ABSTRACT
Kattan et al. at Baylor College of Medicine and D'Amico et al. at Harvard Medical School have each developed preoperative nomograms for prostate cancer recurrence after radical prostatectomy based on readily available clinical variables. Calibration and validation of those tools was achieved using North American patient cohorts, and their validity has not yet been shown in patients from other continents. We investigated the predictive accuracy of these nomograms when applied to European men with localized prostate cancer. Clinical data from patients who underwent radical prostatectomy at the University-Hospital Hamburg and fitted the respective derivation criteria were used for external validation (n = 1003 for the Kattan-Nomogram, n = 932 men for the D'Amico-Nomogram). Nomogram predictions of the probability for 2-years and 5-years freedom from recurrence predicted by the D'Amico-Nomogram and the Kattan-Nomogram respectively were compared with actual follow-up. The predictive accuracy of the nomograms was tested using areas under the receiver-operating-characteristic curves (AUC). The D'Amico-Nomogram AUC predicting 2-years probability of freedom from PSA recurrence was 0.80 vs. Kattan-Nomogram 5-years prediction with an AUC of 0.83. Using the 932 patients who exactly fit the derivation criteria of both nomograms, the predictive accuracy of the Kattan-Nomogram was 0.81. The superiority in predictive accuracy of the Kattan-Nomogram was statistically significant (p = 0.0274) but of unclear clinical significance. The two nomograms predicted recurrence with similar accuracy when applied to men diagnosed with localized prostate cancer in Germany. The high predictive accuracy of both nomograms demonstrates that these predictive tools derived in the U.S. can be applied to non-U.S. patients.
