Influence of local tumour stage and grade on reliability of serum prostate-specific antigen in predicting skeletal metastases in patients with adenocarcinoma of the prostate.

Objective: To determine whether serum prostate-specific antigen (PSA) can be reliably used to predict the absence or presence of skeletal metastases on the bone scan in patients with adenocarcinoma of the prostate.
Methods: We studied 450 consecutive men presenting with adenocarcinoma of the prostate between 1991 and 1995. Serum PSA was measured by the Hybritech Tandem-R monoclonal immunoradiometric assay and bone scanning was performed with 99m-technetium-labelled methylene diphosphonate. In total, 46 patients were excluded for one or more of the following reasons: serum PSA not available; radionuclide bone scan inconclusive; histology of the prostate other than adenocarcinoma; hormonal or other therapy given prior to obtaining the serum PSA and/or bone scan.
Results: Of the 404 patients included, 43% had poorly differentiated (grade 3), 74% had locally advanced (stages T3-4) tumours and 50% had skeletal metastases. The mean and median serum PSA were 348 and 52 ng/ml, respectively, and 77% of the patients had a serum PSA >20 ng/ml. The negative predictive value (for the absence of metastases on bone scan) of a serum PSA <20 ng/ml was 87% for the whole group of patients, 92, 94 and 70% for grade 1, 2 and 3 tumours, and 95, 83 and 50% for stage T1-2, T3 and T4 tumours, respectively. The positive predictive value (for the presence of metastases on bone scan) of a serum PSA >100 ng/ml was 80% for the whole group of patients.
Conclusions: In patients presenting with adenocarcinoma of the prostate, serum PSA alone is not sufficiently reliable to predict the absence or presence of metastases on the radionuclide bone scan. In patients with grade 3 and clinical stage T3-4 tumours, a bone scan should be obtained for accurate staging, regardless of the serum PSA value.