External validation of new prostate cancer risk groups by PSMA-PET (PPP3).

5112 Background: Previously, we proposed novel risk group definitions for prostate cancer patients based on Prostate-Specific Membrane Antigen (PSMA) targeted positron emission tomography (PET). PSMA-PET pan-stage nomograms (PPP3) were developed using the international, multicentre PROMISE registry (NCT06320223) to prognosticate overall survival (OS). Here, we present an external validation of PPP3. Methods: Eligible patients enrolled into our PROMISE registry database after the PPP3 data cap were included into this external validation study. Included patients had histologically proven prostate cancer and underwent PSMA-PET at hospitals in Turkey, Cyprus, Italy, China, South Africa or Germany between 2015 and 2022. PSMA-PET was standardized by PROMISE version 2 (V2); total lesion count, total tumor volume, PSMA expression score and OS follow-up were obtained as per local site practice. PPP3 nomograms were applied to calculate risk groups and Harrell´s C-indices for the external validation cohort. Calibration curves were measured for 5-year OS. Head-to-head comparison between the visual PPP3 nomogram and the simplified risk stratification table was examined by area under the receiver operating characteristics curve (ROC-AUC). Results: 1855 male patients across all disease stages with 179 (9.6%) reported deaths and median OS follow-up of 4.8 years (IQR 3.7-6.1) were analysed. In the external validation cohort C-indices (95% CI) were 0.71 (0.67-0.76) for the visual nomogram and 0.73 (0.69-0.77) for the quantitative nomogram, respectively. By simplified risk stratification table, 77 of 1855 patients (4.2%) were underestimated and 16 (0.9%) were overestimated, when compared with visual nomograms. Prognostic accuracy was comparable using both methods (AUC Nomogram: 0.64 vs. AUC Table: 0.63, p = 0.02). Conclusions: PPP3 nomograms were validated in an external multi-site patient cohort. Prognostication was accurate (C-indices > 0.70) for both PPP3 nomograms. Clinical trial information: NCT06320223 .