ESTRO 2025 - Abstract Book

S1923

Clinical - Urology

ESTRO 2025

1073

Digital Poster Towards a near-automated workflow for CBCT-guided online adaptive radiotherapy in prostate cancer: insights from a prospective evaluation Goda Kalinauskaite 1,2 , Luise Anne Künzel 1 , Anne Kluge 1 , Jakob Dannehl 1 , Celina Mandy Höhne 1 , Christian Robke 1 , Daniel Zips 1 , Carolin Senger 1 1 Department of Radiation Oncology and Radiotherapy, Charité – Universitätsmedizin Berlin, Berlin, Germany. 2 Berlin Institute of Health, Charité – Universitätsmedizin Berlin, Berlin, Germany Purpose/Objective: Online adaptive radiotherapy (online-ART) has the potential to improve target coverage and spare healthy tissue. However, online-ART often requires increased time and human resources. This prospective study (NCT06116019) investigates the impact of workflow optimization to reduce human intervention and streamline processes for efficient delivery of online-ART in prostate cancer. Material/Methods: Online-ART for prostate cancer was prescribed in 20 fractions with a simultaneous integrated boost of 62.0 Gy/57.6 Gy/48.0 Gy to the prostate, periprostatic space, and seminal vesicles (SV). The adaptive workflow comprised seven steps: initial CBCT, influencer (organs influencing location and shape of targets) corrections, target approval, plan selection and QA, second CBCT, and treatment. We evaluated three workflows: 2-Influencer (rectum, bladder; 126 fractions), 3-Influencer (prostate, rectum, bladder; 241 fractions), and 4-Influencer (prostate, rectum, SV, bladder ± small bowel; 246 fractions). In the 4-Influencer workflow, targets were derived from prostate and SV; in other two, deformable image registration was used. Corrections for influencers and targets were categorized as no, minor (<10% slice adjustment), intermediate, or major (extensive re-delineation) (1). Workflow performance was measured by time for each step and correction frequency. Results: Thirty-four prostate cancer patients receiving a total of 613 fractions (95% were treated with adaptive plans), were analyzed. Average treatment time for all fractions was 25 minutes, with no significant differences in total treatment time across workflows (p=0.239). However, the 2-Influencer workflow was faster in influencer approval (median for 2- vs. 3- vs. 4-Influencer: 2.2 vs. 3.7 vs. 4.3 minutes, p=0.002), while the 4-Influencer workflow was faster for target approval (median for 2- vs. 3- vs. 4-Influencer: 5.6 vs. 5.0 vs. 2.5 minutes, p<0.001). Correction rates (no, minor, intermediate, major) varied by structure: bladder (53%, 39%, 5%, 3%), rectum (45%, 48%, 6%, 1%), prostate (18%, 56%, 21%, 5%), and SV (27%, 60%, 9%, 4%). The need for target corrections per fraction in 2-Influencer (61 %) and 3 Influencer (51%) workflows was significantly higher than in 4-Influencer (11%), p<0.001 (Figure 1). The 62 Gy treatment volume increased significantly with prostate influencer corrections, from +3.5% (no corrections) to +9.0% (major corrections) (p = 0.041). Minor, intermediate, and major prostate corrections added 1.3, 3.9, and 8.0 minutes to the average contouring time per fraction, respectively.