Polygenic Risk Score

A PRS depicts a person's genetic risk relative to the average in the general population, and the relative risk derived from it can be used to multiply the absolute risk derived from a clinical risk score. ^[4] For coronary artery disease, the risk among persons in the highest 5% of the PRS distribution is 3 to 5 times that among persons with an average score. ^[4] In a meta-analysis of 979,286 individuals, each 1-standard deviation increment in PRS was associated with higher adjusted coronary artery disease risk (odds ratio 1.67 for metaGRS), and PRS predicted incident disease with a C-index of up to 0.71. ^[12] Current discriminative ability is low in the general population, and translating an individual's percentile into a lifetime disease risk is challenging, so PRS may be most useful where the prior probability of disease is higher. ^[5] Most PRS are derived from European-ancestry populations, with a weaker evidence base in other ancestries; fewer than 3% of participants in the GWAS Catalog were of African ancestry, and differences in linkage disequilibrium and SNP arrays biased to European variants reduce portability. ^{[5] [15]} For colorectal cancer, a European-ancestry PRS reached an adjusted AUC of 0.638 in a European population, but its performance in an East Asian population was limited (adjusted AUC 0.510). ^[15] Multi-ancestry approaches can narrow the gap: integrative stroke polygenic scores built from cross-ancestry and ancestry-specific GWAS strongly predicted ischaemic stroke in European, East Asian and African ancestry populations, and greater population diversity in lipid GWAS improved polygenic prediction of blood lipid levels. ^{[16] [17]}

A clinical consensus statement of the ESC Council on Cardiovascular Genomics, the ESC Cardiovascular Risk Collaboration and the European Association of Preventive Cardiology notes that European Society of Cardiology guidelines do not currently advocate PRS in routine clinical practice, although PRS are commercially available and increasingly sought. ^[7] Adding a coronary artery disease PRS to clinical risk scores gives modest but statistically significant gains in cardiovascular risk prediction for coronary artery disease: 1.5% and 1.6% in a meta-analysis, and a 1% C-statistic improvement with 3.5% net reclassification improvement in prospective Chinese cohorts. ^{[12] [18]} In atrial fibrillation, a PRS based on updated effect estimates improved risk prediction compared with the CHARGE-AF clinical risk score, and stroke genetic risk scores predicted ischaemic stroke independently of clinical risk factors in 52,600 clinical-trial participants with cardiometabolic disease. ^{[19] [16]} Incremental value is disease-dependent: for type 2 diabetes, common-variant genetic risk scores, even with hundreds of loci, only minimally improve prediction over standard clinical risk factors such as age and family history. ^[20] The strongest evidence for PRS in treatment response concerns statins for preventing a first coronary event, where the relative risk reduction has been higher in people at high genetic risk for cardiovascular disease. ^[5] Lifestyle can offset part of a high polygenic risk: in UK Biobank, the excess colorectal cancer risk of a high PRS was largely attenuated by a favorable lifestyle, with a 0.53% reduction in 5-year absolute risk, and in Chinese cohorts people at high genetic risk with a favourable lifestyle had lower gastric cancer risk than those with an unfavourable lifestyle. ^{[15] [21]} Clinical utility has yet to be established: the long-term health and economic benefits of PRS-guided management remain to be determined, and key questions remain on clinical value, cost-effectiveness and implementation. ^{[5] [6] [4]}