Polygenic score informed by genome-wide association studies of multiple ancestries and related traits improves risk prediction for coronary artery disease

Abstract

Accurate stratification of coronary artery disease (CAD) risk remains a critical need. A new polygenic score (GPSMult) incorporates CAD genome-wide association data across five ancestries (>269,000 cases, >1,178,000 controls) with genetic association data for ten CAD risk factors. GPSMultassociates with an OR/SD 2.14, (95%CI:2.10-2.19,P<0.001) for prevalent CAD and HR/SD 1.73 (95%CI 1.70-1.76,P<0.001) for incident CAD. When compared with the previously published GPS2018in external datasets, GPSMultdemonstrated 73%, 46%, and 113% increase in effect size for individuals of African, European, and South Asian ancestry, respectively, and significantly outperformed recently published CAD polygenic scores. GPSMultidentifies individuals with CAD risk extremes, including the top 3% of the population at equivalent risk for a new CAD event as those with prior CAD having a second event. Integrating GPSMultwith the Pooled Cohort Equations results in 7.0% [95%CI:5.9%-8.2%,P<0.001] net reclassification improvement at the 7.5% threshold. Large-scale integration genetic association data for CAD and related traits from diverse populations meaningfully improves polygenic risk prediction.

Polygenic Background Modifies Risk of Coronary Artery Disease Among Individuals With Heterozygous Familial Hypercholesterolemia

“…A previously published genome-wide polygenic score (GPS CADEUR ) of 1.2 million common DNA variants was calculated in participants of both the Dutch screening program and the UK Biobank. 20 This score was derived from a recent genome-wide association study led by the CARDIOGRAMplusC4D Consortium that studied over 1 million individuals. 21 In brief, a range of tuning parameters within the LDPred2 algorithm were used to select the best-performing polygenic score among 116,649 individuals (4,412 CAD cases and 112,237 controls) of European ancestry in the UK Biobank study.…”

Lipoprotein(a) Concentrations Are Independent of Polygenic Score for Coronary Artery Disease

“…A recently developed CAD polygenic score, CAD GPS Mult , was constructed using LDPred2, incorporating the weighted effects of >1.2 million single nucleotide polymorphisms from 58 genome-wide association studies for CAD (>222,000 cases and >914,000 controls), other atherosclerotic diseases, and their risk factors from multiancestry cohorts, external to the UK biobank, and calculated in a holdout population of individuals not included in score training. 4 CAD GPS Mult outperforms other published scores for CAD in external validation datasets. The CAD GPS Mult score was residualized for the first 10 principal components of genetic ancestry and then scaled to a mean of 0 and SD of 1 for each ancestral group (ie, African, East Asian, European, and South Asian) for our analysis.…”

“… 2 Moreover, single nucleotide polymorphisms in the gene encoding Lp(a), LPA , are very strong predictors of high Lp(a) 3 and are included in a new, multiancestry, genome-wide polygenic score for coronary artery disease (CAD) (CAD GPS Mult ). 4 Here, we explored how well high Lp(a) identifies individuals with a high CAD GPS Mult .…”

Enhancing the Prediction Power of Polygenic Risk Scores in Genetically Diverse Coronary Heart Disease

“…Indeed, considering the genetic contribution via PRS of concurrent diseases (diabetes and peripheral vascular disease) and also highly genetic risk factors (obesity, lipids, and blood pressure) improved the predictive power of the multiethnic PRS. 9 The integration of PRS into routine clinical practice remains a significant challenge, both for CHD and…”