PQA 03 - PQA 03 Gynecological Cancer, Pediatric Cancer, and Professional Development Poster Q&A
3533 - Outcomes of Exclusive Use of Magnetic Resonance Image-Guided Brachytherapy in Locally Advanced Cervical Cancer with a Focus on Central Control and Evaluation of an Auto-Segmentation Tool for High-Risk
Y. Landman Gigi1, G. Kim1, C. De Sousa1, E. Schmidt2, K. Sheikh1, M. B. Roumeliotis3, J. Lee1, and A. N. Viswanathan2; 1Johns Hopkins Medicine, Department of Radiation Oncology and Molecular Radiation Sciences, Baltimore, MD, 2Department of Radiation Oncology and Molecular Radiation Sciences, Johns Hopkins University School of Medicine, Baltimore, MD, 3Johns Hopkins School of Medicine, Baltimore, MD
Purpose/Objective(s):Central failure after cervical cancer image-guided brachytherapy (IGBT) is uncommon but a cause of significant morbidity and mortality. In this study we aim to report the outcomes of exclusive magnetic resonance (MR) IGBT in the primary treatment of locally advanced cervical cancer with a focus on central control and evaluate a deep learning auto-segmentation tool (AST) by applying it post hoc on cases with persistent or recurrent central disease. Materials/Methods: Disease outcomes of all consecutive cervical cancer patients treated with primary chemoradiotherapy with a brachytherapy boost in an MR-IGBT exclusive environment in a single center were retrospectively reviewed. Patients were included if at least one post treatment imaging evaluation was available. Continuous parameters were assessed with a Mann-Whitney test and T-test as appropriate. Oncological outcomes were assessed using Kaplan-Meier and Cox regression survival analysis. In cases with a central failure, defined as persistent or recurrent central disease, application of an AST was performed on all available MRIs and contours were reviewed. The mean added computation time for the AST is 12.59±0.79 seconds per MRI. Results: Between 2016 and 2023, 158 patients received radiotherapy with curative intent. Median follow up time was 2.6 years (range 0.3-7.6). The median high-risk clinical target volume (HRCTV) D90% was 77.3Gy [range 56.6Gy-98.8Gy]. The 3-year central control, progression free survival and overall survival (OS) rates were 86.7%,63.3% and 80.3%, respectively. The median D90% was not significantly lower in the 17 patients with central failure (73.75Gy [range 64.8-92.9Gy] versus 77.7Gy [range 56.6-98.8Gy], p=0.14). In 11/17 (64.7%) patients, the central failure was isolated. Only 3 isolated central failures were marginal to the original HRCTV contoured at IGBT, occurring in challenging areas to treat with tandem-based cervical BT (posterior bladder in a stage IVA patient, periurethral in a patient with upper vaginal disease and lateral fundus contiguous with ovary). Only 2 patients with central failures were not in the training set used to build the AST, both failed inside the HRCTV. The AST was applied on their 6 original MRIs. There was no significant difference in the original and new mean HRCTV (70 ml versus 70.7 ml, p=0.93). In 2/6(33%) MRIs, the new HRCTV was deemed better; In 1 the AST avoided contouring around peripheral interstitial catheters, avoiding a bias to select an easier to treat volume since its closer to the applicator, and in 1 it was determined the AST contours fit the anatomy and applicator better. Conclusion: Exclusive MR-IGBT in our cohort offered excellent central control. The use of AST in MR-IGBT may potentially improve selection of target volume in difficult situations and provides an expedited manner for contouring on MR-IGBT cases.
