Essay: Endometriosis

INTRODUCTION
Endometriosis is a benign, estrogen-dependent, inflammatory disease that affects women during their reproductive age. It is defined as the presence of endometrial glands and stroma outside the uterine cavity; the lesions are typically located in the pelvis but can occur at multiple sites, including bowel, diaphragm, and pleural cavity1. Endometriosis is one of the most frequent gynecological disorders, with an estimated prevalence of 5-10% in women in the reproductive age group and while it is a common non-malignant process, ectopic endometrial tissue and resultant inflammation can cause dysmenorrhea, dyspareunia, chronic pain, and infertility2. Symptoms can range from minimal to severely debilitating.
Even though endometriosis was first described in 1860, the pathogenic mechanisms underlying the development and maintenance of this condition remain uncertain. In 1927 Sampson proposed a now well supported and widely accepted pathogenic mechanism, according to which fragments of menstrual endometrium reach the peritoneal cavity by retrograde menstruation, where they deposit and are able to attach to and grow on the peritoneal surfaces3. However, because such reflux of menstrual tissue occurs in nearly all women of reproductive age4, dysregulation of additional factors must contribute to the establishment and maintenance of ectopic endometrium.
Because patients are rarely diagnosed in the early stages of disease and so at the time of clinical presentation endometriosis is already well established in most women, the pathophysiology of this disorder cannot be assessed by experimental evidence in humans. In addition, ethical issues represent an obstacle to the design and the performance of controlled experiments in large populations. A further limitation is that, in order to monitor the disease progression, repeated laparoscopies would be needed. Thus, the design of different strategies is fundamental to allow the identification and clarification of the etiopathogenic mechanisms and the pathophysiology of endometriosis, as well as the development of new therapeutical approaches. For this reason, several animal and in vitro models have been designed for the development of better methods for early diagnosis and potential therapeutical interventions, even though none of them totally mimics all aspects of human disease.
Here we will provide a review of in vitro and animal models commonly used and their relevance to different scientific questions.
IN VITRO MODELS
In vitro models offer the possibility to investigate the signaling pathways at the molecular level, which are at the base of pathophysiology of endometriosis, by using primary cells or cell lines. Furthermore, co-culture models can be used to study the interaction between endometrial cells and other cell types, such as the peritoneal mesothelium, or endothelial cells.
Epithelial and stromal cells
Endometrial epithelial (EEC) and stromal (ESC) cells can be isolated from endometrial biopsies and endometriotic tissue to be cultured in vitro. However, because epithelial cells of endometriotic tissue loose their proliferative capacity during in vitro culture over several days, stromal cells were chosen over epithelial cells in most studies5, 6. Indeed, stromal cells can be cultured more easily and stably for much longer periods than endometrial cells 7. In order to overcome this obstacle and gain further knowledge on the biology of endometriosis, endometriotic epithelial cell lines have been obtained from biopsies of different types of endometriosis. Furthermore, the number of primary cells derived from patient lesions is often limiting for many studies, therefore researchers rely upon cell lines for studying signaling cascades, therapies, and cell-to-cell interaction.
The first endometriotic epithelial cell line to be established was the one obtained from an active peritoneal lesion by transfection with simian vacuolating virus 40 T-antigen (SV40 T), called 12Z 8. These cells express all hormone receptors (ERα, ERβ, and PR) and aromatase 9, and when co-injected with endometrial stromal cells into mice they give rise to peritoneal lesions histologically matching human endometriosis in vivo10. Moreover, immortalized EECs and ESCs display gene expression profiles closely matching those observed in women with endometriosis and endometriotic animal models11. Therefore, immortalized EECs and ESCs represent a useful tool to study the pathophysiology of peritoneal endometriosis.
In order to study other subtypes of endometriosis other than the peritoneal one, Bono et al. purified ovarian endometrioma epithelial cells and immortalized them by transfection of hTERT, cyclin-dependent kinase 4 (cdk4), and human cyclinD1 7. These cells retain the expression of progesterone receptor B and their growth is indeed inhibited by various progestins. On the other hand, the estrogen receptor (ER) is expressed at low levels and aromatase activity is completely lacking. This endometriosis model, being derived from ovarian endometrioma cells, is useful for the study of carcinogenesis of endometriosis, other than the molecular pathogenesis of the disease. An ovarian stromal cell line also exists12, and can be used for the same purposes.
Controls are provided by normal endometrial epithelial cell line, obtained by using human papillomavirus and human telomerase reverse transcriptase (hTERT) for immortalization of endometrial glandular cells 13. These cell lines are also suitable to study the early pathogenesis of endometriosis, however they cannot be used for the generation of therapeutic models, since many characteristics typical of endometriosis are not present.
Because each cell line is derived from individual patients, we cannot be sure that they provide a real representation of the disease. However, they are suitable to study signaling pathways, invasion, migration and proliferation of endometriotic cells, studies that cannot be performed in the few cells isolated from primary tissue.
Mesothelial cells
To investigate the interaction between peritoneum and endometrial cells, mesothelial cell lines14 or primary mesothelial cells15 have been established. A peritoneal mesothelial cell line (LP9) has been obtained through transfection with SV40 T, and it is now commercially available. It has been mostly used in co-culture models with endometrial cells, both stromal and epithelial, to study adhesion16 and invasion17, 18.
Endothelial cells
It is known that a rich vascular supply is needed for the development and maintenance of endometriotic lesions, especially in the peritoneum, which is relatively poor in vascularization compared with the eutopic endometrium19. Therefore, neoangiogenesis plays an important role in the pathogenesis of endometriosis. To explore new potential anti-angiogenic therapies, endothelial cells have been used to develop in vitro models of endometriosis. In these models, endothelial cells are isolated from human endometrium and immortalized by hTERT20; also, they have been proven to retain their phenotype in culture20. To present, there are no studies showing isolation and propagation of endothelial cells obtained from endometriotic lesions, that we are aware of.
Human endometrial endothelial cells (HEECs) have been employed in endometriosis research to investigate the factor involved in the increased angiogenesis capability of ectopic endometrium 21, 22, and the immune system role in the pathogenesis of endometriosis 23, 24.
Alternatively, human umbilical vein endothelial cells (HUVECs) can be used as an endothelial cell model for angiogenesis research25,26. Canosa et al. employed GFP-expressing HUVECs in co-culture with endometrial mesenchymal stroma cells (E-MSC) to analyze the endothelial differentiative ability of eutopic and ectopic E-MSC, in a setting mimicking the pericyte-endothelial interaction in endometriosis25. However, HUVECs are isolated from macrovessels, while endometriotic lesions depend on microvessels; for this reason HUVECs express different extracellular matrix proteins that may influence study results, especially when analyzing the hemostatic balance27.
Immune cells
Although peritoneal fluid is readily accessible during laparoscopy and can provide immune cells for the study of endometriosis pathophysiology28, the cells retrieved may not represent the immune cell phenotype present within endometriosis lesions, as suggested by the difference between peritoneal fluid macrophages and tissue-resident macrophages. Thus, many attempts have been made to retrieve and isolate tissue-resident macrophages from endometriosis lesions by flow cytometry; unfortunately, due to the small size of the biopsy, isolation of these cells revealed to be extremely difficult, limiting also any manipulation ex vivo.
In order to overcome this problem and yield a large number of cells for gene expression and functional studies, peripheral blood monocyte-derived macrophages can be plated and activated by different cytokine and estradiol to generate a phenotype similar to an endometriosis macrophage. 29–31
Nerve fibers
Based on the discovery of a possible role for nerve fibers in the pathogenesis of endometriosis, and to investigate their role neuropathic pain of endometriosis, Meschsners’ group employed a chicken dorsal root ganglia (DRG) model and sympathetic ganglia to investigate the effect of peritoneal fluid from women with endometriosis on nerve outgrowth32.
Rat DRG have also been used to explore the mechanisms underlying endometriosis-associated pain, in particular the possible role of endothelial cell-nerve33 and macrophage-nerve crosstalk29, also by gene expression studies.
Another possible model to study the peripheral nerve mechanisms in endometriosis is the one used by Greaves et al. to explore ER-mediated expression of nociceptive ion channels34. It consists of human embryonic stem cells converted into sensory neurons with a nociceptor-like phenotype using small molecules inhibitors; with this method, neuronal fate acquisition is much faster than during in vivo development, therefore this quick and efficient derivation of nociceptors offers access to studies of human pain35.
Three-dimensional in vitro models
A major limitation in the previous in vitro models is that EECs are cultured as monolayers on tissue culture plastics, hence as two-dimensional cultures, while in vivo EECs exist as a three-dimensional entity, constantly interacting with stroma and forming cell-cell interactions with their entire surface36.
To overcome this issue, three-dimensional cell cultures have been developed; in one model endometrial fragments are cultured on a 3D fibrin-matrix37, mimicking the very early pathogenic events of endometriosis, with invasion, gland and stroma formation, and neoangiogenesis.
With the aim of studying superficial ovarian endometriosis, a different 3D model using endometriotic EEC lines was developed36; cells were cultured in non-adherent conditions using polyHEMA-coated cell culture plastics, forming spheroid structures closely resembling human peritoneal endometriosis lesions, also on the molecular level36.
Very recently, a modular, synthetic, and dissolvable extracellular matrix hydrogel has been developed to foster functional 3D epithelial-stromal co-culture. It can be dissolved on-demand to recover intact cells and paracrine signaling proteins for subsequent analysis38.
In summary, in vitro models represent proper experimental systems for the investigation of the molecular and cellular processes at the bases of the pathogenesis of endometriosis. Primary cells from tissue samples should be chosen over cell lines whenever possible to achieve more certain results. When this is not possible, models should be chosen carefully, since cell lines do not represent the subtypes heterogeneity of endometriosis. The use of a 3D model allows for a more realistic cell-cell interaction evaluation and could be used to develop new treatments targeting pathways, such as cellular hormones synthesis, growth factors, neovascularization signaling, cytokines and interleukins.
ANIMAL MODELS
CHICK EMBRYO CHORIOALLANTOIC MEMBRANE (CAM) MODEL
The chick embryo chorioallantoic membrane (CAM) assay has been used as a model for endometriosis to study the mechanisms involved in adhesion, invasion, and angiogenesis39, 40. The model consists in the grafting of human endometrial tissue on the CAM of fertilized eggs, which serves as a transient gas exchange surface similar to the lung, and is therefore highly vascularized41. This membrane appears at day 3 of incubation and grows up to day 1042. Heterologous tissue can be easily cultured on the CAM without rejection due to the lack of a complete immune system until embryo development day (EDD) 18, even if the presence of T and B cells is detectable at EDD 11-1243.
Endometrial fragments of all cycle stages can invade the mesenchymal layer of CAM and develop endometriosis-like lesions consisting of endometrial glands and stroma within three days after grafting of the human tissue; for this to happen, tissue integrity of the transplanted endometrium is a prerequisite, as single endometrial cells do no not result in the formation of lesions44.
As with tumors, the dense microvascular network of the CAM allows the analysis of angiogenesis in ectopic endometrium and, consequently, of anti-angiogenic drugs effects45. It has been shown that formation of endometriosis-like lesions was significantly impaired after the administration of angiogenesis inhibitors46 and that the peritoneal fluid of women with endometriosis carries a higher angiogenic potential compared to that of non-affected women47.
Moreover, the extracellular matric (ECM) of CAM is similar to the ECM of the human peritoneum renders CAM model suitable to study the invasive properties of endometrioc tissue. Expression of matrix metalloproteases has been thus assessed in several studies, showing that it is increased in endometrial fragments, leading to adhesion and invasion onto the CAM48. By inhibiting MMP activity, invasion capacity of the endometrial tissue was reduced49.
Another endometriosis model in ovo is obtained by grafting endometriotic tissue (deep endomeriosis or ovarian endometriotic tissue) on CAM, instead of endometrial tissue (figure 1A). This model allows studying the effect on treatment on endometriosis. We thus observed that ICI182’780 (estrogen antagonist) topically applied on deep endometriosis transplants significantly reduced the tissue size (figure 1A and B).
This model provides the advantage of working with real endometriotic lesions, yielding a model more accurate and similar to in vivo conditions; moreover, it is suitable for studying treatment feasibility for endometriosis therapies, providing an alternative model to avoid animal suffering and sacrifice before further in vivo evaluation.
Despite CAM model can be used for a restricted amount of time of 14 days (without ethical issue), and is not appropriate to study immunological and inflammatory responses50, it carries several advantages. CAM allows for easy accessibility to ectopic lesions, whose formation can be closely followed during the course of the experiment; it is also a cheap and simple method that is close to the in vivo conditions and consents a rapid implementation and adaptation of the experimental protocol according to the results obtained.
RODENT MODELS
Rodents are the most common animal model used for endometriosis research. This is mainly due to their low cost, short generation time, and the possibility to genetically manipulate them easily. On the other hand, they also present several limitations, such as the lack of spontaneous menstruation; therefore, rodents do not develop endometriosis spontaneously. In order to induce endometriosis in these animals, two main methods have been developed, namely a homo-/autologous and a heterologous models.
Autologous and Homologous murine models
Homologous models consist in surgical transplantation of endometrial fragments in immunocompetent recipients. This has been achieved in different animals and with different methods. The first model of endometriosis has been induced in rats by autotransplantation of uterine fragments into the peritoneal cavity51; the same experiment was replicated years later in mice, with suturing of uterine fragments onto the mesenteric vessels52. However, these models do not offer an ideal representation of human endometriosis, as myometrium was transplanted together with the endometrial tissue; when only endometrium was injected into the peritoneal cavity no lesions developed.51
After this first attempt, new methods were designed in order to create a model with more similar characteristics to human endometriosis. For the first time, Somigliana et al. were able to induce lesions by collecting the endometrium from an ovariectomized and estrogen-treated mouse and injecting it, after mincing it, into the peritoneum of a syngenic mouse, also treated with high-dose exogenous estrogens.53 Although this model was able to overcome the myometrium issue, it carries some important limitations. First, strong estrogen supplementation in recipient animals probably influences development and progression of endometriosis, being an estrogen-dependent disease; secondly, the endometriotic lesions are small and embedded in the murine peritoneum, therefore difficult to identify. To overcome this problem, a donor mouse ubiquitously expressing green fluorescent protein (GFP) was developed; in this way the location and the growth of the transplanted tissue could be observed by fluorescent microscope.54 In the same study, it was also demonstrated that lesions would not grow without estradiol supplementation.
Since the sensitivity of GFP is relatively low, a transgenic donor mouse expressing luciferase was engineered and its uterine biopsies were transplanted in a recipient albino mouse, allowing a higher sensitivity and non-invasive monitoring of the lesions.55 This model is well suited to study the lesion response to anti-angiogenic drugs, but it is limited as both endometrium and myometrium are transplanted.
Recently, Samani et al. developed a mouse model of surgically induced endometriosis to study whether endometriosis-derived cells are able to migrate and form micro-metastases to distant organs, as lung, spleen, liver or brain56. They introduced syngenic endometrial tissue containing a Red Fluorescent Protein (DsRed) into the peritoneum of immunocompetent mice; after 8 weeks they were able to detect fluorescent cells in the extrapelvic organs mentioned above, in all mice tested, by FACS and immunostaining.
Despite autologous models do not truly reflect human endometriosis, they remain a valuable model to study this disease inasmuch they have an intact immune system, which is known to play an important role in the pathophysiology of endometriosis.
Menstrual murine models
With the aim of reproducing the human pathogenesis of endometriosis, several groups tried to develop a “menstruating” mouse model that is a mimicking of the endometrial breakdown seen in women menstruation. All these models follow a similar design, consisting in the use of an ovariectomized mouse treated with estradiol and progesterone, and in the injection of oil in the uterine lumen in order to stimulate endometrial decidualization.
In the first model to be developed, hormonal withdrawn was performed at the same time of oil injection and after 79 hours post-withdrawal, menstrual endometrium was shed.57 External stimulus was needed in order to induce decidualization. To have a more homogeneous endometrial response, the model has been refined by delivering progesterone via an implant, which was removed only 49 hours after oil injection; this allowed a more powerful drop in progesterone levels, mimicking the human cycle.58 The similarity of this model to human menstruation was confirmed by another study, which replicated the experiment and performed molecular analysis of the endometrium, demonstrating that decidualized tissue had the same features of menstrual endometrium observed in women.59
Chang et al. exploited the first menstrual model to induce endometriosis in mice; donor mice were genetically modified for K-ras activation in endometrium, which was transplanted in subcutaneous pockets of wild type recipient animals. Very low frequency of K-ras activation was shown to be necessary to generate endometriotic lesions, without the need for estrogen supplementation.60
The progesterone-implant model of menstrual mouse was used by Greaves et al. to create a new endometriosis model. They kept the progesterone implant for 4 days after the decidualization stimulus in order to yield a greater amount of menstrual endometrium, which was then transplanted in the peritoneum of an estradiol-primed recipient mouse. Estradiol was continued after endometrial transplant in order for the lesions to grow, obtaining endometriotic lesions very similar to human ones.61 As you can imagine, the limitation of this model lies in the fact that there is a strong estrogenic drive, unlike in humans, but it has the advantage of being simpler that the Chang et al model, not requiring genetic modifications.62
Recently, the Greaves model has been used by the same group to study peripheral and central hyperalgesia, typical of endometriosis. It was shown that induction of endometriosis resulted in CNS maladaption consistent with central sensitization, and that by antagonizing EP2 receptors you can reverse primary and secondary hyperalgesia. This proves that this kind of model can be used to test analgesics pre-clinically.63
Heterologous murine model
In order to overcome the limitations associated with the homologous models, and therefore to take into consideration possible causal factors residing in the endometrium itself, new animal models have been developed by xenotransplantation of human endometrial tissue into immunodeficient mice.
In the first models to be designed, human endometrial tissue or ectopic endometriosis biopsies64 were transplanted either by subcutaneous administration65, by minilaparotomy66, or by inoculation into the peritoneal cavity67 in athymic nude mice. These mice were either with intact ovaries66 in some studies, and ovariectomized and treated or not with exogenous estradiol65 in others; hormone supplementation was shown to increase proliferation of endometrial glands, but not the rate of implantation nor the lesion morphology68.
Human endometrial tissue can be collected at any stage of the menstrual cycle, since it does not seems to affect the development of ectopic lesions.66 Attachment of endometrial fragments occurs 2 days after implantation and angiogenesis can be observed since day 4.68 Therefore this model is suitable to investigate the early stages of the biological mechanisms involved in endometriosis development.
Non-invasive in vivo techniques for observation of endometriotic lesions have been developed also for heterologous models. The first group to attempt the establishment of such techniques were Tabibzadeh et al., who marked human endometrial cells with a fluorescent lipophilic dye before intraperitoneal injection into athymic mice69; an important limitation in this method is the inability to differentiate between live and dead cells. Therefore, Fortin et al. exploited the GFP technique to obtain fluorescent human endometrial fragments, but it resulted in poor transfection efficiency70. More recently, the Wang group improved this model by using an isolation-transfection-incubation method with red fluorescent protein (RFP) and obtaining a high rate of transfection and high fluorescence intensity, both in subcutaneous and intraperitoneal lesions71. This technique provides an optimal tool to evaluate and monitor novel drugs efficiency in preclinical animal studies.
In the athymic nude mouse models seen so far the maintenance of intact and well-preserved implanted human endometrial fragments is limited in time, not lasting more than 4 weeks, probably due to a retained activity of B lymphocytes72 and high natural killer (NK) cell compensation.73 Therefore, new recipient mice models have been developed, with an even more defective immune system; severe combined immune deficiency (SCID) mice and non-obese diabetic (NOD) SCID mice, lacking T and B lymphocytes, were proven to be reliable models for endometriosis by several studies 73,74,68, in which lesions were found proliferating up to 10 weeks after transplantation 73. Nonetheless, NK cell activity is to some degrees retained in these models, causing deterioration of transplanted tissues in the long term, as they play a critical role in the inhibition of angiogenesis mediated by IL-12 75, making them inappropriate to study long-term effects of novel therapies. In order to obtain a model suitable to observe such effects, mice lacking T, B, and NK lymphocytes were developed by using different strains, such as RAG-2/γ(c)KO 76, BALB/c-Rag2-/-/IL2rg-/- 77 and NOD/SCID/ γ(c)null 78.
The latter has been used by Masuda et al. to create a model satisfying the requirements of uniform transplanted human tissues, that reproduce the human ectopic and eutopic endometrium characteristics, and of a model available for long-term, noninvasive, and real-time assessment. This was achieved by transplanting a single-cell suspension of human endometrium under the kidney capsule of ovariectomized NOD/SCID/ γ(c)null mice78. Such model is useful for testing potential treatments of endometriosis, but it carries the important limitations of using a severely immunocompromised recipient animal and a non-physiologic site of transplantation, hence peritoneum and immune system roles are not taken into consideration.
Recently, a different heterologous murine model has been proposed, consisting in γ irradiated mice transplanted with human endometrial mesenchymal stromal cells instead of human endometrial tissue79. This method yielded a more efficient gland formation than the conventional model, providing a promising tool to study the responsible stem cells in the pathogenesis of endometriosis and to test new therapies based on human endometrial stem cells.
In summary, as with homologous models, the heterologous ones have the advantage of being cheap and easy to handle; furthermore, they offer the possibility to study the endometrium-intrinsic mechanisms involved in the establishment of endometriosis, to test drugs or antibodies with species-specific effects, and for long-term studies that are not feasible in women.
On the other hand, the heterologous models, being based on immunodeficient mice, do not provide a typical immunological environment, which plays an important role in endometriosis pathogenesis. 80 Also they are extremely heterogeneous in the mouse strain and the tissue source used, making impossible to build a consensus on clinical research priorities.
Rat models
As previously said, rats were the first rodents used to develop a model of endometriosis51. Nowadays the rat model is mainly used to investigate endometriosis-associated pain and pain therapies. Berkley et al. were the first to demonstrate vaginal hyperalgesia together with visceral-visceral referred hyperalgesia in rats81. This study paved the way to several subsequent experiments that clarified pain mechanisms in endometriosis82–86. More recently, innovative techniques were used in the rat model to record visceromotor responses to vaginal distension87, and the electophysiologic responses from sensory neurons innervating an endometriotic lesion on the gastrocnemius muscle of the rat88. To investigate endometriosis-lesion innervation, Silveira et al. employed allogenic transplants of isolated rat endometrium into the anterior eye chamber of female rats, since it is an immunologically-privileged site, which may mimic the immunotolerance observed in the peritoneum of women with endometriosis89. This study proved that the rat anterior eye chamber model is suitable to investigate the factors involved in the reinnervation of ectopic endometrial tissue by different populations of nerve fibers.
Despite the rat model reproduces some aspects of human endometriosis, it relies on the suturing of uterine fragments onto different sites; therefore, it does not simulate the real dissemination of shed endometrial tissue in the pelvic cavity. Another limitation of the rat model is the low number of genetic manipulated animals commercially available, compared with the mouse models.
Other models
As far as what concerns other rodents implied for endometriosis model, the dorsal skin chamber in hamster and rats has been extensively used in the past years, mainly to study angiogenesis and anti-angiogenic drugs. This model consists in implanting two symmetric titanium frames in the dorsal skinfold of the animal, then removing a layer of skin so to reveal the striated muscle, which is subsequently covered with a removable coverslip. To induce endometriosis, endometrial fragments are autologously transplanted onto the striated muscle.50 In this way, angiogenic sprouting, microvessel density and diameter, and lesion growth can be observed and measure in vivo, also in response to anti-angiogenic compound.90 More recently, it has been used also to investigate the effects of the luminal epithelium on the morphological development and vascularization of endometriotic-like lesions.91 Although this model carries several advantages, it has the important limitation of reproducing endometriosis in a non-physiologic microenvironment, being the tissue transplanted on striated muscle and not onto the peritoneum in the abdomen; therefore, the interactions playing between lesions and peritoneum may not be replicated in the skinfold chamber.
PRIMATE MODELS
The first non-human primate model was designed in 1950 by Te Linde and Scott by surgical reposition of the cervix in the rhesus monkey, resulting in increased retrograde menstruation and development of endometriotic lesions. After that, several new non-human primate models have been established, employing different animals, such as the Japanese macaque, the pigtailed macaque, the baboon, and the cynomologus monkey. 92
The advantages of using non-human primate model (NHP) are several, especially when investigating the pathophysiology of endometriosis. NHP are very similar to humans in their endometrial physiology, morphology and menstrual cycles; also they develop spontaneous endometriosis.93
Of all NHP models, the most commonly used is the baboon model, due to its favorable size and similar reproductive anatomy. Their menstrual cycle is similar to women’s in both duration and endometrial remodeling; baboons and women share similar type of placentation and similar changes in the eutopic endometrium during uterine receptivity phase.94 Other advantages offered by the baboon model include non-invasive cycle monitoring by perineal inspection, continuous breeding also in captivity, spontaneous presence of peritoneal fluid, vaginal trans-cervical uterine access, and cross-reactive antibodies and DNA/RNA analytical methods with humans. 95
These characteristics make the baboon with spontaneous endometriosis an excellent model to study the pathophysiology of the disease. Incidence of spontaneous endometriosis in baboons greatly varies between studies. D’Hooghe et al. observed a prevalence of 17 to 25%, which increases to 32% in animals kept in captivity for more than two years96. In other studies, the rate of spontaneous endometriosis was similar if not inferior to the prevalence observed in humans, ranging from 1 to 5% and the captivity did not increase the incidence rare. 97 Anyway, spontaneous endometriosis in these animals is not frequent enough, nor its progression quick enough to perform economically sustainable experiments. Therefore several methods to artificially induce endometriosis have been developed. Based on the previous experience of Te Linde and Scott, D’Hooghe et al. developed two different methods: the first one consisting in the occlusion of the uterine cervix 98 in order to increase the retrograde menstruation, and the second one involving the intraperitoneal inoculation of menstrual endometrium99. Although in the latter study endometrium was retrieved by uterine curettage, thus possibly containing myometrial tissue (not found in menstrual material), Fazleabas et al. obtained the same results by injecting menstrual endometrium obtained using a pipelle. 100
In order to study deep nodular lesions, a new baboon model has been proposed by Donnez et al. in which endometriotic nodules were obtained by laparotomic autografting of full thickness uterine biopsies or endometrium and junctional zone containing biopsies.101 This model was challenged by D’Hooghe, who claims that intrapelvic injection of the total amount of menstrual endometrium is a more physiologic and preclinically relevant model of endometriotic nodules and infiltrative endometriosis. 95
Recently, Nair et al. developed a more efficient baboon model of endometriosis by inducing an unopposed estrogenicity with a potent anti-progestin in baboons surgically inoculated with autologous menstrual endometrium. Endometriosis developed as early as 3 months post-inoculation and it was more extensive and severe in the anti-progestin treated animals. 102
Interestingly, in all cases of induced endometriosis, most lesions were reported to be on peritoneum or uterine surface, but endometriomas were never observed.
Overall, in order to achieve an appropriate experimental setting, induction of endometriosis offers a more controlled and standardized method with respect to spontaneous endometriosis in baboons. 93, 99, 103,
NHP model, and in particular the baboon model, have been used in many studies investigating the pathogenesis of endometriosis, the pathophysiology of the disease but also to test new drugs in a preclinical setting.
Retrograde menstruation as the main risk factor of endometriosis is supported by several studies in baboon: first of all, the ways endometriosis has been induced in this model (obstruction of cervix, menstrual endometrium inoculation 99, 98) prove that retrograde menstruation plays an important role in the pathogenesis of endometriosis. Secondly, in baboons with spontaneous endometriosis there is a higher incidence and recurrence of retrograde menstruation 93; also the extent of endometriosis has been shown to be dependent on the amount of endometrial tissue inoculated intrapelvically.99
Studies on NHP provided clarifications on the link between peritoneal inflammation and endometriosis, since studies of cause-effect relationship in women cannot be conducted due to ethical reasons. In baboons, the volume of peritoneal fluid and the local concentration of WBC and inflammatory cytokines increase after induction of endometriosis, but disappear after 2-3 months.104 This is found also in baboons with spontaneous endometriosis. Furthermore, it has been shown that the percentage of peripheral blood CD4+ and IL-2R+ cells is increased in baboons with long-term or advanced endometriosis (both spontaneous and induced) when compared to animals with normal pelvis or recent endometriosis. 104 In another study it was found that also CA-125 levels were increased in peritoneal fluid after endometriosis induction. 105 Therefore, peritoneal inflammation appears as a consequence, rather than a cause, of endometriosis.
Theory that inherent endometrial dysfunction is associated with infertility of women affected by endometriosis due to implantation failure, is finding confirmation in studies on baboons. It has been shown that the presence of ectopic endometriotic lesions alters the eutopic endometrium gene signature leading to phenotypic changes in both baboons and humans. 94 In particular, during the early stages of the diseases the eutopic endometrium is characterized by an estrogenic dominant phenotype, to which a progesterone-resistant phenotype superimposes in the course of the disease 106, 107. In addition, dysregulation of several signaling pathways, such as EGF, ERK/MAPK, PI3/AKT, and altered gene expression (KRAS, FOS, NODAL) were discovered in the eutopic endometrium.106 In the same way, Gashaw et al. demonstrated an increase in an angiogenic protein, Cyr61, in the eutopic endometrium. 108 Recently, in a new study in baboons it has been shown that a down-regulation of miR-451 leading to an increased expression of YWHAZ could be responsible for the increased cell proliferation and decreased apoptotic response associated with endometriosis. 109
Numerous studies based on baboon model focused on the relationship between the establishment and progression of endometriosis and the immunological status of the animal. Immunosuppression had no effect on disease incidence and extent 110, but selective immunomodulation was proven to be effective: PPAR-γ agonists reduce induced endometriosis 111, while TNF-α antagonists are active on red peritoneal lesions leading to reduction of spontaneous and induced endometriosis112; in particular, TNF-α-binding protein 1 was shown to be able to prevent development of endometriotic lesions.113 Furthermore, an immunological alteration appears to be present also systemically, since induction of endometriosis causes a rapid decrease in Treg cell population, both peripherally and in endometrium, contributing to infertility; also, an increase in Treg cells can be seen in ectopic tissues, thus contributing to the immune tolerance towards the invading lesion.114
NHPs and in particular baboons offer a suitable preclinical model to study candidate therapies for endometriosis. In order to test for preventive strategies a model of artificially induced endometriosis is required; this has been exploited by D’Hooghe when testing TNF-α antagonists, as previously mentioned. 113 Spontaneous and induced endometriosis models can be used to assess therapeutical effects of a drug, since comparisons between before vs after treatment, and treatment vs positive and negative controls can be performed. Examples of this kind of study are the previously illustrated work with PPAR-γ agonists by Lebovic et al.111 and with TNF-α antagonists by Falconer et al.112; recently a JNK inhibitor has also been tested in baboons for pharmacokinetics and efficacy endpoints: the JNK inhibitor combined with medroxyprogesterone acetate was effective in reducing lesion sizes without any significant effect on cycle and endocrine side effects. 115
Among the endocrine agents, GnRH antagonists have been tested in baboons, showing to be effective in regression of endometriotic lesions, but carrying disruption of menstrual cycle 113. Langoi et al., exploiting the knowledge that aromatase inhibitor combined with a progestin was effective in lesion reduction in humans and in rodents, decided to test letrozole (an aromatase inhibitor) alone in baboons obtaining a decrease in lesion volume and in aromatase mRNA levels in lesions, but an increase in ovarian volume.103
Primate models for endometriosis are certainly the most closely related to human physiology and anatomy, allowing us to investigate several aspects of the disease, from the establishment and progression of endometriosis to immunological features and drugs efficacy and toxicity. However, ethical issues and high costs associated to development and maintenance of this model pose some important limitations on its use.
CONCLUSIONS
In vivo and in vitro models of endometriosis represent indispensable tools for the study of pathogenesis and pathophysiology of this very common disease. Since no model replicates all aspects of endometriosis, when choosing which system is to be employed for a targeted research advantages and limitations of each model should be considered.
In vitro models offer a proper experimental system to investigate the signaling pathways at the molecular level, at the base of pathogenesis of endometriosis. Whenever possible, primary cells from tissue samples should be chosen over cell lines.
Rodents are the most common animal models used, due to their low cost, short generation time, and the possibility to genetically manipulate them easily. However they do not reproduce several aspects of endometriosis, which are instead mostly replicated in baboon models. These non-human primates offer the most closely related setting to human physiology and anatomy, but pose important limitations due to their high costs and ethical issues.
Alternatively, CAM models offer a very low cost and simple system, but close to in vivo conditions; It is suitable for studying treatment feasibility for endometriosis therapies, providing thus an alternative model to avoid animal suffering and sacrifice before further in vivo evaluation.