A validated clinical calculator of a brain metastasis risk (BrMR) prediction model in stage I-III breast cancer patients

BACKGROUND: Brain metastasis (BM) significantly affects survival and quality of life in patients with breast cancer. A validated tool to predict BM risk at diagnosis of early breast cancer (eBC) could enable targeted surveillance and improve patient selection for clinical trials. METHODS: Patients with stage I-III eBC diagnosed from January 1, 1997, to May 6, 2020, at The University of Texas MD Anderson Cancer Center were reviewed as a training cohort. After data curation and multiple imputation of missing values, a linear predictive model (Brain Metastasis Risk [BrMR]) was developed and implemented as an online calculator. The model was applied to an independent validation cohort of stage I-III patients referred to Institut Jules Bordet, Belgium, from 2001 to 2021. RESULTS: BM occurred in 2408 of 40 282 training-cohort patients and 35 of 1644 validation-cohort patients. BrMR showed strong discrimination (C-index = 0.79, training; 0.76, validation), with areas under the receiver operating characteristic curve of 0.81, 0.79, 0.76, and 0.75 at 4, 6, 8, and 10 years, respectively. Inflammatory breast cancer, HER2 positivity, higher Ki-67, lymphovascular invasion, adjuvant chemotherapy, grade III, and higher T/N-status were associated with increased BM risk; ER/PR positivity, older age, endocrine therapy, lobular histology, grade II, and lower T-status were associated with decreased risk. CONCLUSIONS: BrMR accurately predicts BM risk in eBC and may support risk stratification and trial enrichment. Prospective validation is needed before clinical use as a surveillance tool.