How are fibroblasts and neuroimmune cells spatially organized in endometriosis lesions?

Two fibroblast zones surround epithelial glands

The paper's organizing result is that endometriosis lesions are not a uniform mix of cells. Using spatial transcriptomics, which measures gene expression while keeping each measurement tied to its position in the tissue section, the authors defined three regions by distance from epithelial glands: Epi (the gland itself), Close Stroma directly around it, and Far Stroma further out.

These zones have distinct fibroblast signatures. Close Stroma is enriched in MME, MMP11 and WNT2, while Far Stroma is enriched in OGN, ASPN and FILIP1L. The authors report this pattern across two spatial transcriptomic platforms and one protein-expression platform, in both ovarian and peritoneal lesions. In the low-resolution Visium V2 data, spots were labeled EPCAM+ for glands, MME+ for close fibroblasts and OGN+ for far fibroblasts; in Visium HD, unsupervised clustering identified the fibroblast compartments, with the gland-adjacent, MME-high clusters labeled Close Stroma and the OGN-high clusters outside them labeled Far Stroma.

Ovarian and peritoneal lesions share architecture

Ovarian and peritoneal lesions differ morphologically and clinically, yet the study found compelling similarities between their spatial clusters. The authors describe this as a shared spatial niche structure, and they used it to look for common features rather than treating each lesion type separately.

The comparison is not a claim that the two lesion types are identical. The paper reports trends and averages across lesions while noting patient-specific heterogeneity in individual tissues. The shared feature is the spatial arrangement — glands, a close fibroblast ring, and a far stromal region — not an identical abundance of every cell type in every patient.

Neurons and macrophages converge near glands

Within this architecture, neurons and macrophages are positioned close to each other. The authors compared distances between macrophages and neurons with distances between macrophages and other cell types, normalized to those other cell types, and found macrophages and neurons sit closer together without necessarily making immediate contact. This was particularly emphasized within the Epi region of ovarian lesions and the Far Stroma of peritoneal lesions. Imaging of CD68+ macrophages and βIII-tubulin+ neurites confirmed close proximity near glands.

Ligand-receptor analysis added a signaling dimension. In Close Stroma, the authors observed extensive bidirectional communication between macrophages and neurons across families including PDGF, SEMA-NRP/PLXNA, MDK-LRP1 and multiple COL-ITG extracellular matrix interactions. In Far Stroma, this network was markedly reduced, which the authors interpret as baseline neuroimmune activity rather than amplified lesion-associated signaling.

Macrophage subtypes also followed the zones. The paper reports a trend toward increased activated macrophages in Epi and Close Stroma for both lesion types, with ovarian lesions showing a significant increase in LYVE1+ and peritoneal macrophages within Close Stroma. These macrophages expressed SPP1, FN1 and SPARC. Independent Xenium samples confirmed increased macrophage transcripts (CD68, CD163, MRC1, CD14, FCGR3A) within Close Stroma.

Nociceptors are enriched in the epithelial niche

The study subclassified neuronal bins by subtype-specific markers. Nociceptors, marked by TRPV1 and NTRK1 and expressing pain receptors such as SCN9A and TRPV1, were significantly enriched within the Epi cluster in both ovarian and peritoneal lesions, and accounted for an average of 18% of neuronal bins across lesions. Mechanoreceptors marked by NTRK2 were the predominant subset at an average of 40% of neuronal bins and showed a trend toward enrichment in Close and Far clusters. Sympathetic neurons were the least represented at roughly 0.1% of the neuronal compartment. About 17% of neuronal bins could not be subclassified despite expressing pan-neuronal markers.

Independent validation supported the localization. Immunohistochemistry showed minimal colocalization between βIII-tubulin and CD10, supporting βIII-tubulin as a neuronal marker, and imaging mass cytometry showed increased neuronal βIII-tubulin in Close Stroma of both lesion types. In the Xenium cohort, nociceptor transcripts (SCN9A, TRPV1, TRPA1, NTRK1, CALCA, TAC1) were increased in Close Stroma.

A 3D co-culture tests epithelial-neuron signaling

Spatial transcriptomics shows where cells sit but not what happens when they interact. To address this, the authors built an in vitro 3D model using peripheral sensory brain organoids co-cultured with human endometriosis epithelial and fibroblast cells. This model was used to validate the epithelial-neuronal interactome suggested by the spatial data.

The paper is explicit about what this model can and cannot show. The authors state that spatial profiling does not provide mechanistic insight, and that the co-culture recapitulates some critical cell types and functional interactions to begin to understand dynamics between neuron and epithelial partners. The organoid result therefore supports the plausibility of the interaction; it does not establish a causal chain in patients.

B-cell aggregates mark the far stromal niche

Beyond neurons and macrophages, the study reports B-cell aggregates in the Far cluster consistently across all ovarian and peritoneal lesions. The authors connect this to prior reports of peripheral and peritoneal fluid B cells in endometriosis and to plasma cells producing autoantibodies against the endometrium.

Their interpretation is that these aggregates likely contribute to a chronic inflammatory environment in the Far Stroma and represent a common immunological mark across lesion subtypes. This is presented as a suggested contribution and a possible therapeutic angle, not as a demonstrated mechanism. The paper notes that most B-cell biology in female reproductive pathologies remains unknown.

Resolution changes what the map can show

A practical point in the paper is that low-resolution spatial transcriptomics did not reliably depict the positioning of immune and neuronal cells and provided only a discontinuous map of lesions. The authors contrast this with their higher-resolution Visium HD and Xenium data, which allowed bin-level assignment of epithelial cells, fibroblasts, vascular cells, neurons, monocytes/macrophages, T cells, NK cells, B cells, dendritic cells and mast cells.

The study also carries boundaries that shape how far the findings travel. Sample sizes differ by platform: Visium V2 covered three samples from two patients, Visium HD covered eight samples from seven patients, and Xenium covered four lesions. Some comparisons, such as macrophage subtype proportions, rest on small numbers (n = 1 ovarian and n = 3 peritoneal lesions in one analysis). The authors describe the spatial data as a snapshot of dynamic interactions, and they note patient-specific heterogeneity in neuronal composition. The follicle analysis linking AMH level to distance from the epithelial gland comes from a single ovarian lesion with 35 follicles.

The close stroma is the priority target

The paper's most decision-relevant distinction is between the two stromal zones. Close Stroma concentrates the features most tied to pain signaling: nociceptor enrichment, activated SPP1/FN1/SPARC macrophages, close macrophage-neuron positioning, and the broadest ligand-receptor communication. Far Stroma retains B-cell aggregates and a reduced neuroimmune network.

For readers comparing lesion types or evaluating therapeutic angles, this means the gland-adjacent fibroblast compartment, not the lesion as a whole, is where the neuroimmune features converge. The authors frame this conserved close-stroma neuroimmune microenvironment as the emerging insight into pathophysiology and the basis for targeted strategies. The evidence supports that priority as a spatial association observed across lesion types and platforms; it does not yet show that targeting this zone changes pain or disease progression.

Sources used in this answer