PulseExploreJournal ClubDebatesTrendingResearchersJournals
Instagram
HomeExploreJournal ClubTrending
Synapse
⌘+K
Synapse
March 4, 2026Journal of Clinical Oncology0 citations

New prostate cancer risk groups by PSMA-PET (PPP3).

View Full Paper
MKMadeleine Josefine KarpinskiBHBoris HadaschikCCCaner Civan

Key Points

  • The aim is to develop novel risk classifications for prostate cancer based on PSMA-PET data to improve prognostication of overall survival.
  • Inclusion of male patients with histologically confirmed prostate cancer from the PROMISE registry
  • Development and validation of risk nomograms using Cox regression models with LASSO penalty
  • Assessment of nomogram performance with calibration curves and C-indices
  • Creation of a simplified risk stratification table
  • Comparison with established clinical risk scores using ROC curve analysis.
  • Analyzed 11,154 patients, with 27.9% recording deaths during the follow-up period.
  • Achieved C-indices of 0.83 for visual and 0.84 for quantitative nomograms in the validation cohort.
  • Identified accurate risk stratification with both PPP3 nomograms compared to existing clinical risk scores.

Abstract

20 Background: PSMA-PET usage in prostate cancer patients is growing rapidly. Thus, novel risk group definitions based on PSMA-PET are urgently needed for guidelines, clinical use and study design. Here, we present improved risk classification based on PSMA-PET PROMISE nomograms (PPP3) to prognosticate 3-, 5- and 7-year overall survival (OS) for the first time. Methods: We included male patients with histologically proven prostate cancer at any disease stage, who underwent PSMA-PET for any indication in the PROMISE registry (NCT06320223). 35 investigator sites from Europe, Asia, Australia, North- and South America were split approximately 2:1 into development and validation cohorts considering equal distribution of site characteristics. Updated PPP3 nomograms were created based on Cox regression models with LASSO penalty for overall survival from the development cohort. We applied calibration curves and Harrell´s C-indices to assess the performance of both PPP3 nomograms. Based on the visual PPP3 nomogram, a simplified risk stratification table was created. Head-to-head comparison of PPP3 nomograms with clinical risk scores separated for each disease subgroup was conducted using area under the receiver operating characteristics curve. Results: We analyzed 11154 patients (n=7253 development and n=3901 validation cohorts) with a median OS follow-up of 4.9 (interquartile range 3.5-6.6) years and 3109 (27.9%) recorded deaths. Clinical disease group and PROMISE metrics (presence of distant metastases, PSMA expression score and total tumor load) were combined into visual and quantitative PPP3 nomograms, respectively. In the validation cohort we reached C-indices of 0.83 (95% confidence interval CI 0.82-0.84) for the visual nomogram and 0.84 (95% CI 0.82-0.85) for the quantitative nomogram, respectively. Both PPP3 nomograms and the simplified risk stratification table (Table 1) were accurate and equal or superior compared to established clinical risk scores (STARCAP, EAU, Gafita, NCCN). Conclusions: We created new risk nomograms by PROMISE along with a simple risk stratification table to prognosticate 3-, 5- and 7-year OS in prostate cancer. PROMISE and PPP3 assessments are available online free of charge (promise-pet.org) for global implementation. Visual PSMA-PET and PROMISE metrics for risk stratification in prostate cancer patients. 1 point 2 points 3 points Disease Group nmCRPC, mHSPC, mCRPC miM1a yes miM1b oligo diss or dmi miM1c yes Total lesion count 6-20 >20 PSMA expression score (highest) 3 Sum of the points results in the following risk groups: i) low risk: 0 points, ii) intermediate risk: 1-4 points, iii) high risk: ≥5 points. nmCRPC=non-metastatic castration resistant prostate cancer. mHSPC=metastatic hormone-sensitive prostate cancer. mCRPC=metastatic castration-resistant prostate cancer. diss=disseminated. dmi=diffuse marrow involvement. PSMA =Prostate Specific Membrane Antigen.

Ask AI
Helpful
Bookmark
Share
View Full Paper

Cite This Study

Karpinski et al. (2026) studied this question.

synapsesocial.com/papers/69a7cd3dd48f933b5eed9729https://doi.org/10.1200/jco.2026.44.7_suppl.20
Ask AI
Helpful
Bookmark
Share
View Full Paper