TFF1/ESR1 Decoupling as a Biomarker for HDAC Inhibitor Benefit in Luminal B Breast Cancer: Discovery in Single-Cell RNA Sequencing, Independent Replication in METABRIC, and a Proposed Correlative Endpoint for NCT07235618

Luminal B (LumB) breast cancer has higher ESR1 (oestrogen receptor) transcript than Luminal A (LumA) yet is more resistant to endocrine therapy. This document reports two original findings from the OrganismCore attractor geometry framework that explain this paradox and generate a directly testable clinical prediction. FINDING 1 (CS-LIT-9): The TFF1/ESR1 ratio — TFF1 is a canonical ER target gene — is specifically lower in LumB than LumA despite LumB having higher raw ESR1 expression. This is ER output decoupling: the ER is present but its transcriptional output is suppressed. Discovered in GSE176078 single-cell RNA sequencing (Wu et al. Nature Genetics 2021; 19,542 cancer cells, 26 patients). Independently replicated in METABRIC bulk RNA sequencing (Curtis et al. Nature 2012; n=1,980; p=0.0019). Two cohorts. Two technologies. Same finding. FINDING 2 (CS-LIT-23): A DNMT3A/HDAC2 co-expression coupling specific to LumB (r=+0.267 in LumB vs r=+0.071 in LumA, p=5.68e-56) is the mechanism of this decoupling. DNMT3A and HDAC2 co-operate in NuRD/Sin3A co-repressor complexes to compact chromatin at ER target gene promoters. This predicts that entinostat (class I HDACi, HDAC1/2/3 selective) will restore ER output efficiency preferentially in LumB patients in whom the lock is present. The TFF1/ESR1 ratio at baseline identifies which patients have the lock. CLINICAL PREDICTION: The E2112 trial (entinostat + exemestane, n=608, no benefit) enrolled unselected HR+ patients including LumA. The framework predicts this was a selection failure, not a drug failure: LumA patients do not have the DNMT3A/HDAC2 lock and derive no benefit; they diluted the LumB signal to null. ENCORE 301 (post-letrozole, inadvertently LumB-enriched) showed OS benefit HR=0.59 (p=0.036). The meta-analysis (n=1,371, May 2025) shows PFS HR=0.80. China NMPA approved entinostat for HR+/HER2- in 2024. NCT07235618 (Phase II entinostat + fulvestrant, post-CDK4/6i failure, Sun Yat-sen University, start January 2026, n=50) is recruiting a population that is >80% LumB by PAM50. Adding TFF1/ESR1 IHC ratio as a correlative endpoint on baseline tissue would directly test this prediction without any protocol amendment, new patients, or additional patient burden. No published paper has computed TFF1/ESR1 as a named ratio, reported it lower in LumB than LumA, described the gap as ER output decoupling, connected it to DNMT3A/HDAC2 co-expression coupling, or proposed it as a patient selection variable for entinostat. This finding is original, two-cohort replicated, and actionable in an actively recruiting trial today. Full computational record: https://github.com/Eric-Robert-Lawson/attractor-oncology (Repository ID: 1172048282)