Acute radiation-induced brain injury in patients with glioma and the construction of a clinical prediction model: a cohort study.

Background: Glioma, which has a high degree of malignancy and mortality, is mainly treated by radiotherapy. Acute radiation-induced brain injury is one of the common complications of radiotherapy and can lead to brain herniation. Identifying risk of acute radiation-induced brain injury can facilitate the improvement of diagnostic and treatment strategies to ultimately improve patient outcomes. The purpose this study was to construct and validate a prediction model for acute radiation-induced brain injury in patients with glioma.

Methods: The data from 420 patients with glioma admitted to the Nanxishan Hospital of Guangxi Zhuang Autonomous Region from January 2020 to December 2024 were retrospectively collected as the training set, while the data from 180 patients with glioma treated at the 940th Hospital of Joint Logistics Support Force of PLA during the same period were collected as the validation set. The differences in the clinical characteristics of patients with acute brain injury (n=112) and non-brain injury (n=308) in the training set were analyzed, as were the risk factors of acute radiation-induced brain injury. According to the relevant risk factors, a prediction model for acute radiation-induced brain injury was constructed and validated in the validation set.

Results: Age, diabetes, size of gross tumor volume, radiation dose of gross tumor volume, and concurrent chemotherapy were independent risk factors for acute radiation-induced brain injury in patients with glioma, with the relative risks being 1.060 [95% confidence interval (CI): 1.030-1.091], 3.080 (95% CI: 1.384-6.852), 1.075 (95% CI: 1.049-1.100), 1.241 (95% CI: 1.176-1.310), and 3.951 (95% CI: 1.877-8.317), respectively. The area under the receiver operating characteristic (ROC) curve of the training set was 0.907 (95% CI: 0.875-0.939), and the area under curve of the validation set was 0.913 (95% CI: 0.861-0.965). The Hosmer-Lemeshow goodness-of-fit test was conducted on the model in the validation set, with a Chi-squared value of 5.135 and a P value of 0.743.

Conclusions: Patients with glioma have a high incidence of acute radiation-induced brain injury during radiotherapy, which can lead to a poor prognosis. The model we developed demonstrated good efficacy and reliability for identifying risk of acute radiation-induced brain injury.