Prostate cancer risk prediction using polygenic hazard scores in Norwegian populations

Abstract More accurate risk prediction is needed for prostate cancer (PCa) to identify individuals at greatest risk of early-onset and clinically significant disease. Current screening paradigms, such as prostate-specific antigen screening, pose risks due to over-diagnosis and over-treatment of indolent disease. Polygenic hazard scores (PHS) can predict age-at-diagnosis of PCa and are being tested in prospective clinical screening trials. We assessed the performance of the latest PHS model for PCa (PHS601) in two Norwegian population-based cohort studies (N = 14,688 and N = 2,850) for predicting age-at-diagnosis of PCa and aggressive PCa. In a subset with whole-genome sequencing (N = 503), we directly compared PHS601-based stratification with screening for rare pathogenic variants. PHS601 effectively stratified participants by risk in both cohorts for both PCa (HR 80/20 = 5.74, 95% CI [5.06, 6.45] and 7.79 95% CI [5.70, 10.59]) and aggressive PCa (HR 80/20 = 4.60, 95% CI [3.19, 6.45] and 3.14 95% CI [1.49, 6.63]). Among individuals with whole-genome sequencing, the top 1.8% of PHS values conferred 8.8-fold higher risk than the median (HR = 8.78 [5.00, 14.40]), exceeding the risk associated with HOXB13 pathogenic variants (HR = 3.77 [1.75, 8.11]; 1.8% carrier frequency). This study provides the first external validation of PHS601 in independent Norwegian population-based cohorts and, within a WGS subset, a direct comparison with rare variant screening, supporting context-specific use of genotyping-based risk stratification in population screening.