Predictors of prostate cancer specific mortality after radical prostatectomy: 10year oncologic outcomes from the Victorian Radical Prostatectomy Registry.

Abstract

To identify the ability of multiple variables to predict prostate cancer specific mortality (PCSM) in a whole of population series of all radical prostatectomies (RP) performed in Victoria, Australia. A total of 2154 open RPs were performed in Victoria between July 1995 and December 2000. Subjects without follow up data, Gleason grade, pathological stage were excluded as were those who had pT4 disease or received neoadjuvant treatment. 1967 cases (91.3% of total) met the inclusion criteria for this study. Tumour characteristics were collated via a central registry. We used competing hazards regression models to investigate associations. At median follow up of 10.3years pT stage of RP (P < 0.001) and high Gleason score of the RP specimen (P < 0.001 for ≥8 [Subhazard ratio (SHR) 11.19] and 4 + 3 = 7 [SHR 7.10]) compared with Gleason score 6 disease were strong predictors of progression to PCSM. Gleason score 3 + 4 = 7 was not at this time a significant predictor of PCSM (P = 0.08, SHR 1.84). Predictors of PCSM, independent of stage and grade, included rural residency (P = 0.003), primary surgeon contributing less than 40 cases (low-volume) to the VRPR (P = 0.025) and the involvement of a trainee surgeon in the operation (P = 0.031). The significant prediction of PCSM by pT cancer stage, Gleason score and primary Gleason pattern at RP in this whole of population study suggests a need to avoid understaging/grading in the process of cancer diagnosis and active surveillance protocols. Multi-modality therapy is likely to have a greater impact on PCSM in higher stage and Gleason grade disease. Identification of increased PCSM with rural residency and with involvement of a trainee urologist, and reduction in PCSM with higher surgeon volume all suggest potential for improved PC outcomes to be achieved with changes to surgical training and service delivery.