Abstract BACKGROUND Up to 50% of patients with advanced-stage breast cancer develop brain metastases (BrMets), a diagnosis associated with a median survival of just 10 months and significant neurologic decline. However, BrMet risk is not uniform among breast cancer patients, and despite subtype-dependent disparities, the underlying susceptibility remains poorly understood, with a critical lack of predictive biomarkers to identify patients at highest risk. METHODS We constructed a clinically annotated tissue microarray (TMA) comprising breast cancer BrMets, adjacent brain tissue, associated primary breast tumors, and non-cancer normal brain samples from patients (n = 149) treated at Stanford Hospital (2008 – 2018). Using genome-wide transcriptome analysis combined with high resolution spatial imaging (GeoMx Digital Spatial Profiling), we profiled tumor, immune, and normal brain cell gene expression in both primary and brain metastatic tumor microenvironments. RESULTS Differential expression analysis identified a 484-gene “Progressor Signature” upregulated in primary tumors of patients who developed BrMets (n = 34) compared to metastatic breast cancer patients who never developed BrMets (MRI-confirmed, n = 29). This signature independently predicted BrMet-free and overall survival (BMFS: log-rank p 0.0001, OS: log-rank p 0.011). Validation in external datasets showed that higher expression of the Progressor Signature correlated with lack of pathological complete response (p = 0.03), increased distant metastasis risk (p = 1.8e07), and shorter time to distant metastasis (log-rank p = 0.0024), underscoring its clinical relevance. Using our novel Equivalent Expression Index, we refined this signature to 154, then 45, genes based on shared expression in BrMets and adjacent brain tissue, enhancing its translational potential. CONCLUSION Our findings provide a foundation for identifying molecular drivers of BrMet progression. Further studies validating the Progressor Signature in independent cohorts, correlating it with primary tumor microenvironment cell composition, and assessing its detectability in cerebrospinal fluid (CSF) could enable early BrMet detection and inform therapeutic strategies.
Garcia et al. (Fri,) studied this question.