External validation and head-to-head comparison of Japanese and Western prostate biopsy nomograms using Japanese data sets

The objective of this study was to perform external validation of a previously developed prostate biopsy nomogram (the CHIBA nomogram) and to compare it with previously published nomograms developed in Japanese and overseas populations. Two different cohorts of patients were used: one from the Chiba Cancer Center (n = 392) in which transperineal 16-core biopsy was performed, and another from Chibaken Saiseikai Narashino Hospital (n = 269) in which transrectal 16-core biopsy was carried out. All patients were Japanese men with serum prostate-specific antigen levels less than 10 ng/mL. The predictive accuracy of our CHIBA nomogram and of four other published nomograms (Finne's sextant biopsy-based logistic regression model, Karakiewicz's sextant biopsy-based nomogram, Chun's 10-core biopsy-based nomogram and Kawakami's three-dimensional biopsy-based nomogram) was quantified based on area under the curve derived from receiver operating characteristic curves. Head-to-head comparison of area under the curve values demonstrated that our nomogram was significantly more accurate than all other models except Chun's (P = 0.012 vs Finne's, P = 0.000 vs Karakiewicz's, and P = 0.003 vs Kawakami's). Our nomogram appears to be more useful for the Japanese population than Western models. Moreover, external validation demonstrates that its predictive accuracy does not vary according to biopsy approach. This is the first report to demonstrate that the predictive accuracy of a nomogram is independent from the biopsy method.