Predictors for biopsy outcome in the European randomized study of screening for prostate cancer (Rotterdam region)

BACKGROUND. In the European Randomized Study of Screening for Prostate Canc er (ERSPC, Rotterdam region), men aged 55-74 years are screened for prostat e cancer by prostate-specific antigen (PSA) sampling, digital rectal examin ation (DRE), and transrectal ultrasound investigation (TRUS). All men with a PSA greater than or equal to 4 ng/ml and/or a suspicious DRE and/or a sus picious TRUS are biopsied. METHODS. Logistic regression analysis was applied to, derive a predictive i ndex that equals the chance to find prostate cancer in a biopsy given the o utcomes of the screening tests. This model was used to assess the number of cancers that could have been detected if all men had been biopsied (extrap olation). Furthermore, the model was used to study the possibilities for im provement of the current screening protocol. RESULTS. PSA was the dominant predictor for prostate cancer in a biopsy, fo llowed by prostate volume, DRE, and TRUS result. It is assessed that 69% (9 5% CI, 52-86%) of cancers that could be identified if all men were biopsied are currently detected. Application of the same methods to screening data obtained in Goteborg (the Swedish ERSPC partner) yielded almost identical r esults. It was found that, in the Rotterdam protocol, a considerable number of men were biopsied according to the screening protocol with an assessed lower chance to have prostate cancer than men who were not biopsied accordi ng to the protocol. CONCLUSIONS. The chance to detect prostate cancer in a biopsy can be modele d quite accurately as a function of serum PSA, prostate volume, DRE, and TR US results. Important improvements in the screening protocol can be achieve d by the application of the predictive index. (C) 1999 Wiley-Liss, Inc.