Multi-marker risk-based screening for prostate cancer

Abstract

Objective To determine prostate cancer screening performance using prostate specific antigen (PSA) along with other markers, expressing markers in age-specific multiples of the median (MoM), and age. Methods A prospective nested case-control study used stored serum from 571 men who died of, or with history of, prostate cancer (cases), and 2169 matched controls. Total, free and intact PSA, human kallikrein-related peptidase 2 (hK2), and microseminoprotein were measured and converted into MoM values. Screening marker distribution parameters were estimated in cases and controls. Monte Carlo simulation used these in a risk-based algorithm to estimate screening performance (detection rates [DRs] and false-positive rates [FPRs]). Results Almost all (99%) cases occurred aged ≥55. Marker values were similar in cases who did and did not die of prostate cancer. Combining age, total PSA and hK2 MoM values (other markers added little or no discrimination) yielded a 1.2% FPR (95% CI 0.2–4.8%) for a 90% DR (59–98%) in men who died of or with a prostate cancer diagnosis within 5 years of blood collection (risk cut-off 1 in 20), two-thirds less than the 4.5% FPR using total PSA alone measured in ng/ml for the same 90% DR (cut-off 3.1 ng/ml). Screening performance over 10 years yielded a 33% (22–46%) FPR for a 90% DR. Conclusion Screening performed up to every 5 years from age 55 using the multi-marker risk-based screening algorithm for future prostate cancer achieves a high DR and a much lower FPR than using PSA alone, resulting in reductions in overdiagnosis and overtreatment.