Three-dimensional patient-derived endometriosis model for drug evaluation

Study design and conceptualization: Wang, L. and Liu, S. P. Drafting the manuscript: Li, Y. Q. and Wang, L. Data curation: Shao, W., Gao, X., and Li, J. Z. Preparation of the tables and figure: Wang, Y. X. and Xing, Y. Y. Revision and editing of the final version of the manuscript: Cheng, Y., Su, W. T., and Zhang, L. S. All authors have read and agreed to the published version of the manuscript.

This study was approved by the Ethics Committee of Jinshan Hospital, Fudan University (no.: JIEC2021‐S49‐03) on May 20, 2022. All the participants provided written informed consent.

This study was supported by the Research Foundation for Clinical Project of Shanghai Municipal Health Commission (no.: 202140366), the “Science and Technology Innovation Action Plan” Basic Research Project of Shanghai Science and Technology Commission (no.: 23JC1403002), the Medical Key Subject of Jinshan District (no.: JSZK2023H01), the Key Subject Development Project of Jinshan Hospital of Fudan University—Obstetrics and Gynecology (no.: ZDXK‐2025–6), the Scientific Research Initiation Foundation for Youth of Jinshan Hospital of Fudan University (no.: JYQN‐JC‐202210), the Medical Engineering Fund of Fudan University (no.: yg2021‐025), the Scientific Research Topics of Shanghai Jinshan District Health Commission (no.: JSKJ‐KTMS‐2022‐08), the Scientific Research Project of Jinshan District Health Commission (no.: JSKJ‐KTYQ‐2023‐04).

It is challenging to isolate and culture gland‐like epithelial organoids from endometriotic lesions because of the lower proportion of gland structures to stromal cells compared with normal uterine endometrial tissues. To reconstruct a biomimetic in vitro endometriomas model, we first isolated and cultured epithelial organoids and stromal cells, respectively. Furthermore, the dissociated cell clusters were mixed with Matrigel and cultured in specialized endometriosis organoid medium for 10–20 days. The irregular gland‐like structures gradually enlarged in the Matrigel, whereas most of the stromal cells adhered to the bottom of the plate (Figure  1A ). Using this culture system, the success rate of gland‐like epithelial organoid establishment was 84.85% (28/33) (Table  3 ). However, we found that the menstrual phase influenced the proliferation of epithelial organoids. Epithelial organoids derived from patients in the secretory and menstrual phases exhibited limited expansibility, rarely propagating beyond three passages. This limited expansibility may be attributable to a state of advanced cellular differentiation acquired during these stages. Establishment and characterization of organoids from patients with endometrioma. (A) Microphotograph of primary epithelial organoids at different days. (B) Microphotograph of different passages of epithelial organoids at different days. (C) Scanning electron microscope (SEM) images of epithelial organoids. Success rate of gland‐like epithelial organoids establishment from different endometriosis patient categories. To purify the organoids from the cellular mixture, we carefully pipetted the epithelial organoids with Matrigel to avoid mingling stromal cells adhering to the bottom of the plate. Leveraging the differences in weight and size between the organoids and single stromal cells, we used the differential speed sedimentation method to separate the organoids from the stromal cells. In the second‐generation Matrigel mixture, only a few stromal cells were observed among the epithelial organoids (Figure  1B ). By the fourth generation, the organoids hardly contained any stromal cells, and the maximum diameter of the organoids reached 500 μm. The SEM images revealed that the organoids possessed a spherical structure with a complex and rough surface topography (Figure  1C ). This roughness was attributable to an abundance of cellular protrusions. We collected smaller fragments retained on the strainer and cultured them on a plate in a 2D manner (Figure  2A‐a ). Moreover, the subjacent stromal cells were collected following the careful removal of the overlying epithelial organoid/Matrigel mixture. To maintain the activity and function of the stromal cells, the third‐ to fifth‐generation cells were used. The stromal cells displayed a spindle shape with high activity, consistent with the findings of a previous study 29 (Figure  2A,B ), which expressed vimentin in the IHC assay (Figure  2A–D ). To mimic the 3D structure of stromal cells, we used a CSwell 600 chamber to form cellular spheroids (Figure  2B ). These uniform stromal cell spheroids were cultured for more than 10 days in this chamber. SEM analysis revealed that the surface of the 3D stromal cell spheroids was smoother than that of the epithelial organoids (Figure  2C ). The IHC assay demonstrated that vimentin, ITGB3, ICAM1, MMP2, MMP9, TIMP1, TIMP2, and MMP12 were positive in the 3D stromal cell spheroids (Figure  2D ). The embedding of stromal spheroids in Matrigel effectively recapitulated their invasive potential within the matrix, with the resultant invasion depicted in Figure  2E . Culture and identification of stromal cells from patients with endometriosis. (A) Microphotograph of stromal cells: (a) primary cells; (b) first‐passage (P1) cells; (c‐d) immunohistochemical (IHC) assay showed that stromal cells expressed vimentin in 2D conditions. (B) Formation of 3D stromal cell spheroids for 3 days. (C) SEM test showed that the surface of 3D stromal cell spheroids cultured for 3 days. (D) IHC assay showed vimentin, ITGB3, ICAM1, MMP2, MMP9, MMP12, TIMP1, and TIMP2 expressed on the stromal cell spheroids cultured for 3 days. (E) Microphotograph of stromal cells invading into the Matrigel. We further identified epithelial organoids and endometriotic tissues using H&E, PAS, and IHC staining with multiple antibodies (Figure  3 ). H&E staining showed that the epithelial organoids were similar to the gland structure of the endometriotic tissue (Figure  3A ), with the formation of a lumen surrounded by tightly packed columnar epithelial cells. PAS staining demonstrated mucus excretion in the lumen of epithelial organoids and epithelial glands in the tissue (Figure  3B ). Identification of epithelial organoids. (A, B) H&E and PAS stainings of the endometriotic tissue and epithelial organoids. (C) IHC staining of the endometriotic tissue and epithelial organoids of endometrioma by ER, PR, CK7, vimentin, E‐cadherin, CD44, MMP2, MMP9, TIMP1, TIMP2, ICAM‐1, and ITGB3 markers. IHC staining indicated that sex hormone‐related receptors such as ERs and PRs were expressed in the epithelial organoids and tissues (Figure  3C ). Cell–cell and cell‐matrix interaction molecules (E‐cadherin, CD44, ICAM1, ITGB3, and CK7) and invasion‐related molecules (MMP2, MMP9, TIMP1, and TIMP2) were also present in the epithelial organoids and tissues. Stromal cell‐related molecules (vimentin) were also expressed in epithelial organoids (Figure  3C ). We cocultured epithelial organoids with stromal cells at different proportions to reconstitute a controllable and biomimetic in vitro model (Figure  4A ). After 9 days of coculture, the stromal cells exhibited robust invasion into the Matrigel and rapid proliferation. Under these conditions, epithelial organoids displayed significant enlargement, with the highest number of organoids observed at a 1:2 (epithelial organoids to stromal cells) ratio compared with other groups. Additionally, epithelial organoids maintained higher viability at ratios of 1:2 and 1:4 than those of the other groups (Figure  4C ). Ki67 expression in glandular organoids exhibited a dose‐dependent decrease in response to increasing ratios of stromal cells, demonstrating an inverse correlation between stromal abundance and epithelial proliferation. This suggests that excessive stromal cells may create a microenvironment that suppresses epithelial organoid proliferation (Supporting Information  S1 ). Coculture of epithelial organoids and stromal cells from a patient. (A) The cocultures of epithelial organoids and stromal cells at various ratios were observed on Day 3 and Day 9 performed in triplicate wells in a single experiment ( n  = 3). (B) The number of epithelial organoids was counted on Day 3 and Day 9 under different coculture ratios, ** p <0.01, *** p  < 0.001 vs. 1:2 group, *** p  < 0.001 vs other groups. (C) The diameters of epithelial organoids were measured on Day 3 and Day 9 for different coculture ratios. This study examined dienogest, a clinically used drug. While previous studies have primarily focused on its role in the stromal cells, 35 , 36 its impact on glandular epithelial cells derived from patients with endometriomas remains unexplored. Therefore, we evaluated its effects on epithelial organoids and stromal cells using a 3D culture system. As shown in Figure  5Aa–d , the number and size of epithelial organoids reduced significantly at a concentration of 20 μM. The ATP levels, measured by the CellTiter‐Glo ® assay, demonstrated a dose‐dependent decline in cell viability. Dienogest (20 μM) exhibited an anti‐proliferative effect on the stromal cells in a 2D culture system derived from Patient 007 (Supporting Information  S2 ). Furthermore, the size of stromal cell spheroids from Patient 010, with 2000 cells/spheroid, administered the same concentration gradient of dienogest in the microwell, decreased in a dose‐dependent manner (Figure  5Ba,b ). However, 3D stromal cell spheroids derived from Patient 018 did not respond to dienogest (Figure  S3A,B ). We further evaluated the role of dienogest in the coculture of epithelial organoids and stromal cells in Patient 014, 016, and 017. Dienogest treatment revealed potential patient‐specific heterogeneity. A dose‐dependent inhibition of cell activity was observed in Patient 014 (at 20 and 40 μM) (Figure  6A ), whereas Patient 016 and 017 remained non‐responsive across the tested concentration range (Figure  6B,C ). This contrast provides preliminary evidence of inter‐individual variation in drug responses that must be validated in larger cohorts and with correlative clinical data. Drug evaluation of epithelial organoids and stromal cell spheroids with different concentrations of dienogest, respectively. (A) Drug testing on epithelial organoids from Patient 007 performed in triplicate wells in a single experiment ( n  = 3): (a) microphotograph of organoids with different concentrations of dienogest at Days 0, 2, 4, and 6; (b) cell viability of epithelial organoids after drug testing; (c) the average area of the epithelial organoids after drug treatment; (d) the average diameter of the epithelial organoids after drug testing. (B) Drug testing of stromal cell spheroids from Patient 010 performed in triplicate wells in a single experiment ( n  = 3): (a) Microphotograph of stromal cell spheroids with different concentrations of dienogest at Days 0, 3, and 6; (b) cell viability of stromal cell spheroids after drug treatment. * p  < 0.05, ** p <0.01, *** p <0.001, **** p <0.0001 vs. 0 μM. Drug evaluation of cocultures of epithelial organoids and stromal cell spheroids. (A) Drug testing on cocultures at a 1:10 (epithelial organoids to stromal cells) ratio from Patient 014 performed in triplicate wells in a single experiment ( n  = 3): (a) microphotograph with different concentrations of dienogest at Days 0, 2, 4, and 6; (b) cell viability after drug testing, ** p <0.01, **** p <0.0001. (B, C) Cell viability of Patient 016 and Patient 017 performed in triplicate wells in a single experiment at the same ratios to Patient 014.

Endometriosis is a complex and dynamic disorder characterized by the retrograde flow of endometrial tissue, its subsequent adhesion to and invasion of pelvic organs (such as the ovaries), and its hormone‐responsive feature, which often leads to recurrence and metastatic potential. Previous in vitro models mainly involved single‐cell lines originating from normal uterine tissue under 2D culture conditions, which cannot accurately replicate the intricate pathological mechanisms of endometriosis. Exfoliated microtissues and endometriotic lesions mostly comprise epithelial and stromal cells. Cell–cell crosstalk plays a significant role in the pathological progression of endometriosis. A reliable and biomimetic endometriosis in vitro model, which incorporates epithelial and stromal cells, has great potential for elucidating the underlying mechanisms of the disease and developing therapeutic drugs. Here, we first generated patient‐derived epithelial organoids and stromal cells from endometriomas and established a viable biobank for pharmacological screening. These primary models closely recapitulated the histological characteristics of native tissues. Furthermore, we developed a coculture system wherein both cell types were combined at an optimized ratio. Notably, this platform demonstrated that both individual cultures and their cocultures exhibit specific responses to dienogest treatment. Construction of endometriotic epithelial organoids represents a significant advancement in endometriosis research. However, studies on endometriosis patient‐derived epithelial organoids remain limited. To comprehensively understand and confirm the disease‐specific features of these epithelial organoids, we performed H&E, PAS, and IHC staining using multiple markers. The glandular organoids exhibited a lumen‐bordering cell layer, with some showing luminal invasion. ER and PR were normally expressed in endometriotic glandular epithelial organoids, consistent with findings in tissues and previous studies on endometrial organoids. 29 Cell–cell and cell‐matrix interaction molecules, including E‐cadherin, CD44, ICAM1, ITGB3, and CK7, are expressed on epithelial organoids and cells in the tissue. E‐cadherin and CD44 were found to be overexpressed in endometriotic cells, where they regulate cellular adhesion, cell–cell interactions, and tissue integrity. 37 , 38 , 39 , 40 The sCD44 was upregulated in the peritoneal fluid of patients with endometriosis; however, the source of sCD44 in the peritoneal fluid remains unclear. Our patient‐derived endometrioma model may serve as a foundational tool for initial mechanistic studies of sCD44, and its potential relevance to other endometriosis subtypes, such as peritoneal or pelvic lesions, remains a subject for future validation. ICAM‐1, a member of the integrin adhesion protein family expressed on various cell types, plays significant roles in inflammatory and immune responses, and in the adhesion and survival of the endometrium to the peritoneum. 41 , 42 Kuessel et al. reported that ICAM‐1 expression was significantly higher in ectopic endometriotic tissues than in eutopic endometrial and control samples. 43 Similarly, ICAM‐1 was expressed in glandular epithelial organoids, stromal cells, and endometriotic lesions in this study. In our previous studies, we observed that ITGB3 was overexpressed in endometriotic lesions and stromal cells derived from patients. 44 , 45 , 46 As an extracellular matrix receptor, ITGB3 plays a significant role in mediating interactions between cells and their surrounding environment. In this study, we also identified the expression of ITGB3 in epithelial organoids. During implantation, vascularization, and growth of endometrial tissues within the host, ECM degradation is a key step in angiogenesis and tissue remodeling. 47 MMPs play crucial roles in this complex biological cascade. 48 , 49 Previous studies have demonstrated that the expression profile of MMPs in the endometrium of patients with endometriosis differs significantly from that in healthy individuals. 50 , 51 , 52 In this study, we comprehensively evaluated the MMP2 and MMP9 expression levels, along with their natural inhibitors TIMP2 and TIMP1, respectively, in epithelial organoids, stromal cells, and original tissues (Figures  2 and 3 ). Our findings revealed that MMP9 exhibited higher expression levels than MMP2. Conversely, TIMP2 was more prominently expressed than TIMP1, leading to inverse expression ratios of MMP2/TIMP2 and MMP9/TIMP1. Notably, Collette et al. reported that MMP9 secretion, as measured by zymography and enzyme‐linked immunosorbent assay (ELISA), was elevated in patients with endometriosis relative to healthy controls. 52 In contrast, no statistically significant differences in MMP2 secretion were detected between patients with and without endometriosis. Our study confirmed an imbalance between MMPs and their natural inhibitors. Additionally, vimentin was detected in epithelial organoids and stromal cells, suggesting that this phenomenon is a pathologically relevant feature of endometriosis, potentially related to epithelial–mesenchymal transition (EMT). Therefore, CD10, a more specific marker for stromal cells, 53 is an important molecule to further validate the cell populations and function in endometroma. Endometriotic tissues are predominantly composed of epithelial and stromal cells, both within the host tissues and ectopic lesions. We conducted coculture experiments with varying ratios of epithelial‐to‐stromal cells to establish an in vitro biomimetic endometriomas model (Figure  4 ). Our findings indicate that epithelial organoids demonstrated high viability and robust proliferation at epithelial‐to‐stromal cell ratios of 1:2 and 1:4. Epithelial organoid survival and luminal structure formation are adversely affected by increased stromal cell abundance. Notably, consensus on the optimal epithelial‐to‐stromal cell ratio for 3D coculture using primary endometriotic cells is lacking. Song et al. successfully generated endometriotic spheroids (ES) by culturing immortalized endometriotic epithelial (12Z) and stromal (iEc‐ESC) cell lines at a ratio of 1:50. Their study revealed 4522 differentially expressed genes when comparing ES cells to spheroids containing uterine stromal cells. 54 However, the 3D spheroid model failed to recapitulate the formation of epithelial gland‐like structures. Griffith et al. cocultured epithelial organoids and stromal cells from normal endometrial tissues. They used a density of 10 000 stromal cells and 10 intact endometrial epithelial organoids (EEOs), which resulted in an approximate epithelial‐to‐stromal ratio of 1: 1. 31 Single‐cell sequencing studies further underscore the disparity in the ratio of these two cell types between endometriotic and healthy endometrial tissues. 55 In adenomyotic tissues, the ratio of epithelial‐to‐stromal cells is approximately 3:10. 56 In contrast, ectopic ovarian endometriosis exhibits a significantly lower epithelial‐to‐stromal ratio compared with the normal endometrium, eutopic endometrium, ectopic peritoneal lesions, and ectopic adjacent tissue. 57 In this study, we further screened various epithelial‐to‐stromal cell ratios to optimize conditions for maintaining the high viability of both cell types. Our results showed that in Matrigel‐based coculture systems, high‐density epithelial organoids grew rapidly and efficiently formed luminal structures. In contrast, a high density of stromal cells significantly hampered the proliferation of epithelial organoids and the development of their luminal architecture (Figure  4 , Supporting Information  S1 ). These findings provide an alternative foundation for modeling diverse endometriotic lesions in vivo, consistent with the observations of Courtois et al. 57 Currently, there is a lack of reliable in vitro models for developing drugs to treat endometriosis. Existing normal and endometriotic cell lines are limited and often fail to accurately reflect in vivo drug effects owing to individual biological variations. Consequently, patient‐derived endometriotic models are highly significant for drug‐screening development. In this study, we successfully established 28 patient‐derived epithelial organoids (a success rate of 28 of 33) and 33 stromal cell lines (100% success rate). To assess the responsiveness of this novel model to pharmacological interventions, we used dienogest, a synthetic progestin currently used clinically. 58 , 59 Multiple clinical trials have demonstrated that dienogest is an effective and safe therapeutic option for managing adenomyosis and endometriosis. It has been shown to enhance the patients' quality of life, alleviate pelvic pain, reduce the size of lesions, and decrease the recurrence rate. 60 , 61 , 62 A published article noted that 2 mg of dienogest once daily is the concentration of this drug for clinical use. 32 Mechanistically, dienogest has been found to inhibit interleukin‐1β ‐induced C‐C motif chemokine ligand 20 (CCL20) upregulation in endometriotic epithelial cells. 63 Hayal Uzelli Şimşek et al. further demonstrated that dienogest suppresses the proliferation, telomerase activity, and migration of endometrial mesenchymal stem cells (E‐MSC) derived from healthy and diseased tissues. 64 Another study reported that dienogest decreases cell proliferation and invasion while promoting apoptosis by upregulating endoplasmic reticulum (ER) stress. 36 Collectively, these findings suggest that dienogest may play a complex role in treating endometriosis by targeting different pathways across various cell types. Therefore, we conducted a preliminary investigation into the effects of dienogest on stromal cell spheroids, epithelial organoids, and their cocultured mixtures. Our results showed that dienogest inhibited the proliferation of epithelial organoids and stromal cell spheroids in a dose‐dependent manner, with an IC50 of 20–40 μM. Notably, stromal cells and epithelial organoids exhibited different drug responses. Using endometrial explants treated with or without different concentrations of dienogest, Japarath Prechapanich et al. reported a dose‐dependent inhibition of cell outgrowth with an IC50 of 0.1–1 μM. 65 Fu L et al. reported that dienogest inhibited 2D‐cultured patient‐derived stromal cells, inducing G0/G1 arrest at 0.1–1 μM for 24 and 48 h. 66 Okada et al. demonstrated that dienogest inhibited 2D‐cultured normal endometrium stromal cells in a dose‐dependent manner, with IC50 >10 μM. 35 These prior studies, in conjunction with our current findings, indicate that dienogest can inhibit normal and endometriosis‐derived stromal and epithelial cells in vitro and in 3D‐cultured organoids. However, the effective concentration ranges varied significantly. Its anti‐proliferative effect may be mediated through its role as a progesterone receptor partial agonist, which can downregulate estrogen receptor signaling and thereby suppress estrogen‐dependent growth in endometriotic lesions. Future research should focus on conducting more comprehensive drug evaluations by comparing in vitro results with in vivo models, thereby further clarifying the therapeutic potential and optimal dosing of dienogest for endometriosis treatment. The in vitro model developed in this study has some limitations. First, the target organs/tissues implanted by normal endometrial tissues, such as the ovary, peritoneum, and colon, usually contain different ECM and various cell types. However, the Matrigel matrix is extracted from Engelbreth–Holm–Swarm (EHS) mouse sarcomas and comprises laminin, collagen IV, entactin/nidogen, and a number of growth factors. The mechanical properties of different concentrations of Matrigel could affect the biological functions of organoids and stromal cells, as well as the crosstalk between them. Therefore, Matrigel can maintain high endometrial cell viability; however, it fails to accurately mimic the complexity of the target tissues. This discrepancy may lead to inaccurate drug response profiles and the misinterpretation of disease mechanisms. To address this issue, potential solutions include replacing Matrigel with substitutes derived from target tissues or organs, such as ovarian ECM, peritoneal tissue extracts, or synthetic hydrogels engineered with defined chemical molecules that replicate the properties of the target tissues. Second, the current model, composed mainly of epithelial organoids and stromal cells, lacks key components, such as immune cells and vascular structures. These elements play crucial roles in the implantation and invasion of the endometrial tissue within target organs. To overcome this limitation, organ‐on‐a‐chip (OOC) or microfluidic technologies can be employed to construct vascular structures by perfusing fluids containing immune cells, thereby closely recapitulating the in vivo environment. Third, we explored cocultures of epithelial organoids and stromal cells at different ratios; however, the crosstalk between them was not fully investigated. The specific mechanisms of coculture with varying ratios of invasion and proliferation in target tissues/organs should be studied over long‐term dynamic menstrual cycles. In future studies, we aim to investigate these interactions between epithelial organoids and stromal cells at various ratios by mimicking the dynamic hormonal changes in the menstrual cycle through the administration of estrogen and progestin. In detail, subsequent studies will utilize this model to: (1) characterize the proliferative response to estradiol alone, (2) investigate the inhibitory effects of progesterone and progestins like dienogest on estrogen‐driven growth, and (3) fully map the hormone‐response profile of patient‐derived endometriotic cells. It will also be crucial to verify the model and its utility in testing a broader range of hormonal agents, including estrogen and progestin. Fourth, this study identified some protein markers of endometriomas based on previous studies; however, a more comprehensive analysis is warranted. Future investigations should incorporate genomic, transcriptomic, and proteomic analyses to enable a more detailed comparison between the model and native endometriotic tissues. Lastly, a systematic investigation of drug responses across a larger cohort of patients is the immediate next step and a primary goal of our ongoing research. Despite these limitations, this study presents a promising in vitro model for investigating endometriomas. By integrating OOC technologies with other novel biotechnologies, this model can significantly enhance our understanding of the pathogenesis of endometriosis and facilitate the screening of new and effective therapeutic agents.

In this study, we successfully isolated and cultured endometrial glandular epithelial and stromal cells from patients with endometriomas and generated 3D epithelial organoids and stromal cell spheroids exhibiting high histological and molecular fidelity to native endometriotic lesions. Moreover, we established a coculture system integrating epithelial organoids and stromal cells to recapitulate the intricate cellular interactions within the disease microenvironment, providing a more relevant in vitro representation of the disease. Upon evaluation with dienogest, a clinically employed therapeutic agent for endometriosis, the patient‐derived models exhibited heterogeneous drug responses, underscoring interpatient variability in pharmacological sensitivity and highlighting the complexity of endometriosis treatment. The patient‐specific in vitro model system developed in this study represents a major advancement in this field. It provides a platform for investigating the pathogenic mechanisms underlying endometriosis and also holds immense promise for advancing drug development. Specifically, it may facilitate targeted drug discovery, thereby paving the way for more effective treatments for patients with endometriosis.

Endometriosis is a chronic gynecological disorder in which functional endometrial‐like tissue develops outside the uterine cavity, such as the ovaries (known as endometriomas) and the peritoneal cavity. It is a common benign gynecological disorder that results in distressing symptoms, including severe pelvic pain and infertility, in approximately 10% of women of reproductive age. 1 The most widely accepted hypothesis for the development of endometriomas is the menstrual retrograde theory 2 ; however, many studies have demonstrated that various cells, including macrophages, bone marrow stem cells, and natural killer cells, are involved in the migration, adhesion, invasion, and implantation of tiny endometriotic tissue derived from the uterus. 3 , 4 , 5 , 6 , 7 Therefore, its pathogenesis is complex and remains unclear. The current therapeutic strategies for endometriosis include surgery, medication, and assisted reproductive technology. 8 , 9 For patients without indications for surgery, medication is preferred to alleviate the symptoms. The therapeutic drugs include progestins, non‐steroidal anti‐inflammatory drugs (NSAIDs), gonadotropin‐releasing hormone agonist (GnRH‐a), and oral contraceptives 9 , 10 , 11 , 12 ; however, these drugs do not completely cure endometriosis. Moreover, some drugs exhibit serious adverse effects in patients with endometriosis, including ovulation inhibition and estrogen deficiency symptoms. 13 , 14 , 15 Thus, current medications are unsuitable for patients with fertility needs or for long‐term administration. Owing to these limitations, there is an urgent need to develop new therapeutic drugs that can effectively improve endometriosis symptoms without affecting fertility. Some in vivo and in vitro models are available for studying intricate pathological mechanisms and developing novel therapeutic agents. Animals cannot spontaneously develop endometriosis, except for some nonhuman primates. 16 , 17 Rodent models have been used to mimic human endometriosis by grafting diseased tissues or cell lines into the peritoneum of immunocompetent mice. 18 , 19 , 20 These models are valuable for studying the pathological processes and drug evaluation of endometriosis; however, it is critical to consider the differences between humans and animals. For example, human interferon alpha‐2b reduced the size of experimental endometriosis in rats, but increased the recurrence of endometriosis in patients. 21 , 22 In vitro models derived from human tissues have been used to study the biology of endometriosis, including primary endometriotic cells and immortalized cell lines, such as epithelial (EEC12Z, EEC16), and stromal cells (ESC22B). 23 , 24 , 25 , 26 These cells are usually cultured as a monolayer on tissue culture plates and are called two‐dimensional (2D) cultures. In contrast, epithelial and stromal cells in endometriotic tissue exist within a complex 3D microenvironment and continuously interact with one another. The 3D culture model closely recapitulates the constitutive and functional characteristics of endometriotic lesions in vivo. Organoid is a promising technology that displays 3D histological and pathological features of the original organs. 27 , 28 Boretto et al. generated epithelial organoids from ectopic lesions, healthy endometria, and matched eutopic endometria of patients with endometriomas. 29 These epithelial organoids showed substantial proliferation and genomic stability during long‐term culture. Additionally, epithelial organoids recapitulate the endometriotic phenotype in vitro and in vivo; for example, ectopic organoids contain a much thicker lumen‐bordering cell layer than organoids derived from healthy and eutopic endometrium. The epithelial organoids can be expanded for a long time in vitro; nonetheless, the endometriotic lesions also contain other important cells, such as stromal cells, which are related to the migration, adhesion, invasion, and implantation of tiny endometriotic tissues. Therefore, the current organoid models have limitations in mimicking the pathological structures and functional features of endometriosis. Wiwatpanit et al. successfully generated the first complete multicellular endometrial organoids comprising epithelial and stromal cells, 30 whereas Gnecco et al. developed a fully synthetic extracellular matrix (ECM) that supports stable coculture of endometrial glandular epithelial and stromal cells, 31 in 2020 and 2023, respectively. Despite these advances, multiple cell‐based endometriomas (EO) models for investigating endometriomas remain underdeveloped, representing a critical research gap. Moreover, patient‐derived organoids and 3D stromal cell models have not yet been used to evaluate drug efficacy. In this study, we developed 3D patient‐derived organoids, stromal cell spheroids, and a coculture model of both cell types to mimic endometriotic lesions. They were identified and characterized using morphological and IHC methods. We further evaluated the effects of a type of clinical medicine (dienogest) 32 on stromal cell spheroids, epithelial organoids, and cocultures. These models are promising drug‐screening platforms.

The authors declare no conflicts of interest. The authors alone are responsible for the content and writing of this manuscript.

Endometriotic tissue samples were obtained from premenopausal women who underwent laparoscopic surgery for endometrial cyst removal following a diagnosis of endometriomas. Patients were consecutively enrolled from April 2023 onward. After approval by the Ethics Committee of Jinshan Hospital Affiliated to Fudan University and informed consent from the patients themselves, ectopic endometrial tissues were collected from women who underwent surgery for endometriomas. The inclusion criteria were as follows: (i) diagnosis of endometriomas after pathological confirmation, (ii) regular menstruation, and (iii) the patients gave informed consent. To minimize potential confounding factors, the following exclusion criteria were applied: (i) hormonal therapy receipt within the previous 6 months; (ii) history of pregnancy and lactation within 6 months; (iii) concurrent diagnosis of hormone‐dependent disorders (including uterine fibroids, adenomyosis, or breast hyperplasia); and (iv) presence of hepatic/renal dysfunction, endocrine disorders, or autoimmune diseases. The patient information is presented in Table  1 . Patients' information. Abbreviations: BMI, body mass index, calculated as the weight in kilograms divided by the square of the height in meters (kg/m 2 ); CA125, normal Cancer antigen 125; cm, centimeter; h, hour; rASRM stage, stage was classified according to the revised Classification of the American Society for Reproductive Medicine; VAS, visual analog scale. All surgical specimens were immediately transported on ice to the laboratory and processed within 6–48 h. All participants signed an informed consent form granting permission for tissue sampling and data collection. This study was approved by the Ethics Committee of Jinshan Hospital, Fudan University (no.: JIEC2021‐S49‐03). Fresh surgical specimens were preserved in tissue storage solution at 4°C for 6–48 h before processing (B001; Suzhou Jiyan Biotech. Co. Ltd., Suzhou, China). The specimens were further rinsed two to three times with tissue rinsing buffer (B002; Suzhou Jiyan Biotech. Co. Ltd., Suzhou, China) to remove blood and debris, and minced into small pieces using sterile surgical scissors. The suspension was centrifuged at 500 rpm for 5 min to collect tissue fragments. To maximize the yield of epithelial cells derived from severe fibrotic tissues with limited glands, we employed a two‐step enzymatic method, including matrix‐cell dissociation and cell–cell dissociation using different enzymes. After removing the supernatants, the fragments were transferred to a 60‐mm dish and dissociated in 3–5 mL of dissociation solution I (D001‐1; Suzhou Jiyan Biotech. Co. Ltd., Suzhou, China) containing supplement‐I (S001; Suzhou Jiyan Biotech. Co. Ltd., Suzhou, China) under gentle shaking conditions for 40–120 min in a 37°C incubator, with the specific incubation time guided by the extent of tissue fibrosis. During dissociation, the fragments were periodically triturated with a pipette every 15 min and monitored under an inverted phase‐contrast microscope (model: Nikon Eclipse Ts2; Nikon Instruments Inc., Tokyo, Japan) to assess tissue fragment dissociation. The dissociation process was completed when the tissues loosened and the epithelial glands were visibly released from the tissues. Subsequently, the dissociated tissues were collected and digested in TrypLE (12 604 013; Thermo Fisher Scientific, USA) for 10–15 min with intermittent shaking (two to three times every 5 min). Prolonged digestion broke down most glandular structures into smaller epithelial clusters and cell aggregates. The suspension was then diluted with five times of organoid wash solution (B004; Suzhou Jiyan Biotech. Co. Ltd., Suzhou, China) and filtered through a 100‐μm cell strainer. During filtration, most epithelial and stromal cells passed through the sieve; however, predominantly undigested fibrous tissue fragments were retained. The filtered cell pellet was collected, and red blood cells were removed using red blood cell lysis buffer (B005; Suzhou Jiyan Biotech. Co., Ltd., Suzhou, China). The cell pellet was resuspended in 100–300 μL human endometriosis organoid culture medium (ODM001; Suzhou Jiyan Biotech. Co. Ltd., Suzhou, China) and kept on ice for 3–5 min to cool the cells. The pellets were mixed with an equal volume of Matrigel (356 231; Corning, NY, USA) on ice. Fifty microliters of mixtures were deposited in a prewarmed 24‐well untreated plate and incubated in a 37°C incubator for 5 min. Subsequently, the plate was inverted for an additional 15 min in a 37°C incubator. Then, complete medium (0.5 mL of complete medium) (ODM001 supplemented with supplement‐I) was gently overlaid onto each Matrigel droplet. The culture plates were maintained in a humidified incubator at 37°C with 5% CO₂ for 48 h, and the complete medium was replaced with organoid medium without supplement‐I. Epithelial organoids typically form within 10–20 days, with medium changed every 2 days. When the organoids reached above 150 μm in diameter, they were enzymatically dissociated and passaged for further expansion. To culture stromal cells, the undigested tissues containing stromal cells retained on the strainer were collected and resuspended in enriched DMEM/F‐12 (OPM bioscience, Shanghai, China) with 10% fetal bovine serum (FBS, A5256701; Gibco, USA). These fragments were further plated in a culture dish and maintained for 5–10 days to allow stromal cell outgrowth. Additionally, stromal cells that migrated beneath the epithelial organoid/Matrigel mixture were harvested after careful removal of the overlying matrix. All stromal cells derived from the same patient were pooled and suspended in enriched DMEM/F‐12 containing 10% FBS. Once the stromal cells reached 70%–80% confluence, they were enzymatically dissociated and passaged for further expansion. We cultured stromal cells in a CSwell 600 chamber to form 3D stromal cell spheroids (KIT000‐0001; Suzhou Jiyan Biotech. Co., Ltd., Suzhou, China). The epithelial organoids (150–200 μm diameter) were harvested using organoid recovery solution (B008; Suzhou Jiyan Biotech. Co., Ltd., Suzhou, China) and dissociated with TrypLE for 5–10 min at 37°C. The organoids were digested into small clusters (20–30 μm diameter) comprising approximately three to five cells each cluster. The epithelial clusters and stromal cells (derived from the same patient) were counted using a cell counter (Halo counter HD‐4FL; Hiscore Inc., Beijing, China) and cocultured at varying ratios (1:2, 1:4, 1:10, 1:50, 1:100, and 1:200), such that one epithelial cell cluster (containing three to five cells) was mixed with 2, 4, 10, 50, 100 or 200 stromal cells, respectively. The cell mixture was embedded in Matrigel and cultured for 9 days in a custom coculture medium, consisting of a 1:1 ratio of organoid culture medium to stromal cell medium, to support both epithelial and stromal components. Images were acquired on Days 3 and 9. The diameter and number of organoids were manually measured for each condition using ImageJ (NIH, USA). The experiments were performed as described in a previous publication. 33 Briefly, the endometriomas tissues, epithelial organoids, and stromal cell spheroids were fixed in 4% paraformaldehyde (PFA, cat no.: 158127; Sigma, USA) dissolved in phosphate‐buffered saline (PBS) at 4°C for 12–24 hrs. Epithelial organoids and stromal cell spheroids were embedded in Organoid Embedding Gel (SE‐Gel, cat. no.: S‐001; Suzhou Jiyan Biotech. Co. Ltd., Suzhou, China) for paraffin embedding. Paraffin sections (5 μm) were stained with H&E, Periodic acid Schiff (PAS), and IHC methods. We performed H&E staining to characterize the morphology of the tissue and organoids, according to the manufacturer's instructions (C0105M; Beyotime Biotech, China). PAS staining was used to visualize mucin according to the manufacturer's instructions (C0142M; Beyotime Biotech, China). The sections were deparaffinized and sequentially incubated in periodic acid solution and Schiff's reagent, followed by counterstaining with H&E before dehydration. We used IHC staining to identify the biological markers of epithelial organoids and stromal cells. In brief, the sections were sequentially immersed in xylene and a graded series of ethanol (100%, 95%, and 70%) to remove paraffin and rehydrate the sample. Subsequently, epitope retrieval was performed in an EDTA solution (pH 9.0) or sodium citrate buffer (10 mM, pH 6) in an autoclave for 2 min. After inactivating endogenous peroxidase activity and blocking non‐specific binding sites with 5% bovine serum albumin (BSA), the sections were incubated with a specific primary antibody diluted in an antibody diluent overnight at 4°C. After washing with PBS to remove unbound primary antibody, a biotinylated secondary antibody (specific to the host species of the primary antibody) was applied to the samples and incubated for 60 minutes at room temperature. Detailed information on the primary and secondary antibodies used according to the manufacturer's instructions is shown in Table  2 . Nuclei were counterstained with hematoxylin. The pictures of stained sections were taken and analyzed using an Olympus #CX31 and a Hamamatsu#NanoZoomer S360. The surface microstructures of the epithelial organoids and 3D stromal cell spheroids were visualized using a HITACHI SU8010 cryo‐scanning electron microscope (cryo‐SEM), based on a previous method. 34 Antibodies information used for immunohistochemistry examination. Abbreviations: BMI, body mass index, calculated as the weight in kilograms divided by the square of the height in meters (kg/m 2 ); CA125, normal Cancer antigen 125; cm, centimeter; h, hour; rASRM stage, stage was classified according to the revised Classification of the American Society for Reproductive Medicine; VAS, visual analog scale. In vitro models that recapitulate intercellular crosstalk are significant for elucidating the pathophysiological mechanisms of endometriomas and for screening potential therapeutic agents. Therapeutic targets for endometriomas may be expressed in different cell types or play a crucial role in inhibiting interactions among various cell types within endometriotic lesions. Subsequently, we examined the responses of epithelial organoids, 3D stromal cell spheroids, and cocultures of these two cell types to dienogest (a selective PR agonist). The coculture ratio of epithelial organoid to stromal cell was 1:10 based on our initial optimization experiments. The cells were mixed with Matrigel at a density of 1000 cells/clusters per 10 μL drop and then cultured in a 96‐well plate. After seeding for 2 days, the cells were treated with dienogest at concentrations of 1, 5, 10, 20, and 40 μM in their corresponding culture media, as detailed in the section Methods 2.2–2.3. The medium was changed every 2 days with dienogest. The cultures were monitored on Days 0, 2, 4, and 6 by taking images. To determine the response of the stromal cell spheroids to dienogest, we first constructed 3D stromal cell spheroids using a Gel‐free 3D Cell Culture & Detection Plate (cat. no.: GF996; Suzhou Jiyan Biotech. Co. Ltd., Suzhou, China). Seeding of 2000 stromal cells per well yielded nine uniform spheroids, with each spheroid comprising approximately 220 cells. To assess drug‐induced morphological changes, the spheroids were monitored by bright‐field microscopy, and images were documented on Days 0, 3, and 6. The same concentration gradient of dienogest was applied for cocultures of epithelial organoids and stromal cells, which were prepared by culturing the cells embedded in Matrigel at a ratio of 1:10 according to method of 2.3 . Upon completion of the drug treatment, cell viability was quantified using the CellTiter‐Glo ® Luminescent Cell Viability Assay (Promega, USA) in accordance with the manufacturer's protocol. We performed statistical analyses using GraphPad Prism 8. A two‐way ANOVA with Dunnett's test for multiple comparisons (95% confidence intervals) was used in this study. All experiments were performed independently at least three times. Statistical significance was defined as p  < 0.05. All data were presented as mean ± sd.

Appendix S1.