RESUMO
BACKGROUND: To test discriminant ability of the 2014 ISUP Gleason grade groups (GGG) for prediction of prostate cancer specific mortality (PCSM) after radical prostatectomy (RP), brachytherapy (BT), external beam radiation (EBRT) or no local treatment (NLT) relative to traditional Gleason grading (TGG). METHODS: In the Surveillance, Epidemiology, and End Results (SEER)-database (2004-2009), 2,42,531 non-metastatic prostate cancer (PCa) patients were identified, who underwent local treatment (RP, BT, EBRT only) or NLT. Follow-up endpoint was PCSM. Biopsy and/or pathological Gleason score (GS) were categorized as TGG ≤6, 7, 8-10 or GGG: I (≤6), II (3 + 4), III (4 + 3), IV (8), and V (9-10). Kaplan-Meier plots, multivariable Cox regression analyses and receiver operating characteristics (ROC) area under the curve analyses (AUC) were used. RESULTS: Median follow-up was 76 months (IQR: 59-94). For the four examined treatment modalities, all five GGG strata and all three TGG strata independently predicted PCSM. GGG yielded 1.5-fold or greater HR differences between GGG II and GGG III, and twofold or greater HR differences between GGG IV and GGG V. Relative to TGG, GGG added 0.4-1.1% to AUC. CONCLUSIONS: This large population-based cohort study confirms the added discriminant properties of the novel GGG strata and confirms a modest gain in predictive accuracy. Prostate 77: 686-693, 2017. © 2017 Wiley Periodicals, Inc.