The mean probability of finding cancer in transrectal ultra sound guided prostate biopsy: An update

Abstract

4613 Background: We provide an update to our previously presented study resulting in a useful table that provides estimated probability of a positive transrectal ultrasound (TRUS) guided prostate biopsy based on patient age, digital rectal exam (DRE) findings, and baseline serum prostate specific antigen (PSA) level. We validated the tables with two datasets from two different institutions. Methods: A step-wise logistic regression model was developed using data from 3,814 men participating in the Tyrol Screening Project (Europe) from January 1993 to April 2001. Variables used in the model were age, DRE findings, and serum PSA level. Bootstrap validation was used and 95% confidence intervals were calculated for each mean probability estimate. The resulting model was used to generate a look up table and then validated with prospective data from 491 Virginia Mason Medical Center (VM) patients and 474 Stanford University (SU) patients. Results: In the Tyrol data used to build the base model, the mean age was 63 and mean PSA was 9.6 ng/mL. DRE was suspicious in 581 (15.2%) of men. Carcinoma was confirmed by biopsy in 1,022 (26.8%) men. The model achieved an area under the receiver operating characteristic curve of 0.68 on the VM data and 0.63 on the SU data. At a 90% sensitivity, the model had a specificity of 12% and 10% for the VM and SU validations, respectively. Conclusions: We have developed a simple look-up table from a large screening database that provides estimates of the probability of finding cancer by TRUS guided biopsy utilizing readily available clinical variables that is validated on two independent data sets. Such a table can be used by practitioners and patients when assessing diagnostic options for suspected cancer of the prostate. Author Disclosure Employment or Leadership Consultant or Advisory Stock Ownership Honoraria Research Funding Expert Testimony Other Remuneration Institute for Clinical Research, Veterans Affairs Medical Center