3196 - Reduced Margin Daily AI-Assisted Online Adaptive Radiotherapy for Bladder Cancer
Presenter(s)
D. Levy1, P. A. Blumenfeld1, J. Feldman2, Y. Hillman2, A. Salhab1, M. Fang1, A. Popovtzer1, and M. R. Wygoda1; 1Hadassah Medical Center, Hebrew University of Jerusalem, Jerusalem, Israel, 2Hadassah Medical Center, Jerusalem, Israel
Purpose/Objective(s): Radiotherapy for bladder cancer requires large margins (LM) to ensure target coverage and account for changes in daily bladder filling. These LM cause an undesirable increase in dose to organs at risk (OARs).We report our experience using an AI-assisted, online adaptive radiotherapy (oART) platform with cone-beam CT (CBCT) to optimize treatment through margin reduction.
Materials/Methods: Thirty-seven consecutive patients with muscle-invasive bladder cancer (89.2% male, median age 73, stages II-IVA) received 55Gy in 20 fractions with oART. Target volumes included bladder: proximal urethra (CTVp) and pelvic nodes (CTVn) were included as indicated. Two margin cohorts were studied: Cohort 1 (26 patients) used Small Margin (SM) PTV (0.5-1.0cm) for both adaptive and scheduled plans; Cohort 2 (11 patients) used LM PTV for scheduled plans and SM PTV for adaptive plans. Treatment workflow involved CBCT acquisition, deformable registration to generate synthetic CT, AI-based auto-contouring with manual review, and creation of both scheduled and adaptive plans, with the superior plan delivered. Toxicity was assessed weekly during treatment (CTCAE v5.0) and at follow-up visits. Post-treatment cystoscopy and imaging were performed every 3–6 months. We compared target coverage and dose to OARs in both cohorts and hypothesized that daily oART would improve PTV coverage compared to non-adapted treatment, allowing for smaller margins while reducing doses to OARs. In addition, local control, cystectomy-free survival (CFS), progression-free survival (PFS), and overall survival (OS) were assessed.
Results: A total of 740 adaptive and scheduled plans were generated, with the adaptive plan selected in 96.9% of fractions, median delivery time 19.3 minutes. PTV coverage (D95%) significantly improved with oART in both cohorts: 99.2% vs. 91.6% (p<.0001) in Cohort 1 and 100.4% vs. 93.8% (p<.0001) in Cohort 2. Average daily PTV volume change was 14.1% (range 0%-157.3%). In Cohort 2, oART significantly decreased radiation to rectum (V45, V75, V91, V100 had a relative decrease of 66.5%, 88.2%, 89.8%, 91.9%, respectively; all p<.0001) and bowel (V68, V75, V91, V100 had a relative decrease of 20%, 26.1%, 56.6%, 57.8%, respectively; all p<.0001). With median follow-up of 12 months, local control was 94.4%, three patients developed metastatic disease, and four patients died. The 1-year actuarial CFS was 97.2%, PFS was 94.4% and OS was 97.3%. Only grade 1-2 acute toxicities occurred, with long-term grade 3 radiation cystitis in 5.4% of patients.
Conclusion: This largest cohort of bladder cancer patients treated with reduced margin daily AI-assisted oART using CBCT demonstrates superior target coverage and OAR sparing compared to conventional treatment, with excellent oncologic outcomes and toxicity profiles.