Clinical Outcomes of Proton Beam Re-Irradiation for Recurrent Gliomas

Abstract

Despite improved molecular subtyping, overall prognosis of recurrent gliomas remains poor. Salvage re-irradiation (reRT) for recurrence can result in significant local control and proton beam radiotherapy (PBRT) offers a potential advantage due to lower integral dose to previously irradiated tissue. There is limited data on PBRT reRT outcomes. Twenty-two consecutive patients (median age 49, range 24-73) who received PBRT reRT for recurrent glioma between 2013 and 2018 were analyzed. The majority underwent PBRT reRT for recurrent Grade IV glioma (50%, 41% for WHO Grade III, and 9% for WHO Grade II). Prior to PBRT reRT, patients had a median of 1 surgery (range 1-4) and 2 systemic therapy regimens (range 1-5); 82% and 18% had 1 and 2 prior courses of RT, respectively, with a median duration of 27 months (range 7-232) prior to PBRT. Median total prior RT dose was 60 Gy (range 50.4-119.4). PBRT reRT was delivered using 3D conformal (50%), pencil beam (45%) scanning or a mix (5%). Planned reRT dose was a hypofractionated regimen in 45% of patients with doses of 30-40 Gy (RBE) in 10-15 fractions while the remaining 55% had doses of 45-60 Gy (RBE) in 25-33 fractions. Kaplan-Meier was used to determine local failure (LF), intracranial distant failure (DF), and overall survival (OS) from start of PBRT. Proportional hazards modeling was used to assess clinical predictors of these outcomes. The median OS post PBRT reRT was 11 months (6 months for Grade IV, 11 months for Grade III). Grade IV tumors had significantly poorer local control with a median time to LF after reRT of 6 months (p=0.02). Median time to DF was 7 months with no significant differences by WHO grade. Tumors with 1p19q codeletion (HR=0.1, p=0.02) and increased duration from prior RT (HR=0.98, p=0.04) were associated with improved OS post PBRT reRT while IDH1 mutant status trended toward better OS (HR=0.3, p=0.06). No molecular tumor markers were found to be associated with LF or DF, including MGMT methylation status. The number of prior lines of systemic therapy was also not associated with OS, LF, or DF. Patients had large reRT volumes with median PTV of 190 cc (range 3-676). PTV size was not prognostic of LC but hypofractioned courses trended towards poorer LC (HR 4.2, p=0.08) with no difference in DF (p=0.97) or OS (p=0.12). Two patients (10%) terminated PBRT early, one due to a mechanical fall and subsequent subdural hematoma; the other progressed during PBRT and required urgent debulking surgery. Three patients (14%) developed symptomatic radiation necrosis at 2, 2, and 12 months post reRT that were treated with bevacizumab and high dose steroids. PBRT reRT was generally well tolerated with low rate of radiation necrosis and promising local disease control for recurrent glioma. While tumor grade and 1p19q codeletion status were predictive of OS, local control after reRT was associated with tumor grade and not tumor molecular markers. Further study is needed to elucidate which patients would most benefit from proton reRT.