Introduction and Objectives
The external validity of predicting pathological stage in clinically localized prostate cancer between the Partin Tables of 2001 and their updated version of 2007 was compared.
Material and Methods
Clinical and pathological data of 687 consecutive patients who underwent radical prostatectomy for clinically localized prostate cancer between January 2003 and December 2008 were used to compare the Partin Tables of 2001 and 2007 in their external validity. Receiver operating characteristic (ROC) curve were performed to compare the observed and predicted rates of the Partin Tables for organconfined disease (OC), extracapsular extension (ECE), seminal vesicle invasion (SVI) and lymph node involvement (LN).
Mean age of patients was 62.1 (±6.4) years, and mean PSA was 8.2 (±5.2) ng/ml. An unsuspicious digital rectal examination (T1c) was assessed in 71.5% of patients. Of the whole cohort, 76.2% of men were presented with OC, 17.8% had ECE, 6.0% showed SVI and 1.2% had lymph node involvement in the obturator region. The area under the receiver operating characteristic curve (AUC) of the Partin Tables of 2001 and 2007 were 0.727 and 0.722 for OC, and 0.662 and 0.650 for ECE. The Partin Tables of 2001 and 2007 showed an AUC of 0.788 and 0.779 for SVI, and 0.786 and 0.746 for LN, respectively.
Our external validation shows a good accuracy of the updated Tables to predict OC, SVI and LN. However, the predictive accuracy for ECE was only modest for both versions of the Partin Tables. Overall, the newer version of the Partin Tables could not outbalance the version of 2001 in their predictive accuracy for any pathological stage, and they failed to demonstrate a clear advantage. Our results underline the necessity to perform an external validation before the implementation of a new predicting tool.
© 2009 European Association of Urology. Published by Elsevier Inc. All rights reserved.