Prospective use of free PSA to avoid repeat prostate biopsies in men with elevated total PSA.

Published

Journal Article

BACKGROUND: Prostate-specific antigen (PSA) is a most valuable tool for the early detection of prostate cancer; however, it has a high false-positive rate as presently used in prostate cancer screening programs. Patients with persistent PSA elevations, normal digital rectal examinations, and multiple negative biopsies present a clinical dilemma. We prospectively evaluated whether free PSA improves the specificity of PSA and can be useful as a clinical guide to avoid repeat prostate biopsies in a group of such patients. METHODS: Sixty-seven men with persistent PSA elevations (mean 9.6 ng/mL; range 4.1-24.8 ng/mL), normal digital rectal examinations, and two or more prior sextant biopsies (mean 2.8), had serum collected for measurement of total and free PSA. All patients were rebiopsied to determine the receiver-operating characteristics (ROC) of total PSA vs. percent free PSA for prostate cancer detection. RESULTS: This study by biopsy identified 11 new prostate cancer cases. The median percent free PSA was significantly higher at 18.1% among men without prostate cancer, compared to 6.4% in men with prostate cancer (P < 0.00005). When sensitivity was plotted against I-specificity, the area under the receiver-operating curve (ROC) for percent free PSA was 0.95, compared to 0.75 for free PSA density, 0.59 for PSA density, and 0.54 for PSA. In patients with elevated total PSA levels, normal digital rectal examinations, and two prior sets of negative sextant prostate biopsies, a cutoff of 10% free PSA would maintain sensitivity at 91% with a corresponding specificity of 86%. CONCLUSIONS: Selective measurement of percent free PSA in cases of uncertain diagnosis can significantly improve the specificity of prostate cancer detection compared to total PSA alone. A low percent free PSA (< 10%) appears to be a significant predictor of prostate cancer even after two or more negative prostate biopsies.

Full Text

Duke Authors

Cited Authors

  • Morgan, TO; McLeod, DG; Leifer, ES; Moul, JW; Murphy, GP

Published Date

  • January 1, 1996

Published In

Volume / Issue

  • 7 /

Start / End Page

  • 58 - 63

PubMed ID

  • 8950365

Pubmed Central ID

  • 8950365

International Standard Serial Number (ISSN)

  • 1050-5881

Language

  • eng