Abstract:
Objective : To validate the predictive accuracy of the 2007 Partin tables for prostate cancer in the Chinese population and to compare the 2007 and 1997 versions.
Methodse : Data from 116 patients with prostate carcinoma who underwent radical prostatectomy from January 2003 to September 2007 in our hospital were reviewed. All cases were pathologically confirmed as prostate adenocarcinoma and had complete data of preoperative serum prostate specific antigen (PSA), clinical stage, and biopsy Gleason score. Pathological features of the radical prostatectomy specimens were divided into four types: organ confinement (OC), extraprostatic extension (EPE), seminal vesicle invasion (SVI) and lymph node involvement (LNI). Of all the patients, 93 patients met the 2007 Partin tables' prediction inclusion criteria and 113 patients met the 1997 Partin tables' prediction inclusion criteria. Both Partin tables were used to predict pathological features in each patient, and then Receiver operating characteristics (ROC) curves were used to assess the predictive value of each table.
Results : According to the 2007 Partin tables inclusion criteria, the incidence of OC, EPE, SVI and LNI were noted in 51%, 15%, 13%, and 22%, respectively, of the cases; according to the 1997 inclusion criteria the incidence of OC, EPE, SVI and LNI were noted in 50%, 16%, 12%, and 23%, respectively. The area under the curve (AUC) for OC, EPE, SVI and LNI disease prediction of the 2007 Partin tables was 0.786, 0.613, 0.666 and 0.811, respectively. The area under the curve (AUC) for OC, EPE, SVI and LNI disease prediction of the 1997 Partin tables was 0.834, 0.605, 0.716, and 0.808, respectively.
Conclusion : Both Partin tables have a reasonable predictive value for OC and LNI (AUC>0.7). In predicting SVI, the 1997 version is superiorto the 2007 version (AUC: 0.716 vs 0.666), but for EPE, both Partin tables show low predictive value.