Beyond one-size-fits-all: single-cell transcriptomic signatures predict drug efficacy and reveal responder subgroups in endometriosis

Abstract

Endometriosis affects [~]10% of reproductive-age women, yet targeted non-hormonal therapies remain unavailable, and treatment response is highly variable. Here, we apply a single-cell framework to resolve therapeutic heterogeneity at a resolution previously unattained in drug development efforts. Using scRNA-seq profiles from eutopic and ectopic tissues, combined with a machine learning-based drug response model, we identified compounds predicted to revert disease-associated transcriptional states and map cell-type-specific vulnerabilities across patients and tissues. Our analysis revealed pronounced tissue-specific and inter-patient heterogeneity in predicted responses. Stromal, endothelial, and stem cell populations emerged as the dominant therapeutic targets, collectively revealing selective sensitivity to two recurrent drug classes, histone deacetylase and tubulin polymerisation inhibitors. Transcriptomic comparison of predicted responders and non-responders to these drugs pointed to conserved molecular programmes involving extracellular matrix remodelling, angiogenesis, and proliferative activation. These signatures were shared between eutopic and ectopic stromal compartments, supporting the feasibility of assessing therapeutic response using readily accessible eutopic tissue. Our findings show that this single-cell framework can dissect therapeutic heterogeneity in endometriosis, support the development of precision non-hormonal therapies and identify responder subgroups relevant for patient stratification. Together, these results highlight that underlying molecular diversity in endometriosis necessitates therapeutic approaches beyond a one-size-fits-all model.