A robustness-inclusive comparison of proton- versus photon-based whole-pelvic radiotherapy for prostate cancer within a randomised clinical trial

Background and purpose: Proton therapy offers potentially improved normal tissue sparing compared to photon-based whole-pelvic radiotherapy (WPRT) for high-risk prostate cancer, but its sensitivity to anatomical and setup variations raises concerns about robustness. The aim of this study was – within the setting of a multicentre randomised clinical trial – to explore whether the dose-volume advantages of proton therapy persisted when subject to inter-fractional variation. Materials and methods: Five patients treated with WPRT at a national proton centre were included. Comparative photon plans were created independently by five photon therapy centres. Nominal proton and photon plans were evaluated alongside recalculated plans on two repeat computed tomography scans per patient and with robustness scenarios simulating geometric uncertainties. Dose-volume metrics for target volumes and normal tissues were compared between the two modalities using linear mixed effects models accounting for patient and centre variability. Results: Target coverage was consistently robust for both modalities across all plan types. Proton therapy resulted in significantly reduced bowel V_35Gy by 11.2 percentage points (95% CI [4.1:18.4], p = 0.01) and bowel mean dose by 13.9 Gy (95% CI [9.5:18.4], p < 0.001). Bladder mean dose was also lower with proton therapy (reduced by 18.4 Gy, p = 0.02). These advantages remained consistent across nominal, recalculated, and uncertainty scenario plans. No consistent modality-related differences were observed for high-dose normal tissue metrics. Conclusion: Within this robustness-inclusive multicentre comparison study, proton-based WPRT maintained target coverage comparable to photon therapy and consistently reduced low- and intermediate-dose exposure to normal tissues, while demonstrating preserved robustness under the influence of inter-fractional variation.