3404 - Post-Treatment PSMA PET as a Predictor of PSA Relapse in Primary Prostate Cancer

Purpose/Objective(s): Prostate-specific membrane antigen (PSMA) PET is increasingly used for staging and detection of metastatic prostate cancer. However, its prognostic value for predicting relapse at the primary tumor site remains unclear. We evaluated whether quantitative PSMA PET metrics, combined with radiomics and clinical features, can predict long-term PSA relapse following definitive radiation therapy.

Materials/Methods: We retrospectively analyzed 204 patients with biopsy-proven prostate cancer who underwent both pre-radiation and post-radiation PSMA PET/CT. All primary tumor volumes were contoured by experienced radiation oncologists. Quantitative features including clinical variables (e.g. the Gleason score and risk groups), PET SUVs, PET/CT radiomics and their pre-post radiation changes (delta features) were derived. A support vector machine (SVM) based framework was applied for feature selection and outcome prediction. Patients were randomly split into a training set (75% with five-fold cross-validation) and an independent test set (25%). Training was repeated 500 times to compute averaged performance metrics with confidence intervals. Median follow-up time for PSA tests exceeds 4 years (range: 10 – 122 month). Model performance was assessed primarily using the area under the ROC curve (AUC), with the balanced accuracy reported as a complementary metric.

Results: Clinical variables alone demonstrated limited predictive value (AUC close to 0.5). Pre-treatment PET SUV showed minimal association with PSA relapse. In contrast, post-treatment PET SUV and pre-post treatment change in SUV demonstrated improved discrimination (AUC ~0.63). Incorporation of CT radiomics modestly improved performance for pre-treatment models (AUC ~0.68). Notably, post-treatment PET radiomics alone achieved an AUC of 0.72, with marginal gains when combined with CT radiomics and delta features (AUC ~0.75).

Conclusion: This is, to our best knowledge, the largest cohort to date evaluating PSMA PET for primary tumor relapse prediction. Post-treatment PSMA PET within the prostate demonstrated strong prognostic value that outperforms traditional clinical factors. These findings suggest that quantitative post-radiation PSMA PET imaging may serve as a clinically meaningful biomarker for early risk stratification and personalized surveillance following radiation therapy.