Peritoneal endometriosis and "endometriotic" nodules of the rectovaginal septum are two different entities*

FERTILITY AND STERILITY Vol. 66, No.3, September 1996 Copyright 1996 American Society for Reproductive Medicine Printed on acid-free paper in U. S. A. Peritoneal endometriosis and "endometriotic" nodules of the rectovaginal septum are two different entities* Jacques Donnez, M.D., Ph.D.t Michelle Nisolle, M.D. Pierre Smoes, M.D. Nadine Gillet, M.D. Sandrine Beguin, M.D. Franroise Casanas-Roux, B.S. Department of Gynecology, Universite Catholique de Louvain, Cliniques Universitaires St. Luc, Bruxelles, Belgium Objective: To compare histologically and stereologically the endometriotic nodule of the rectovaginal septum to peritoneal endometriosis. Design: Morphometric investigation, cytokeratin and vimentin content, and steroid receptor evaluation were performed on endometriotic tissue from the peritoneum (n = 52) and rectovaginal nodules (n = 68). Setting: An academic teaching hospital. Patients: Biopsies were taken from 120 patients undergoing a laparoscopy for infertility and! or pelvic pain (52 from typical black peritoneal endometriotic implants and 68 from endometriotic nodule of the rectovaginal septum). None of the patients were treated. Results: Mitotic activity was found to be significantly different in peritoneal and rectovaginal endometriosis. The evaluation suggested that the stroma is not mandatory for the invasion of glandular epithelium in the rectovaginal nodule, which is, like a adenomyoma, a circumscribed nodular aggregate of smooth muscle and glandular elements. Cytokeratin and vimentin content as well as the estrogen receptor (ER) and Preceptor (PR) content were significantly lower in both types oflesion when compared with eutopic endometrium. But vimentin immunoreactivity in epithelium, as well as the ER and PR content, were significantly lower in nodules when compared with black peritoneal. Conclusion: It is suggested that the rectovaginal endometriotic nodule is a different disease from peritoneal endometriosis and must be called rectovaginal adenomyosis or rectovaginal adenomyoma. Its histopathogenesis probably is not related to the implantation of regurgitated endometrial cells but to the metaplasia of Mullerian rests. Fertil Steril 1996;66:362-8 Key Words: Endometriosis, rectovaginal, peritoneum, adenomyoma Typical and subtle endometriotic are characterized histologically by both epithelium and stroma of the endometrial type (1-6). The threedimensional evaluation of peritoneal endometriosis recently demonstrated two different types of peritoneal endometriotic, according to the degree of ramification (7). The stereometric evaluation demonstrated a higher vascularization and mitotic activity in red, which have a polypoid or vesicular Received November 7,1995; revised and accepted April 3, 1996. * Supported by Fonds de la Recherche Scientifique, grant no. 3.4587.90, Brussels, Belgium and a grant from Ipsen Biotech, Paris, France. t Reprint requests: Jacques Donnez, M.D., Ph.D., Department of Gynecology, Universite Catholique de Louvain, Cliniques Universitaires St. Luc, Avenue Hippocrate 10, B-1 Brussels, Belgium (FAX: 32-2-764-95-07). aspect, and we suggested that such were the first stage of implantation, later becoming an enclosed implant (black or bluish appearance) (8). The endometriotic nodule of the rectovaginal septum has been described by Koninckx and Martin (9) as the consequence of deep-infiltrating endometriosis. According to the depth of infiltration, the rectovaginal septum endometriotic nodule was considered as the deepest form of endometriosis. But, on the other hand, some authors (6, 10, 11) have suggested that the rectovaginal endometriotic nodule is an adenomyotic nodule. The aim of this study is to compare histologically and morphometrically the rectovaginal septum endometriotic nodule to peritoneal endometriosis, using stereographic computer technology (7, 11-14) in order to differentiate these two entities. 362 Donnez et ai. Peritoneal and rectovaginal endometriosis Fertility and Sterility

MATERIALS AND METHODS In a first series of 52 women undergoing laparoscopy for infertility, peritoneal biopsies of 3 to 5 mm in size were taken from typical black endometriotic implants (n = 52) with a biopsy punch forceps (26,175 DH; Storz, Tuttlingen, Germany). Twentythree were biopsied during the follicular phase and 29 were biopsied during the luteal phase. In a second series of 68 women complaining of pelvic pain and/or infertility, an endometriotic nodule of the rectovaginal septum was diagnosed by palpation. In this second series, 33 were removed during the follicular phase and 35 were removed during the luteal phase. A laparoscopy was carried out and the nodule removed using the surgical technique described previously (11). The rectovaginal endometriotic nodule was defined in our series as a large and deep nodule (>2 cm in size) whose largest area was under the peritoneal surface. The visible on the peritoneal surface through the laparoscope are often minimal. Morphometric Study Peritoneal endometriotic biopsies and rectovaginal septum nodules were fixed in formaldehyde and embedded in paraffin. Six-micrometer serial sections were stained with Gomori's trichrome and examined on a blind basis with an Orthoplan microscope (Leitz, Wetzlar, Germany). In all cases, the epithelial height and the mitotic index were calculated, as previously described (12). Immunocytochemical Study The quantitative evaluation of cytokeratin and vimentin content as well as the immunostaining quantitative analysis of steroid receptor content were performed by an image analysis program set on a Vidas 21 computer (Kontron Bildanalyse GmBH, Eching, Germany). Both programs were set up to allow the interactive selection of the objects and/or structures for subsequent automatic measurement. All samples were analyzed field by field using an Axioskop light microscope (Zeiss, Oberkochen, Germany) through a CCD 72 E camera (Dage-MTI, Michigan City, IN). The image features were displayed on a Red Green Blue monitor and stored for processing by the image analysis program. Before image analysis started, the camera had to be adjusted interactively to define the amplification factor and the reference voltage in order to make possible the result comparison. Data management and evaluation were checked according to specific search criteria on the Videoplan (Kontron Bildanalyse GmBH, Eching, Germany) and dis- Vol. 66, No.3, September 1996 played on a Video Graphic Array monitor and printed. Cytokeratin and Vimentin Content Six-micrometer sections were stained by immunoperoxidase complex according to the procedures described previously (13) and recommended by Becton Dickinson (Becton Dickinson Immunocytometry Systems, San Jose, CA) for the detection of cytokeratin peptides 8 and 18 and by Dakopatts (Dakopatts, Glostrup, Denmark) for the detection of vimentin. The quantitative evaluation of cytokeratin and vimentin content was carried out by the image analysis program previously described (13). At least 10 fields of each biopsy specimen were analyzed field by field using the 40x objective of the Axioskope light microscope. The image analysis program allowed the interactive selection of the positive and/ or negative cellular surface areas at the glandular epithelium and stroma levels. The percentage of cytokeratin and vimentin surface area was calculated in the epithelium cytoplasm, taking into account the negative- and positive-staining cells. Estrogen and PReceptors Estrogen receptor and PR content within the stroma and glandular epithelium of the endometriotic was determined by immunoperoxidase techniques using the peroxidase-anti peroxidase complex (14). Ten-micrometer cryostat sections were stained according to the immunocytochemical assay kit procedure recommended by Abbott Laboratories (Diagnostic Division, North Chicago, IL) for the detection ofer (ER-ICA) and PR (PR-ICA). A negative control was prepared in each case. The quantitative evaluation of steroid receptor content was carried out by the image analysis program described previously (14). All sections were examined and analyzed field by field using the x objective of the Axioskop light microscope. In each case, more or less positive and negative nuclei of the stroma and glandular epithelium respectively were selected on a blind basis and evaluated. The only selection criterion was the absence of superimposition of the nuclei in order to obtain the real optical density. All current records concerning the cytokeratin, vimentin, ER, and PR content were appended in a database at the end of the file and compared with the values obtained in eutopic endometrium and reported previously (13, 14). The distribution of specific staining was evaluated according to an optical density scale using the values of all the positive nuclei of a given receptor; the quantitative H-score (QH-score) was calculated as follows: QH-score: L Pi, where i is the optical density degree: from 0 (neg- Donnez et at. Peritoneal and recto vaginal endometriosis 363

Figure 1 Rectovaginal adenomyoma (Gomori's trichrome). Scanty endometrial-type stroma and glandular epithelium are disseminated in muscular tissue (x60). ative nuclei) to 3 (high optical density) and P is the percentage of stained nuclei for each given i (from 0% to %). This method is a modification of the semiquantitative H-score analysis. All data are reported as means ± SEM. The t-test, the X 2 test, and the one-way analysis of variance (Fisher's test) were used for statistical analysis. RESULTS Biopsies taken from typical puckered black or bluish peritoneal showed the presence of endometrial elements (glands and stroma) in all cases (%). During the luteal phase, the ectopic glandular epithelium was never found to be in phase with the eutopic endometrium. In some cases, subnuclear vacuoles were observed, but in <20% of the cells. More than 80% of glandular cells demonstrated pseudostratification and the stroma rarely demonstrated any typical progestational changes, such as marked stromal edema, decidualization, or infiltration by lymphocytes. In all cases of recto vaginal nodules, endometrial elements (glandular epithelium and stroma) were also observed. Histologically, scanty endometrial-type stroma and glandular epithelium are disseminated in muscular tissue (Fig. 1). Cellular differentiation in phase with the eutopic endometrium was never observed. Infiltration of surrounding fibromuscular tissue by endometrial glands with some signs of hyperplasia can be found. Very often, endometriotic glands and stroma were discovered by serial section up to the vaginal mucosa, which was sometimes replaced by endometrial epithelium. In some cases (Fig. 2), Mullerian rests and endometriotic were visible clearly beneath the vaginal mucosa. It is obvious (Fig. 3) that the invasion process of the smooth muscle by glandular epithelium did not require the presence of 364 Donnez et al. Peritoneal and recto vaginal endometriosis Figure 2 Rectovaginal adenomyoma (Gomori's trichrome). Mullerian remnant (t) presence of a layer of cuboidal epithelium and endometriotic lesion (it) consisting of glandular epithelium and stroma beneath the vaginal mucosa (a, x 25; b, x 60). stroma. Indeed, in all cases, the glandular epithelium was observed in some areas without any surrounding stroma, very deep in the muscle. Areas of oviduct-like epithelium with ciliated cells were demonstrated in 63% of black and 71% Figure 3 Rectovaginal adenomyoma (Gomori's trichrome). Invasion of the smooth muscle by only the glandular epithelium without stroma (x240). Fertility and Sterility

Table 1 Morphologic Characteristics of Peritoneal and Rectovaginal Nodule Endometriosis Throughout the Menstrual Cycle Peritoneal black lesion Rectovaginal nodule lesion Proliferative phase (n = 23) Secretory phase (n = 29) Proliferative phase (n = 33) Secretory phase (n = 35) Mitotic index (%) Epithelial height (!-lm) 1.01 :+: 1.39 15.6 :+: 3.4 0.38 :+: 0.62t 15.0 :+: 3.2 0.71 :+: 0.94 17.4 :+: 5.6 0.09 :+: 0.25t 16.2 :+: 4.3 * Values are means:+: SEM. t Significantly different from the proliferative phase (P < 0.05). of nodules. The mitotic activity was calculated in glandular epithelium and its value was 0.7% in the nontreated black and 0.3% in the nodules, which is significantly lower (P < 0.05). When compared with the mitotic index observed in peritoneal endometriosis according to the phase of the cycle (Table 1), it was found to be significantly lower (P < 0.05) in the glandular epithelium of nodules. In each group, a significant (P < 0.05) decrease in the mitotic activity was observed during the luteal phase when compared with follicular phase. The epithelial height was similar in both groups. Cytokeratin and Vimentin Staining During the follicular phase, the cytokeratin H score in the glandular epithelium (Fig. 4) was significantly (P < 0.01) lower in nodules when compared with eutopic endometrium and black. But the pattern throughout the cycle was similar. No cytokeratin staining was ever observed in the stromal cells. During the follicular and luteal phases, the vimentin H-score (Fig. 5) in the glandular epithelium was significantly lower in nodules than in the eutopic endometrium or black (P I)... 300 o :;: 1 ± endometrium peritoneal rectovaginal black nodul e Figure 4 Cytokeratin. Mean:+: SEM H-score of the epithelial cells in eutopic endometrium and in peritoneal and rectovaginal nodule endometriosis throughout the cycle: during the proliferative phase (ill!) and the secretory phase (_). Vol. 66, No.3, September 1996 < 0.05). It is important to note that vimentin was never expressed in the glandular epithelium and stroma of nodules during the follicular phase. The vimentin H-score in the stroma was similar during the secretory phase in eutopic endometrium, in black, and in nodules. Estrogen Receptor and PR Content The ER content of the glandular epithelium of black and nodules was similar (Fig. 6). The pattern throughout the cycle was similar to that observed in eutopic endometrium. Indeed, there is a significant decrease in ER during the secretory phase in the three tissues. In the nodules, the ER content of the stroma was significantly (P < 0.05) lower during the follicular phase and significantly (P < 0.05) higher during the luteal phase when compared with eutopic endometrium and to black endometriotic. A significant (P < 0.05) increase in stromal ER content was observed throughout the cycle, although a significant (P < 0.05) decrease was observed in eutopic endometrium and in peritoneal. When compared with peritoneal, the PR content of the glandular epithelium and stroma of nodules was significantly (P < 0.05) lower than that of eutopic endometrium and black during the follicular phase (Fig. 7). The pattern throughout the cycle was also different. Indeed, although a significant decrease (P < 0.05) was observed in the eutopic endometrium and in peritoneal, a significant increase was observed in the nodules throughout the cycle. DISCUSSION It generally is believed that endometriosis is caused by the implantation of retrograde menstrual endometrial cells or by metaplasia. In the pelvis, three different forms of endometriosis must be considered: peritoneal, ovarian, and rectovaginal septum (6). In the present study, two different types of endometriosis (peritoneal endometriosis and rectovaginal endometriosis) were analyzed morphologically in Donnez et al. Peritoneal and rectouaginal endometriosis 365

300 GLANDULAR EPITHELIUM STROMA endometrium peritoneal rectovaginal endometrium peritoneal rectovaginal black nodul" black nodul e 1 o-'--_...l.i.",,",,", Figure 5 Vimentin. Mean :!: SEM H-score of the epithelial and stromal cells in eutopic endometrium and in peritoneal and rectovaginal nodule endometriosis throughout the cycle: during the proliferative phase CI!l!) and the secretory phase C.). order to determine their individual characteristics. The so-called typical or black lesion is considered as an enclosed implant surrounded by fibrosis (8). The rectovaginal nodular mass, the first description of which was given by Sampson (15) in 1922, often involves the posterior vaginal fornix (11, 16) and resembles an adenomyoma (11). In the uterus corpus, an adenomyoma is a circumscribed nodular aggregate of smooth muscle and endometrial glands without any secretory changes under the influence of P (17). The endometrial glands present in a uterine adenomyoma may not respond to physiological levels ofp, and secretory changes frequently are absent or incomplete during the second half of the cycle. Similar histologic observations are made at the level of the "endometriotic" rectovaginal nodule. In our study, the ectopic endometrium of black rarely demonstrated typical progestational changes, although some authors (18, 19) reported that 70% of foci with a cycle pattern underwent changes that were considered synchronous (::!::3 days) with the eutopic endometrium. This probably is due to the fact that, in some cells (but <20%), some basal vacuoles can be observed. Nevertheless, this sign cannot be interpreted as a typical secretory change because it has to be revealed in >% of the epithelial cells to be considered as a secretory change and because basal vacuoles can also be observed in anovulatory cycles or in the absence of P (20). In the nodules, the glandular epithelium never demonstrated any progestational changes. All cases were "out of phase" when compared with the eutopic endometrium. When compared with eutopic endometrium (13), the presence of a lower expression of cytokeratin in the epithelium of peritoneal endometriosis and rectovaginal endometriosis could be interpreted as a lower degree of differentiation or as a delay in the differentiation. Indeed, in hyperplastic endometrial (21), the loss of vim en tin expression and the absence of secretory changes give rise to suspicion regarding their benign process. It has also been interpreted (13) as a low degree of cell differentiation. The loss of vimentin in eutopic endometrium during the secretory phase could be responsible for the cell disorganization associated with an ischemic process 300 GLANDULAR EPITHELIUM STROMA endometr I um per I toneal r ectovagl nal endometr i urn per I toneal r ectov agi nal black nodul e black nodul e ~ 0 1 u If' :I: 0 Figure 6 Estrogen receptors. Mean:!: SEM quantitative H-score in nuclei of glandular epithelium and stroma of eutopic endometrium and in peritoneal and rectovaginal nodule endometriosis throughout the cycle: during the proliferative phase CI!l!) and the secretory phase C.). 366 Donnez et al. Peritoneal and rectouaginal endometriosis Fertility and Sterility

GLANDULAR EPITHELIUM STROMA endometrium peritoneal rectovaglnal endom etr I um per I toneal r ectovagl nal black nodul e black nodul e 300 Figure 7 Progesterone receptors. Mean::!: SEM quantitative H-score in nuclei of glandular epithelium and stroma of eutopic endometrium and in peritoneal and rectovaginal nodule endometriosis throughout the cycle:during the proliferative phase C~) and the secretory phase C-). ~ 0 u.. ± 1 0 and could playa role in menstruation and implantation. It may also reflect a change in glandular function (13). A similar decrease throughout the cycle was observed in eutopic endometrium as well as in ectopic peritoneal endometrium. The very low vimentin immunoreactivity and the absence of any decrease throughout the cycle in the glandular epithelium of nodules account for the low degree of differentiation and their unresponsiveness to endogenic hormonal variations. Nevertheless, the coexpression of vimentin and cytokeratin indicates a close relationship with their mesodermal Mullerian origin. The significantly lower vimentin expression in the epithelium of nodules (when compared with black and eutopic endometrium) has led us to suggest, as we did for red (13), that the low expression of vimentin can be related to a trend toward hyperplasia. Indeed, the invasion of glandular cells without stroma in the smooth muscle of nodules was very often clearly seen. In black, the content of ER was lower than in eutopic endometrium but the variation throughout the cycle was similar (14,22). The variations of ER and PR in the nodules throughout the cycle suggest that they are probably not regulated by steroids. In our study, the very low glandular epithelium and stroma ER content during the follicular phase can explain the absence of any secretory change in the nodule glandular epithelium. In a recent study (23), it was suggested that a low ER level was the key factor in explaining the out-of-phase endometrium despite normal P levels, but the reduction in PR also could cause resistance to P action and result in inadequate secretory transformation. The absence of response to P levels suggests that the different regulatory mechanisms of endometriotic steroid receptors result in deficient endocrine dependency (14,24) or that the receptors are present but biologically inactive. Vol. 66, No.3, September 1996 The differences between peritoneal endometriosis and endometriosis of the rectovaginal septum observed in our study and the very similar histologic descriptions of endometriotic nodules and uterine adenomyoma have led us to suggest that the socalled endometriotic nodule of the rectovaginal septum is not the consequence of deep-infiltrating endometriosis. It is, in fact, an adenomyoma or an adenomyotic nodule consisting essentially of smooth muscle with active glandular epithelium and scanty stroma. This nodule can develop from Mullerian rests. Indeed, Mullerian rests were found in the rectovaginal tissue. Moreover, the origin of adenomyotic nodule by a process of metaplasia from Mullerian rests is consistent with the putative MUllerian potential of the peritoneum, which has been referred to as the secondary Mullerian system (25). Mullerian remnants are usually not apparent, although they may be recognizable as small cysts measuring several millimeters in diameter. They are most commonly lined with a single layer of cuboidal cells or columnar endosalpingiotic epithelium with a mixture of ciliated, secretory, and intercalated cell types (25). Because of the poor differentiation (lower vimentin and cytokeratin content) and the hormonal independence of these (lower vimentin content, lower ER and PR content), their growth in the rectovaginal septum tissue is not regulated by circulating steroids. The low mitotic activity and the low steroid receptor content observed in this pathology can account for the relatively slow evolution of the adenomyoma. REFERENCES 1. Jansen RPS, Russel P. Non-pigmented endometriosis: clinical laparoscopic and pathologic definition. Am J Obstet Gynecol 1986;155:1154-8. 2. Redwine DB. The distribution of endometriosis in the pelvis by age groups and fertility. Fertil Steril 1987;47:173-5. Donnez et ai. Peritoneal and recto vaginal endometriosis 367

3. Stripling MC, Martin DC, Chatman DL, Vander Zwaag R, Poston WM. Subtle appearances of pelvic endometriosis. Fertil Steril 1988;49:427-31. 4. Martin DC, Hubert GD, Vander Zwaag R, EI-Zeky FA. Laparoscopic appearances of peritoneal endometriosis. Fertil Steril 1989;51:63-7. 5. Nisolle M, Paindaveine B, Bourdon A, Berliere M, Casanas Roux F, Donnez J. Histologic study of peritoneal endometriosis in infertile women. Fertil Steril 1990;53:984-8. 6. Donnez J, Nisolle M, Casanas-Roux F, Bassil S, AnafV. Rectovaginal septum, endometriosis or adenomyosis: laparoscopic management in a series of 231 patients. Hum Reprod 1995;10:630-5. 7. Donnez J, Nisolle M, Casanas-Roux F. Three-dimensional architectures of peritoneal endometriosis. Fertil Steril 1992; 57:980-3. 8. Nisolle M, Casanas-Roux F, Anaf V, Mine JM, Donnez J. Morphometric study of the stromal vascularization in peritoneal endometriosis. Fertil Steril 1993;59:681-4. 9. Koninckx PR, Martin DC. Deep endometriosis: a consequence of infiltration or retraction or possibly adenomyosis externa? Fertil Steril 1992;58:924-8. 10. Brosens IA. Is mild endometriosis a progressive disease? Hum Reprod 1995;9:2209-11. 11. Donnez J, Nisolle M, Casanas-Roux F, Brion P, Da Costa N. Stereometric evaluation of peritoneal endometriosis and endometriotic nodules of the recto-vaginal septum. Hum Reprod 1995;11:224-8. 12. Donnez J, Casanas-Roux F, Caprasse J, Ferin J, Thomas K. Cyclic changes in ciliation, cell height, and mitotic activity in human tubal epithelium during reproductive life. Fertil Steril 1985;43:554-9. 13. Nisolle M, Casanas-Roux F, Donnez J. Coexpression of cytokeratin and vimentin in eutopic endometrium and endometriosis throughout the menstrual cycle: evaluation by a computerized method. Fertil Steril 1995;64:69-75. 14. Nisolle M, Casanas-Roux F, Wyns C, de Menten Y, Mathieu P-E, Donnez J. Immunohistochemical analysis of estrogen and progesterone receptors in endometrium and peritoneal endometriosis: a new quantitative method. Fertil Steril 1994;62:751-9. 15. Sampson JA. Intestinal adenomas of endometrial type. Arch Surg 1922;5:217-80. 16. Reich H, McGlynn F, Salvat J. Laparoscopic treatment of cul-de-sac obliteration secondary to rectocervical deep fibrotic endometriosis. Reprod Med 1991;36:516-22. 17. Zaloudek C, Norris HJ. Mesenchymal tumors of the uterus. In Kurman R, editor. Blaustein's pathology of the female genital tract. New York: Springer-Verlag, 1987:373-480. 18. Roddick JW, Conkey C, Jacobs EJ. The hormonal response of endometrium in endometriotic implants and its relationship to symptomatology. Am J Obstet GynecoI1960;79:1173-7. 19. Bergqvist A, Ljungberg 0, Myhre E. Human endometrium and endometriotic tissue obtained simultaneously: a comparative histological study. Int J Gynecol PathoI1984;3:135-45. 20. Mazur T, Kurman RJ. Normal endometrium and infertility evaluation. In: Mazur T, Kurman RJ, editors. Diagnosis of endometrial biopsies and curettings. A practical approach. New York: Springer-Verlag, 1995:7-32. 21. Nakopoulou L, Minaretzia D, Tsionou C, Mastrominas M. Value of immunohistochemical demonstration of several epithelial markers in hyperplasia and neoplastic endometrium. GynecoIOncoI1990;37:346-53. 22. Bouchard Ph, Marraoui J, Massai MR, Medalie DA, De Ziegler D, Perrot-Applanat M, et al. Immunocytochemical 10- calization of oestradiol and progesterone receptors in human endometrium: a tool to assess endometrial maturation. Baillieres Clin Obstet GynaecoI1991;5:107-15. 23. Hirama Y, Ochiai K. Estrogen and progesterone receptors of the out-of-phase endometrium in female infertile patients. Fertil Steril 1995;63:984-8. 24. Metzger DA. Cyclic changes in endometriosis implants. In: Brosens la, Donnez J, editors. The current status of endometriosis. Research and management. Carnforth, United Kingdom: Parthenon Publishing, 1993:89-108. 25. Clement PE. Endometriosis, of the secondary Mullerian system, and pelvic mesothelial proliferations. In: Kurman RS, editor. Blausteins's pathology of the female genital tract. New York: Springer-Verlag, 1987:516-59. 368 Donnez et al. Peritoneal and recto vaginal endometriosis Fertility and Sterility