The Accuracy Of Partin Tables Derived From U.S. To Predict Pathological Features Of Chinese Prostate Cancer

Objective To compare the predictive value of Partin tables and their update version for postoperative pathologic features of prostate cancer in China. Methods A cohort of 167 patients who underwent radical prostatectomy for clinically localized prostate cancer between June 2000 and December 2005 at Changhai hospital were evaluated. Predictions made with Partin tables and Partin tables update using pretreatment PSA, biopsy Gleason score and clinical stage were compared with actual pathologic features. The predictive accuracies of Partin tables and their update were assessed and compared by receiver operating characteristic (ROC) analysis. A nonparametric approach was used to compare the areas under two correlated receiver operating characteristic curves. Results For the Partin tables, the area under the ROC (AUC) for organ confined disease, established capsular penetration, seminal vesicle involvement and lymph node involvement was 0.685, 0.655, 0.798 and 0.742 respectively. These values were virtually low in the Partin tables update, namely 0.639, 0.603, 0.713 and 0.641. Statistic analysis with nonparametric compare approach revealed that the predictive value of Partin table for seminal vesicle involvement was significantly superior to Partin tables update (p<0.05). Conclusions Our inter-racial external data validation and comparative study confirmed good predictive accuracy of Partin tables in Chinese prostate cancer, but not Partin Tables Update.