ESTRO 2025 - Abstract Book

S3090

Physics - Inter-fraction motion management and offline adaptive radiotherapy

ESTRO 2025

885

Digital Poster Offline adaptive MR-guided proton therapy improves plan quality compared to online adaptive MR-LINAC for ultra-hypofractionation in prostate cancer Egon Hübner 1 , Emilie Alvarez-Andres 1,2 , Fabian Funer 1,2 , Franciska Lebbink 1,3 , Annika Lattermann 1,2 , Tobias Hölscher 1,2 , Stefan Menkel 1,2 , Aswin L. Hoffmann 1,2,4 , Esther G.C. Troost 1,2,4 1 OncoRay-National Center for Radiation Research in Oncology, Faculty of Medicine and University Hospital Carl Gustav Carus, Technische Universität Dresden, Dresden, Germany. 2 Department of Radiotherapy and Radiation Oncology, Faculty of Medicine and University Hospital Carl Gustav Carus, Technische Universität Dresden, Dresden, Germany. 3 National Center for Tumor Diseases (NCT/UCC), ; Helmholtz-Zentrum Dresden-Rossendorf (HZDR), Dresden, Germany. 4 Institute of Radiooncology-OncoRay, Helmholtz-Zentrum Dresden-Rossendorf, Dresden, Germany Purpose/Objective: Magnetic resonance guided photon therapy (MRgXT) enables online plan adaptation based on daily high-contrast soft-tissue imaging, making it suitable for hypofractionation of prostate cancer. Proton therapy (PT) reduces the radiation dose to healthy tissues, such that a combination of MR-guidance and PT is gaining increased attention. In this study, we compared target coverage and organ at risk (OAR) doses for online adaptive MRgXT and offline adaptive MR-guided proton therapy (MRgPT) for ultra-hypofractionated radiotherapy in prostate cancer patients. Material/Methods: Data of ten prostate cancer patients having undergone online ultra-hypofractionated intensity-modulated radiotherapy (IMRT) on the 1.5T MR-LINAC (Elekta, Sweden) using an ‘adapt-to-shape’-workflow with 7 x 6.1Gy were included in this study. The original IMRT dose distributions for every patient and each fraction were compared to de novo generated MRgPT-plans. For the latter, the T2w MR images acquired before each MR-LINAC fraction were converted into synthetic stopping-power-ratio (sSPR) maps for accurate proton dose calculation using a deep learning-based method [1]. OAR dose-volume-histogram metrices for rectum and bladder were used for a fraction wise comparison of the online MRgXT- and offline MRgPT-plans. For dose coverage of the clinical target volume (CTV), the homogeneity index (HI = (D 5% -D 95% ) / D 50% ) was compared. The Wilcoxon-test was employed for statistical analysis with a significance level of 0.05. Results: The median reduction for the HI was 60.4% (range 54.0% - 67.9%). The rectal dose was reduced when comparing offline MRgPT versus online MRgXT, with a median of 32.4% (14.5% - 91.4%) for the V 38.4Gy , 47.6% (29.7% - 85.0%) for the V 32Gy , and 52.0% (28.8% - 81.7%) for the V 28Gy . The V 40Gy for the bladder decreased by a median of 30.1% (-4.6% - 91.1%) (Figure 1). The reduction was statistically significant for all parameters.