Clinical features and multidimensional prognostic model of EGFR-TKI acquired resistance in NSCLC patients with or without brain metastases.

OBJECTIVE: To analyze the clinical features and related factors of acquired resistance to EGFR-TKI in patients with non-small cell lung cancer (NSCLC) and brain metastases, and to construct a prognostic prediction model. METHODS: A total of 186 NSCLC patients were retrospectively enrolled and divided into four groups based on EGFR-TKI resistance and brain metastasis status: resistant with brain metastases (n = 65), resistant without brain metastases (n = 65), sensitive with brain metastases (n = 28), and sensitive without brain metastases (n = 28). Univariate and multivariate logistic regression analyses were used to identify independent risk factors for resistance. A Cox proportional hazards model was applied to evaluate prognostic factors, and ROC curves were used to assess the predictive performance of the model. RESULTS: Multivariate analysis showed that non-classical EGFR mutations, TP53 co-mutations, MET amplification, CEA >5 μg/L, and third-generation EGFR-TKI therapy were independently associated with resistance (all P < 0.05), while anti-angiogenic therapy significantly reduced the risk of resistance. The median overall survival (14.0 vs. 38.0 months) and progression-free survival (9.0 vs. 26.0 months) were significantly shorter in the resistant group (P < 0.001). The number of brain metastases ≥3, ECOG PS ≥2, and T790M negativity were independent predictors of poor prognosis. The combined prediction model demonstrated good discriminative ability, with an AUC of 0.81 (95% CI: 0.76-0.86). CONCLUSION: This study clarifies the factors associated with EGFR-TKI resistance and prognostic characteristics in NSCLC patients with brain metastases. The constructed prediction model helps identify high-risk patients and provides a reference for formulating individualized treatment strategies.