Serum prostate-specific antigen as surrogate for the histological diagnosis of prostate cancer

To determine whether there is a cut-off value of serum prostate-specific antigen (PSA) which can be used confidently to make the diagnosis of prostate cancer, thereby obviating the need for biopsy.

During the period October 1991 to March 1998 the Department of Chemical Pathology at Tygerberg Hospital performed a total of 6,733 serum PSA assays on 3,960 patients. The histopathological and clinical diagnoses of these patients were obtained from records in the departments of Anatomical Pathology, Urological Oncology and Radiation Oncology. The serum PSA levels were correlated with the histopathology reports, using different PSA cut-off values ranging from 5 to 500 ng/ml, to calculate the sensitivity, specificity, and positive and negative predictive values of each cut-off value of PSA in predicting the presence of prostate cancer.

In total, 3,837 (57%) of the 6,733 serum PSA assays were < or = 4 ng/ml, 1,045 (15.5%) of the assays were > or = 50 ng/ml, and 798 (11.9%) were > or = 100 ng/ml. Of the total of 3,960 individual patients, 531 (13.4%) had a serum PSA > or = 50 ng/ml and 423 (10.7%) had a PSA > or = 100 ng/ml. A serum PSA of > or = 30 ng/ml had a positive predictive value (PPV) of 90% at a specificity of 87% and sensitivity of 78%, while a PSA > or = 60 ng/ml had a PPV of 98% at a specificity of 98% and sensitivity of 65% for the presence of prostate cancer. The PPV reached 99% at a PSA > or = 100 ng/ml and 100% at a PSA > or = 500 ng/ml, with a specificity of 99% and 100%, but sensitivity of only 53% and 19%, respectively.

A serum PSA > or = 60 ng/ml has a PPV of 98% for the presence of adenocarcinoma of the prostate, and may be used as a surrogate for histological diagnosis where facilities for obtaining prostatic biopsies are not readily available, thus decreasing costs and patient morbidity.