e16066 Background: Armstrong et al. (Clin Cancer Res. 2007;13:6396–403) recently published a nomogram predicting the probability of survival in patients with hormone-refractory prostate cancer. The initial report showed a predictive accuracy of 0.69. So far, this nomogram was not yet externally validated. We validated this nomogram in a cohort of patients with hormone-refractory prostate cancer participating in five phase II trials. Methods: In our institution, 84 patients received chemotherapy for metastatic symptomatic hormone-refractory prostate cancer from September 1999 to November 2006. Follow-up was stopped in March 2008 and median follow up for all patients was 15.7 month. 75% of patients received docetaxel based chemotherapy. At the time of inclusion, median age was 69.0 years, median PSA was 40ng/ml, 93% had known bone metastases and 19% had visceral metastases. The area under the receiver operating characteristics curve was used to estimate the predictive accuracy of the nomogram and calibration plots were used for comparison between predicted and observed probabilities. Results: Of all patients, median survival was 15.5 month. The nomogram predicted median survival was 19.9 month, overestimating the true survival by roughly 30%. The predictive accuracy of the predictions at 1 year, 2 years and 5 years was 0.76, 0.69 and 0.64, respectively. The calibration plots showed departures from ideal predictions at 1 year (up to 20% of over estimation of true survival) and 5 years ( up to 19% of under estimation of true survival), whereas the predictions at 2 years showed almost perfect calibrations. Conclusions: The Armstrong et al1 nomogram provides accurate survival predictions in patients with hormone-refractory prostate cancer. The current external validation of this nomogram provides evidence that the nomogram might generally be applied to those patients in daily practice. No significant financial relationships to disclose.
Read full abstract