Background and purpose: Prognostic stratification of lung cancer brain metastases helps to develop a rational treatment plan. The current prognostic scoring system for lung cancer with brain metastasis includes: recursive partitioning analysis (RPA) classes

diagnosis-specific graded prognostic assessment (DS-GPA) index

basic score for brain metastases (BSBM) and lung-molecular graded prognostic assessment (lung-mol GPA). This study aimed to predict the prognosis of epidermal growth factor receptor (EGFR) gene mutant lung cancer with brain metastases using the four scoring systems. Methods: Patients with EGFR-mutant lung cancer and brain metastases diagnosed at the Fudan University Shanghai Cancer Center from Mar. 2008 to Mar. 2018

and patients who received tyrosine kinase inhibitors and brain radiotherapy were analyzed retrospectively. Independent prognostic factors of EGFR mutation brain metastasis were explored using the univariate analysis of log-rank test and COX regression model. The predictive effects of the 4 prognostic scoring systems on overall survival were analyzed by log-rank test

and the area under the curve (AUC) of each scoring system was calculated using the receiver operating characteristic (ROC) curve. Results: A total of 126 patients were enrolled in the study. The median survival time was 39 months. Independent prognostic factors for EGFR-mutant lung cancer with brain metastases included KPS score

extracranial metastasis

initial brain metastasis

EGFR locus and number of brain metastases. The P values of log-rank test for PRA

DS-GPA

BSBM and lung-mol GPA scoring systems were 0.032

0.124

0.002 and 0.003

respectively; the AUC values were 0.619

0.608

0.615 and 0.621

respectively. Conclusion: Except DS-GPA

the PRA

BSBM and lung-mol GPA scoring systems have a good predictive effect. Lung-mol GPA is the best among the four scoring systems. How to optimize the prognostic prediction model of EGFR-mutant lung cancer with brain metastasis is worth exploring further.