Abstract A187: A transitional molecular state predicts early mortality and is differentially represented across racial groups in ER-positive breast cancer

Abstract

Black women with ER-positive breast cancer experience higher mortality than White women despite comparable diagnoses, yet standard receptor-based classification treats ER+ tumors as biologically uniform. We hypothesized that a coordinated transitional molecular state, rather than any single gene, is differentially represented across racial groups and associated with adverse clinical outcomes. Using TCGA-BRCA (N=786; 157 Black, 539 White), we integrated DNA methylation (principal component analysis of 1,127 CpG probes) with RNA-seq to characterize tumor states along a continuous EMT–luminal (ESR1) landscape. We then quantified each tumor's position along the epithelial–mesenchymal transitional continuum using a reference-anchored, four-gene score (the Hybrid Lock Score, HLS). We characterized the molecular programs distinguishing these states, tested racial enrichment using regression adjusted for tumor stage and ER status (Benjamini-Hochberg FDR), and evaluated the HLS as a prognostic marker in the independent METABRIC cohort (n=383 high-risk ER+, 180 events) using multivariable Cox regression. The methylation axis distinguished three molecular states along a directional gradient (ANOVA F=196.97, p<0.0001): Stable ER+, a Transitional ER+ intermediate, and Complete Loss (TNBC). The Transitional ER+ state was characterized by partial loss of luminal identity, elevated EMT, epigenetic remodeling (EZH2 elevation, GREB1 suppression), and attenuated circadian regulation; CLOCK showed the largest race-associated effect among 18 candidate genes (Cohen's d=-0.83; adjusted β=-0.58, q<0.001), with concordant suppression of CRY1 and ARNTL/BMAL1. Black patients were disproportionately represented in the Complete Loss (TNBC) state (race distribution across states χ2=26.98, p<0.0001); moreover, within the Transitional ER+ compartment, they exhibited higher HLS (Hedges' g=0.71, p=0.0018; adjusted β=0.49, p=0.006), indicating greater luminal-identity loss even among transitional tumors. In METABRIC, higher HLS independently predicted mortality (HR=1.28 per standard deviation, p=0.004; adjusted for age, stage, PR, HER2), with risk concentrated in the first five years (0–60 months HR=1.80, p≈10-5) and absent thereafter (HR=1.00). These findings identify a coordinated Transitional ER+ state, marked by epigenetic, circadian, and EMT program shifts, that is differentially represented across racial groups and independently associated with early mortality. Rather than a single-gene account of disparities, this state-centric framework provides a reproducible foundation for studying coordinated biological programs associated with cancer health disparities and for biologically informed pre-treatment risk stratification. Generative AI was used to assist with language editing of this abstract; all data, analyses, interpretations, and final content were reviewed and approved by the author. Audrey Guo. A transitional molecular state predicts early mortality and is differentially represented across racial groups in ER-positive breast cancer [abstract]. In: Proceedings of the 19th AACR Conference on the Science of Cancer Health Disparities; 2026 Oct 18-21; Atlanta, GA. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2026;35(10 Suppl):Abstract nr A187.