Serum prostate-specific antigen discriminates weakly between men with benign prostatic hyperplasia and patients with organ-confined prostate cancer.

To determine if serum prostate-specific antigen (PSA) can identify those patients harboring an organ-confined prostate cancer prior to treatment for symptomatic benign prostatic hyperplasia (BPH), we examined retrospectively the preoperative serum PSA level for two groups of men. Group 1 consisted of 187 consecutive patients with a histologic diagnosis of BPH as determined from complete pathologic analysis of the transurethral resection of the prostate (TURP) specimen. Group 2 included 198 consecutive patients with histologically confirmed organ-confined prostate cancer as determined from step-section analysis of the retropubic radical prostatectomy specimen. The median serum PSA value for group 1 was 3.9 ng/ml (range 0.2-55 ng/ml), whereas the median serum PSA level for group 2 was 5.9 ng/ml (range 0.4-58 ng/ml). Although this difference was statistically significant (p < 0.001), the distribution of serum PSA values for group 1 overlapped considerably with the distribution for group 2. For both groups, there was a clustering of PSA values below 10.0 ng/ml (group 1: 90%; group 2: 73%). The area under the receiver operating characteristic curve for log PSA values to discriminate men with BPH from patients with organ-confined prostate cancer was 0.66 (95% confidence interval: 0.60-0.72); a 'perfect' test has an area of 1.0, whereas a test with 'no information value' has an area of 0.5. These findings suggest that serum PSA has only a modest ability to distinguish men with BPH from patients with organ-confined prostate cancer. As a result, some symptomatic BPH patients choosing a non-TURP therapy may harbor a clinically significant prostate cancer despite being evaluated with a serum PSA determination.