Open Access

Madeleine J. Karpinski, Kambiz Rahbar, Martin Bögemann, Laya Rahbar Nikoukar, Michael Schäfers, Sebastian Hoberück, Matthias Miederer, Tobias Hölscher, Sazan Rasul, Marcin Miszczyk, Francesco Lanfranchi, Matteo Bauckneht, Christian H. Pfob, Felix Kind, Karolien Goffin, Anika Hüsing, Claudia Kesch, Ken Herrmann, Martin Stuschke, Andrei Gafita, Johannes Hüsing, Jeremie Calais, Michael S. Hofman, Thomas A. Hope, Jonathan Miksch, Timo F. W. Soeterik, Andrea Di Giorgio, Andrea Farolfi, Anders Bjartell, Elin Trägårdh, Lena M. Unterrainer, Adrien Holzgreve, Gabriel T. Sheikh, Isabel Rauscher, Matthias Eiber, Boris A. Hadaschik, Wolfgang P. Fendler

• Background and objective We established prognostic nomograms incorporating prostate-specific membrane antigen (PSMA) positron emission tomography (PET) parameters standardised by Prostate Cancer Molecular Imaging Standardized Evaluation (PROMISE; PPP1). Here, we develop an updated PPP2 risk score from a large international multicentre registry study. Methods We included 6128 prostate cancer patients who underwent PSMA-PET at 20 hospitals in Europe, USA, and Australia between 2013 and 2022. Investigator sites were split 2:1 into the development (4044 patients) and validation (2084 patients) cohorts. We created nomograms of version 2 (PPP2) based on Cox regression models with the least absolute shrinkage and selection operator penalty for overall survival (development cohort). Performance of both nomograms was measured using Harrell’s C-index and calibration plots and a head-to-head comparison with the National Comprehensive Cancer Network (NCCN) risk score by receiver operatingBackground and objective We established prognostic nomograms incorporating prostate-specific membrane antigen (PSMA) positron emission tomography (PET) parameters standardised by Prostate Cancer Molecular Imaging Standardized Evaluation (PROMISE; PPP1). Here, we develop an updated PPP2 risk score from a large international multicentre registry study. Methods We included 6128 prostate cancer patients who underwent PSMA-PET at 20 hospitals in Europe, USA, and Australia between 2013 and 2022. Investigator sites were split 2:1 into the development (4044 patients) and validation (2084 patients) cohorts. We created nomograms of version 2 (PPP2) based on Cox regression models with the least absolute shrinkage and selection operator penalty for overall survival (development cohort). Performance of both nomograms was measured using Harrell’s C-index and calibration plots and a head-to-head comparison with the National Comprehensive Cancer Network (NCCN) risk score by receiver operating characteristic curves (validation cohort). Key findings and limitations Predictors were distant metastases (extrapelvic nodal metastases [M1a], bone metastases [M1b], and visceral metastases [M1c]), PSMA expression score, and total lesion count (visual PPP2) or total tumour volume (quantitative PPP2). C-indices (95% confidence interval) in the validation cohort were 0.80 (0.78–0.82; visual) and 0.80 (0.79–0.82; quantitative), respectively. Accuracy of both the PPP2 nomograms was superior to the NCCN risk score (n = 1034, area under the curve 0.84 vs 0.76; p < 0.001). The retrospective design represents a limitation of the study. Conclusions and clinical implications PPP nomograms were improved in an international multicentre study to predict accurately the 3- and 5-yr overall survival probabilities of prostate cancer. PPP2 yielded superior accuracy to the NCCN risk score. A free software tool has been created for PROMISE and PPP2 assessments (promise-pet.org).…