Background: Several tools have been developed to predict the outcome of prostate biopsies performed to diagnosis prostate cancer (PCa). However, few studies have focused on the comparative accuracy of these predictive tools. We aim to establish the predictive accuracy of three commonly used nomograms by comparing their prostate biopsy outcome predictions with actual pathological results.
Methods: From January 2008 to December 2010, 708 consecutive patients with an elevated serum PSA level and/or abnormal DRE were referred to our institution. All data were collected prospectively. All patients underwent a TRUS 12-core biopsy. Probability of a positive biopsy was predicted using three online risk calculation nomograms. The discriminative ability of the nomograms was assessed via AUC and the most accurate model was calibrated and compared to actual biopsy results.
Results: Of 667 patients fulfilling all three nomograms criteria, 384 (57.5%) had PCa and 283 (42.5%) did not. AUC for the PCPT-CRC, SWOP-PRI, and Montreal nomograms was 0.68 (95% CI, 0.63-0.72), 0.72 (95% CI, 0.68-0.76), and 0.79 (95% CI, 0.76-0.82), respectively. A comparison of the three models' performance showed that the Montreal model provided the greatest predictive accuracy (P = 0.03).
Conclusions: External validation of three commonly used nomograms designed to predict the likelihood of a positive prostate biopsy reveals the Montreal model was more accurate than either the PCPT-CRC or SWOP-PRI models. The Montreal nomogram achieves a diagnostic accuracy of 79% and is superior to PSA alone though we await further research to define the probability (of cancer) threshold above which a prostate biopsy would be advised.
Copyright © 2011 Wiley Periodicals, Inc.