1994 Volume 85 Issue 10 Pages 1543-1551
Diagnostic utility of serum markers and their relative values to prostatic volume were evaluated using Receiver Operator Characteristics Analysis (ROC analysis) in 173 patients who underwent ultrasound guided biopsy of the prostate gland. Seventy cases (40.5%) of prostate cancer were detected. As a whole, prostate specific antigen density (PSAD) and prostate specific antigen (PSA) were more useful than gammaseminoprotein density (GSMD), gammaseminoprotein (GSM) and prostatic acid phosphatase in diagnosing cancer judged by the area under the ROC curve denoting a test's diagnostic accuracy (p<0.05). No significant difference was noted, however, between PSAD and PSA (p>0.05). Prostate specific antigen density was more predictive for prostate cancer than PSA in a subgroup of patients with PSA levels of 2.0-10.0ng/ml (p<0.05). No advantage of PSAD was obtained in patients with intermediate PSA levels of 2.0-5.0ng/ml or benign-feeling glands (p>0.05).
Higher sensitivity could be achieved by using Eiken PSA 2.0ng/ml as a cutoff rather than the recommended value of 3.0ng/ml. This helped to diagnose 5 more cases of prostate cancer who otherwise might have been missed if PSA cutoff of 3.0ng/ml had been used. A PSAD cutoff of 0.15 has a sensitivity of 81.4%, a specificity of 87.4% and an accuracy of 85.0%. However, use of this cutoff for biopsy could result in unacceptable numbers of undiagnosed cases, including many potentially curable cancers. Though PSAD may enhance sensitivity and specificity in a certain group of patients, this gain is not sufficient to reliably define the group at highest risk of prostate cancer. Indication of biopsy should still be determined based upon PSA concentration rather than PSAD value.