Clinical usefulness of serum prostate specific antigen for the detection of prostate cancer is preserved in men receiving the dual 5alpha-reductase inhibitor dutasteride

Abstract

Purpose: We determined whether the decrease in serum PSA seen with 5alpha-reductase inhibitors affects the clinical usefulness of PSA for prostate cancer screening using data from 2 dutasteride benign prostatic hyperplasia studies.

Materials and methods: A total of 2,802 men 50 years or older with a clinical diagnosis of benign prostatic hyperplasia, no history of prostate cancer, PSA 1.5 to 10 ng/ml, prostate volume 30 cc or greater, an American Urological Association symptom score of 12 or greater and peak urinary flow rate 15 ml per second or less were randomized to 0.5 mg dutasteride daily or matching placebo for 24 months. Increases in PSA from baseline and the maximum increase from nadir to month 24 were compared between the groups and analyzed by prostate cancer status, as determined by PSA driven biopsy and an advised cutoff of more than 4 ng/ml after doubling to correct for dutasteride treatment with sensitivity and specificity calculated for each.

Results: In placebo treated men without prostate cancer there was an 8.3% median increase in PSA at month 24 compared with -59.5% in those who received dutasteride, using doubled values to correct for dutasteride treatment. In those with prostate cancer these changes were 23.8% and -37.2%, respectively. Using the upper PSA limit of 4 ng/ml sensitivity for prostate cancer in men receiving dutasteride vs placebo was 0.737 vs 0.804, while specificity was 0.671 vs 0.578. Using a PSA increase from nadir of 0.8 ng/ml the sensitivity of dutasteride was 0.548 and its specificity was 0.795.

Conclusions: A doubling factor is effective for maintaining the sensitivity and specificity of PSA for prostate cancer detection in men on dutasteride. Increases in serum PSA in men receiving dutasteride should be considered suspicious and serial PSA measurements should be used to evaluate changes from nadir.