Radiation-Induced Brain Injury after Proton Radiotherapy Is Linked to Increased Distal Edge Linear Energy Transfer (LET) and Anatomically Variable Radiation Sensitivity

Abstract

The incidence of radiation necrosis (RN) after proton beam therapy (PBT) has been discussed particularly in the context of a potentially enhanced biological effectiveness (RBE) towards the distal field edge. RN is commonly identified as radiation-induced contrast enhancing (RI-CE) lesions in magnetic resonance images (MRI), which need to be distinguished from post-operative changes and gliosis. We set out to (i) determine the safety of PBT by evaluating the incidence of RI-CE lesions and (ii) investigate the possibility of spatially variable radiation sensitivity, either caused by enhanced distal RBE or engrained in the anatomy of the brain. We reviewed 110 patients (median age 35 years) who had received actively scanned proton radiotherapy between 2010 and 2015 for histologically proven low grade glioma. Median dose was 54 Gy (RBE = 1.1) (range 50.4 – 60 Gy RBE). Every treatment plan was re-calculated with Monte Carlo (Fluka) to yield a dose and dose-averaged linear energy transfer (LETD) distribution. Regular follow-up amounted to more than 1200 MRI examinations (median follow up 39 months), on which RI-CE lesions were delineated. Every volume element of 1 mm3 in the brain receiving a dose of more than 40 Gy(RBE) was tagged with (i) the proximity to the ventricular system (PVS) and (ii) the LETD. The relative risks for RI-CE lesions were assessed by logistic regression with these parameters, without further biological model assumptions. To account for the growth dynamics of lesions between MRI scans, voxel weights were modified to avoid statistical bias of large lesions at the time of discovery. PBT for low grade glioma results in overall survival of 95 % after two years and 90 % after seven years. Median progression free survival has not yet been reached with surviving fractions of 85 % after two years and 54 % after seven years. The incidence of temporary or clinically silent radiation induced brain injury was significantly higher than previously assumed and was observed in 31 (28 %) patients with a median latency of 15 months. RI-CE lesions with clinical symptoms ≥ CTCAE °II were observed in seven (6.3 %) patients. Of all RI-CE lesions, 26 (51 %) were in a high LETD (LETD >= 5.0 keV/mum) volume and 32 (63 %) were in proximity to the ventricular system (distance voxel-ventricle <= 4mm). Consequentially, both high LETD: OR = 2.6 (95% CI: 1.4–4.8, p = 0.003) and PVS: OR = 8.7 (95% CI: 5.0–15.4, p < 0.001) are highly predictive risk factors. PBT is a safe and efficacious option for patients with low grade glioma. Clinically relevant and symptomatic radiation necrosis was as rare as known from conventional photon treatment. The spatial analysis of the lesion occurrence pattern hints at a moderately increased RBE at the distal edge of the proton beam. Surprisingly, a dose > 40 Gy(RBE) in proximity to the ventricles increases the risk of a lesion 3-fold more than LETD, which hints at a narrowly defined region of elevated radiation sensitivity.