In-Silico Trial of Simulation-Free Hippocampal-Avoidance Whole Brain Radiotherapy Using Diagnostic MRI-Based and Online Adaptive Planning

Abstract

<h3>Purpose/Objective(s)</h3> Hippocampal-avoidance whole brain radiotherapy (HA-WBRT) has been shown to better preserve cognitive function with no difference in intracranial progression-free and overall survival. One limiting factor for its wider implementation is the plan complexity and considerably longer planning time required. We hypothesized that, <i>in silico,</i> a simulation-free workflow using diagnostic magnetic resonance (MR)-based radiation planning and online adaptation to account for differences in head position would expedite time to HA-WBRT delivery without compromising plan quality. <h3>Materials/Methods</h3> Seven patients treated for brain tumors who had cone beam computed tomography (CBCT) imaging available that sufficiently visualized the brainstem and ventricles were selected for this <i>in silico</i> study. Structures were contoured on the diagnostic brain MR, and a baseline HA-WBRT plan was generated. Then, to mimic simulation-free first fraction treatment, contours were transferred onto a daily setup CBCT for each patient. Within a commercially available, CT-based adaptive platform, contours were manually adjusted for anatomical differences; differences in head tilt between the diagnostic brain MR and CBCT were accounted for by triangulating with the brainstem, ventricles, and cranium and by online adaptation. Constraints were based on CC-001 protocol. Workflow tasks were timed, and baseline and adapted plans were compared using paired t-test on statistical software. <h3>Results</h3> All plans required adaptation at first fraction to account for head tilt differences and to meet constraints. An adaptive hippocampal avoidance planning template was applied to each patient with no modifications resulting in all objectives being met in the initial plan and adaptive plan. Structures that benefitted the most from first-fraction adaptation included hippocampi and optic chiasm; 71% and 57% of cases, respectively, required adaptation to meet constraints at first fraction. On pairwise comparison, there was a significant improvement in all dosimetric parameters for target coverage and with limiting hotspots in the hippocampal avoidance zone after adaptation (<i>p≤</i> .02); planning target volume (PTV) coverage increased by a median of 5% (IQR: 4%). Median duration of tasks was as follows: structure propagation and alignment on CBCT (8.5 min); contour adjustment (22.4 min); dose calculation (12.4 min); plan evaluation (0.9 min); total process time was 44.2 min. <h3>Conclusion</h3> Simulation-free HA-WBRT plans were successfully created using diagnostic brain MR imaging and first-fraction online adaptation to account for positional differences between diagnostic and on-table imaging. A Phase I clinical trial (NCT05096286) evaluating the feasibility and safety of simulation-free HA-WBRT using this methodology is underway.