Refining prognostic stratification in salvage radiotherapy for prostate cancer: a retrospective multicenter cohort study using PSMA-PET and machine learning

Background: Salvage radiotherapy (SRT) is the standard treatment for biochemical recurrence (bREC) after radical prostatectomy (RP), yet optimal radiation dose, field size, and the role of advanced imaging like PSMA-PET remain unclear. This study assessed the impact of SRT dose and the prognostic role of PSMA-PET on 2-year biochemical relapse-free survival (bRFS) in patients with localized disease. Methods: In this retrospective multicenter study, 255 patients treated with SRT across 11 centers were selected from a database of 1,201 cases. Inclusion criteria included PSA persistence/recurrence (≥0.1 ng/mL) post-RP and negative PSMA-PET for nodal or distant metastases. Patients receiving androgen deprivation therapy or with PSA > 0.5 ng/mL pre-SRT were excluded. Prognostic factors were identified using LASSO analysis and modeled with CART analysis. The primary endpoint was 2-year bRFS. Results: With a median follow-up of 33 months, 2-year bRFS was 88.2 %. Lower pre-SRT PSA (<0.2 ng/mL), PSMA-PET negativity, longer PSA doubling time, older age, favorable ISUP grades (1-2), and shorter intervals from surgery to bREC were associated with improved bRFS. CART analysis demonstrated that PSMA-PET findings significantly influenced prognosis in patients with PSA levels below 0.2 ng/mL, while PSA-doubling time was predictive in patients with PSA 0.2-0.5 ng/mL. SRT dose and elective nodal irradiation did not significantly affect bRFS. Conclusions: This study underscores the complexity of prognostic modeling in SRT for prostate cancer, highlighting the value of PSMA-PET and CART for refined risk stratification.