BACKGROUND: Patients with glioblastoma (GBM) often have disease progression after standard temozolomide-based chemoradiation. The benefits and optimal use of re-irradiation (re-RT) following re-operation (re-OP) in recurrent GBM (rGBM) remain uncertain. In this study, we assessed the efficacy and safety of post-operative re-RT in patients with isocitrate dehydrogenase-wild-type rGBM, aiming to identify survival benefits and determine clinicogenetic criteria for patient selection. PATIENTS AND METHODS: Data from the Korean Radiation Oncology Group 21-02 retrospective study were evaluated, including 531 patients with rGBM from 2013 to 2019. A subset of 164 patients undergoing re-OP were analyzed for survival and benefits of post-operative re-RT. Additionally, 206 patients receiving re-RT, irrespective of re-OP, were evaluated for risks of radiation necrosis. The overall survival (OS) after re-OP was the primary endpoint. Statistical analyses included the Kaplan-Meier method and log-rank test for OS, Cox proportional hazards regression model for univariate and multivariate analyses, and the Fine-Gray competing risk model for assessing the risk of brain necrosis. RESULTS: The median OS after re-OP was 13.4 months. Kaplan-Meier analysis revealed significantly better OS for those who received re-RT (17.6 months) than for those who did not (11.0 months; P = 0.002). Factors associated with improved OS included higher Karnofsky performance status scores, post-operative re-RT, and additional systemic therapy after re-OP. Factors associated with adverse outcomes included recurrence outside the initial RT field and homozygous deletion of CDKN2A/B. The incidence of grade 2 or higher RT necrosis was 5.8% among those undergoing both re-OP and post-operative re-RT. CONCLUSION: Post-operative re-RT appears to be associated with enhanced survival and minimal toxicity in patients with rGBM following temozolomide chemoradiation. Our study suggests a novel clinicogenetic criterion for re-RT after re-OP in rGBM, which requires further validation.
Park et al. (Mon,) studied this question.