The Pathogenesis of Peritoneal Endometriosis

The pathogenesis of peritoneal endometriosis is still subject to debate, although the condition has been known since 1860. The implantation theory explains the pathogenesis by deposition and subsequent growth of retrogradely shed viable endometrial cells. The scientific data to support this theory will be discussed. Transformation of mesothelium to endometrium-like tissue under the influence of products of regurgitated endometrium (induction) is a plausible alternative. In both models, cell adhesion molecules may play an important functional role.

In peritoneal endometriosis, a delicate equilibrium seems to exist between the 'attacking forces' (retrograde menstruation) and the defence mechanisms. On the one hand, the amount and the nature of the regurgitated menstrual debris is important in the development of the disease. On the other hand, the active intra-abdominal environment probably reduces regurgitated endometrial tissue into single cells deprived of their functional cell adhesion properties. If this active milieu is impaired, or if the number of regurgitated cells is too large, the surviving cells can adhere to exposed extracellular matrix (ECM) of a damaged peritoneal lining. An intact peritoneal lining may be an important additional line of defence. If all defence mechanisms fail, endometriosis will develop.

In endometriosis, at least three different forms must be defined (1). These three forms are: peritoneal, ovarian and recto-vaginal. The first histological description of a lesion consistent with endometriosis was given by Von Rokitansky (2) in 1860. It was Cullen (3,4) who suggested in 1896 that endometriomas, or adenomyomas as he called these lesions, resembled the mucosa of the uterus. The pathogenesis of this enigmatic disease is still poorly understood and remains controversial.

The theories of the pathogenesis of endometriosis, in particular of peritoneal endometriosis, can be divided into three main concepts. The oldest concept is that endometriosis develops at the site where it is identified (in situ development). This may occur from the remnants of the Wolffian or Müllerian ducts, or alternatively, from metaplasia of the peritoneal or ovarian tissue (5,6).

A second concept is based on the assumption that endometriosis results from differentiation of mesenchymal cells, activated by substances released by degenerating endometrium that arrives in the abdominal cavity (the induction theory) (7,8).

A third concept is based on the transplantation and subsequent implantation of endometrial tissue (9,10). This would imply transport of viable endometrial cells during menstruation through the Fallopian tubes into the abdominal cavity, implantation of these cells onto the peritoneum and the development of these cells into endometriosis (the transplantation or implantation theory).

In-situ development

The theories concerning development of endometriosis from either the Wolffian or Müllerian tissue have been met with considerable opposition over the years and have been largely abandoned. The finding of endometriosis at sites such as the serosal surface of the colon and the small intestines make a pure embryonic origin too restrictive. The theory of coelomic metaplasia still has some support, because it can explain the origin of endometriosis, regardless of the sites or the conditions of its occurrence (11). It does not, however, explain why endometriosis occurs exclusively in women, and typically during the reproductive years, or why endometriosis mainly affects the pelvic organs, or why it only occurs in women with functional endometrium. Proof of this theory is lacking, either experimentally or clinically. There is only some circumstantial evidence in case reports of endometriosis occurring in young girls, even before menstruation, and in reports of endometriosis at rare localisations, such as pleura or diaphragma.

Induction

In 1955, Levander and Normann (7) proposed the induction theory. This theory is based on the assumption that specific substances, which are released by degenerating endometrium, induce the development of endometriosis from omnipotent stem cells present in connective tissue. The suggestion was made, based on experiments in rabbits, that cell-free endometrial products were capable of inducing endometrial metaplasia (8). However, the changes do not meet the criteria for endometriosis, since endometrial stroma was absent. Lauchlan (6) introduced the term "secondary Müllerian system," which refers to all Müllerian type epithelium located outside the original Müllerian ducts. This layer of cells particularly those on the surface of the ovary could then, through metaplasia, develop into four cell types, one of which is endometrial. This could occur before or after invagination. The fact that both serous and mucinous epithelium can be found in or around endometriotic lesions appears to argue in favour of this concept (6).

Implantation

The conditions that have to be met to support the implantation theory are threefold, firstly, retrograde menstruation has to occur; secondly, retrograde menstruation should contain viable endometrial cells; and finally, adhesion to the peritoneum has to occur with subsequent implantation and proliferation. The implantation theory was neglected for a long time, because menstrual effluent was considered to contain only non-viable endometrial tissue and retrograde menstruation was thought to be a rare phenomenon, although the theoretical concept was recognised by some authors (12,13,14,15). Retrograde menstruation and peritoneal adhesion of endometrial tissue is an essential element in the pathogenesis of endometriosis according to Sampson's theory (9,10,16). Menstruation is almost unique to man and a few other primates. Only recently menstruation and menstrual shedding have been associated with disorganisation of the site-specific distribution of desmoplakin I/II, E-cadherin, and a- and § catenins (17). Menstrual effluent is composed of blood elements, endometrial cells and extracellular fluid. It also contains viable endometrial cells as shown by the classical study of Keettel and Stein (18) in 1951. Cron and Gey (19) had tried earlier to prove the viability of cast-off menstrual endometrium in culture, but they had used a curette to remove the endometrium. Geist (20) had suggested that desquamation of endometrium was not due to local necrosis, because menstrual effluent contained viable endometrial cells that remained alive for at least one hour. Ridley and Edwards (21) demonstrated in 1958 that endometrial cells obtained from the menstrual effluent could be implanted into the abdominal wall fascia. However, only in one of eight cases did they succeed in finding endometriosis developing at the site of injection.

Retrograde menstruation

At laparotomy, Watkins (22), like Sampson before him, observed that blood may drip from one or both fallopian tubes during menstruation. The presence of blood in peritoneal fluid (PF) by visual documentation has since been reported by several other authors (23,24). Passage and transfer of endometrial fragments into the peritoneal cavity through the fallopian tubes also has been shown (25). Peritoneal fluid has been found to contain endometrial tissue in up to 59% of patients, with or without endometriosis, undergoing laparoscopy at various stages of the menstrual cycle (26-30).  Kruitwagen and co-workers (28) have demonstrated viable endometrial cells in PF. These authors succeeded in culturing cells in vitro, and their data strongly suggest an endometrial origin of epithelial cells in PF. Furthermore, the anatomic distribution of endometriosis correlates very well with principles of transplant biology (31). It has been suggested that the presence of blood in the pouch of Douglas at laparoscopy is inadequate proof of retrograde menstruation, as there is only a weak correlation between blood staining of PF and the presence of endometrial cells (32). Demonstration of the presence of endometrial cells in PF may be an objective way to assess retrograde menstruation, although endometrial glands have also been reported in the peritoneal cavity after dilatation and curettage or after uterotubal irrigation (25,26,33,34). In most studies, Papanicolaou staining was used to demonstrate the presence of endometrial cells in PF. (29,30,32). This has the disadvantage that only rather large clusters of cells, resembling endometrial glandular and stromal tissue, are detected but single cells are not recognised. Van der Linden and co-workers (35) demonstrated the presence of endometrial cells in PF using immunohistochemistry. They compared the immunohistochemical staining properties of these fragments to those of cells present in endometrium, menstrual effluent, peritoneum and endometriotic lesions. The staining characteristics, based on the application of monoclonal antibodies against various epithelial markers in cells from menstrual effluent, endometrium, peritoneal fluid, and endometriotic lesions were remarkably similar.  Their study showed that in women with patent tubes peritoneal fluid contains single epithelial cells, rather than endometrial tissue fragments. It is plausible that endometrial epithelial cells shed by the endometrium, are modified in the peritoneal cavity prior to progressing to an endometriotic lesion.  Although specific epithelial markers were demonstrated in cells obtained from menstrual effluent, endometrium, peritoneal fluid, as well as endometriotic lesions, this is not conclusive evidence that endometriosis originates from endometrium by retrograde shedding of viable tissue fragments.

Adhesion

If retrograde menstruation is important in the pathogenesis of endometriosis then, at some point, endometrial tissue, either glands or stroma, should adhere to the peritoneum. Unfortunately, in vivo studies on the initial contact between single or a group of endometrial cells and the peritoneal lining are still lacking. In theory, either the glandular epithelial cells, stromal cells or both cell types may be directly involved in the contact with the mesothelium of the peritoneum. Alternatively, both cell types may be mutually influencing each other to allow this initial contact. Or, there may be direct contact between endometrial cells and the submesothelial.  Both implantation of viable endometrial tissue fragments and induction of coelomic metaplasia by these fragments will require adhesion of endometrial cells to the peritoneal lining. In studies of van der Linden and co-workers (36,37) members of the integrin and cadherin family, important cell adhesion molecules, have been reported to be expressed in endometriotic lesions and in cells and tissues that are potentially involved in the development of endometriosis. These authors focused their attention on cadherins and integrins. Cadherins are considered the most important cell adhesion molecules involved in cell-cell adhesion and integrins for cell-extracellular matrix interactions. Cadherins belong to a group of calcium-dependent transmembrane glycoproteins (38-42). Cells adhere preferentially to cells which express the same cadherin (homophylic cell-cell interaction) (43). Integrins are a family of cell membrane glycoproteins consisting of an a and a § subunit that mediate cell-cell and, primarily, cell-matrix adhesion (42,44-46). Integrins a2§1, a3§1, a4§1, a5§1, and a6§1 and E-cadherin were demonstrated to be expressed in endometriotic lesions as well as in cells and tissues that are potentially involved in the development of endometriosis (36). Regurgitated cells obtained from peritoneal fluid showed expression of cell adhesion molecules, particularly E-cadherin and some §1-integrins, but to a lesser extent than the cells from the tissues they are supposed to stem from. The expression pattern of cell adhesion molecules suggests that the loss of cell adhesion properties could be involved in the shedding of endometrial tissue during menstruation and the attachment of endometrial tissue fragments to the peritoneum. All cells potentially involved in the pathogenesis of endometriosis express members of the integrin and cadherin families of cell adhesion molecules.  However, the demonstration of cell adhesion molecules in menstrual effluent, endometrium, peritoneal fluid, as well as in endometriotic lesions, is no strict evidence that endometriosis originates from endometrium by retrograde shedding of viable tissue fragments. Co-ordination of the action of various cell adhesion molecules is required for effective cellular adhesion. It is, therefore, not plausible that in the pathogenesis of endometriosis the processes of adhesion of shed endometrial tissue can be explained by the function (presence or absence) of one single cell adhesion molecule.  E- and P-cadherin are presumably functionally involved in the maintenance of epithelial structures in endometrium and endometriosis, both during the proliferative and the secretory phase of the cycle (37,42). Both E- and P-cadherin expression were detected in all cycle phases in endometrial samples and did not vary throughout the menstrual cycle (47). If these adhesion molecules are functionally involved in the cyclic menstrual shedding, the loss of expression is limited to a short period of time. Of the §1 integrins, only a2§1 expression was modulated during the menstrual cycle, as it was only absent in the midluteal phase (47). Since cadherins and §1-integrins could be detected in late luteal phase endometrium, these cell adhesion molecules could be involved in the attachment of endometrial fragments to the peritoneal lining as a result of retrograde menstruation. In an in vitro model using amniotic membranes and isolated components of extracellular matrix (ECM) coated on glass, stripped amniotic membranes proved to be a valid model to study interactions between endometrial cells and ECM (48). One important observation was the loss of cell adhesion molecule expression, paralleled by loss of adhesion to stripped amniotic membranes, of endometrial cells after collagenase digestion.  Apparently collagenase destroys both structure and function of cell adhesion molecules at the cellular surface. This strongly suggests a functional role of one or more of these cell adhesion molecules in endometrial cell-ECM interactions. The results of this study were in accordance with the observations that in peritoneal fluid single cells are more often observed than cell clumps and that expression of cell adhesion molecules on these cells is very weak or absent (35,47). Proteolytic activity in peritoneal fluid may affect structure and function of adhesion molecules in regurgitated endometrial cells. Loss of expression of cell adhesion molecules and subsequent loss of intercellular adhesion may be one of the initial defence mechanisms in the peritoneal cavity that affect endometrial cells in peritoneal fluid. This may have a twofold effect: it prevents their adhesion to the peritoneum, and facilitates at the same time their removal and destruction by peritoneal macrophages. In an in vitro model to investigate the adhesion between endometrial fragments and cells to an ECM covered by an intact epithelium, intact amniotic membranes were used (49). No adhesion of fragments of normal endometrium to intact epithelium was found, whereas these fragments readily adhered to amniotic membranes which were denuded of their epithelium.  Peritoneum and amniotic membrane show a great similarity in structure and in morphological and immunohistochemical features. It was therefore suggested that an intact peritoneal mesothelium prevents adhesion between endometrial cells shed into the peritoneal cavity and the peritoneum (49). Carcinoma cell lines did show adhesion to intact epithelium. This suggests that the adhesive behaviour of endometrial carcinoma cells in the process of metastasis is different from that of normal, shed endometrial fragments. Disruption of the peritoneal lining seems to be a prerequisite for adhesion between endometrial cells and the peritoneal wall. This would be in accordance with the fact that endometrial tissue growing on the peritoneal surface has never been described (50). The findings of these studies support the contention that in endometriosis, in particular in peritoneal endometriosis, a delicate equilibrium exists between attacking forces (retrograde menstruation) and defence mechanisms. On the one hand, the amount and the nature of the regurgitated menstrual debris is important in the development of the disease (51). On the other hand, one of the first lines of defence is the collagenase-like activity of the peritoneal fluid and the active intra-abdominal milieu, characterised by activated macrophages (52). This milieu probably reduces endometrial tissue into single cells, that, in addition, have lost their functional cell adhesion properties. If this active peritoneal fluid is impaired in disposing of the regurgitated cells, or if the number of regurgitated cells is too large, the surviving cells can adhere to exposed ECM of damaged peritoneal lining. An intact peritoneal lining may be an important additional line of defence. If all defence mechanisms fail, endometriosis will develop. Recent concepts on the further development of endometriosis consider minimal endometriosis as a normal condition occurring intermittently in normal women, in contrast to endometriotic disease occurring as deeply infiltrating endometriosis, and cystic ovarian endometriosis (53). Future research should be directed towards finding how the processes involved in the pathogenesis of endometriosis can take place, i.e. by focusing on interfering with the adhesion process, for instance by specific inhibition of cell adhesion molecules at the protein level using antibodies. The role of an intact epithelium in preventing adhesion of endometrial fragments to the mesothelium needs to be elucidated.