Risk Prediction of Prostate Cancer with Single Nucleotide Polymorphisms (SNPs) and Prostate-Specific Antigen (PSA)

Abstract

PurposeCombined information on single nucleotide polymorphisms (SNPs) and prostate-specific antigen (PSA) offers opportunities for improving the performance of screening by risk stratification. We aimed to predict the risk of prostate cancer (PrCa) based on PSA together with SNPs information. Materials and MethodsProspective study of 20,575 men with PSA test and 4,967 men with polygenic risk score for PrCa based on 66 SNPs from the Finnish population-based screening trial for PrCa and 5,269 samples on seven SNPs from the Finnish PrCa DNA study. Bayesian predictive model was built for estimating the risk of PrCa by sequentially combining genetic information with PSA in comparison with PSA alone among study subjects limited with 4 ng/mL or above. ResultsThe posterior odds for PrCa based on seven SNPs together with the PSA level ranged from 3.7 at 4 ng/mL, 14.2 at 6 ng/mL, 40.7 at 8 ng/mL, to 98.2 at 10 ng/mL. The area under receiver operating characteristic curve was elevated to 88.8% (95% CI: 88.6%-89.1%) with PSA in combination with the risk score based on seven SNPs in comparison with 70.1% (95% CI: 69.6%-70.7%) with PSA alone. It was further escalated to 96.7% (95% CI: 96.5%-96.9%) when all prostate cancer susceptibility polygenes were combined. ConclusionsExpedient use of multiple genetic variants together with information on PSA levels better predicts the risk of PrCa than PSA alone and allows higher PSA cut-offs. Combined information also provides a basis for risk stratification that can be used for optimizing the performance of PrCa screening.