Background and Objective:
Polygenic scores (PGS) for coronary artery disease (CAD) measure an individual’s genetic liability for CAD. Multiple PGSs for CAD have been published and whilst they are generally considered predictive in individuals of European ancestry, performance is attenuated in non-European ancestry groups (AG). Here we collate publicly available CAD PGS and benchmark their performance across several AGs to evaluate their utility for personalized prevention.

Methods:
We queried the “PGS Catalog” to extract standardized odds/hazard ratios (OR/HR) for published CAD PGSs across multiple AGs (European, African, Hispanic, South-Asian, East-Asian, Greater-Middle-Eastern). We restricted our analysis to PGSs specifically developed for prevalent and/or incident CAD and analyzed AG-specific PGS performance to identify the five best-performing PGSs in each AG to include in a comparative analysis.

Results:
In general, PGS applied to the European AG had the highest OR/HRs per standard deviation of the PGS (min: 1.74[1.61-1.89]; max: 1.89[1.75-2.03]), while the African AG had the lowest performances overall (min: 1.05[0.94-1.17]; max: 1.40[1.30-1.52]). PGSs applied to Hispanic, East-Asian and Greater-Middle-Eastern AGs exhibited highly heterogeneous performance overall, although their absolute best performance was comparable to those in Europeans with OR/HRs of 1.93[1.67-2.22], 1.84[1.74-1.94] and 1.81[1.66-1.98] respectively. PGS000018 demonstrated the best performance in both Africans (1.40[1.03-1.52]) and Hispanics (1.93[1.67-2.22]), while PGS000013 was the best performing PGS in Europeans (1.89[1.75-2.03]) and ranked second in Hispanics (1.52[1.43-1.62]), East-Asians (1.66[1.47-1.86]) and South-Asians (1.58[1.42-1.76]). Furthermore, PGS000337 was best in East-Asian (1.84[1.74-1.94]) and Greater-Middle-Eastern AGs (1.81[1.66-1.98]), while PGS000296 performed best in South-Asians (1.66[1.53-1.81]).

Conclusion:
There is currently no gold-standard CAD PGS that can be applied with equal predictive performance across all AGs. However, performance of CAD PGS in different AGs can be optimized by applying different ancestry-specific CAD PGSs to achieve the best possible predictive performance in every individual, an essential prerequisite towards implementing PGSs in CAD personalized prevention protocols.