Efficient Clinical Implementation of an MRI-Guided Adaptive Radiation Therapy Program

Abstract

<h3>Purpose/Objective(s)</h3> Magnetic Resonance Image-guided Adaptive Radiation Therapy (MRgART) is ideal for intra-abdominal tumors as it accounts for daily anatomical changes in target volumes and nearby organs at risk (OAR) caused by soft-tissue deformation, respiration, organ filling/emptying and patient weight changes. The dosimetric gains associated with MRgART may be undermined by longer treatment times, resulting in unaccounted for intra-fractional OAR changes and patient motion. In this study, we aim to evaluate each step in our MRgART workflow for optimal resource utilization and potential treatment time savings. <h3>Materials/Methods</h3> This IRB approved retrospective review included pancreas cancer patients treated at our institution between 2018 and 2020 with MR guidance. We examined MRgART prescription details, and steps in the adaptive workflow, including MR simulation, image fusion, contouring, re-plan/QA and tumor tracking for resource allocation and clinical workflow efficiency. A generalized estimating equation (GEE) model with an exchangeable working correlation structure and Wilcoxon signed rank test were used for statistical time analysis. <h3>Results</h3> The study included 17 eligible pancreas cancer patients treated consecutively on our 0.35 T MR-Linac with MRgART. We examined 115 treated fractions (77 adapted). Two of the patients received conventional fractionation, and 15 were treated with stereotactic body radiation therapy (SBRT). SBRT prescription dose was 40 Gy to PTV in 5 fractions with 6MV FFF X-rays. Non-adaptive fractions averaged 38.3 ± 3.2 min (mean ± s.d.) compared to significantly longer adaptive treatments lasting 57.8 ± 4.3 min (an increase of 19.6 min or 51.3%, <b>P</b> < 0.01). Within each adapted treatment, the greatest time was spent on OAR contouring (13.6 min), followed by patient setup (10.6 min), imaging (9.1 min), re-plan (7.3 min), and QA (2.2 min). Beam-on time was 26% of overall treatment duration. Major contributors to increased treatment time with MRgART were contouring, physics review, replan and QA. Other sub-categories, such as, patient setup, imaging and beam-on times remained unchanged when compared to non-adaptive treatments. On average, treatment times decreased by 13.6% (61.4 min to 53 min, <b>P</b> = 0.06) as patients progressed through their treatment course reflecting improved patient confidence, compliance and comfort with their treatments and breathing instructions. Gains in treatment time also resulted from limited OAR contouring around the PTV and audio/visual coaching of patients leading to longer duty cycle. <h3>Conclusion</h3> Efficient resource utilization is important for safe and effective implementation of MRgART. Online OAR contouring, imaging and patient setup are potential targets for further time improvements and could be achieved <b>via</b> offline staff training, emphasis on robust AI based imaging, contouring and planning strategies, and in-room monitor device for improved active patient participation, comfort and compliance.