Validation of a Genome-Wide Polygenic Score for Coronary Artery Disease in South Asians

Genome-wide polygenic scores (GPS) integrate information from many common DNA variants into a single number. Because rates of coronary artery disease (CAD) are substantially higher among South Asians, a GPS to identify high-risk individuals may be particularly useful in this population. We used summary statistics from a prior genome-wide association study to derive a new GPS CAD for South Asians. We validated this GPS CAD in 7,244 South Asian UK Biobank participants and tested it in 491 individuals from a case-control study in Bangladesh. Next, we built a static ancestry and GPS CAD reference distribution using whole genome sequencing from 1,522 Indian individuals, and tested a framework for projecting individuals onto this static ancestry and GPS CAD reference distribution using 1,800 CAD cases and 1,163 controls newly recruited in India. The GPS CAD , containing 6,630,150 common DNA variants, had odds ratio per standard deviation (OR/SD) of 1.58 in South Asian UK Biobank participants and 1.60 in the Bangladeshi study (p < 0.001 for each). We next projected individuals of the Indian case-control study onto static reference distributions, observing an OR/SD of 1.66 (p < 0.001). Compared to the middle quintile, risk for CAD was most pronounced for those in the top 5% of the GPS CAD distribution - ORs of 4.16, 2.46, and 3.22 in the South Asian UK Biobank, Bangladeshi, and Indian studies, respectively (p < 0.05 for each). We developed and tested a new GPS CAD using three distinct South Asian studies, and provide a generalizable framework for ancestry-specific GPS assessment.