Blood–Brain Barrier Repair of Bevacizumab and Corticosteroid as Prediction of Clinical Improvement and Relapse Risk in Radiation-Induced Brain Necrosis: A Retrospective Observational Study

Background: Blood-brain barrier (BBB) disruption after endothelial damage is crucial part of radiation-induced brain necrosis (RN), but little is known of BBB disruption quantification and its role in evaluation of therapeutic effect and prognosis for drug treatment. In this retrospective study, BBB repair by bevacizumab and corticosteroid and the correlation between BBB permeability and treatment response and relapse were evaluated by dynamic contrast-enhanced MRI (DCE-MRI). Methods: 41 patients with RN after radiotherapy for nasopharyngeal carcinoma (NPC) (28 treated with bevacizumab and 13 with corticosteroid), 12 patients with no RN after NPC radiotherapy and 12 patients with no radiotherapy history were included as RN, non-RN and normal group respectively. DCE-MRI assessed BBB permeability in white matter of bilateral temporal lobe. DCE parameters were compared in baseline among three groups. DCE parameters after treatment were compared and correlated with RN volume decrease, neurological improvement and relapse. Results: Extent of BBB leakage in baseline increased from normal group, non-RN group and to RN necrosis lesions, especially Ktrans (Kruskal-Wallis test, P＜0.001). In RN group, bevacizumab induced Ktrans and ve decrease in radiation necrosis lesions (both P＜0.001) while corticosteroid showed no obvious effect on BBB. The treatment response rate of bevacizumab was significantly higher than corticosteroid (30/34 [88.2%] vs 10/22 [45.4%], P＜0.001). Spearman analysis showed baseline Ktrans、Kep、vp positively correlated with RN volume decrease, improvement of cognition and quality of life in bevacizumab treatment. After six-month follow-up for treatment response cases, relapse rate of bevacizumab and corticosteroid was 10/30 (33.3%) and 2/9 (22.2%) with no statistic difference. Post-bevacizumab Ktrans level predicted relapse in six months with AUC 0.745 (P < 0.05, 95% CI 0.546-0.943, sensitivity= 0.800, specificity= 0.631) Conclusions: Bevacizumab improved BBB leakage in RN necrosis. DCE parameters may be useful to predict therapeutic effect and relapse after bevacizumab.