Introduction: Accurate prediction of extraprostatic extension (EPE) of prostate cancer (PCa) is the keystone for deciding whether to perform a neurovascular bundle sparing (NVBs) radical prostatectomy or not, which will subsequently affect the postoperative functional outcomes especially potency. Partin tables are the most commonly used predictive tools (PTs) for prediction of EPE. Moreover, they are the most commonly externally validated. In these settings, the aim of our work is to perform a systematic review of the literature and a meta-analysis for the discriminative performance of the different versions of Partin tables for EPE prediction.
Evidence acquisition: A systematic search of Medline, Scopus and Cochrane library was performed to include all the external validation (EV) studies that reported the discriminative performance (area under the curve [AUC]) of the different versions of Partin tables as a PT for EPE. Different versions of Partin tables (1997, 2001, 2007, 2010, and 2013) were included in separate meta-analyses. The pooled AUC with 95% CI were calculated to determine the weighted summary AUC using the random effect model.
Evidence synthesis: Twenty-six studies carried out in different countries including the USA, Korea, Germany, Ireland, China, Austria, France, Italy, the UK, and India were included. Considering the small number and generally low quality of the EV studies in literature, most of the included studies showed some sort of bias especially in the sample size & missing data domain. The pooled EPE AUC were 0.642 (95% CI; 0.601-0.682), 0.672 (95% CI; 0.617-0.727), 0.659 (95% CI; 0.623-0.695), 0.669 (95% CI; 0.623-0.715) and 0.644 (95% CI; 0.545-0.742) for the 1997, 2001, 2007, 2010 and 2013 versions, respectively.
Conclusions: Despite being the most commonly used predictive tool for prediction of EPE, the pooled EPE AUC for different versions of Partin tables showed poor discriminative performance. Thus, surgeons must be cautious when referring to Partin tables for prediction of EPE. Further EV studies are required to confirm these results.