Radiation Associated Brain Necrosis following Proton Therapy for Head and Neck Skull Base and Intracranial Tumors

Abstract

To characterize the patterns and outcomes of radiation associated brain necrosis (RABN) after proton therapy (PT) for selected head and neck cancers (HNCs) with base of skull and/or intracranial involvement. Outcomes of patients with HNC treated with PT between 2010 and 2018 participating in a prospective clinical study who had one or more follow-up MRIs (≥ 6 months after treatment) and who had received at least 40 Gy(RBE) to brain were analyzed. Those who received re-irradiation were excluded. Diagnostic radiology reports and available images were reviewed for RABN defined as contrast enhanced lesions on T1- and/or edema on T2-weighted sequences. MRIs showing the RABN lesions were co-registered (rigid) with initial treatment planning CT and lesions manually contoured and volume and spatial location determined. Time to RABN was calculated from end of PT to first imaging occurrence using Kaplan-Meier method. 105 patients formed the cohort; 53 (50%) were male. Regarding disease site, 23 (22%) were nasopharynx, 19 (18%) were paranasal sinus, and 17 (16%) were peri-orbital. 62% had also received concurrent chemotherapy. Median follow-up time (from end of PT to latest MRI) was 33 months (range: 6-92). Overall, 18 patients (17.1%) developed RABN after a median latency of 24 months (range: 9-33). Actuarial RABN-free rate was 98.9%, 89.4 %, and 80.9% at 1, 2 and 3 years, respectively. All lesions were incidentally discovered on surveillance imaging and asymptomatic (grade 1) and initially observed. All lesions had a contrast enhancing component on T1 MRI. Location of lesions were temporal lobe in 12, frontal lobe in 4 and cerebellum in 2 patients. Of those with RABN, 13 (72%) had been treated for T4 or unresectable disease and 14 (78%) had intracranial tumor extension and the remainder skull base involvement. Median prescription dose to the most cranial/proximal extent of disease was 69 Gy(RBE) (range: 60-70). Median RABN lesion size (T1) was 0.2cc (range: 0.0 [punctate] to 1cc), and 12 (67%) were inside or overlapping with the high dose target volume (TV). In 5 patients (28%) RABN proved to be transient and completely resolved w/o therapy over a median time of 33 months from onset (range: 8-42), 4 (22%) show regression (2 w/ and 2 w/o medical therapy), and 3 (17%) stability (1 w/ and 2 w/o medical therapy). Three patients (17%) had progression on latest imaging, yet remain asymptomatic; each had received 70 Gy(RBE) to the most cranial TV with concurrent chemotherapy. RABN lesions following PT for skull base/intracranial disease were small, asymptomatic and regressed or completely resolved in half of the patients affected. Observation with surveillance imaging without surgical intervention is recommended for these asymptomatic radiographic lesions.