Feasibility of Therapist-Driven MR-Guided Adaptive Radiotherapy for Oligometastatic Disease: Geometric Accuracy and Dosimetric Impact.

Abstract

Introduction: MRI-guided adaptive radiation therapy (ART) is the resource-intensive process of daily treatment plan modification. This study aims to demonstrate the feasibility of radiation therapist (RT)-led MRI-guided ART for oligometastatic disease (OMD) by comparing geometric accuracy and dosimetric differences between RT and radiation oncologist (RO) re-contouring.

Methods: Five RTs and five ROs retrospectively re-contoured gross target volumes (GTVs) and organs-at-risk (OARs) for eight OMD cases. RT and RO contours were compared against consensus RO Simultaneous Truth and Performance Level Estimation (RO-STAPLE) contours using the Dice similarity coefficient (DICE), mean distance to agreement (MDA), planning target volume (PTV) D95 and OAR D0.5cc using the Wilcoxon signed-rank test. Moreover, an RO qualitatively scored all contours using a 5-point Likert scale.

Results: We found very good geometric accuracy with average (±standard deviation) GTV DICE of 0.82 ± 0.06 for RTs and 0.85 ± 0.09 for ROs and MDA of 0.88 ± 0.03 mm for RT and 0.75 ± 0.05 mm for ROs relative to the RO-STAPLE. Qualitative GTV Likert scores were excellent, 4.8/5 for RTs and 4.7/5 for ROs. Mean percent difference in PTV D95 compared to RO-STAPLE was small but significantly higher for RTs (0.5% ± 1.5%) compared with ROs (-0.7% ± 1.9%, p < 0.05). Mean relative change in OAR D0.5cc results was small with -1% ± 6% for RTs and -1% ± 12% for ROs.

Conclusions: Here we provide the first report of geometric and dosimetric contouring uncertainty for MR-guided online ART for OMD. Our results show that RT re-contouring maintains similar performance for eligible targets and OARs compared with RO contours, establishing the initial feasibility of an RT-led workflow.