Daily artificial intelligence-assisted adaptive small margin radiotherapy during trimodal therapy of muscle-invasive bladder cancer.

OBJECTIVES

Adaptive radiotherapy (ART) enables daily plan modification based on anatomy and may improve outcomes in muscle-invasive bladder cancer (MIBC) treated with bladder-preserving chemoradiation. This retrospective study evaluated the dosimetric and clinical outcomes of daily ART.

METHODS

Between March 2021 and August 2024, 37 patients with T2-T4a MIBC received 55 Gy in 20 fractions with or without chemotherapy. Patients were stratified by margin strategy: small-margin (SM, n=26) adaptive versus scheduled plans, and small-margin adaptive versus large-margin (SM-LM, n=11) scheduled plans. Daily cone beam CT-guided ART with AI-assisted contouring, physician review, and plan selection was performed. Planning target volumes (PTVs) were generated by expanding the clinical target volume (CTV) by 0.5-1.0 cm in SM plans; 1.0-1.5 cm in LM plans. Toxicity was graded using CTCAE v5.0.

RESULTS

Across 740 fractions, adaptive plans were chosen in 96.9%. ART improved coverage in both cohorts. In SM patients, mean PTV V95 increased from 92.9% to 99.3% and CTV V98 from 96.7% to 99.8% (p < 0.01). In SM-LM comparison, PTV V95 improved from 95.5% to 99.8% (p < 0.01), with rectum and bowel dose reductions up to 91.9% and 57.8% (rectum V100%, bowel V100% respectively, both p < 0.01). At 12 months, cystectomy-free, progression-free, and overall survival were 95.8%, 82.7%, and 92.8%. No grade ≥3 acute toxicities occurred.

CONCLUSIONS

Daily ART improves target coverage, reduces organ-at-risk exposure, and enables margin reduction in bladder-preserving radiotherapy. This represents one of the largest series of daily online ART for MIBC, demonstrating that margin-reduced ART is feasible and improves dosimetry compared with non-adaptive workflows.