In the Prostate Cancer Prevention Trial (PCPT), men receiving finasteride had a 24.8% lower risk of prostate cancer than men receiving placebo but a higher risk of high-grade cancer. We examined the impact of finasteride on the sensitivity and area under the receiver operating characteristic curve (AUC) of prostate-specific antigen (PSA) for detecting prostate cancer.Methods
We studied men in the placebo and finasteride groups of the PCPT who had a prostate biopsy and concurrent PSA tests during the 7-year study. We compared the placebo and finasteride groups for sensitivity and AUC of PSA for the detection of all prostate cancer, of Gleason grade 7 or higher prostate cancer, and of Gleason grade 8 or higher prostate cancer. All statistical tests were two-sided.Results
Of 5112 men in the placebo group, prostate cancer was detected in 1111. Gleason tumor grade was available for 1100 men, of whom 240 had grade 7 or higher and 55 had grade 8 or higher. Of 4579 men in the finasteride group, 695 had prostate cancer. Gleason grade was available for 686 men, of whom 264 had grade 7 or higher and 81 had grade 8 or higher. The AUC of PSA for all outcomes was greater for the finasteride group than the placebo group. For detecting prostate cancer versus no cancer, the AUCs were 0.757 and 0.681, respectively (P<.001); for detecting Gleason grade ≥7 versus ≤6 or no cancer, the AUCs were 0.838 and 0.781, respectively (P=.003); and for detecting Gleason grade ≥8 versus ≤7 or no cancer, the AUCs were 0.886 and 0.824, respectively (P=.071). The sensitivity of PSA was higher for men in the finasteride group than in the placebo group at all PSA cutoffs matched by specificity.Conclusions
PSA had statistically significantly better sensitivity and AUC for detecting prostate cancer in the finasteride arm of the PCPT than in the placebo arm. This bias would be expected to contribute to greater detection of all grades of prostate cancer with finasteride.