Neurologic Events and Outcomes in Patients Receiving Proton and Photon Reirradiation for High Grade Non-Codeleted Gliomas

Abstract

Patients undergoing reirradiation (ReRT) for high grade glioma are at risk for tumor progression, pseudoprogression, and radiation necrosis. We investigated factors associated with neurologic events and disease control after re-irradiation with protons and photons at a single academic center. We reviewed records and MRIs of patients receiving scanning beam proton (since center opening in 2016) and photon (since 2015) reirradiation in ≥10 fractions for grade 3 anaplastic astrocytoma (AA) and grade 4 glioblastoma (GBM), excluding 1p19q co-deleted oligodendrogliomas and extensive multifocal/leptomeningeal disease. The primary endpoint was time from ReRT to ≥ grade 2 pseudoprogression or radiation necrosis (PsP/RN, grade 2: moderate symptoms requiring outpatient steroids/bevacizumab, grade 3: severe symptoms leading to admission or surgical intervention). Dose was converted to EQD2 using a/b = 3. Cox proportional hazards model was used to calculate survival and time to PsP/RN. A total of 53 patients were included (26 protons, 27 photons, median KPS 80). Patients receiving protons had more favorable features. Compared to the photons, the proton group was younger (48 vs. 58) and more likely to have AA (46% vs. 22%) and resection within 3 months (42% vs 26%). The proton group also had a longer interval from prior RT (57 vs. 39 months) and were less likely to receive bevacizumab at reRT (15% vs. 59%). CTV was 130 cc for protons vs 99 cc for photons, and most had active disease at time of ReRT identified on planning MRI (76% protons, 85% photons). Median OS was 10.5 months (14.1 months protons, 8.1 months photons), with time from initial RT the only significant factor on multivariate analysis. Median PFS was 9.4 months (9.8 months protons, 6.2 months photons). 9 patients (18%) had ≥ grade 3 PsP/RN (8 proton, 1 photon) and 21 patients (41%) had ≥ grade 2 PsP/RN (16 proton, 5 photon). Grade 3 events included 1 seizure (photon group), 1 hemorrhage, 1 thalamic stroke, 1 shunt placement, 1 re-resection, and PSP4 4 PsP/RN requiring admission. Protons were associated shorter time to ≥ grade 2 PsP/RN (4 months vs. not reached, p = 0.027). When accounting for bevacizumab use at time of reRT, the association between protons and PsP/RN lost significance but there remained a trend (grade 2, p = 0.095, HR 2.4; grade 3, p = 0.105, HR 5.8). CTV, MGMT status, EQD2, and interval from prior RT were not associated with PsP/RN. High grade neurologic events were common in patients with predominantly active, unresected high grade gliomas receiving ReRT. Though ascertainment and survival bias are significant limitations, pseudoprogression and necrosis appeared to be more prominent in patients receiving protons. These results contribute to ongoing efforts to both optimize ReRT for high grade glioma and investigate biologic effects of proton therapy.