Sensitivity and specificity of prostate-specific antigen for prostate cancer detection with high rates of biopsy verification

In this article we explain findings from the Prostate Cancer Prevention Trial (PCPT) concerning the operating characteristics of PSA for biopsy-detectable prostate cancer, with special emphasis on a subpopulation of men with PSA less than 4 ng/ml, what is often regarded as the "normal" level of PSA in healthy men. The PCPT enrolled 18,882 healthy men 55 years of age or older, with a PSA value less than 3 ng/mL and a normal digital rectal exam (DRE); 9,459 of these men were randomized to the placebo arm and 9,423 to the finasteride arm In this report we summarize the operating characteristics of PSA only for the placebo arm of the PCPT; operating characteristics of PSA on the finasteride arm are more complicated to assess since finasteride approximately halves the PSA value (4) and will be reported only briefly. In our first analysis, we focused on a group of 2,950 men on the placebo arm who had had an end-of-study biopsy and a normal DRE and PSA < 4 ng/mL for all 7 years of the study. For prostate cancer, the standard PSA cut-off of 4 ng/mL has low sensitivity: with this cut-off only 20.5% of the prostate cancer cases test positive-nearly 80% of prostate cancer cases are missed. The specificity at this cut-off is high (93.6%) meaning only 6.2% of men who do not have prostate cancer falsely test positive. Lowering the PSA threshold for screening increases detection of aggressive cancer at an earlier stage, but has the unavoidable tradeoff of increased detection of the biologically irrelevant cancers.