Purpose: Only a small percentage of Hispanic patients have been included in studies that developed prognostic models for breast cancer and brain metastases. Therefore, there is a clear need for tools tailored to this demographic. This study assesses the efficacy of common prognostic tools in a Hispanic population.
Methods and materials: We retrospectively analyzed a data set of Hispanic patients with breast cancer and newly diagnosed brain metastases from 2009 to 2023 at a single referral center. For each prognostic tool, Kaplan-Meier curves were built. The performances of the models were compared using the area under the curve (AUC), C-statistic, and Akaike information criteria (AIC).
Results: Of 492 patients, the median time from breast cancer to brain metastasis diagnosis was 22.7 months (IQR, 12.1-53.3). The median overall survival was 11.6 months (95% CI, 9.9-13.4). All models were validated as prognostic tools (P < .001). The model with the better performance was the breast graded prognostic assessment (GPA; AIC, 402; AUC, 0.65), followed by the modified GPA (AIC, 406; AUC, 0.64), the disease-specific GPA (AIC, 407; AUC, 0.62), recursive partitioning analysis (AIC, 421; AUC, 0.62), and GPA (AIC, 422; AUC, 0.60).
Conclusions: The breast GPA demonstrated superior accuracy in prognosticating outcomes for Hispanic patients with breast cancer and brain metastases. This underscores the critical importance of incorporating racial and ethnic diversity in creating and validating medical prognostic tools.
Copyright © 2024 Elsevier Inc. All rights reserved.