Strategies for modelling endometrial diseases

Abstract

Each month during a woman’s reproductive years, the endometrium undergoes vast changes to prepare for a potential pregnancy. Diseases of the endometrium arise for numerous reasons, many of which remain unknown. These endometrial diseases, including endometriosis, adenomyosis, endometrial cancer and Asherman syndrome, affect many women, with an overall lack of efficient or permanent treatment solutions. The challenge lies in understanding the complexity of the endometrium and the extensive changes, orchestrated by ovarian hormones, that occur in multiple cell types over the period of the menstrual cycle. Appropriate model systems that closely mimic the architecture and function of the endometrium and its diseases are needed. The emergence of organoid technology using human cells is enabling a revolution in modelling the endometrium in vitro. The goal of this Review is to provide a focused reference for new models to study the diseases of the endometrium. We provide perspectives on the power of new and emerging models, from organoids to microfluidics, which have opened up a new frontier for studying endometrial diseases.

The human endometrium is the uterine lining, which first comes into contact with an embryo for pregnancy to be established. There is only a specific window of time during the secretory phase of the menstrual cycle in which the tissue is receptive to embryo implantation and, to achieve this state, the endometrium grows and differentiates in response to ovarian sex hormones^1,2 (FIG. 1 a). When no embryo is present, the upper part of the endometrial lining, called the functionalis, degenerates and is sloughed off, resulting in menstruation^1,3. The remaining endometrial layer, called the basalis, allows for the regrowth of the endometrium as oestrogen levels rise and the cycle repeats itself, on average, every 28 days throughout a woman’s reproductive years. This highly regulated and controlled regeneration of the uterine lining is subject to dysregulation in diseases.

Fig. 1 |. The human endometrium and its most common pathologies.

a | The healthy endometrium consists of epithelial, stromal, vascular and immune cells that are distributed in the two main layers of the endometrium, the stratum basalis and stratum functional. Every 28 days, on average, the functional layer grows out from the basal layer, sloughs off and is regenerated. This process is controlled by female sex steroids produced by the ovary during the proliferative and secretory phases of the menstrual cycle. Adapted from ‘Uterine Cycle’, by BioRender. com (2022). Retrieved from https://app.biorender.com/biorender-templates. b | Asherman syndrome is an acquired pathology presenting with intrauterine adhesions in the uterine cavity, which has little to no vascularization or glands. These adhesions replace the functional luminal epithelium. c | Endometriosis is a disease in which endometrium-like tissue is found outside the uterus, growing within the peritoneal cavity. d | In adenomyosis, endometrial glands and stroma invade the underlying myometrial layer, which can result in an enlarged uterus. e | Endometrial hyperplasia and cancer. Endometrial hyperplasia is commonly described as the pathological proliferation of endometrial glands and is considered to be a precursor of endometrial carcinoma.

When the endometrial lining is damaged, as in Asherman syndrome, hormone-driven regeneration cannot occur^4,5 (Fig. 1 b). In situations in which the endometrium is misplaced within the myometrium or outside the uterine cavity, as is the case for adenomyosis^6,7 or endometriosis⁸, respectively, the hormone response is altered. These conditions are often associated with chronic inflammation and pain (FIG. 1 c,d). When there is a chronic imbalance of hormones, endometrial proliferation and differentiation are offset, resulting in endometrial hyperplasia and even cancer⁹ (FIG. 1 e). These diseases commonly affect cis women (hereafter referred to as women) of reproductive age with no cure and limited options for relief.

Studies using in vivo (Box 1) and in vitro models of these diseases have yielded insight into mechanisms related to development, growth and survival of diseased tissues; however, there is still much work to be done. Much of the progress is hindered by the limitations of current model systems. We have seen new advancements in endometrial organoid technology, reviewed by others^10–14, as well as in microfluidic culture platforms^15,16. These tools are now beginning to improve our understanding of the mechanisms of endometrial diseases. In this Review, we focus on the in vitro models and microfluidics that have been used to study endometrial diseases, including endometriosis, adenomyosis, Asherman syndrome and endometrial hyperplasia and cancer (TABLE 1).

Box 1 |. In vivo models of endometrial disease.

Currently, neither in vitro nor in vivo models of endometrial diseases completely reproduce disease physiology. In vitro models lack important components of endometrial disease, including the immune system, endometrial endothelial cells and pericytes, ovarian hormones and the ability to observe metastasis in endometrial cancer. Equally, animal models cannot exactly reproduce human disease. Nevertheless, much has been learned from animal models of endometrial disease, which have been extensively reviewed^{144,161–163}. Common models are highlighted below.

Asherman syndrome

The characteristic fibrosis of Asherman syndrome is often modelled in animals by physically traumatizing the endometrium^160,164. Pregnant rats have also been subjected to surgical abortion, resulting in endometrial fibrosis and reproducing the iatrogenic cause of Asherman in humans¹⁶⁵.

Endometriosis

Spontaneous endometriosis occurs only in humans and non-human primates, including baboons, whose reproductive physiology is similar to that of women^{166 167}. Endometriosis can be induced in baboons by seeding menstrual endometrium in the peritoneal cavity¹⁶⁸. Although rodent models do not display spontaneous endometriosis, they are fairly cost-effective and can be genetically manipulated⁸⁹. Endometriosislike lesions can be formed in rodents by transplanting endometrial tissue into the peritoneal cavity^161,169,170 or by xenografting human endometriotic explants into immunocompromised mice^161,171.

Adenomyosis

Spontaneous adenomyosis occurs in non-human primates, mice, rats and rabbits^162,172. Adenomyosis has been induced in mice by oestradiol and progesterone treatment, neonatal exposure to tamoxifen or prolactin, or by inducing hyperprolactinaemia^162,172.

Endometrial cancer

Endometrial cancer is frequently modelled using genetically engineered mouse models containing a knockout of Pten^144. Conditional deletion of Pten in the uterus induces endometrial cancer in 100% of mice by 3–4 weeks¹⁷³. Other genetically engineered mouse models have been reported to develop endometrial cancer^146,174. Patient-derived xenograft models, in which a fragment of human endometrial tumour is engrafted into an immunocompromised mouse, represent tumour heterogeneity and preserve the genetic and histological characteristics of the original tumour^175–178.

Table 1 |.

Advantages and limitations of models of endometrial disease

Disease	Model	Advantages	Limitations	Key references
Asherman syndrome	Rodent model of mechanically damaged endometrium	Replicates the entire lesion with all cell types. Enables study of regeneration	Costly. Variations between animal and human physiology	Alawadhi et al. 2014 (REF.160)
	2D culture: wound healing assay	Simple and inexpensive. Can easily test compounds to determine regenerative effect	Cannot reproduce an adhesion. Does not include all cell types involved	de Miguel-Gomez et al. 2021 (REF78)
Endometriosis	Tissue explant	Preserves 3D tissue structure. Retains all cell types and physical interactions between cells. Does not require optimization of tissue digestion protocols	Reduced viability after 24 h. Limited ability to manipulate cell types of study	Fasciani et al. 2003 (REF.91)
	Co-culture model	Includes two or more cell types. Easily adaptable to different cell types. Study paracrine or direct interactions. Can be 2D or 3D cultures	Does not necessarily replicate tissue structure. Extended culture can be difficult	Chen et al. 2021 (REF104), Mei et al. 2014 (REF110)
	Organoid	Replicates epithelial cell polarity and hormone responsiveness. Reproduces characteristics and gene expression of eutopic and ectopic endometrium. Can be established from menstrual flow. Non-invasive. Amenable to drug testing and long-term expansion and biobanking	Current endometriosis organoids contain only one cell type. Does not reproduce interactions between endometrial cells and microenvironment. Extended experimental culture can be challenging	Boretto et al. 2019 (REF120)
	Microfluidic system	Can be used to precisely seed cells for 2D culture. Replicates uterine peristaltic movement	Current systems do not include culture medium circulation. Currently contain only endometrial cells	Chen et al. 2012 (REF125), Elad et al. 2020 (REF131)
Adenomyosis	Co-culture model	Includes two or more cell types. Easily adaptable to different cell types. Can be used to study paracrine or direct interactions. Can be 2D or 3D cultures	Does not necessarily replicate tissue structure. Long-term cultures can be difficult	An et al. 2017 (REF.135)
	3D model	Can examine invasion in vitro. Can study interaction between cell types. Can be used for drug testing	Current models do not contain epithelial cells. Relies on an exogenous scaffold	Mehasseb et al. 2010 (REF137)
	Microfluidic system	Mimics peristaltic shear stresses	Current systems do not include culture medium circulation	Elad et al. 2020 (REF131)
Endometrial cancer	Organoid	Reproduces histology and genetics of original tumours. Long-term expansion and biobanking. Amenable to drug testing and predicts patient responses	Current endometrial cancer organoids contain only one cell type. Does not recapitulate interactions between tumour and microenvironment	Boretto et al. 2019 (REF120)

The endometrium

Clinical features

A woman’s reproductive lifespan is marked by more than 400 menstrual cycles over the course of about 40 years (FIG. 1 a). Every month, growth and differentiation of the endometrium occurs in response to controlled fluctuations in levels of the ovarian steroid hormones oestrogen and progesterone, in preparation for the implantation of an embryo. If there is no embryo, the endometrium is no longer needed and is shed as ovarian hormone levels fall. There are two layers in the human endometrium. The functionalis layer is the upper two-thirds of the tissue, which is shed during menstruation and the basalis layer is the lower third of the endometrium closest to the myometrium, which remains during menses^1–3,17.

Multiple cell types make up the endometrium. The epithelial layer is composed of luminal epithelial cells, which line the endometrium, and glandular epithelial cells, which are invaginations of the epithelial cells extending from the lumen into the myometrium^18,19. Stromal cells as well as the extracellular matrix (ECM) provide structural and endocrine support. Blood vessels are remodelled into spiral arteries and perivascular cells and immune cells play important roles in endometrial remodelling throughout the cycle^1–3,17. Interestingly, the proportions of these cell populations change during the menstrual cycle^20,21. In the past 2 years, 3D imaging of human endometrial tissue demonstrated that the endometrial glands form a unique network in the basalis layer that expands horizontally along the muscular layer. These structures are reminiscent of rhizomes, with glands vertically emanating from the rhizomes into the functionalis layer^22,23.

In vitro models

Cells in monolayer.

Endometrial epithelial and stromal cells are two major cell types of the endometrium that respond to ovarian steroid hormones (FIG. 2a). Immortalized versions of these cells have been used to study endometrial biology and disease^24–27; however, immortalized cells can differ genetically and phenotypically from their tissue of origin after prolonged propagation. Because of this limitation, cell lines have remained less favoured than primary epithelial and stromal cells directly obtained from human endometrial tissues as a model of the endometrium^28,29.

Fig. 2 |. In vitro models of endometrial diseases.

a | 2D, adherent, monolayer cell culture is the traditional method of growing cells in vitro. The method is simple and cost-effective, but endometrial epithelial cells lose their polarity and hormone response during culture. b | Endometrial explants are short-lived in culture, but they preserve native cell-cell interactions, diverse cell populations and tissue organization. c | 2D and 3D endometrial co-cultures can include one cell type in a monolayer on the bottom of a culture plate, with another cell type or organoids in a transwell tissue culture insert (left). The cells communicate through paracrine factors released into the culture medium. Other common 3D endometrial co-cultures consist of endometrial epithelial organoids or whole glands plated in Matrigel in the bottom of a culture well, with stromal cells cultured in two dimensions on the surface of the Matrigel (right). d | Endometrial epithelial organoids consist of epithelial cells that form a hollow, gland-like structure in a 3D matrix, typically made out of Matrigel. Epithelial cells preserve their polarity, with their apical side oriented towards the lumen of the organoid. e | Multicellular endometrial organoids are made up of both endometrial epithelial cells and stromal cells. The stromal cells, located in the interior of the organoid, serve as a scaffold for the epithelial cells, which are polarized with their apical side facing the exterior.

Primary endometrial stromal cells maintain many of the important markers and functions seen in vivo, and can be readily decidualized in response to oestrogen, progestin and a cAMP agonist^29–32. This hormonal combination induces decidualization in endometrial stromal cells cultured as monolayers and has been used as a powerful tool to elucidate molecular mechanisms associated with hormone-driven differentiation of the stromal cells. Markers of decidualization such as insulin-like growth factor binding protein 1 and prolactin have been used as the gold standard for successful decidualization^30,32. Although decidualization can be induced in vitro, the hormonal conditions used for this purpose are not entirely physiological. For example, pharmacological concentrations (much higher concentrations than those found in vivo) of progestins and cAMP agonist are used, with medroxyprogesterone acetate as the progestin of choice, which also acts on the glucocorticoid receptor^33,34, and length of treatment varies from 72 h to 15 days^31,35,36.

Primary epithelial cells cultured in monolayer lack polarity, have a short lifespan and often lose responsiveness to hormones¹⁰. Additionally, epithelial glands have complicated winding and branching patterns in the tissue, which are not observed in primary epithelial cells cultured in two dimensions. Thus, endometrial cells cultured as monolayers do retain certain physiological characteristics and respond to exogenous hormones but lack the 3D architecture and lose their characteristics the longer they are propagated.

Three-dimensional cultures.

A decade ago, endometrial explants were used to model the endometrium in culture, as they maintained important characteristics of the native endometrial tissue, especially tissue architecture, tissue composition and direct contact between different cell types³⁷ (FIG. 2b). However, explants have limited viability and undergo substantial tissue degradation and glandular disintegration after 24 h in culture. As a result, organotypical models of the endometrium, typically consisting of both endometrial epithelial and stromal cells embedded in matrix, were developed to mimic the structure and scaffold of the native endometrium (FIG. 2c). These models can reproduce endometrial hormone responses and have revealed that paracrine and endocrine signalling between endometrial cell types are important in the endometrial response to hormones^38,39. A 2005 study embedded benign endometrial epithelial cells in Matrigel inside a tissue culture insert, with benign endometrial stromal cells growing on the plate beneath the insert³⁸. In this model, oestradiol stimulated proliferation of the epithelial cells, while the progestin medroxyprogesterone acetate decreased epithelial proliferation, demonstrating that primary epithelial cells grown in a 3D organotypic model retained their hormone responsiveness. A 2019 study sought to model angiogenesis in the endometrium and recreated the endometrial perivascular niche⁴⁰. This study used a gelatin hydrogel matrix to create an organotypic model consisting of endometrial stromal cells, human umbilical vein endothelial cells and a layer of endometrial epithelial cells. However, this model lacked perivascular cells, which have a mesenchymal stem cell-like regeneration potential and also contribute to the regulation of angiogenesis^41–43.

The emergence of organoid technology has transformed in vitro culture models in the research community (FIG. 2d). In 2017, two groups^44,45 introduced long-term expandable endometrial epithelial organoids embedded in Matrigel. These organoids were derived from human or mouse endometrium and cultured in a cocktail of growth factors and signalling molecules. The organoids were genetically stable, maintained polarity and reproduced the physiology of the glands of the native endometrium. Hormone response to oestradiol and progesterone was preserved in the organoids over 14 days. Progesterone treatment induced mucin secretion into the organoid lumen, which also occurs in the secretory endometrium in vivo⁴⁵. Microarray analysis revealed that transcriptional expression was very similar between primary endometrial glands and the corresponding endometrial organoids. Additionally, these organoids showed upregulation of epithelial markers and glandular secretion genes compared with primary stromal cells, demonstrating that the organoids maintained a distinct epithelial identity⁴⁴. In 2019, single-cell RNA sequencing of endometrial epithelial organoids treated with hormones⁴⁶ analysed the cell types present in these organoids in response to hormone treatments. Organoids treated with oestradiol for 2 days consisted of five cell clusters: ciliated, epithelial, proliferative, stem and unciliated, and treatment for an additional 6 days with oestradiol and progestin resulted in the appearance of an additional cell cluster, the secretory cells.

In 2020, a new, scaffold-free organoid model was described, containing both endometrial epithelial and stromal cell types⁴⁷ (FIG. 2e). When epithelial and stromal cells were plated on low-adhesion agarose micromoulds, polarized epithelial cells localized to the outer surface of the organoid, with stromal cells on the interior. The organoids expressed functional markers, including oestrogen receptor, progesterone receptor and androgen receptor, and responded to menstrual cycle hormones. Similarly, a 2021 study demonstrated the construction of the assembloid, bringing together epithelial organoids surrounded by a bed of stromal cells, which responded to hormones and kinase inhibitors⁴⁸. This assembloid is the most anatomically similar to the native endometrium out of all in vitro 3D models thus far, as it consists of stromal cells with embedded epithelial cells surrounding a hollow lumen.

Organoids have also been derived from endometrial fragments in menstrual flow, a non-invasive method that could facilitate gathering organoids from a diverse range of patients to study endometrial pathologies such as endometriosis^49,50 (BOX 2). Menstrual flow-derived organoids could also be used for precision medicine, such as determining endometrial sensitivity to chemotherapeutic agents. There have been many reviews on endometrial organoid technology in the past couple of years and the technology itself is readily used by many in the field^10–14,51.

Box 2 |. Challenges associated with human endometrial samples.

Endometrial samples for research usually come from endometrial biopsies or hysterectomies performed for endometrial cancer or benign uterine pathologies such as fibroids or endometriosis. There are many obstacles to collecting these specimens. Researchers must obtain an approved Internal Review Board protocol; set up a collaboration with a hospital, pathologist and pathology centre; hire and train patient coordinators; obtain consent from patients to collect their tissues and ensure the tissues reach the lab in a timely manner.

Screening patients for a study presents a particular challenge. Many researchers wish to enrol only premenopausal women whose endometrium is cycling, limiting the patients they can approach for consent. Use of combined oral contraceptive¹⁷⁹ or progestin-containing intrauterine devices¹⁸⁰ can result in endometrial atrophy and a very low cell yield. An individual’s current menstrual cycle stage also affects the thickness of the endometrium¹⁸¹ and how the cells respond to hormones. In addition, an estimated 20–30% of women undergoing hysterectomy are diagnosed with adenomyosis¹⁸², while approximately 15% are confirmed to have endometriosis¹⁸³. Endometrial samples from these women might not be suitable to study the healthy endometrium. Inter-patient variation represents another challenge, and it can be difficult to obtain enough samples to reach a meaningful conclusion. Finally, obtaining endometrial samples via biopsy or hysterectomy is invasive for the patient. Advances from the past year in establishing endometrial organoids from menstrual fluid provide a non-invasive way to easily sample from many patients^49,50. Differentiation of induced pluripotent stem cells (iPSCs) into endometrial stromal cells offers another method of obtaining endometrial cells through less-invasive procedures, such as a blood draw^184,185. iPSCs could be readily biobanked and expanded as needed. Future developments could enable the differentiation of other endometrial cell types, allowing the generation of entire endometrial models from one patient source, which could also be used in precision medicine approaches.

Microfluidics technology.

Standard tissue cultures consist of cells or explants placed in a cell culture dish with the appropriate medium. Between medium changes, cell cultures deplete the nutrients present while remaining submerged in increasing amounts of their metabolic byproducts for extended periods of time⁵². Fluidic perfusion systems automatically move medium through a culture, dynamically replenishing the medium, which is reflective of in vivo physiology and thereby promotes healthier in vitro cultures. Furthermore, shear stresses that are necessary for the physiological behaviour of certain cell types (principally endothelial cells) are also introduced. These microfluidic devices are often referred to as ‘organs-on-a-chip’ or microphysiological systems (MPSs)⁵³ and have been applied to model reproductive processes¹⁶.

To design and create a MPS of the healthy human endometrium, selection of materials is a key consideration. Polydimethylsiloxane (PDMS) is a widely used elastomer for MPSs because of its ease-of-use, optical transparency and oxygen permeability. Nonetheless it has some undesirable properties, such as the absorption of small hydrophobic molecules, including certain drugs and steroid hormones, which makes it less suitable for drug discovery and especially reproductive biology research⁵⁴. To avoid this issue, models have been moving towards nonabsorbent elastomeric compounds⁵⁵, biocompatible thermoplastics such as polysulfone⁵⁶, cyclic olefin (co)polymer⁵⁷ or polystyrene⁵⁸ and 3D printed materials. However, extensive biocompatibility testing is necessary as some plastics can leach endocrine-disrupting chemicals, such as bisphenol A, or toxic compounds into the culture medium⁵⁹.

MPSs have been used to recreate endometrial environments (FIG. 3a,b) such as the perivascular niche. A 2017 study demonstrated the creation of a dual-chamber PDMS-based microfluidic device designed to investigate the vascular-stromal interface in the human endometrium^60,61. The authors showed that the increased differentiation of stromal cells into decidual cells occurred owing to the flow-induced shear stress on endothelial cells. Despite the MPS not containing other important cell types such as pericytes or smooth muscle cells, the model demonstrated the importance of flow as well as paracrine interactions between endothelial and stromal cells. This model could reproduce endometrial hormone response in the proliferative and secretory phases over 28 days, making it highly relevant to the study of endometrial physiology. In 2021, a group created a vascularized endometrium-on-a-chip that incorporated the epithelium, stroma and blood vessels⁶² and demonstrated the effect of the emergency contraception drug levonorgestrel in increasing endometrial permeability and blood vessel regression. However, the translational value of this model in particular is limited owing to the use of high levels of hormones, immortalized cell lines and human umbilical vein endothelial cells, which are not specific to the endometrium. Nonetheless, cell-specific behaviour was maintained in long-term culture (14 days) for both models.

Fig. 3 |. Current approaches to model the endometrial niche and pathologies using microfluidic technologies.

a | Microfluidic device recreating the endometrial perivascular niche using a perfused endothelial chamber (composed of human umbilical vein endothelial cells (HUVECs)) and a static stromal chamber to study the role of the endometrial vasculature during decidualization. b | The design of a multichannel microfluidic device able to recapitulate the vascularized endometrium. Two channels with stromal fibroblasts generate a pro-angiogenic gradient that influences the angiogenesis and morphological changes of blood vessels residing in a third channel. The blood vessels in this system are composed of HUVECs. c | The multi-organ microfluidic platform (MFP) called EVATAR can be used to recreate the female reproductive tract in vitro. The Quintet MFP has been used for five different tissues. d | A co-culture device created by 3D printing a mould for polydimethylsiloxane (PDMS) enables the culture of an endometrium construct and ovarian cells (granulosa and theca cells). The endometrial well consists of endometrial stem cells, stromal fibroblasts and HUVECs supported in a hyaluronic acid, collagen and agarose hydrogel. e | Microfluidic channels are used to selectively pattern healthy or diseased endometrial stromal cells and human peritoneal mesothelial cells to study the peritoneal niche found in endometriosis. Here, cells were loaded into microfluidic channels (1), which were then removed (2), leaving the micropatterned cells (3). f | Microfluidic device that recreates healthy and diseased peristaltic flow patterns, introducing selective wall shear stresses on endometrial tissue-engineered constructs. PTFE, polytetrafluoroethylene. Part a reprinted from REF⁶¹, CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Part b reprinted with permission from REF⁶², Oxford University Press. Part c reprinted from REF⁶⁴, CC BY 4.0 (https://creativecommons.org/1icenses/by/4.0/). Part d reprinted with permission from REF⁶⁵, IOP Publishing. Part e reprinted with permission from REF¹²⁵, Oxford University Press. Part f adapted from REF¹³¹, CC BY 4.0 (https://creativecommons.org/1icenses/by/4.0/).

To recreate the endometrial lumen, another 2021 study used a similar microfluidic design that contained monolayers of primary endometrial epithelial and stromal cells in culture for 72 h⁶³. Here, culture medium flowed over the epithelial cells for 3 days, creating in vitro-derived uterine luminal fluid. Although this study did not investigate the effect of flow or hormone fluctuations, the effect of changes in maternal metabolic stressors such as levels of glucose and insulin on endometrial functions was assessed, including the effects on transcriptional response and protein secretion.

The endometrium has been included in a few multiorgan devices. The EVATAR system integrated the mouse ovary, human fallopian tube, endometrium, ectocervix and liver, which were connected together by microfluidic channels with dynamic media flow⁶⁴ (FIG. 3c). A 2018 study described a ‘10-way’ multi-MPS platform accommodating tissues that include the liver, gut, lung, heart, pancreas, brain, skin, kidney, skeletal muscle and the endometrium⁵⁶. Finally, a 2020 study used microfluidic manufacturing techniques to produce a static ‘dual reproductive organ-on-chip’ that recapitulated the crosstalk between multiple cellular components of the human endometrium and the ovarian follicle⁶⁵ (FIG. 3d). These devices incorporated a model of the healthy endometrium in communication with other organs. Such platforms could be applied or redesigned to study how diseases of the endometrium interact with other organ systems. The ideal endometrial MPS remains a theoretical exercise for now, missing key features such as scarless healing, immune response and interaction with the myometrium in the uterine junctional zone.

Asherman syndrome

Clinical features

Asherman syndrome is an acquired endometrial disorder characterized by intrauterine adhesions or adhesions of the endocervix. The prevalence of Asherman syndrome can vary substantially; depending on the subpopulation, it is reported in 0.8–45.5% of women with infertility^4,5. The opposing uterine walls are partially or completely fused by fibrous adhesions, obliterating the uterine cavity⁶⁶. The degree and localization of intrauterine adhesions lead to the complete or partial dysfunction of the endometrium, resulting in menstrual abnormalities, reduced fertility and recurrent pregnancy loss. Asherman syndrome is considered to be mainly caused by iatrogenic trauma to the endometrium due to postpartum or miscarriage curettage, but could also be induced by intrauterine infections^4,66. Because the endometrium is not able to regenerate without scarring, as it normally does following menstruation or parturition, it is probable that the trauma reaches the basalis layer or further, resulting in a poorly vascularized and thin endometrium and a limited response to oestrogen and progesterone^67–71.

In vitro models

Much of the current research related to Asherman syndrome is focused on the regeneration of the endometrium, restoring its regular thickness and proliferation rate. The most common models for endometrial regeneration are mouse or rat animal models (BOX 1) in which endometrial damage is induced mechanically via scraping^72,73 or excision⁷⁴ or by injecting ethanol or other cytotoxic solutions^75,76. Notably, some studies of Asherman syndrome did use 3D organoids or spheroids, but only as a method to aid regeneration, and not as a tool to study the pathology itself^76,77. The in vitro models that have been used to study Asherman syndrome directly were standard 2D wound healing assays to investigate the regenerative capabilities of certain compounds such as platelet-rich plasma⁷⁸ or media conditioned by mesenchymal stem cells⁷⁹. Advances in technologies could potentially enable the development of in vitro models that can reproduce scarring, fibrosis or even adhesions between two endometrial surfaces, enabling the use of in vitro models to study the pathology of Asherman syndrome. In addition, incorporation of perivascular cells (vascular smooth muscle cells and pericytes) would enable researchers to examine the regenerative properties of these cells^41–43 in the context of Asherman syndrome.

Endometriosis

Clinical features

Endometriosis is a disease in which endometrial tissue grows outside the uterus in the peritoneal cavity and ovaries. Symptoms include severe pain and infertility in approximately 10% of women who are of reproductive age globally⁸. The treatment of endometriosis is estimated to cost approximately $22 billion each year in the USA alone⁸⁰ and there is currently no cure. Theories for the development of endometriosis include Sampson’s retrograde menstruation theory, as well as metaplasia of cells that abnormally transdifferentiate into endometrial cells at ectopic sites^8,81. Hormone responses in endometriosis in both the ectopic lesions and the eutopic endometrium are aberrant, with a dependence on oestrogen and a suboptimal response to progesterone^82–85. It remains unclear whether the presence of ectopic lesions influences the eutopic endometrium or whether there are other inherent differences in the eutopic endometrium that promote endometrial tissue survival at ectopic sites⁸⁶. With the exception of some non-human primates^87,88 (BOX 1), animals do not develop endometriosis spontaneously and thus in vitro models using human tissues have been used to study the pathogenesis of this disease. Primary endometriotic cells are difficult to culture and have a limited lifespan, leading to the development of other in vitro culture systems⁸⁹.

Explant models

Simple endometrial explant culture systems have been used to model important steps in endometriosis development. To study the interactions between ectopic endometrial fragments and fibrin, a protein that is deposited on damaged mesothelium and which is implicated in the formation of peritoneal adhesions⁹⁰, a 2003 study grew endometrial explants in a 3D fibrin matrix⁹¹. Endometrial glands and stroma invaded the matrix and formed tubular structures by 2 weeks of culture, and after 5 weeks, CD31⁺ capillaries had sprouted. Follow-up studies identified glycodelin and COX2 as molecules that can promote neovascularization in endometriosis⁹² and showed that lovostatin, a VEGF inhibitor, inhibited cell growth and angiogenesis in a dose-dependent manner⁹³. The original endometriotic explants used in the 2003 study became necrotic at 5 weeks of culture; however, by that time, the outgrowth from the explant exceeded the size of the original tissue⁹¹. Another study of endometrial explants found that treatment with the progestin dienogest decreased outgrowth of the explants in a dose-dependent manner, suggesting that dienogest might reduce early endometriosis invasion⁹⁴. Thus, although explants have limited viability, they can be useful to study the outgrowth and invasion of endometriosis explants through a matrix, as the growing tissue continues to be viable.

Co-culture models

Co-culture of endometriotic cells with other cell types of interest in endometriosis has revealed much about the important interactions between cells and their microenvironment. Co-cultures have provided an important tool to observe early steps in endometriosis establishment, including the beginning of adhesion, invasion and angiogenesis, which are difficult to observe in vivo^94,95.

Chick embryo chorioallantoic membrane.

Early assays to study angiogenesis in the endometrium included co-culturing endometrial explants directly on chick embryo chorioallantoic membrane (CAM)⁹⁶. CAM-endometrial co-cultures revealed that angiostatic agents significantly decreased angiogenesis in endometriosis-like lesions in CAM, showing that this assay could be used for testing anti-angiogenesis drugs for endometriosis⁹⁷. A 2008 study also showed that transplanting endometrium from women taking combination oral contraceptives onto CAM resulted in fewer endometriosis-like lesions compared with transplanting endometrium from women not taking oral contraceptives⁹⁸. These studies demonstrated the utility of the CAM-endometrial co-culture model in studying the early stages of endometriosis, including lesion establishment and angiogenesis, as well as its use in drug testing.

Peritoneal or amniotic membrane.

Co-culture of endometrial explants in direct contact with peritoneal explants were carried out to demonstrate that the endometrium mostly adhered in regions where mesothelial cells were missing, suggesting that a damaged mesothelium would allow endometriosis lesions to become established^99–101. Similarly, endometrial explants adhered to stripped amniotic membranes, showing that this model could be used to study interactions between endometrial cells and the ECM^100,102. Co-culture of endometrium with amniotic or peritoneal membranes thus offers the ability to observe the interactions between ectopic endometrium and the mesothelium or matrix at the implantation site.

Bone marrow-derived stem cells.

Bone marrow-derived stem cells (BMSCs) have been shown to migrate into the eutopic endometrium and are speculated to contribute to endometrial regeneration¹⁰³. BMSCs are also known to associate with endometriotic lesions; however, their impact remains unclear¹⁰⁴. Co-culture of commercially purchased BMSCs with endometriotic cells using transwell inserts revealed that BMSCs promote the proliferation of endometriotic stromal cells but not healthy stromal cells, possibly owing to stem cell-derived factors¹⁰⁴. Reciprocally, co-culture of primary endometrial stromal cells from mice with BMSCs isolated from mouse bone marrow in a transwell system promoted the migration of BMSCs¹⁰⁵. However, much controversy remains surrounding the role of BMSCs in the endometrium. It is still unclear whether BMSCs have a key role in endometrial physiology or whether they simply acquire endometrial characteristics after migrating into the uterus¹⁰⁶. As bone marrow contains a heterogeneous mixture of cells¹⁰⁷, and because there are no established markers for BMSCs in the endometrium¹⁰⁶, future studies could be improved by ensuring that the BMSC types used display the same markers as the BMSCs that associate with endometriotic explants, or by examining BMSCs isolated from explants themselves.

Immune cells.

It has been suggested that immune dysfunction in women with endometriosis could allow ectopic endometrial cells to escape destruction by the immune system¹⁰⁸. Co-culture of endometriotic cells and immune cells has revealed important paracrine interactions between these two cell types. Studies have shown that co-culturing macrophages with endometriotic cells altered the characteristics of the macrophages towards the more tolerant M2 phenotype^109,110. Additionally, macrophages co-cultured with endometriotic stromal cells either directly or within a tissue culture insert were less phagocytic than those cultured with normal stromal cells^109,110. Treatment with oestradiol and the dioxin TCDD synergistically promoted M2 activation in macrophages that were cultured in direct contact with endometriotic stromal cells¹¹¹. It is clear that the presence of endometriotic cells shifts macrophages towards a more permissive phenotype, which could make the establishment of endometriotic lesions easier. Reciprocally, immune cells can alter the characteristics of the endometrial cells, resulting in a more endometriosis-like phenotype. Studies have shown that co-culture with macrophages resulted in increased proliferation, clonogenicity and invasion of endometrial stromal cells as well as secretion of the pro-inflammatory cytokine IL-6 (REFS.^112–115). These co-culture studies demonstrate that immune cells can influence endometrial cell behaviour.

Natural killer (NK) cells have been shown to affect macrophages and endometriotic cells as well. Conditioned medium from co-cultures of endometriotic stromal cells and NK cells promoted increased macrophage migration compared with conditioned medium from endometriotic stromal cells alone¹¹⁶. Co-culture of endometriotic stromal cells with NK cells added directly to the culture wells caused a decrease in CD16 expression on NK cells compared with NK cells cultured alone, possibly decreasing NK cell activity. IL-15 secreted by endometriotic stromal cells decreased the expression of NK cell functional molecules such as granzyme B and interferon-y (IFNy)¹¹⁷. Another study showed that NK cell viability, cytotoxicity and functional marker expression were decreased by co-culturing endometriotic stromal cells and macrophages in direct contact with NK cells in a transwell insert¹¹⁸.

These studies support both possibilities, first, that endometriotic cells in ectopic lesions dysregulate immune cell function, promoting immune escape, and second, that immune cells in endometriosis are pro-inflammatory and promote ectopic endometrial cells to establish themselves and become pathological. Although co-cultures are valuable models to study endometriosis-immune cell interactions, the in vivo situation involves more than two cell types. Future co-culture studies could include a diverse immune cell mixture of macrophages, NK cells and T cells, along with both epithelial and stromal endometriotic cells. Various combinations of these cells from healthy and diseased tissues could reveal whether immune cells are actively promoting lesion development or are compromised and unable to clear ectopic lesions.

Spheroids and organoids

The first in vitro spheroid endometriosis model was developed in 2014 by culturing both new and established endometriosis cell lines in low-adhesion plates¹¹⁹. Cells in 3D cultures more closely recapitulated the histological and molecular characteristics of endometriotic lesions than did those in 2D cultures. A 2019 study generated organoids from endometriotic lesions, healthy endometrium and matched eutopic endometrium of patients with endometriosis¹²⁰. Ectopic organoids were morphologically different from eutopic and disease-free organoids, with a thicker, stratified epithelial layer, and they could produce endometriosis-like lesions when engrafted into mice. Bulk RNA sequencing of these three organoid types revealed differential gene expression in ectopic organoids compared with healthy organoids, while eutopic organoids and healthy organoids displayed similar gene expression. Ectopic organoids showed altered expression of ECM-receptor interaction genes, adhesion and invasion genes and PI3K-AKT pathway genes, among others. Other studies have confirmed the ability of the endometrial epithelial organoid model to faithfully reproduce the characteristics of ectopic endometriotic lesions, eutopic endometrium of women with endometriosis and healthy endometrium, enabling the comparison of these tissue types^121–123.

Organoids derived from ectopic lesions and eutopic endometrium preserve the methylation patterns found in ectopic and eutopic biopsy samples, respectively, and ectopic and eutopic organoids have methylation patterns that differ from each other¹²¹. Organoids from the eutopic endometrium of women with endometriosis display a different expression of glycodelin A compared with healthy organoids¹²³. Additionally, an epithelial organoid model revealed that progesterone receptor B is downregulated in both ectopic and eutopic endometrium compared with healthy endometrium, but through different mechanisms¹²².

Endometriotic stromal cell spheroids are also useful to study the early stages of endometriosis in the context of endometriotic lesion invasion of the ECM. In a 2021 study, endometriotic stromal spheroids, with or without endometriotic epithelial cells, were grown in either collagen I matrix or Matrigel¹²⁴. Endometriotic stromal spheroids, but not single stromal cells or spheroids co-cultured with endometriotic epithelial cells, directionally migrated through the matrix while remodelling it. Thus, spheroid and organoid models of endometriosis are useful in reproducing and identifying differences in ectopic lesions compared with normal endometrium, as well as differences between the eutopic endometrium of women with and without endometriosis. Spheroids and organoids can be used to study gene expression, hormone response, DNA methylation and invasion of endometriosis.

Microfluidic models

Microfluidic technology has been used to understand the interactions between endometriotic cells and peritoneal mesothelial cells. A 2012 study used microfluidic channels made from PDMS to create a 2D micropatterned static co-culture of endometrial stromal cells and human peritoneal mesothelial cells (HPMCs) taken from patients with endometriosis and from healthy control individuals¹²⁵. Here, the microenvironment found in peritoneal endometriosis was recreated by seeding one channel with endometrial stromal cells and the other with HPMCs (FIG. 3e). Then, the channels were removed and cell interaction and migration speed were monitored dynamically. Interestingly, when healthy HPMCs were co-cultured with both healthy and endometriotic stromal cells they were able to withstand the invasion of the stromal cells. Diseased HPMCs lost cell-cell contact and died when co-cultured with both types of endometrial stromal cell, suggesting that peritoneal health could be an important factor in the establishment of endometriosis.

Different microfluidic platforms were also used to create sensitive methods that characterized the matrix metalloproteinase activity and mechanical properties of endometrial cells from patients with endometriosis and were able to distinguish them from endometrial cells from healthy individuals^126,127. These findings demonstrate the versatility of this technology to provide detailed insights into the pathophysiology of endometriosis. Studies have attempted to recapitulate the peristaltic movement of the smooth uterine muscle layer and its effects on the endometrium^128–131 (FIG. 3f). Developing such systems would be especially useful for modelling endometriosis, in which patients experience a higher frequency of uterine contractions compared with women without endometriosis¹³². Microphysiological systems could potentially be used to support long-term co-cultures of endometrial cells with other disease-relevant cell types including immune cells, mesothelial cells and perivascular cells.

Adenomyosis

Clinical features

Adenomyosis is characterized by the invasion of endometrial tissue into the myometrium of the uterus. The burden oflesions can range from several small lesions dispersed throughout the myometrium to larger nodules or cysts^6,7. As a result, this condition can cause an enlarged uterus, heavy uterine bleeding and severe pain with one study reporting approximately 82% of patients with adenomyosis undergoing a hysterectomy and 37.6% reporting chronic use of pain medication¹³³. The causes of adenomyosis are unknown although theories have been proposed, including invasion of endometrial cells through a damaged junctional zone or de novo generation of endometrium-like tissue owing to differentiation of misplaced embryonic Mullerian remnants. Similar to endometriosis, growth of the displaced tissue is dependent on oestrogen. A 2021 study used 3D visualization of the endometrial glands within adenomyotic lesions, revealing an ‘ant colony’-like organization with direct invasion of endometrial glands into the myometrium²³. Such visualization of native tissues has provided novel insights into the structural pathology of adenomyosis.

In vitro models

Co-cultures.

Similar to the pathogenesis of endometriosis, the immune cell component of the endometrium might have an important role in the development of adenomyosis. This relationship has been interrogated using co-cultures of endometrial epithelial cells with macrophages, which demonstrated epithelial-to-mesenchymal transition in the epithelial cells, regardless of whether the epithelial cells were derived from healthy women or women with adenomyosis, and showed increased M2 polarization of the macrophages^134,135. Co-culture of endometrial stromal cells with M2-polarized macrophages resulted in decreased apoptosis and increased proliferation and invasiveness of the stromal cells compared with stromal cells cultured without macrophages¹³⁶. Thus, adenomyotic endometrial cells promoted tolerance in macrophages, whereas M2 macrophages promoted a more invasive phenotype in endometrial cells. Future studies could involve more complex co-cultures including adenomyotic stromal cells, epithelial cells and immune cells, which could reveal much about how this crosstalk is dysregulated in adenomyosis. Finally, co-culture of epithelial and stromal cells from adenomyosis with myometrial cells could reveal mechanisms associated with invasion and survival.

Three-dimensional models.

A 2010 study demonstrated the development of a 3D model of adenomyosis in which myometrial cells were suspended in a collagen matrix, with stromal cells seeded in a 2D layer on top of this matrix, which allowed the invasion of the stromal cells into the collagen matrix with myometrial cells to be measured¹³⁷. Adenomyotic stromal cells invaded statistically significantly more than control stromal cells, and co-culture with adenomyotic myometrial cells increased the invasiveness of both adenomyotic and healthy stromal cells. Using this model, the same group showed that treatment with oestradiol or the selective oestrogen receptor modulator tamoxifen increased the invasiveness of stromal cells, showing the potential of this model to quantify invasion, study the role of hormones and test drugs in adenomyosis¹³⁸. Future 3D models that include adenomyotic epithelial cells could help elucidate their role in myometrial invasion.

Microfluidic systems

Uterine contractions create peristaltic flows within the uterus, and hyperperistalsis can lead to the development of adenomyosis or endometriosis¹³¹. A 2020 study created a microfluidic model to mimic the peristaltic wall shear stresses that occur during peristalsis by culturing endometrial epithelial cells on top of Matrigel containing myometrial smooth muscle cells¹³¹. By deforming the cover of a fluid-filled chamber where these cells were covered, the authors were able to generate shear stress¹³¹. F-actin expression in both epithelial and myometrial cells increased based on the time the cultures were exposed to peristaltic wall shear stresses. These microfluidic models offer an opportunity to study peristaltic stresses that would be difficult to observe or control in vivo. Nonetheless, many technological advancements need to take place to enable the recreation of adenomyotic lesions¹³⁹.

Endometrial hyperplasia and cancer

Clinical features

More than half a million new cases of endometrial cancer occur per year worldwide¹⁴⁰. The USA has the highest incidence of endometrial cancer in the world, and disease rates continue to increase globally¹⁴¹. Endometrial cancer is typically divided into two major subtypes, with type 1 being the most prevalent and associated with risk factors such as marked obesity, diabetes mellitus and hypertension^9-142, whereas type 2 endometrial cancers are more aggressive and less common, carrying a different mutation profile. The Cancer Genome Atlas study of endometrial carcinoma revealed four molecular subtypes of endometrial cancer: polymerase epsilon ultramutated; microsatellite instability hypermutated; copy number low; and copy number high¹⁴³. Polymerase epsilon ultramutated tumours are characterized by somatic mutations in polymerase epsilon accompanied by a very high mutation rate (232 × 10⁻⁶ mutations per Mb), whereas hypermutated tumours are characterized by mutations in the DNA mismatch repair pathway, resulting in microsatellite instability and a high mutation rate (18 × 10⁻⁶ mutations per Mb). Copy number high and copy number low tumours are distinguished on the basis of the number of copy number alterations they contain. These genomic features permit a reclassification that could better inform post-surgical adjuvant treatment for women with aggressive tumours.

Most cases of endometrial cancer are strongly associated with continuous sex steroid imbalance due to excessive oestrogen and insufficient progesterone action in endometrial tissue⁹. Endometrial cancer develops when benign cells of the endometrium grow in an uncontrolled manner, accumulating mutations and transforming into malignant cells. Oestrogen promotes genotoxicity and proliferation of the endometrium, whereas progesterone antagonizes oestrogen-driven growth by promoting differentiation⁹. Depending on the grade and stage of the cancer, hysterectomy is the first line of treatment, which can be accompanied by radiation or chemotherapy. Preclinical studies for drug testing have been limited to cell lines^144,145 and animal models¹⁴⁶ (BOX 1), and many targeted drugs have had a modest impact in clinical trials. Given the rising incidence of endometrial cancer, the risk factors associated with lifestyle and behaviour, and the lack of therapeutic solutions, additional tools for mechanistic investigation as well as drug testing are needed at the level of both prevention and treatment ofthis disease.

In vitro models

Organoids.

Studies from the past 5 years have demonstrated that organoids derived from endometrial tumours recapitulate important histological, genetic and gene expression features of the original tumours^{44,120,147–149}. A 2017 study from one of the first groups to develop endometrial organoids found that endometrial cancer organoids recreate characteristics of the original tumour, including a disorganized epithelial structure and breaching of the basement membrane⁴⁴. Following up their initial study, Boretto et al.¹²⁰ then showed that organoids derived from endometrial tumours could be expanded long term, displayed long-term genomic stability and recapitulated tumour histology and morphology. Endometrial cancer organoids preserved the microsatellite instability, somatic copy number alterations and mutations of the original tumour and recreated the tumour when orthotopically transplanted in vivo. In addition, organoids established from other pathologies, such as endometrial hyperplasia and Lynch syndrome, maintained the mutations that occurred in the original tissue¹²⁰.

Patient-derived endometrial cancer organoids can also be subcutaneously injected or orthotopically engrafted into the uterus of immunocompromised mice, which provides a physiological tumour microenvironment¹²⁰ (BOX 1). Xenografted organoids reproduced tumour histology and biomarkers, including loss of hormone receptor expression, and metastasis from the uterus to the peritoneum occurred in high-grade cancer organoids¹²⁰. The accuracy with which endometrial cancer organoids reproduce patient tumours suggests that these organoids provide an excellent opportunity for testing drugs to treat endometrial cancer. In the first study to examine endometrial cancer cell lines in three dimensions, cells grown in 3D culture were more resistant to the common chemotherapeutic agents doxorubicin and cisplatin than those grown in 2D culture, indicating that the results of drug testing studies could be substantially different in 3D models¹⁵⁰. Multiple studies have revealed that endometrial cancer organoids successfully recapitulate patient-specific responses to chemotherapeutic drugs^120,147,151. In fact, a 2021 study tested common chemotherapeutic drugs in endometrial cancer organoids and successfully predicted that one patient’s tumour was resistant to several drugs¹⁵¹. Demonstrating the feasibility of using endometrial cancer organoids to screen drugs on a larger scale, a 2020 study used mouse endometrial cancer organoids to test 276 small molecules that target epigenetic factors in vitro. This study identified a menin-mixed-lineage leukaemia (menin-MLL) inhibitor and validated this inhibitor in an orthotopic xenograft mouse model of endometrial cancer¹⁵². Endometrial cancer organoids reliably recreate tumour histology, genetics and drug response, and have enormous potential for biobanking and use in precision medicine. Future endometrial cancer therapeutic strategies could involve creating endometrial cancer organoids from drug-resistant tumours and screening chemotherapeutic agents to predict which drugs a patient is likely to respond to.

Microfluidics

Although microfluidics systems are being used in endometriosis and adenomyosis research, they have not yet been used to study endometrial cancer pathogenesis. These studies are surely to come, as microfluidic models have been used to study breast cancer¹⁵³ and ovarian cancer¹⁵⁴. Future microphysiological models of endometrial cancer could provide valuable information about how endometrial tumours metastasize to specific target organs by conditioning the pre-metastatic niche¹⁵⁵. For example, endometrial cancer organoids could be grown in microfluidic co-culture with liver spheroids, allowing observations about how secreted factors from the cancer organoids alter the liver spheroids to make them more receptive to colonization. The inclusion of endothelial cells in these microphysiological models could determine how endometrial tumours might alter distant vasculature via secreted factors to promote the permeability of blood vessels to cancer cells. Finally, the inclusion of immune cells in microphysiological models of endometrial cancer would incorporate the role of inflammation as well as provide insights into responses to immunotherapy.

Perspectives for the future

New technologies facilitate new discoveries. There have been great advancements in modelling the endometrium as well as improving culture systems to provide dynamic flow. As these technologies are adapted to understand the pathogenesis of endometrial diseases by the research community, the challenge now lies in knowing how to apply these new technologies in a meaningful way.

The overarching goal is to mimic endometrial tissue physiology in vitro. The human endometrium is one of the most challenging tissues to model ex vivo given the intricate cyclic changes of the tissue, each with its morphologies and functions. The ideal endometrial model, as depicted in FiG. 4, would represent the architecture of the endometrium, including the lumen, functionalis and basalis layers and the myometrium. The model would contain all the important cell types of the endometrium, including epithelial, stromal, endothelial, perivascular and immune cells, and the appropriate matrix, all within an MPS that promotes dynamic flow of medium. The medium could include circulating immune cells, and biosensors could be integrated within the cultures to measure activities in real time. Additionally, multiple MPS units can be connected to study inter-organ communication. Components of this model can be changed to mimic the endometrial disease of interest or to determine the major drivers of disease.

Fig. 4 |. Idealized design of MPS of the healthy endometrium and its diseases.

An idealized microfluidic device incorporates various cell types and niches of the natural endometrium, including a myometrial layer and both basalis and functional layers of the endometrium. The myometrium ideally contains inner myometrial smooth muscle cells of the junctional zone and outer smooth muscle cells that form bundles and can contract with certain frequencies. This contractile frequency is important to recapitulate certain aspects of endometriosis and particularly adenomyosis, in which lesion invasion can also be assessed. Cell health and behaviour are monitored continuously by on-chip imaging and incorporated biosensors. The basalis region contains rhizome-like epithelial structures that are vascularized and perfused with physiological capillary structures that contain endothelial cells and pericytes, while the functional layer contains vascularized organoids generated from menstrual flow. The addition of new cell types to endometrial organoid models will allow researchers to interrogate new aspects of endometrial diseases in vitro. In particular, immune cells, including macrophages, natural killer (NK) cells and T cells, play an important part in the pathogenesis of diseases such as endometriosis. Endothelial cells are important in the growth and remodelling of the endometrium, and angiogenesis is essential to the establishment and progression of endometrial pathologies, including endometriosis and cancer. Furthermore, fibrosis and scarring, modelled with stromal fibroblasts, would help in investigation of intrauterine adhesions. This idealized microphysiological system (MPS) can be integrated with other organs-on-a-chip for toxicity screening, drug metabolism, pharmacokinetics and studying diseases within a system. Finally, new technologies that allow 3D printing of both matrix proteins and live cells can offer improved precision and adaptability in generating complex bioengineered structures.

A complete in vitro model of endometrial disease must include immune cells. Immune cells such as T cells, NK cells and macrophages can be dysfunctional during the establishment of endometriosis, adenomyosis and endometrial cancer and can promote immune evasion. Future models could include various immune cells from healthy or pathological endometrium to reveal the complex interplay between endometrial cells and immune cells. Vascular remodelling and angiogenesis have a key role in normal endometrial function and promote growth of diseases when unregulated, especially in endometriosis¹⁵⁶ and endometrial cancer¹⁵⁷.

Incorporation of a vascular network would involve endometrial endothelial cells, as well as an appropriate niche for vessel formation and branching. Studies from the past year have shown that pericytes have mesenchymal stem cell properties and display a limited ability to differentiate into endometrial stromal-like cells^41,43. In addition to their role in regulating angiogenesis, pericytes could be included in pathological models to examine their potential for regenerating the endometrium in diseases such as Asherman syndrome⁴¹.

The matrix that organoids or 2D cultured cells are in contact with is an important aspect of the pathological microenvironment and could come from the specific niche of the diseased tissues, such as ECM derived from myometrium for adenomyosis studies, or ECM from peritoneal tissues for endometriosis. The ability to 3D print matrix materials and cells could enable easy creation of customized scaffolds and structures. Human endometrial ECM can be obtained through decellularization of endometrial tissue¹⁵⁸ or synthetic hydrogels can be used to generate matrices with similar properties to the human endometrial ECM¹⁵⁹. The molecular and biophysical properties of the matrices can be modulated to recreate specific healthy or pathological niches to examine how diseased cells interact with these microenvironments.

It will be important to improve our understanding of how the endometrium regenerates after menses, especially for pathologies such as Asherman syndrome, in which regeneration does not occur. 3D models have not yet reproduced the degeneration of the functionalis during menses. Future models could seek to incorporate organoids derived from both functionalis and basalis endometrium. Notably, menstrual flow-derived organoids would by definition consist of the functionalis tissue. Finally, there is an undeniable dysregulation of hormones in all of the diseases mentioned in this Review. Hormones are essential for endometrial function and must be included in cultures. Fine tuning hormonal conditions to mimic physiology is needed, as current concentrations to induce decidualization in vitro are very high^30–32, probably because other cell types that contribute to the decidualization process are absent. It was interesting to note that many studies included in this Review on in vitro models of endometrial diseases did not include or describe hormonal stimulation. Inclusion of hormones will be essential as we strive towards representing the endometrial diseases in vitro.

Microfluidic platforms offer advancements from the standard static cultures in terms of mechanical flow dynamics, constant replenishment of new medium and elimination of toxic metabolites. Improvements in microfluidic technology that would facilitate wider use of these systems would include fabrication materials that do not absorb sex steroid hormones, improved affordability and a user-friendly platform. Depending on the needs of the study, integration of biosensors for important signalling molecules or hormones such as oestradiol and progesterone and bioanalytical tools within the microfluidic platform would allow researchers to collect data in real time (FIG. 4). Microfluidic platforms that allow extended culture time of endometrial models, for example, through multiple menstrual cycles with cyclic levels of menstrual hormones, would enable researchers to observe alterations in the endometrium that might accumulate over time in response to risk factors such as unopposed oestrogen action.

Microfluidic platforms also allow the recreation of mechanical influences from peristaltic movements in the myometrium, which can continue to improve our understanding of how mechanical stresses affect neighbouring endometrial cells or endometrial tissue within the myometrium, as in adenomyosis. With regard to effective drug testing platforms, a robot-assisted high-throughput microfluidic system would be essential. MPSs could include multiple different tissue types relevant to the disease in question. As we consider the use of multiple organs in one platform, the notion of deriving all organ models from the same patient is most attractive and is becoming possible with induced pluripotent stem cell technologies. This approach would permit organs with the same genetics to communicate with each other and enables researchers to test drugs in a culture system that is personalized.

Conclusions

It is apparent that the new advancements in culture models and platforms hold incredible potential to unravel unknown mechanisms that drive endometrial disease development and growth. As previously mentioned, the challenge lies in how best to implement these technologies to obtain the most meaningful data. It is tempting to use the newest tool, as it grows in popularity, and try to fit it into our research. However, specific biological questions should be answered using the correct approach that is customized to reflect the disease and the niche of interest. For now, a ‘one-size-fits-all’ approach does not exist, and biology should serve as the blueprint for technological innovation. In the past few years, many researchers have used endometrial organoids and organotypic cultures derived from healthy endometrium and have shown these systems to reflect in vivo characteristics of the tissue. Now, efforts should pivot towards generating reliable disease models using the appropriate cells that interact in a physiological manner using organoids and microfluidics, among others.

Key points.

• Endometrial diseases, including Asherman syndrome, endometriosis, adenomyosis and endometrial cancer, represent a major health burden in reproductive-age women and do not have effective therapies.

• Traditional in vitro culture techniques have long been used to study endometrial diseases, but they are limited by using single cell types and static cultures.

• Emerging technologies, such as organoids and microphysiological systems, are physiologically relevant and can reproduce many characteristics of native tissues and disease states.

• Although multiple microphysiological systems that model the endometrium have been developed, most of these have not yet been applied to endometrial diseases.

• Further innovations in organoid and microfluidic models will enable more accurate endometrial disease modelling and drug testing.