A new study from French biotech endogene.bio leveraging single-cell technology has mapped the profound molecular heterogeneity within endometriosis, revealing why patients respond so differently to treatments and offering a new framework for precision drug development.

The research moves beyond viewing the disease as a single entity and instead identifies distinct cellular programs that determine therapeutic vulnerability.

Endometriosis, affecting roughly 10% of reproductive-age women, is notorious for its variable treatment outcomes. Current hormonal therapies provide satisfactory results in only about half to two-thirds of patients, with advanced disease often showing poor response.

This clinical unpredictability has been a major barrier, contributing to high rates of clinical trial failure. The new study posits that this variability stems from fundamental biological differences at the cellular level, which have been obscured by traditional analysis methods.

A Single-Cell Blueprint of the Disease

To dissect this complexity, researchers integrated publicly available single-cell RNA sequencing (scRNA-seq) datasets from 110 samples, encompassing 672,590 high-quality cells from both eutopic (uterine) and ectopic (lesion) tissues. Using a machine learning classifier trained on a human endometrial cell atlas, they harmonized data across four studies to create a unified cellular map of the disease.

Their analysis confirmed stromal cells as the largest population, particularly in the eutopic endometrium (~40% of cells). Crucially, disease-associated transcriptional changes were concentrated in stromal, endothelial, and stem-like cell (eMSC) populations. These cells showed strong enrichment for pathways linked to epithelial-mesenchymal transition (EMT), hypoxia, and angiogenesis—key processes in lesion invasion, fibrosis, and survival.

Predicting Drug Response at Cellular Resolution

The core of the study applied a computational tool called scTherapy, which matches disease-associated gene expression signatures with large-scale drug perturbation databases to predict cell-type-specific drug sensitivities. This approach moves beyond bulk tissue analysis to pinpoint therapeutic vulnerabilities within the specific cell populations driving pathology.

The model identified several non-hormonal compound classes with predicted efficacy. The tubulin polymerization inhibitor, ABT-751 (formerly Eisai’s E7010), was the most frequently prioritized, predicted to be effective in approximately 85% of patient samples across cell types. Histone deacetylase (HDAC) inhibitors, particularly panobinostat, also emerged as recurrent candidates. Together, ABT-751 and panobinostat were predicted to target about 51.6% of stromal cells derived from peritoneal lesions.

Tissue and Patient-Specific Landscapes

A key finding was that drug response is not uniform. Predictions varied significantly by tissue origin:

• Eutopic Endometrium: Showed sensitivity to a broader range of compounds, especially HDAC inhibitors.
• Ectopic Peritoneum: Most responsive to ABT-751 and panobinostat.
• Ovarian Lesions: Exhibited a more selective pattern, with proteasome inhibitors like bortezomib being more prominent.

Furthermore, clustering analysis of individual patient predictions revealed that heterogeneous responses converged into a limited number of shared therapeutic profiles. This suggests patients can be stratified into subgroups with conserved molecular vulnerabilities, providing a rationale for stratified clinical trials.

A Path to Non-Invasive Monitoring

Perhaps one of the most translational findings is the conserved molecular response between hard-to-access lesions and readily available tissue. The study demonstrated that the transcriptomic signatures distinguishing drug responders from non-responders in peritoneal lesions were significantly mirrored in stromal cells from the eutopic endometrium.

This correlation was strong for both panobinostat and ABT-751. Since eutopic endometrial cells are shed during menstruation, this strongly supports the potential of using menstrual blood as a non-invasive liquid biopsy for patient stratification and longitudinal monitoring of treatment response.

Implications for Drug Development and Care

The study argues for a paradigm shift in endometriosis research and development, akin to the transformation in oncology. By defining diseases by underlying cellular programs rather than anatomical features, drug developers can design smarter trials.

• For Pharma: Stratifying trial populations by molecular subtype could reduce late-stage failure rates and development costs.
• For Clinicians: A biology-driven framework may eventually guide personalized treatment choices.
• For Patients: It promises clearer explanations for treatment outcomes and reduces reliance on invasive surgical biopsies for monitoring.

The authors note limitations, including the need for experimental validation of computational predictions and the current bias of perturbation databases toward cancer cell lines. However, the biological overlap between cancer and endometriosis progression supports the relevance of the approach.

This research provides a scalable framework to turn the long-standing challenge of endometriosis heterogeneity into a definable feature, possibly paving the way for a new era of precision medicine in women’s health.