Levels of prostate specific antigen that predict skeletal spread in prostate cancer.

The ability of serum prostate specific antigen (PSA) and serum acid phosphatase (SAP) to identify skeletal spread was evaluated in untreated patients with prostatic cancer. Twenty patients with scintigraphic evidence of metastatic disease in bone (M1) at diagnosis were compared with 50 untreated patients in whom scans were repeatedly negative during long-term surveillance. Using the present laboratory upper limit of normal (ULN) of 3 iu/l, the sensitivity and specificity of SAP for M1 disease were 80 and 86% respectively. Stepwise discriminant analysis demonstrated that SAP was able to stage patients correctly (bone scan positive or negative) with 81% predictive accuracy at an optimum cut-off limit of 4.6 iu/l. By contrast, whilst PSA (Hybritech) was 100% sensitive for skeletal disease at 10 ng/ml--at the expense of poor (36%) specificity--analysis determined that an optimum cut-off limit of 58 ng/ml led to 79% predictive accuracy for disease in bone. It was concluded that PSA levels > 58 ng/ml are highly indicative of spread to the skeleton, even in the absence of radiological or scintigraphic evidence of metastases.