BackgroundTreatment of brain metastases (BMs) in non-small cell lung cancer (NSCLC) patients, especially those with non-sensitive genetic mutations, is hindered by limited drug delivery through the blood–brain barrier (BBB). This retrospective study explores the efficacy of systemic treatments during brain metastasis to radiotherapy evaluation window in improving patient survival.MethodsIn this retrospective cohort study, we evaluated 209 NSCLC patients with non-sensitive mutations and BMs, treated between 2016 and 2023 at two tertiary medical centers (Chongqing University Cancer Hospital and Guangxi Medical University Cancer Hospital). The patients were divided into three groups, namely chemotherapy alone (C; n = 95), chemotherapy plus immune checkpoint inhibitors (ICIs) (C + I; n = 62), and chemotherapy with ICIs and antiangiogenic therapy (A) (C + I + A; n = 52). Statistical analyses were performed using R software, version 4.3.3. Categorical variables were compared using Fisher’s exact test, and survival curves were estimated with the Kaplan–Meier method and compared via the log-rank test. Univariate and multivariate Cox regression models were used to assess factors associated with overall survival (OS). Bayesian model averaging (BMA) was employed to address model uncertainty and improve result robustness. Subgroup analyses evaluated treatment-related mortality risk.ResultsFrom an initial cohort of 658 NSCLC patients with BMs, 209 were analyzed with a median age of 59; the majority were male (80.9%) and diagnosed with adenocarcinoma (78.9%). Univariate analysis identified significant variables influencing outcomes, including BMs radiotherapy EQD2, BMs count, local thoracic treatment, BMs radiotherapy field, intracranial response, and systemic treatment post-BMs diagnosis. The C + I + A regimen significantly improved median OS to 23.6 months compared to 11.4 months with C and 16.2 months with C + I, with a hazard ratio (HR) of 0.60 (95% CI: 0.43–0.82; P < 0.0001). The two-year OS rate was highest in the C + I + A group at 38.5%, versus 10.5% in C and 20.4% in C + I (P < 0.001). Cox regression and BMA analyses confirmed the stability of BMA in providing HR estimates, yielding area under the curve (AUC) values of 0.785 for BMA and 0.793 for the Cox model, with no significant difference in predictive performance. Subgroup analysis revealed a 71% mortality risk reduction with C + I + A (HR: 0.29; 95% CI: 0.18–0.47; P < 0.0001), showing consistent benefits regardless of patient sex, BMs count, extracranial metastases presence, and local thoracic treatments. Treatment sequence analysis indicated a median OS of 33.4 months for patients starting with A, though not statistically significant (HR: 0.59; P = 0.36). The overall incidence of radiation-induced brain injury was low at 3.3%, with rates in the C, C + I, and C + I + A groups being 3.2%, 4.8%, and 1.9%, respectively (P = 0.683).ConclusionOur study demonstrates the significant benefit of the C + I + A combination therapy in improving OS and reducing mortality risk in NSCLC patients with non-sensitive gene-mutated BMs. The sequential administration of A followed by ICIs shows a promising synergistic effect with cranial radiotherapy, highlighting the potential for optimized treatment sequencing. These findings emphasize the efficacy of tailored combination therapies in complex oncological care and suggest that our approach could lead to meaningful improvements in clinical outcomes for this challenging patient population.