Adaptive MRI-Guided Reirradiation for High-Risk Recurrent Grade 4 Gliomas: Early Clinical Outcomes and Volumetric Response.

PURPOSE: Recurrent WHO grade 4 gliomas have poor outcomes and limited salvage options. Reirradiation (re-RT) can provide local control in selected patients but is constrained by cumulative dose and toxicity. Personalized ultrafractionated stereotactic adaptive radiation therapy (PULSAR) delivers re-RT as high-dose "pulses" spaced weeks apart, creating opportunities for interim magnetic resonance imaging (MRI) assessment and adaptive replanning on an MRI-linear accelerator METHODS AND MATERIALS: We retrospectively analyzed 45 patients with recurrent WHO grade 4 gliomas treated with MRI-guided PULSAR re-RT on a 1.5-T MRI-linear accelerator (5 planned pulses; typical prescription 25-35 Gy in 5 fractions with dose painting). A 0-5-mm expansion from either or both GTVs was used to generate the treated CTV without an additional PTV margin. On-treatment MRI (T1 postcontrast ± T2-FLAIR) was used for volumetric assessment and adaptive replanning when indicated. Endpoints included overall survival (OS), progression-free survival (PFS), local failure (LF), and grade ≥3 toxicity (cerebral edema or hematologic events). Univariable regression explored associations between clinical factors and outcomes.

RESULTS: Median age was 54.2 years; 78% of tumors were IDH wild type, with frequent corpus callosum involvement (68.9%) and multifocal/multicentric disease (40.0%). Thirty-six patients (80%) underwent ≥1 adaptive replan, and 30 (66.7%) completed the planned course. Median follow-up was 16.7 months. Median OS and PFS from re-RT start were 6.9 and 5.1 months, respectively. The 12-month cumulative incidence of LF was 56.0% (68.0% in a sensitivity analysis accounting for cause-specific mortality). The 12-month cumulative incidence of grade ≥3 cerebral edema was 40.0%, and grade ≥3 hematologic toxicity was 38.0%. Concurrent bevacizumab was associated with a lower incidence of grade ≥3 cerebral edema (20% vs 60%; HR = 0.32; P = .04). ECOG performance status ≥2 and IDH wild type status were associated with worse OS and PFS. Volumetric response on T1 postcontrast or T2-FLAIR imaging was not significantly associated with OS or PFS.

CONCLUSIONS: MRI-guided PULSAR re-RT with adaptive replanning was feasible in a high-risk recurrent WHO grade 4 glioma cohort. Interim MRI assessment supported real-time treatment decision-making, but the clinical benefit of this adaptive strategy relative to conventional re-RT remains unclear without a control arm. Prospective studies are needed to standardize imaging/adaptation workflows, optimize pulse spacing, refine margin strategies, and determine whether MRI-guided adaptive PULSAR improves clinically meaningful outcomes.