Incorporation of polygenic risk scores with PSA testing for prostate cancer risk stratification in men undergoing prostate biopsy

Purpose: While prostate-specific antigen (PSA) screening reduces prostate cancer (PC) mortality, limited discriminatory ability leads to overdiagnosis of indolent disease and missed clinically significant PC. Genetically informed screening strategies utilizing polygenic scores may improve early PC detection. We hypothesized that incorporating polygenic scores for PC risk (PHS601) and benign elevated PSA (PRS447) with traditional risk factors would improve prediction of clinically significant PC on prostate biopsy in a diverse population. Methods: We identified veterans who underwent their first prostate biopsy from October 1999 to September 2021 using the VA Corporate Data Warehouse, with genetic data from the Million Veteran Program. We excluded patients with prior PC diagnosis, those without pre-biopsy PSA within 90 days, and PSA levels >50 ng/mL. The cohort included 25,222 men: 19,376 (76.8%) White, 4,891 (19.4%) Black or African American, and 955 (3.8%) other race/ethnicities. The primary outcome was clinically significant PC (Gleason score ≥7). We evaluated two genetic scores (PHS601 and PSA447, both standardized to z-scores) along with pre-biopsy PSA, age at first biopsy, race/ethnicity, and smoking status. We performed multivariable logistic regression using three progressive models: clinical features alone, clinical features plus PHS601, and clinical features plus both genetic scores. Results: Among 25,222 patients, 8,319 (32.9%) were diagnosed with clinically significant PC. Both genetic scores showed strong independent associations: elevated PHS601 increased risk (OR=1.73, 95% CI: 1.68-1.79, P<0.001), while higher PSA447 decreased risk (OR=0.69, 95% CI: 0.67-0.72, P<0.001). At PSA of 4 ng/mL, predicted probabilities of clinically significant cancer ranged from 9.1% to 56.9% across PHS601 percentile groups (≤2nd to ≥98th). Model discrimination improved from 0.635 (clinical features alone) to 0.714 (plus PHS601) to 0.737 (both genetic scores, p<0.001). Conclusion: Polygenic scores for PC risk and benign PSA elevation significantly improve prediction of clinically significant PC on biopsy compared to traditional risk factors alone. These findings support genetically informed PC screening strategies. Men with low genetic PC risk could defer biopsy until higher PSA values, reducing overdiagnosis, while those with high genetic risk could undergo biopsy at lower PSA thresholds, reducing missed clinically significant PC.