Abstract
PURPOSE: To evaluate prognostic factors in breast cancer brain metastasis (BCBM) patients receiving stereotactic radiotherapy (SRT) and to develop an effective prediction model for overall survival (OS). PATIENTS AND METHODS: We retrospectively collected clinical and pathological data from BCBM patients treated with SRT. Prognostic factors were assessed using univariate and multivariate Cox regression analyses. A nomogram was developed based on Cox models and validated internally using bootstrap and cross-validation methods. Model discrimination was evaluated through calibration plots and the concordance index (C-index), while clinical utility was assessed using clinical decision curve analysis (DCA). RESULTS: Among 101 BCBM patients who received SRT, 96 patients were included in the analysis. Multivariate analysis identified several prognostic factors for OS, including the number of brain metastases (BM), molecular subtypes of breast cancer, BM as the site of first metastasis, Karnofsky Performance Status (KPS), and receiving systemic therapy after SRT. The final model, selected by the Akaike information criterion (AIC), included age, KPS, molecular subtype, number of BM, BM as the site of first metastasis, planning target volume (PTV), liver metastases, albumin levels, and neutrophil count. The calibration curve demonstrated good consistency between the nomogram predictions and actual observations. The C-index for the nomogram was 0.823 (95% CI, 0.760–0.885), and the bias-corrected C-index, generated via bootstrap validation with 1,000 resamples, was 0.772. This performance was significantly better than that of the Recursive partitioning analysis (RPA: C-index: 0.627), the Graded Prognostic Assessment (GPA: C-index: 0.637), and Breast-specific Graded Prognostic Assessment (breast-GPA: C-index: 0.699) systems. Kaplan-Meier survival curves showed excellent discrimination among four groups classified by this nomogram. CONCLUSION: The proposed nomogram provides more accurate predictions for BCBM patients undergoing SRT, outperforming traditional prognostic models.