A meta-analysis of the performance characteristics of the free prostate-specific antigen test.

Overview

OBJECTIVES: To conduct a meta-analysis of the diagnostic performance of the percent free prostate-specific antigen (%fPSA) test in determining prostate cancer status and to assess its value in helping to decide whether to biopsy the prostate. METHODS: Articles identified through a MEDLINE search were included if they presented adequate original primary data to calculate a receiver operating characteristic (ROC) curve in subjects possessing histopathologically verified diagnoses. Articles containing subjects with concurrent, non-prostate-related genitourinary conditions were excluded. Percent free PSA sensitivity, specificity, ROC curves, and positive likelihood ratios were calculated for all PSA ranges and for the reflex range of PSA between 4 and 10 ng/mL. RESULTS: Among the 41 studies, containing 19,643 subjects, area under the curve for %fPSA was 0.70 for all PSA levels, decreasing to 0.68 in the reflex range. A test cutoff of 20% would lead to 92% sensitivity and 23% specificity. Positive likelihood ratios ranged from 1.0 to 4.0, exceeding 2.0 at %fPSA of 15% or less. Within the reflex range, however, likelihood ratios exceeded 2.0 only at %fPSA of 7% or less. CONCLUSIONS: Percent free PSA can be a useful adjunct to PSA for primary prostate cancer screening only under certain defined situations. In the "gray zone," or reflex range, of PSA testing, %fPSA improves clinical information only when levels reach extreme values.