, with a mean follow-up of 12.1 years. In the training set, PRS-meta and PRS-combine exhibited comparable predictive performance, with odds ratios per standard deviation (SD) of 2.21 (95% confidence interval [CI]: 2.04, 2.38) and 2.23 (95% CI: 2.06, 2.41), respectively. Both trans-ancestry PRSs showed approximately 60% and 90% increases in pseudo-R 2 relative to the top-performing East Asian-specific and European-specific PRSs, respectively. In the CKB cohort, both scores were similarly associated with incident CAD, with hazard ratios (HRs) per SD of 1.33 (95% CI: 1.27, 1.41) for PRS-meta and 1.31 (95% CI: 1.24, 1.38) for PRS-combine. Individuals in the top versus bottom quintile had HRs of 2.06 (95% CI: 1.75, 2.42) for PRS-meta and 1.84 (95% CI: 1.56, 2.16) for PRS-combine. The trans-ancestry PRS constructed via PRS-CSx markedly improved CAD risk prediction in the Chinese population. This finding highlights an efficient and generalizable approach to developing more equitable PRSs for underrepresented populations, with PRS-meta demonstrating particular promise.