UroToday - Negative Influence of Changing Biopsy Practice Patterns on the Predictive Value of Prostate-Specific Antigen for Cancer Detection on Prostate Biopsy - Abstract

[Guest guest]

Monday, 24 March 2008

Department of Urology, New York Presbyterian Hospital, Weill Medical College

of Cornell University, New York, New York.

A correlation between prostate specific antigen (PSA) level and positive

prostate biopsy rate was established in an era when biopsy practice patterns

were different from what they are today. We evaluated if changes in biopsy

practice patterns have affected the ability of PSA to predict cancer

detection on prostate biopsy in the current era.

Of 3634 prostate biopsies performed from 1993-2005, 1607 met criteria for

analysis. Biopsy data were divided into 3 time-cohorts (1993-1997,

1998-2001, and 2002-2005) to assess for practice patterns shifts and

correlation between PSA and biopsy results.

Significant changes in biopsy practice patterns included an increase in

biopsy cores and more frequent use of PSA 2.5-3.99 ng/mL as a biopsy

indication. In men with normal DRE, a moderate correlation between PSA and

positive biopsy rate did exist from 1993-1997, but was subsequently lost. On

multivariate analysis, PSA was not a significant predictor of biopsy result

in men with normal DRE.

Early in the PSA era, the predictive power of PSA depended on multiple

factors: high prevalence of disease, higher prevalence of high-grade

disease, and low likelihood of prostate cancer diagnosis in men with low

PSA. Now, beyond the culling effect of increased biopsy incidence and with

shifted biopsy practice patterns, the correlation between PSA and biopsy

result is lost in men with normal DRE. Diagnosing a higher proportion of

tumors in men with a PSA between 2.0-4.0 ng/mL has negatively influenced the

predictive value of PSA for cancer detection.

Written by

Schwartz MJ, Hwang DH, Hung AJ, Han J, McClain JW, Shemtov MM, Te AE, Sosa

RE, Vaughan ED Jr, Scherr DS.

Reference

Cancer. 2008 Mar 10. Epub ahead of print.

doi:10.1002/cncr.23353

PubMed Abstract

PMID:18330908