{"paper_id":"e6ff922e-3fb2-49e2-9e93-c27a591ddf1f","body_text":"Abstract\nBackground:\nStromal heterogeneity, regulated-cell-death-associated stress, and chronic inflammation contribute to endometriosis, but their relationships within specific stromal populations remain incompletely understood. This study integrated clinical characterization with single-cell and spatial transcriptomics to investigate a TCF21-associated stromal death–inflammation signature.\nMethods:\nA retrospective cohort of 150 patients with histopathologically confirmed endometriosis was characterized clinically. Public single-cell datasets (GSE179640 and GSE213216) were analyzed to identify stromal populations, reconstruct transcriptional trajectories, evaluate death-associated and inflammatory programs, and infer intercellular communication. Spatial transcriptomic analysis assessed TCF21-associated expression patterns in endometriotic lesions. Multiplex immunofluorescence, RT-qPCR, TCF21 silencing, EdU incorporation, Transwell assays, and three-dimensional spheroid culture were used for experimental evaluation.\nResults:\nSingle-cell analysis identified 13 major cell classes and a fib-TCF21-associated stromal state characterized by TCF21, NEXN, IFI44L, WNT2B, and SPRR2F. Cell-death-associated and inflammatory transcriptional programs varied across stromal states, with fib-TCF21 exhibiting selective rather than uniformly elevated program activity. Independent single-cell analysis identified a related but anatomically heterogeneous stromal expression pattern. Spatial transcriptomics demonstrated localized TCF21 expression and heterogeneous distributions of stromal, stress-associated, and inflammatory programs. Tissue immunofluorescence supported stromal-associated TCF21 expression, while lesion-derived ihESCs exhibited higher basal TCF21 expression than the eutopic stromal-cell model. TCF21 silencing was associated with reduced EdU incorporation, Transwell membrane traversal, and three-dimensional spheroid growth.\nConclusion:\nIntegration of clinical, single-cell, spatial, and experimental findings identifies a TCF21-associated stromal transcriptional signature linked to heterogeneous death-related stress and inflammatory programs in endometriosis. These findings support further investigation of TCF21-associated stromal biology and its potential relevance to molecularly targeted interventions.\n1 Introduction\nEndometriosis is a chronic, estrogen-dependent inflammatory disorder characterized by the presence of endometrial-like tissue outside the uterus. Affecting approximately 5%–10% of women of reproductive age, it is associated with pelvic pain, infertility, and impaired quality of life (Zondervan et al., 2018; ; ). Ectopic lesions develop in diverse anatomical sites and contain epithelial, stromal, immune, and vascular components. Their persistence involves complex interactions among hormonal dysregulation, immune dysfunction, chronic inflammation, and extracellular matrix (ECM) remodeling ( ; ). Fibrosis is particularly prominent in ovarian endometriomas and deep endometriosis, where repeated tissue injury promotes collagen deposition and altered stromal-cell behavior ( ). However, the cellular programs underlying these pathological changes remain incompletely understood.\nSingle-cell RNA sequencing (scRNA-seq) has substantially advanced our understanding of cellular heterogeneity in endometriosis. Studies of normal, eutopic, and ectopic tissues have identified diverse stromal and immune populations with distinct transcriptional programs related to angiogenesis, immunotolerance, fibroblast differentiation, and ECM remodeling (\n; ; ). Stromal cells, in particular, exhibit considerable heterogeneity in contractile, inflammatory, and matrix-remodeling properties. Recent integration of single-cell and spatial transcriptomics has further revealed anatomically distinct stromal populations and signaling microenvironments within ovarian endometriomas ( ). These findings highlight the importance of stromal heterogeneity and spatial organization in lesion biology. Nevertheless, how specific stromal states interact with cell-death-associated stress responses and inflammatory signaling remains poorly characterized.Regulated cell death (RCD) is increasingly recognized as an important component of inflammatory diseases. Apoptosis, ferroptosis, pyroptosis, and necroptosis involve distinct molecular mechanisms and can influence tissue homeostasis, immune responses, and inflammatory signaling (Wu et al., 2023). Similarly, pyroptotic T-cell-derived interleukin-16 and prostaglandin E2-associated HMGB1 signaling have been implicated in endometriosis-related inflammation and cellular behavior (Zhang et al., 2024; ). These findings suggest that cell-death-associated injury, stress adaptation, and inflammatory signaling may coexist within ectopic lesions, although their contributions are likely to vary across cellular populations.\n; ). Endometriotic lesions are repeatedly exposed to hemorrhage, iron accumulation, and oxidative stress, creating conditions that may alter both cell-death susceptibility and adaptive survival responses. Experimental evidence indicates that ferroptotic endometrial stromal cells can promote angiogenic and inflammatory mediator release, whereas macrophage-derived interleukin-33 enhances stromal resistance to ferroptotic stress through ATF3/SLC7A11 signaling ( ;The stromal compartment may represent an important interface between cell-death-associated responses and inflammatory remodeling. Distinct stromal populations may exhibit different responses to oxidative injury, inflammatory cytokines, and extracellular signals, while interactions with macrophages, endothelial cells, and epithelial cells may further shape the local microenvironment. However, conventional transcriptomic approaches cannot adequately resolve these cell-state-specific relationships. Combining single-cell profiling, transcriptional trajectory reconstruction, functional enrichment, intercellular communication inference, and spatial transcriptomics provides an opportunity to characterize stromal programs associated with cellular stress and inflammation. Importantly, enrichment of RCD-related transcriptional signatures reflects associated molecular programs rather than direct evidence of a particular cell-death process. Understanding their cellular specificity and spatial distribution may therefore reveal biologically relevant stromal states and potential molecular targets.\nIntegrating clinical characteristics with molecular analyses may provide additional context for understanding disease heterogeneity. Endometriosis exhibits considerable variation in anatomical distribution, lesion severity, pain burden, infertility, and inflammatory characteristics (Zondervan et al., 2018; ; ; ). These features describe the clinical manifestations of the disease but cannot independently resolve the cellular and molecular organization of ectopic lesions. A biologically informed stromal signature may therefore complement conventional clinical characterization by identifying cellular programs associated with lesion remodeling and inflammatory activity.\nIn the present study, we integrated clinical phenotyping with single-cell and spatial transcriptomic analyses to investigate stromal heterogeneity and death–inflammation-associated programs in endometriosis. Clinical characteristics were evaluated in 150 patients with surgically and histopathologically confirmed endometriosis. Public transcriptomic datasets were analyzed to characterize cellular composition, stromal subpopulations, inferred trajectories, functional programs, and intercellular communication. Particular attention was directed toward a TCF21-associated stromal state and a five-gene signature comprising TCF21, NEXN, IFI44L, WNT2B, and SPRR2F. Spatial transcriptomics was used to investigate the tissue distribution of TCF21-associated programs, while tissue-based expression analysis and targeted cellular experiments were performed to evaluate the biological relevance of TCF21. Together, these analyses aimed to characterize a stromal death–inflammation signature and provide insight into the cellular and spatial relationships among stromal remodeling, stress-associated responses, and inflammation in endometriosis.\n2 Materials and methods\n2.1 Study design and participants\nThis study integrated clinical phenotyping, public single-cell and spatial transcriptomic analyses, and TCF21-focused tissue and cellular experiments. The retrospective cohort included 150 patients who underwent surgery for histopathologically confirmed endometriosis at Dalian Women and Children’s Medical Center (Group). The study was approved by the institutional Ethics Committee (FEJT-KY-2026-58), conducted in accordance with the Declaration of Helsinki, and granted a waiver of individual consent for retrospective use of de-identified records and archived biospecimens.\n2.2 Clinical variables\nVariables included age, body mass index, reproductive history, infertility, dysmenorrhea visual analogue scale, chronic pelvic pain, deep dyspareunia, dyschezia, and EHP-30 pain-domain scores (Tables 1 and 2.\n). Laboratory measures included anti-Müllerian hormone, antral follicle count, CA125, neutrophil-to-lymphocyte ratio (NLR), systemic immune-inflammation index (SII), and coagulation indices. NLR was calculated as neutrophils divided by lymphocytes; SII was calculated as platelets multiplied by neutrophils and divided by lymphocytes. Anatomical and operative characteristics included ovarian endometrioma, laterality, maximum lesion diameter, deep endometriosis and involved sites, adenomyosis, rASRM stage, procedures, operative time, estimated blood loss, and complications ( ; ). Multiple deep-lesion sites in the same patient were permitted. The cohort was characterized descriptively inTABLE 1\n| Characteristic | Overall (N = 150) |\n|---|---|\n| Demographic and reproductive characteristics | |\n| Age, years | 34.0 ± 5.9 |\n| Body mass index, kg/m2 | 22.8 ± 2.7 |\n| Age at menarche, years | 12.7 ± 1.1 |\n| Prior endometriosis surgery, n (%) | 23 (15.3) |\n| Infertility, n (%) | 68 (45.3) |\n| Nulligravidity, n (%) | 57 (38.0) |\n| Pain and symptom burden | |\n| Symptom duration, years | 5.3 [3.5–7.4] |\n| Dysmenorrhea VAS score | 5.8 ± 1.5 |\n| Chronic pelvic pain, n (%) | 48 (32.0) |\n| Deep dyspareunia, n (%) | 42 (28.0) |\n| Dyschezia, n (%) | 28 (18.7) |\n| Urinary-tract symptoms, n (%) | 10 (6.7) |\n| EHP-30 pain-domain score | 35.9 ± 13.3 |\n| Preoperative ovarian-reserve and laboratory profile | |\n| Anti-Müllerian hormone, ng/mL | 2.73 [2.05–3.73] |\n| Antral follicle count | 10 [8–13] |\n| CA125, U/mL | 59.2 [44.4–78.8] |\n| Neutrophil-to-lymphocyte ratio | 2.2 [1.7–2.8] |\n| Systemic immune-inflammation index | 518.2 [378.6–717.0] |\n| Fibrinogen, g/L | 3.2 ± 0.5 |\n| D-dimer, mg/L | 0.3 [0.2–0.4] |\n| Activated partial thromboplastin time, s | 29.0 ± 2.3 |\nBaseline clinical, symptom, and laboratory characteristics of the 150-patient endometriosis cohort.\nValues are mean ± SD, median [IQR], or n (%), as appropriate.\nVAS, visual analogue scale; EHP-30, Endometriosis Health Profile-30; AMH, anti-Müllerian hormone; NLR, neutrophil-to-lymphocyte ratio; SII, systemic immune-inflammation index.\nTABLE 2\n| Characteristic | Overall (N = 150) |\n|---|---|\n| Anatomic disease phenotype | |\n| Ovarian endometrioma, n (%) | 125 (83.3) |\n| Laterality among patients with ovarian endometrioma | |\n| Unilateral | 80/125 (64.0) |\n| Bilateral | 45/125 (36.0) |\n| Largest lesion diameter, cm | 5.2 ± 1.7 |\n| Deep endometriosis, n (%) | 53 (35.3) |\n| Deep-disease site | |\n| Uterosacral-ligament involvement | 32 (21.3) |\n| Rectovaginal/vaginal involvement | 17 (11.3) |\n| Bowel involvement | 20 (13.3) |\n| Bladder/ureteral involvement | 6 (4.0) |\n| Coexisting adenomyosis, n (%) | 33 (22.0) |\n| Disease severity and operative phenotype | |\n| rASRM stage, n (%) | |\n| Stage I–II | 32 (21.3) |\n| Stage III–IV | 118 (78.7) |\n| Surgical and perioperative characteristics | |\n| Complete excision, n (%) | 120 (80.0) |\n| Bilateral ovarian cystectomy, n (%) | 38 (25.3) |\n| Ureterolysis, n (%) | 16 (10.7) |\n| Bowel shaving/disc excision/segmental resection, n (%) | 17 (11.3) |\n| Operative time, min | 107.8 ± 35.5 |\n| Estimated blood loss, mL | 56.8 [39.2–86.6] |\n| Postoperative length of stay, days | 4.1 ± 1.1 |\n| Clavien–Dindo grade ≥II complication, n (%) | 6 (4.0) |\nAnatomic phenotype, disease severity, and surgical characteristics of the 150-patient endometriosis cohort.\nValues are mean ± SD, median [IQR], or n (%).\nrASRM, revised American Society for Reproductive Medicine.\n2.3 Transcriptomic datasets and preprocessing\nGSE179640 served as the single-cell discovery dataset, comprising an analyzed subset of 16 specimens (three normal, nine eutopic and four ectopic tissues) (\n). GSE213216 was used for independent assessment ( ). Expression matrices were linked to GSM identifiers, donor information and tissue origin. Analyses were performed in R version 4.3.2 with Seurat version 4.4.0. Cells with fewer than 200 or more than 6,000 detected genes, or a mitochondrial transcript proportion above 20%, were excluded; potential doublets were screened with DoubletFinder version 2.0.3. Following quality control, sample-specific retained-cell counts were recorded in the analysis metadata. Data were log-normalized using a scale factor of 10,000, 2,000 highly variable genes were selected, and datasets were integrated using Seurat’s canonical correlation analysis (CCA) anchor-based workflow. Principal-component analysis was followed by UMAP using the first 30 principal components; graph-based clustering used a resolution of 0.5 and random seed 1234.2.4 Cell identities and stromal-state analysis\nCell identities were assigned from canonical markers and published annotations. DCN/COL1A1 indicated stromal cells; RGS5/ACTA2 marked mural cells; PECAM1 marked endothelial cells; EPCAM and keratins marked epithelial cells; and lineage markers identified myeloid and lymphoid cells. MPZ-associated cells were provisionally annotated as Schwann-like. Cell fractions and observed/expected enrichment ratios were summarized by biological specimen. Stromal and mural cells were extracted and reclustered to distinguish eF-HOXA10, eF-COL7A1, eF-myo, eF-HOXA13, fib-TCF21, vascular smooth-muscle and perivascular states. State-associated markers were identified using Seurat’s Wilcoxon rank-sum test (FindAllMarkers; min.pct = 0.25, |log2 fold change| > 0.25, Benjamini–Hochberg-adjusted P < 0.05).\n2.5 Pseudotime and candidate signature\nStromal and mural cells were subjected to pseudotime analysis using Monocle 2 with DDRTree (\n). The trajectory was rooted in a region enriched in normal or eutopic stromal cells. Tissue and cell-state annotations were projected onto the trajectory, and pseudotime-associated genes were evaluated using Monocle’s differentialGeneTest with a natural-spline model of pseudotime and a false-discovery-rate threshold of q < 0.05. UCell ( ) was used to compare signature activity across cell states and tissue groups. Diffusion pseudotime served as an independent ordering analysis ( ).2.6 Functional pathways and death–inflammation programs\nGene Ontology and KEGG enrichment of stromal-state-associated genes was performed with clusterProfiler version 4.10.1 and org.Hs.eg.db version 3.18.0, using Benjamini–Hochberg-adjusted P < 0.05. UCell (\n) was applied to curated gene sets covering apoptosis, ferroptosis, pyroptosis and necroptosis, with autophagy and oxidative stress scored as separate stress-associated processes. Inflammatory programs included TNF/NF-κB, IL6/JAK/STAT3, interferon response, TGF-β, complement and chemokine signaling. Scores were summarized across cell types, stromal states and tissues; standardized values were displayed in heatmaps. Candidate-gene associations with pathway scores were examined using Spearman correlation.2.7 Cell communication and regulon analyses\nCell–cell communication was inferred separately for eutopic and ectopic samples using CellChat version 1.6.1 and CellChatDB.human (\n). Cell groups containing fewer than 10 cells were excluded; communication probabilities were estimated by the triMean procedure with 100 bootstrap iterations. Transcription-factor regulatory programs were inferred using SCENIC version 1.3.1, GENIE3-based coexpression, human cisTarget motif resources and AUCell activity scoring ( ).2.8 External single-cell and spatial analyses\nGSE213216 was processed separately to identify stromal populations sharing features of the discovery-defined fib-TCF21 state. Marker expression, positive-cell fractions and five-gene signature scores were assessed by anatomical tissue group while retaining donor identifiers. Spatial transcriptomic analyses used Visium sections GSM6690475 (BEME_346) and GSM6690476 (BEME_355G), including their count matrices, spot coordinates and hematoxylin-and-eosin tissue images. Spot quality was assessed using library depth, genes detected and mitochondrial fraction. Spatial domains were defined using Seurat graph-based clustering on the first 30 principal components (resolution 0.4). Normalized TCF21 expression, the five-gene signature and selected death/stress and inflammatory programs were mapped onto tissue images; spatial structure was assessed using Moran’s I, with within-section Spearman correlations used for descriptive co-distribution.\n2.9 Tissue multiplex immunofluorescence\nFormalin-fixed, paraffin-embedded eutopic endometrial and ectopic endometriotic tissues were sectioned at 4 μm. Slides underwent deparaffinization and heat-induced antigen retrieval in 10 mM sodium citrate buffer (pH 6.0; 95 °C for 20 min). Sequential multiplex immunofluorescence with tyramide signal amplification was performed using anti-TCF21 (SAB,C56612; 1:100), anti-vimentin (Cell Signaling Technology, #5741; 1:200) and anti-CD10 (Cell Signaling Technology, #65534; 1:250). The respective channels were displayed in yellow (TCF21), green (vimentin) and orange/red (CD10), with DAPI nuclear counterstaining. Separate fluorescence channels and merged images were obtained using a Leica TCS SP8 confocal microscope with LAS X acquisition software under comparable imaging settings. Morphologically defined stromal regions were examined using Fiji/ImageJ version 2.14.0 in five nonoverlapping fields per specimen; field-level measurements were averaged within each tissue specimen.\n2.10 Cell models, RT-qPCR and TCF21 silencing\nT HESCs, hEM15A and ihESCs were used as immortalized endometrial stromal models from different donor and tissue contexts (\n; ). T HESCs were maintained in DMEM/F12 supplemented with 10% charcoal/dextran-treated fetal bovine serum, 1% ITS+ and 500 ng/mL puromycin; hEM15A and ihESCs were maintained in DMEM/F12 with 10% fetal bovine serum at 37 °C and 5% CO2. Cells were screened for mycoplasma. RNA was extracted with RNAiso Plus (Takara, #9108); 500 ng of total RNA was reverse-transcribed using PrimeScript RT reagent kit (Takara, #RR047A). RT-qPCR was performed using TB Green Premix Ex Taq II (Takara, #RR820A) on a Roche LightCycler 480. TCF21 primers were 5′-TCCTGGCTAACGACAAATACGA-3′ (forward) and 5′-TTTCCCGGCCACCATAAAGG-3′ (reverse); GAPDH was used with primers 5′-CAGCCTCAAGATCATCAGCA-3′ and 5′-GTCTTCTGGGTGGCAGTGAT-3′. Relative expression was calculated using 2−ΔΔt after confirming comparable amplification efficiencies ( ). For functional experiments, ihESCs were transfected at 70% confluence with 50 nM TCF21-targeting siRNA (TCF21 sense sequence 5′-ATGATCAGCTGGGCCAAGAA-3′) using Lipofectamine RNAiMAX (Invitrogen). Medium was renewed after 6 h and TCF21 knockdown evaluated at 48 h.2.11 EdU, transwell and spheroid assays\nFor EdU analysis, transfected ihESCs were exposed to 10 μM EdU for 3 h, followed by click-chemistry detection, DAPI staining and quantification of EdU-positive nuclei. Invasion was assessed using 24-well, 8-μm-pore Transwell inserts (Corning) precoated with Matrigel diluted 1:8. 2 × 105 transfected cells were seeded in serum-free DMEM/F12 in the upper chamber, and DMEM/F12 containing 10% fetal bovine serum was added below. After 48 h, membranes were fixed and stained; invaded cells were counted in five randomly selected fields per insert. Three-dimensional spheroids were formed by seeding 5 × 103 cells per well in ultra-low-attachment, round-bottom 96-well plates. Bright-field images were obtained on days 1, 2 and 3, and spheroid projected area was measured using Fiji/ImageJ after scale calibration.\n2.12 Statistical analysis\nStatistical analyses were performed using GraphPad Prism version 10.2.3 and R version 4.3.2. Clinical variables were expressed as mean ± standard deviation, median (interquartile range) or n (%). Two-group comparisons used Student’s t-test or Mann–Whitney U test and comparisons involving more than two groups used one-way analysis of variance or Kruskal–Wallis testing. Two-sided P < 0.05 indicated statistical significance.\n3 Results\n3.1 Clinical features of the endometriosis cohort\nThe clinical cohort included 150 patients with surgically and histopathologically confirmed endometriosis. The mean age was 34.0 ± 5.9 years and mean body mass index was 22.8 ± 2.7 kg/m2. Infertility was recorded in 68 patients (45.3%), 57 (38.0%) were nulligravid, and 23 (15.3%) had undergone previous endometriosis surgery. Symptoms had been present for a median of 5.3 years (interquartile range [IQR], 3.5–7.4). Mean dysmenorrhea visual analogue scale (VAS) score was 5.8 ± 1.5; chronic pelvic pain, deep dyspareunia, and dyschezia were documented in 32.0%, 28.0%. The median CA125 level was 59.2 U/mL (IQR, 44.4–78.8), neutrophil-to-lymphocyte ratio was 2.2 (IQR, 1.7–2.8), and systemic immune-inflammation index was 518.2 (IQR, 378.6–717.0) (Table 1).\nOvarian endometrioma was present in 125 patients (83.3%); 80 of these patients (64.0%) had unilateral and 45 (36.0%) had bilateral disease. The mean maximum lesion diameter was 5.2 ± 1.7 cm. Deep endometriosis was documented in 53 patients (35.3%), and adenomyosis in 33 (22.0%). Advanced anatomical disease (rASRM stage III–IV) accounted for 118 patients (78.7%). Complete excision was recorded in 120 (80.0%), with mean operative time of 107.8 ± 35.5 min and median estimated blood loss of 56.8 mL (IQR, 39.2–86.6). These findings illustrate the substantial symptomatic and anatomical burden represented in the cohort (Table 2).\n3.2 Single-cell profiling reveals remodeling of the endometriotic cellular landscape\nReanalysis of the GSE179640 discovery subset, comprising three normal, nine eutopic, and four ectopic tissue specimens, resolved 13 major cell classes (Figures 1A–C). These included stromal, pericyte, putative Schwann-like, endothelial, and epithelial populations; T, natural killer, B, and plasma cells; macrophages, mast cells, plasmacytoid dendritic cells, and cycling cells. Canonical markers supported the assigned cellular identities, while sample-colored and tissue-split UMAPs illustrated the distribution of these populations across specimens (Figures 1A–D).\nFIGURE 1\nMarked differences in captured cell composition were apparent among tissue categories (Figures 1E–H). Epithelial cells contributed a larger fraction of normal samples and were minimally represented in the ectopic subset. In contrast, stromal and putative Schwann-like populations were more prominent among captured ectopic cells. Observed-to-expected ratios for stromal cells increased from 0.36 in normal tissue to 0.83 in eutopic tissue and 1.63 in ectopic lesions; the corresponding value for putative Schwann-like cells was 3.01 in ectopic tissue (Figure 1G). The specimen-level plots reported differences across tissue groups for epithelial (P = 0.007), Schwann-like (P = 0.005), and stromal fractions (P = 0.004) (Figure 1H). Together, the data indicated substantial changes in stromal and other non-epithelial compartments, motivating more detailed stromal-state analysis.\n3.3 Stromal trajectory reconstruction delineates a fib-TCF21-associated state\nStromal and mural cells were analyzed separately to characterize their transcriptional heterogeneity. The reconstructed DDRTree trajectory contained multiple branches populated by normal, eutopic, and ectopic cells (Figures 2A,B). Seven annotated stromal-related states were projected onto this trajectory: eF-HOXA10, eF-COL7A1, eF-myo, eF-HOXA13, fib-TCF21, vascular smooth muscle cells, and perivascular cells (Figures 2C,D). The fib-TCF21 population was concentrated toward one branch enriched in ectopic cells, whereas other states occupied different regions of the inferred transcriptional ordering.\nFIGURE 2\nExpression profiling across trajectory bins revealed changing programs related to stromal identity and remodeling, including HOXA10, WNT2B, COL7A1, MMP7, CTHRC1, HOXA13, CXCL14, TCF21, CCN5, ACTG2, and MYH11 (Figure 2E). Individual trajectory plots showed the spatial distribution of TCF21, NEXN, IFI44L, WNT2B, and SPRR2F, and their smoothed expression trends were examined across pseudotime (Figures 2F,G).\n3.4 Stromal functional programs and predicted intercellular communication\nFunctional enrichment of stromal-associated transcriptional features highlighted terms related to muscle-system and tissue development, cytoskeletal organization, endothelial proliferation, and epithelial proliferation regulation (Figure 3A). Pathway-level annotations included cytoskeleton-related processes, complement and coagulation cascades, Ras signaling, and ovarian steroidogenesis (Figure 3B). CellChat analyses of eutopic and ectopic datasets yielded interconnected predicted communication networks involving stromal, epithelial, endothelial, immune, and mural populations (Figures 3C,D). The displayed aggregate network scores were 0.678 for eutopic cells and 0.815 for ectopic cells; the compared inputs contained 17,160 and 19,365 cells, respectively. Sender–receiver summaries showed pronounced stromal-associated contributions to the inferred interaction matrices, with different interaction patterns between tissue groups (Figures 3E,F).\nFIGURE 3\n3.5 Cell-death-associated and inflammatory programs differ across stromal states\nScoring of regulated-cell-death-associated and stress-response gene sets revealed heterogeneous transcriptional profiles across the major cell classes (Figure 4A). Macrophages showed relatively high scores for several death- and stress-related programs, whereas epithelial and other populations displayed distinct signatures. Within the stromal compartment, the distributions differed markedly among the separately annotated fib-TCF21 and fib-State1–12 populations (Figures 4B,C).\nFIGURE 4\nThe fib-TCF21 state showed a more selective pattern: its standardized scores were 0.35 for ferroptosis-related genes, 0.92 for autophagy-associated genes, and 0.41 for oxidative-stress-related genes, compared with −0.27 for apoptosis- and −0.70 for pyroptosis-related genes (Figure 4B). Among inflammatory pathways, relatively positive scores were observed for IL6/JAK/STAT3 (1.19) and complement-associated signaling (0.93), while the broad inflammatory-response score was −0.50 (Figure 4C). Patient-paired displays showed shifts in mean death-associated and inflammatory program scores between eutopic and ectopic samples, with interpatient variability (Figures 4D,E). An exploratory diffusion-pseudotime analysis indicated nonlinear profiles for TCF21 and the examined death/inflammation-related scores, rather than a common, continuously increasing pattern (Figure 4F). Spearman heatmaps showed both positive and negative candidate-gene/program associations, with coefficients generally of modest magnitude (Figures 4G,H). These observations support transcriptional heterogeneity at the interface of stromal stress and inflammation.\n3.6 Spatial domains exhibit regionally variable stromal signature and pathway scores\nTwo endometriotic lesion sections from GSE213216 (GSM6690475/BEME_346 and GSM6690476/BEME_355G) provided tissue-level spatial context (Figures 5A,B). In GSM6690475, transcriptome-based segmentation distinguished four spatial domains, including a domain annotated as TCF21-associated fibrotic stroma and other stromal or margin-associated regions (Figure 5C). NEXN transcripts were visible across multiple spatial spots, whereas IFI44L and WNT2B showed sparse or focal detection within the same section (Figures 5D–F). The fib-TCF21 signature and the cell-death-associated and inflammatory transcriptional program scores showed regionally heterogeneous spatial distributions (Figures 5G-I).\nFIGURE 5\n3.7 Independent single-cell data support a recognizable, heterogeneous TCF21-associated pattern\nThe GSE213216 single-cell dataset was used to assess whether the discovery-associated stromal pattern could be recognized in an independent atlas. Broad cell populations were identifiable in the validation UMAP, including stromal, mural, endothelial, epithelial, lymphoid, myeloid, and mast-cell compartments (Figure 6A). Within the stromal embedding, a subset was annotated as fib-TCF21-like, whereas remaining stromal cells showed distinct distributions (Figure 6B). The five-gene score showed variation across the stromal manifold (Figure 6C), and the dot plot displayed higher TCF21, NEXN, IFI44L, WNT2B, and SPRR2F expression within the fib-TCF21-like annotation than in the comparison stromal group (Figure 6D).\nFIGURE 6\nThe five-gene score also varied by tissue source, with heterogeneous distributions across ovarian ectopic, eutopic, and peritoneal ectopic samples (Figure 6E). A cross-cohort forest display showed positive standardized estimates for the five candidate genes in both datasets, with TCF21 having the largest displayed estimates (Figure 6F). TCF21 regulon activity was positively associated with death-associated scores in the GSE213216 and GSE179640 plots (r = 0.68 and 0.61, respectively) and with inflammatory scores (r = 0.66 and 0.60, respectively; Figures 6G,H).\n3.8 Spatial expression of TCF21 is detected across independent lesion sections\nExpanded spatial visualization illustrated the histological structures of both Visium lesions and the sequencing-quality distributions in GSM6690475 (Figures 7A,B). TCF21 transcripts were detected in localized regions of both GSM6690475 and GSM6690476, with differing spatial extent and signal intensity (Figure 7C). Within GSM6690475, spatial expression maps also identified NEXN-positive spots, whereas IFI44L and WNT2B were less frequently detected (Figure 7D). Death-associated and inflammatory scores each occupied nonuniform, partly distinct spatial regions (Figures 7E,F). These observations provided additional spatial support for heterogeneity in the TCF21-associated transcriptional context.\nFIGURE 7\n3.9 TCF21 is enriched in ectopic endometriotic stromal compartments\nMultiplex immunofluorescence revealed TCF21 expression in both eutopic and ectopic endometrial tissues, with more extensive TCF21-positive staining in stromal-rich regions of ectopic lesions, as delineated by vimentin and CD10 expression (Figure 8A). Quantitative analysis showed a higher proportion of TCF21-positive stromal cells in ectopic lesions than in eutopic endometrium (Figure 8C), accompanied by increased TCF21 mRNA expression (Figure 8D). RT-qPCR analysis of T-HESC, hEM15A, and ihESCs demonstrated relatively elevated expression of TCF21, NEXN, IFI44L, and WNT2B in ihESCs, whereas SPRR2F showed less pronounced variation among the 3 cell lines (Figure 8B). Further comparison confirmed higher basal TCF21 mRNA expression in ihESCs than in hEM15A cells (Figure 8E). Together, these findings supported the enrichment of TCF21 in ectopic stromal contexts and provided a rationale for selecting ihESCs for subsequent functional experiments.\nFIGURE 8\n3.10 TCF21 silencing is associated with reduced proliferation, membrane traversal, and spheroid growth\nFollowing si-TCF21 treatment in ihESCs, representative EdU fluorescence images showed fewer EdU-positive nuclei than in si-NC-treated cells, consistent with the lower EdU-positive percentage displayed in the accompanying summary (Figures 9A,B). Transwell images similarly showed fewer stained cells on the opposite membrane surface after si-TCF21 treatment, with a reduction in the corresponding cell-count summary (Figures 9C,D). These findings indicate differences in proliferative activity and membrane traversal under the experimental conditions shown.\nFIGURE 9\nIn nonadherent three-dimensional cultures, spheroids were observed in both groups over days 1–3. Control spheroids increased in size over time, whereas the si-TCF21 group showed smaller structures at each recorded time point (Figure 9E). The diameter summary likewise displayed lower spheroid diameters in the si-TCF21 group on days 1, 2, and 3 (Figure 9F). Together, the displayed assays are consistent with reduced proliferation, membrane traversal, and spheroid growth following TCF21-targeting treatment.\n4 Discussion\nThe present study integrated clinical characterization, single-cell and spatial transcriptomics, and targeted experimental validation to investigate the relationship between stromal heterogeneity, cell-death-associated programs, and inflammation in endometriosis. Three principal findings emerged. First, single-cell analysis identified a heterogeneous stromal compartment containing a fib-TCF21-associated state characterized by five candidate genes: TCF21, NEXN, IFI44L, WNT2B, and SPRR2F. Second, regulated-cell-death-associated and inflammatory transcriptional programs differed across stromal states and exhibited heterogeneous spatial distributions. Third, tissue localization and TCF21 silencing experiments supported the involvement of TCF21 in endometriotic stromal-cell behavior. Together, these findings identify a TCF21-associated stromal transcriptional signature within the broader context of cellular stress responses and inflammatory remodeling.\nAn important observation was the extensive heterogeneity of stromal populations across endometriotic tissues. Although previous studies have demonstrated the importance of stromal cells in lesion establishment and fibrosis, recent high-resolution analyses indicate that stromal remodeling involves multiple transcriptionally distinct populations rather than a uniform fibroblast response. A 2026 study integrating single-cell and spatial transcriptomics demonstrated marked differences between peritoneal and ovarian endometriosis, with the former characterized by chemokine-associated inflammatory niches and the latter showing prominent fibromuscular differentiation (\n). Another large-scale single-cell investigation identified altered inflammatory, adhesion, and cell-survival programs in the eutopic endometrium of patients with endometriosis ( ). These findings support the concept that both cellular identity and anatomical location contribute to disease-associated transcriptional heterogeneity. Our identification of a fib-TCF21-associated state adds further resolution to this framework by examining a specific stromal transcriptional program alongside its death- and inflammation-associated characteristics.The five-gene signature provides a means of characterizing this stromal state at the transcriptional level. TCF21, NEXN, IFI44L, WNT2B, and SPRR2F exhibited distinct expression patterns across the inferred stromal trajectory, with relatively higher expression in portions of the fib-TCF21-associated branch. Previous single-cell and spatial analyses identified CXCR4-positive fibroblasts associated with FN1-mediated signaling and immune–fibrotic remodeling (\n). Similarly, a 2025 spatial multi-omics study of ovarian endometriomas highlighted ECM–receptor interactions, focal adhesion, and metabolically distinct stromal and perivascular populations ( ). Compared with these studies, the present analysis focuses on the relationship between a TCF21-associated stromal state and transcriptional programs related to cell death and inflammation. Importantly, the five-gene signature should be considered a cell-state descriptor rather than evidence that all five genes are directly regulated by TCF21.Recent experimental studies support the context-dependent importance of ferroptosis in endometriosis. Iron overload-induced ferroptosis has been shown to impair CD8-positive T-cell function and contribute to an altered immune environment (Wang et al., 2025). Conversely, increased FZD7 expression can protect ectopic endometrial cells against ferroptosis through β-catenin/SLC7A11 signaling ( ). Another study demonstrated that HSD11B1 promotes ferroptosis resistance through a JUND/IL-10-dependent mechanism in ectopic stromal cells (Yang et al., 2026). More recently, experimental induction of ferroptosis using FIN56 was shown to suppress endometriotic lesion development through mitochondrial oxidative stress and lipid peroxidation (Zeng et al., 2026). Together, these studies suggest that ferroptosis may exert different biological effects depending on the affected cell population and molecular context. Our findings are consistent with this complexity. Nevertheless, the transcriptional scores analyzed here do not establish whether fib-TCF21 cells undergo ferroptosis, resist ferroptotic injury, or activate compensatory stress responses. These possibilities require direct functional evaluation.\nInflammatory heterogeneity was also evident within the stromal compartment. The fib-TCF21-associated state displayed selective enrichment of IL6/JAK/STAT3- and complement-associated transcriptional features rather than uniformly elevated inflammatory signaling. This observation is consistent with emerging evidence that endometriotic stromal populations possess distinct inflammatory properties. A recent study identified a FOSL2-associated stromal population exhibiting a senescence-associated secretory phenotype, in which inflammatory cytokine secretion promoted macrophage recruitment and polarization (\n). Such findings emphasize that stromal cells may actively shape inflammatory microenvironments through cytokine production and interactions with immune populations. However, the presence of an inflammatory transcriptional program does not necessarily imply increased secretion of every associated cytokine. In the present study, IL6/JAK/STAT3-related enrichment therefore provides a biological context for the TCF21-associated state.Spatial transcriptomic analysis provided another important dimension to the interpretation of the stromal signature. TCF21 expression was observed in localized regions of two endometriotic lesion sections, whereas the five-gene signature and death/inflammation-associated programs exhibited nonuniform spatial distributions. These findings suggest that stromal transcriptional features are organized within specific tissue regions rather than evenly distributed throughout the lesion. Previous spatial investigations have similarly demonstrated compartment-specific molecular organization in endometriosis. A study using GeoMx spatial transcriptomics identified altered communication between lesion epithelium and macrophages, while observing relatively limited transcriptional differences in some stromal compartments compared with matched eutopic endometrium (\n). These observations indicate that the contribution of stromal and immune compartments may vary according to lesion type, anatomical location, and the spatial regions examined.Of particular interest, the spatial distributions of TCF21-associated and inflammatory programs were not completely overlapping. TCF21 was selected for further investigation because of its association with the identified stromal state and its established role in endometriotic stromal biology. Importantly, recent work has substantially advanced understanding of TCF21 function in endometriosis. A 2026 Nature Communications study demonstrated that TCF21 regulates cytoskeletal reorganization, adhesion, and invasion through transcriptional activation of LIMK2 and downstream cofilin signaling (Zhu et al., 2026). The authors also provided in vivo evidence that modulation of TCF21 activity influences endometriotic lesion development. These findings establish TCF21 as a functionally relevant regulator rather than a previously unidentified disease-associated gene. Accordingly, the contribution of the present study lies primarily in placing TCF21 within a broader stromal transcriptional context involving cellular stress, inflammation, and spatial heterogeneity. The correlations between inferred TCF21 regulon activity and death/inflammation-related scores provide additional hypotheses concerning its potential regulatory relationships.\nOur experimental observations further supported the biological relevance of TCF21 in endometriotic stromal cells. Tissue immunofluorescence demonstrated TCF21 expression within stromal-associated regions, while cellular analysis showed relatively higher basal expression in lesion-derived ihESCs. Following TCF21 silencing, ihESCs exhibited reduced EdU incorporation, fewer cells traversing Transwell membranes, and smaller spheroids during three-dimensional culture. These findings are consistent with a role for TCF21 in cellular proliferation and invasive or migratory behavior, in agreement with previously reported TCF21-dependent cytoskeletal functions (Zhu et al., 2026). The three-dimensional findings additionally suggest that TCF21 may influence stromal-cell aggregation or spheroid growth. Previous three-dimensional endometriosis models have demonstrated that stromal-cell invasion and matrix remodeling require assessment within an appropriate extracellular matrix environment ( ). Furthermore, reduced Transwell cell counts may partly reflect differences in proliferation or cell viability. Thus, the present functional assays support TCF21-associated cellular phenotypes. These findings also have potential implications for molecular characterization and precision targeting. The five-gene signature may help distinguish a subset of stromal cells within the heterogeneous endometriotic microenvironment, while TCF21 represents a candidate factor for targeted functional investigation.\nSeveral limitations should be acknowledged. First, the retrospective clinical cohort originated from a single center and contained a high proportion of ovarian endometriomas and advanced-stage disease, potentially limiting its generalizability. Second, the discovery single-cell analysis included a relatively small number of ectopic specimens, while the independent dataset exhibited anatomical and transcriptional heterogeneity. Differences in tissue location, donor characteristics, menstrual-cycle phase, and preprocessing may influence the representation of stromal states. Third, pseudotime analysis and regulon inference provide computational descriptions of transcriptional relationships but cannot establish lineage progression or direct transcriptional regulation. Fourth, spatial analyses were limited to two Visium sections, and spot-level resolution restricted precise cellular attribution. Finally, although TCF21 silencing altered several stromal-cell phenotypes, the experiments did not directly examine regulated-cell-death execution or establish a causal relationship between TCF21 activity and inflammatory cytokine secretion. Additional rescue experiments, independent perturbation strategies, cell-death-specific functional assays, and primary-cell validation would strengthen the mechanistic interpretation.\nOverall, our findings support a model in which TCF21-associated stromal cells represent a transcriptionally distinct component of the heterogeneous endometriotic microenvironment. Their association with selected death-related stress and inflammatory programs highlights the importance of considering cell-state identity and spatial organization when investigating endometriosis biology. By integrating cellular, spatial, and experimental observations, this study provides a framework for examining TCF21-associated stromal remodeling and its potential relationships with inflammatory tissue responses. Future investigations should determine whether these transcriptional associations correspond to functionally defined stress-adaptation mechanisms and whether targeting TCF21-associated pathways can modify pathological stromal behavior.\n5 Conclusion\nThis study identifies a TCF21-associated stromal death–inflammation transcriptional signature in endometriosis through the integration of clinical characterization, single-cell and spatial transcriptomics, and targeted experimental evaluation. The findings reveal substantial stromal heterogeneity, with the fib-TCF21-associated state exhibiting selective relationships with cell-death-associated stress and inflammatory programs. Spatial analyses further demonstrate regionally heterogeneous TCF21-associated expression patterns, while tissue localization and functional experiments support the involvement of TCF21 in endometriotic stromal-cell behavior. These findings provide a cellular and spatial for understanding stromal remodeling in endometriosis and identify TCF21-associated pathways as candidates for further mechanistic investigation.\nStatements\nData availability statement\nThe original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding authors.\nAuthor contributions\nWL: Conceptualization, Data curation, Validation, Writing – original draft, Writing – review and editing. XW: Data curation, Writing – original draft, Writing – review and editing. AA: Software, Supervision, Writing – original draft, Writing – review and editing. JF: Conceptualization, Formal Analysis, Writing – original draft, Writing – review and editing. CH: Conceptualization, Data curation, Funding acquisition, Writing – original draft, Writing – review and editing.\nFunding\nThe author(s) declared that financial support was received for this work and/or its publication. This study was supported by Ningxia Hui Autonomous Region Key Research and Development Program Project (Grant No. 2023BEG01001 and 2023BEG02037); National Natural Science Foundation of China, No. 82460304; Dalian City Outstanding Young Science and Technology Talent Program, No. 2023RY020; Dalian Science and Technology Innovation Fund, No. 2025JJ13PT056; Liaoning Provincial Natural Science Foundation, No. 2025-MS-368.\nConflict of interest\nThe author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.\nGenerative AI statement\nThe author(s) declared that generative AI was not used in the creation of this manuscript.\nAny alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.\nPublisher’s note\nAll claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.\nReferences\n1\nAibarS.González-BlasC. B.MoermanT.Huynh-ThuV. A.ImrichovaH.HulselmansG.et al (2017). SCENIC: single-cell regulatory network inference and clustering. Nat. Methods14, 1083–1086. 10.1038/nmeth.4463\n2\nAmerican Society for Reproductive Medicine (1997). Revised American society for reproductive medicine classification of endometriosis: 1996. Fertil. Steril.67, 817–821. 10.1016/S0015-0282(97)81391-X\n3\nAndreattaM.CarmonaS. J. (2021). UCell: robust and scalable single-cell gene signature scoring. Comput. Struct. Biotechnol. J.19, 3796–3798. 10.1016/j.csbj.2021.06.043\n4\nBeckerC. M.BokorA.HeikinheimoO.HorneA.JansenF.KieselL.et al (2022). ESHRE guideline: endometriosis. Hum. Reprod. Open2022, hoac009. 10.1093/hropen/hoac009\n5\nBurnsG. W.FuZ.VegterE. L.MadajZ. B.GreavesE.FloresI.et al (2025). Spatial transcriptomic analysis identifies epithelium-macrophage crosstalk in endometriotic lesions. iScience28, 111790. 10.1016/j.isci.2025.111790\n6\nChenY.YiY. H.HanH. J.CuiH.ShenD. H.WeiL. H.et al (2016). Establishment of human endometriosis-derived immortalized eutopic endometrium stromal and epithelial cell lines. Int. J. Clin. Exp. Med.9 (8), 16450–16458.\n7\nDindoD.DemartinesN.ClavienP. A. (2004). Classification of surgical complications: a new proposal with evaluation in a cohort of 6336 patients and results of a survey. Ann. Surg.240, 205–213. 10.1097/01.sla.0000133083.54934.ae\n8\nDuempelmannL.SheppardS.McKinnonB.DuoA.SkrabalovaJ.AndrieuT.et al (2026). Endometriosis-related alterations in the endometrium revealed by integrated single-cell and AI-powered approaches. Nat. Commun.17, 6688. 10.1038/s41467-026-73020-4\n9\nFonsecaM. A. S.HaroM.WrightK. N.LinX.AbbasiF.SunJ.et al (2023). Single-cell transcriptomic analysis of endometriosis. Nat. Genet.55, 255–267. 10.1038/s41588-022-01254-1\n10\nGalluzziL.VitaleI.AaronsonS. A.AbramsJ. M.AdamD.AgostinisP.et al (2018). Molecular mechanisms of cell death: recommendations of the nomenclature committee on cell death 2018. Cell Death Differ.25, 486–541. 10.1038/s41418-017-0012-4\n11\nGarcia GarciaJ. M.VannuzziV.DonatiC.BernacchioniC.BruniP.PetragliaF. (2023). Endometriosis: cellular and molecular mechanisms leading to fibrosis. Reprod. Sci.30, 1453–1461. 10.1007/s43032-022-01083-x\n12\nHaghverdiL.BüttnerM.WolfF. A.BuettnerF.TheisF. J. (2016). Diffusion pseudotime robustly reconstructs lineage branching. Nat. Methods13, 845–848. 10.1038/nmeth.3971\n13\nHeF.ZuS.LinY.ShiJ.LinS. (2026). Integrated single-cell and spatial transcriptomics reveal divergent immunological and stromal programs in peritoneal versus ovarian endometriosis. BMC Womens Health26, 278. 10.1186/s12905-026-04456-5\n14\nHuangZ. X.WuR. F.MaoX. M.HuangS. M.LiuT. T.ChenQ. H.et al (2020). Establishment of an immortalized stromal cell line derived from human endometriotic lesion. Reprod. Biol. Endocrinol.18, 119. 10.1186/s12958-020-00669-x\n15\nHuangY.LiR.HuR.YaoJ.YangY. (2022). PEG2-induced pyroptosis regulates the expression of HMGB1 and promotes hEM15A migration in endometriosis. Int. J. Mol. Sci.23, 11707. 10.3390/ijms231911707\n16\nHuangJ.ShiW.LiaoZ.WangM.ZhongQ.QiuL.et al (2026). N6-methyladenosine regulated FZD7 inhibits ferroptosis in endometriosis via β-catenin/SLC7A11 pathway. Free Radic. Biol. Med.243, 54–66. 10.1016/j.freeradbiomed.2025.11.020\n17\nJinS.Guerrero-JuarezC. F.ZhangL.ChangI.RamosR.KuanC. H.et al (2021). Inference and analysis of cell–cell communication using CellChat. Nat. Commun.12, 1088. 10.1038/s41467-021-21246-9\n18\nJonesG.KennedyS.BarnardA.WongJ.JenkinsonC. (2001). Development of an endometriosis quality-of-life instrument: the endometriosis health Profile-30. Obstet. Gynecol.98, 258–264. 10.1016/S0029-7844(01)01433-8\n19\nLiG.LinY.ZhangY.GuN.YangB.ShanS.et al (2022). Endometrial stromal cell ferroptosis promotes angiogenesis in endometriosis. Cell Death Discov.8, 29. 10.1038/s41420-022-00821-z\n20\nLiuS.LiX.GuZ.WuJ.JiaS.ShiJ.et al (2025). Single-cell and spatial transcriptomic profiling revealed niche interactions sustaining growth of endometriotic lesions. Cell Genom5, 100737. 10.1016/j.xgen.2024.100737\n21\nLivakK. J.SchmittgenT. D. (2001). Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) method. Methods25, 402–408. 10.1006/meth.2001.1262\n22\nMaJ.ZhangL.ZhanH.MoY.RenZ.ShaoA.et al (2021). Single-cell transcriptomic analysis of endometriosis provides insights into fibroblast fates and immune cell heterogeneity. Cell Biosci.11, 125. 10.1186/s13578-021-00637-x\n23\nQiY.ChenX.ZhengS.WuT.LiZ.ChengJ.et al (2025). Single-cell and spatially resolved omics reveal transcriptional and metabolic signatures of ovarian endometriomas. Nat. Commun.16, 11539. 10.1038/s41467-025-66706-8\n24\nQiuX.MaoQ.TangY.WangL.ChawlaR.PlinerH. A.et al (2017). Reversed graph embedding resolves complex single-cell trajectories. Nat. Methods14, 979–982. 10.1038/nmeth.4402\n25\nSaundersP. T. K.HorneA. W. (2021). Endometriosis: etiology, pathobiology, and therapeutic prospects. Cell184, 2807–2824. 10.1016/j.cell.2021.04.041\n26\nShaoW.JuH.XiahouZ.FangS.YanR.LiC.et al (2025). Fibroblast heterogeneity and FN1-mediated signaling in endometriosis revealed by single-cell and spatial transcriptomics. Front. Immunol.16, 1680849. 10.3389/fimmu.2025.1680849\n27\nShiW.TangX.YangF.YinH.ZhouQ.SunF.et al (2026). FOSL2-driven SASP in endometrial stroma promotes the inflammation of endometriosis. NPJ Aging12, 101. 10.1038/s41514-026-00447-w\n28\nStejskalováA.FinckeV.NowakM.SchmidtY.BorrmannK.von WahldeM. K.et al (2021). Collagen I triggers directional migration, invasion and matrix remodeling of stroma cells in a 3D spheroid model of endometriosis. Sci. Rep.11, 4115. 10.1038/s41598-021-83645-8\n29\nTanY.FlynnW. F.SivajothiS.LuoD.BozalS. B.DavéM.et al (2022). Single-cell analysis of endometriosis reveals a coordinated transcriptional programme driving immunotolerance and angiogenesis across eutopic and ectopic tissues. Nat. Cell Biol.24, 1306–1318. 10.1038/s41556-022-00961-5\n30\nTangD.KangR.Vanden BergheT.VandenabeeleP.KroemerG. (2019). The molecular machinery of regulated cell death. Cell Res.29, 347–364. 10.1038/s41422-019-0164-5\n31\nTaylorH. S.KotlyarA. M.FloresV. A. (2021). Endometriosis is a chronic systemic disease: clinical challenges and novel innovations. Lancet397, 839–852. 10.1016/S0140-6736(21)00389-5\n32\nWangS.XuL.JiaoX.DongX.ZhongZ.RenQ.et al (2025). Iron overload-induced ferroptosis in CD8+ T cells leads to functional abnormalities that promote endometriosis progression. BMC Med.23, 588. 10.1186/s12916-025-04369-4\n33\nWuQ.LiangZ.JiangJ.FengX.LiuJ.ZhangZ.et al (2023). Macrophages originated IL-33/ST2 inhibits ferroptosis in endometriosis via the ATF3/SLC7A11 axis. Cell Death Dis.14, 668. 10.1038/s41419-023-06182-4\n34\nYangY.HuangY.WangY.LiR.LiH. (2026). HSD11B1 suppresses ferroptosis in endometrial stromal cells through the JUND/IL-10 axis to promote endometriosis progression. Reprod. Biol. Endocrinol.24, 67. 10.1186/s12958-026-01565-6\n35\nZengC.ZhuJ.LuR.WuP.LiX.LiF.et al (2026). FIN56-induced ferroptosis suppresses the development of endometriosis by augmenting mitochondrial ROS and lipid peroxidation via the ACACA/ARID5A/NOX4 axis. Chin. Med. J. Engl.139, 2650–2665. 10.1097/CM9.0000000000003770\n36\nZhangJ.ZhaoW.ZhouY.XiS.XuX.DuX.et al (2024). Pyroptotic T cell-derived active IL-16 has a driving function in ovarian endometriosis development. Cell Rep. Med.5, 101476. 10.1016/j.xcrm.2024.101476\n37\nZhuJ.WuP.MaY.PengZ.FaH.QinL.et al (2026). TCF21 promotes epithelial-to-mesenchymal transition and cytoskeleton reorganization in uterine development and endometriosis. Nat. Commun.17, 3420. 10.1038/s41467-026-69551-5\n38\nZondervanK. T.BeckerC. M.KogaK.MissmerS. A.TaylorR. N.ViganòP. (2018). Endometriosis. Nat. Rev. Dis. Prim.4, 9. 10.1038/s41572-018-0008-5\nSummary\nKeywords\nendometriosis, inflammation, regulated cell death, single-cell RNA sequencing, spatial transcriptomics, stromal heterogeneity, TCF21\nCitation\nLi W, Wang X, Aziz AR, Feng J and Ha C (2026) Clinical and single-cell integration identifies a TCF21-associated stromal death–inflammation signature in endometriosis. Front. Cell Dev. Biol. 14:1979976. doi: 10.3389/fcell.2026.1979976\nReceived\n27 August 2026\nRevised\n26 October 2026\nAccepted\n27 October 2026\nPublished\n05 November 2026\nVolume\n14 - 2026\nEdited by\nZui Zou, Naval Medical University, China\nReviewed by\nXinpei Deng, Sun Yat-sen University Cancer Center (SYSUCC), China\nMuhammad Akram, University of Management and Technology, Pakistan\nUpdates\nCopyright\n© 2026 Li, Wang, Aziz, Feng and Ha.\nThis is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.\n*Correspondence: Wangshu Li, doctor_lws@outlook.com; Aziz ur Rehman Aziz, azizjatoi@hotmail.com; Jiuxiang Feng, dlfcyyfjx@163.com; Chunfang Ha, hachunfang@163.com\nDisclaimer\nAll claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.","source_license":"CC0","license_restricted":false}