Comparison of prostate specific antigen with prostate specific antigen density for 3 clinical applications.

We compared prostate specific antigen (PSA) to PSA density for 3 clinical uses: detection of nonpalpable prostate cancer, staging of clinically localized prostate cancer and prediction of PSA detectability following radical prostatectomy. Of 153 men with normal digital rectal examinations undergoing prostate biopsy 25% had prostate cancer. Analysis of receiver operator characteristic curves demonstrated that PSA density predicted the outcome of biopsy significantly better than did PSA (p = 0.0013). Pathological examination of 155 radical prostatectomy specimens revealed that 56% had pathologically uncontained disease. There was no difference between the ability of PSA and PSA density to predict the pathological findings (p = 0.2379). PSA data more than 1 year postoperatively were available in 96 of the 155 prostatectomy patients. Of these men 41% had postoperative PSA levels of 0.1 ng/ml. or more. Preoperative PSA and PSA density values were almost identical in the ability to identify these patients (p = 0.6776). While PSA density is superior to PSA for the detection of prostate cancer in patients with a palpably normal prostate, it does not offer any improvement over PSA for staging of prostate cancer or for the prediction of PSA detectability after radical prostatectomy.