A simplified online adaptive workflow for long-course magnetic resonance-guided radiotherapy in esophageal cancer

Abstract

Background and purpose: Online adaptive magnetic resonance-guided radiotherapy (MRgRT) enables high-precision radiotherapy for esophageal cancer patients but is less feasible due to long on-table times in combination with long-course treatment. In this study, we conducted an in-silico assessment of a simplified online adaptive workflow, Adapt-To-Shape-lite (ATS-lite), in which deformable propagated contours are not modified, and assessed its feasibility.

Materials and methods: The ATS-lite workflow was simulated for all fractions of nine esophageal cancer patients who had previously received full online adaptive MRgRT with manual contour corrections if needed. The deformable propagated contours were not adjusted. A dose of 41.4 Gy in 23 fractions was prescribed. Intra- and interfraction dose accumulation were performed to evaluate target coverage per fraction and across the entire treatment. For individual fractions, coverage of the manually corrected clinical target volume (CTV) was considered adequate if V95% > 98 % and V90% > 99.5 %. Feasibility was assessed by recording treatment times in the first patients treated with ATS-lite.

Results: The ATS-lite workflow provided adequate target coverage over the entire treatment for all patients, with sufficient coverage in 90% of the 177 fractions analyzed. Closer inspection revealed that inadequate target coverage in individual fractions was primarily attributed to enlargement of the manually corrected CTV, rather than poor contour propagation in the ATS-lite workflow. In seven patients, the ATS-lite workflow achieved a median time per fraction of 23 min.

Conclusions: The ATS-lite workflow provides adequate target coverage and is feasible for online adaptive MRgRT in long-course esophageal cancer treatments.

Keywords: Esophageal cancer; MR-guided radiotherapy; MR-linac; Online adaptive radiotherapy.

Fig. 1

Schematic overview of the in-silico assessment for the ATS-lite workflow. For each fraction, the clinical target volume of the ATS-lite workflow simulation (CTV_ATS-lite), expanded with the planning target volume margin, was used for treatment plan optimization. The planned dose distribution was convolved with the 3D tumor motion trajectory to derive the estimated delivered dose distribution. Individual fraction coverage by the estimated delivered dose distribution was evaluated for the manually corrected CTV during the clinical ATS procedure of that fraction (CTV_ATS). Interfraction dose warping to the pre-treatment reference scan was performed to estimate the total treatment dose. Target coverage across the entire treatment was evaluated based on the pre-treatment CTV (CTV_ORG). The full-ATS workflow simulation followed the same methodology, except for the evaluation of CTV_ATS coverage in individual fraction.

Fig. 2

Examples of fractions with inadequate coverage of the clinical target volume (CTV) defined in the clinical ATS procedure (CTV_ATS) for patients 2 (top), 8 (middle), and 3 (bottom). For each fraction, the unadjusted propagated CTV on which the treatment plan was optimized (CTV_ATS-lite, orange) and the CTV_ATS (blue) are depicted on the online MRI (left) and the estimated delivered dose distribution (middle). The reference MRI (right) shows the reference CTV used for deformable contour propagation (CTV_ORG, green). Notable differences between the CTV_ORG and the CTV_ATS can be observed for all three fractions.

Fig. 3

Median duration of each step in the ATS-lite workflow for online adaptive MR-guided radiotherapy for esophageal cancer patients (n = 132 fractions). IQR: interquartile range.