Progress in Cardiovascular Diseases · Nov 7, 2026
Integrating polygenic and classical risk factors for cardiovascular risk stratification in a polish primary-care cohort
By Banach M, Stempor P, Osadnik T et al.
Background Polygenic risk scores (PRS) promise improved cardiovascular risk stratification, but most validation derives from UK or Nordic biobanks. The transferability of these scores to Central European populations and their incremental value over classical risk factors in routine care remain unestablished. Objectives To determine whether a coronary PRS derived from external European data transfers to a Polish primary-care population and adds discrimination for prevalent cardiovascular disease beyond classical risk factors. Methods In the nationwide Polish LIPIDOGEN2015 primary-care cohort, we derived PRS for coronary heart disease, myocardial infarction (MI) and stroke from FinnGen data with continuous shrinkage priors and applied them to imputed genotypes. We compared allele frequencies of 1397 unrelated participants with a non-Finnish European (NFE) reference and, in 1328 participants with clinical data, related the PRS to coronary artery disease (CAD), MI and stroke recorded at baseline and to a composite including death during follow-up, with and without classical risk factors or SCORE2. Results Allele frequencies correlated with the NFE reference (r = 0.97). Per standard deviation (SD), the coronary PRS was associated with prevalent CAD (odds ratio 1.36, 95% CI 1.11 to 1.67) and the composite endpoint (1.26, 95% CI 1.04 to 1.51), as were two validated genome-wide scores. Adding it to classical risk factors changed the area under the curve (AUC) for the composite endpoint by 0.0017 (95% CI −0.0005 to 0.0076), from 0.860 to 0.861; results with SCORE2 were similar. The study had 80% power for odds ratios per SD of 1.31 (CAD) and 1.27 (composite). Conclusions A coronary PRS derived from external European data retains its association with prevalent CAD without re-derivation in this Polish primary-care cohort but, in this cross-sectional, event-limited sample, adds little discrimination to classical risk factors. Its value for predicting future events in this setting remains to be tested against incident outcomes.