Abstract
BACKGROUND: Estrogen receptor-negative breast cancers are clinically heterogeneous. Two subtypes, Luminal Androgen Receptor and Molecular Apocrine Breast Cancers, are both defined by androgen signaling but identified through distinct transcriptomic classifiers. Their relationship remains unclear despite overlapping features. METHODS: We analyzed transcriptomic and genomic data from public and institutional cohorts using two classification systems. Gene set enrichment and immune deconvolution analyses were performed to characterize differences between overlapping and discordant tumors. A four-gene signature was developed and validated using RNA sequencing and RT-qPCR on both fresh-frozen and formalin-fixed samples. RESULTS: Substantial overlap was observed between the two subtypes, with shared activation of androgen and PI3K/AKT/mTOR signaling pathways. Discordant tumors were enriched in immune and stromal signals. A four-gene signature (AR, FOXA1, SPDEF, TFF3) identified overlapping tumors with high sensitivity and specificity across datasets and remained accurate in RT-qPCR assays on FFPE material. CONCLUSION: Luminal Androgen Receptor and Molecular Apocrine Breast Cancers constitute a single molecular entity within estrogen receptor-negative breast cancers. The validated four-gene signature enables robust clinical identification and may guide future therapeutic stratification.