Large ovarian endometriomas
J. Donnnez, M. Nisolle, St. Gillerot, M. Smets, S. Bassil, F. Casanas-Roux
The pathogenesis of typical ovarian endometriosis is a source of controversy. The original paper of Sampson on this condition reported that perforation of the so-called chocolate cyst led to spillage of adhesions and the spread of peritoneal endometriosis. The findings of Hughesdon (1957) contradicted Sampson’s (1921, 1927) hypothesis and suggested that adhesions are not the consequence but the cause of endometriomas (Hughesdon, 1957). In 93 % of typical endometriomas, the pseudocyst is formed by an accumulatation of menstrual debris from the shedding and bleeding of active implants located by ovarioscopy (Brosens et al., 1994, 1995) at the site of inversion, resulting in a progressive invagination of the ovarian cortex. Some other authors (Nezhat et al., 1992, 1994) have suggested that large endometriomas may develop as a result of secondary involvement of functional ovarian cysts in the endometriotic process. According to our opinion (Donnez et al., 1993, Nisolle and Donnez,), the haemorrhagic cysts are the consequence of metaplasie of epithelial inclusions in the ovary.
The aim of this chapter is to describe the management and the histological data from a series of 814 patients with endometriomas of more than 3 cm in size.
From January 1982 to March 1994, 814 patients underwent laparoscopy for ovarian endometriomas larger than 3 cm in size. Bilateral endometriomas were found in 639 cases (79 %). The largest endometriomas were > 15 cm (n=8). All peritoneal lesions were noted and the ovarian endometriomas which were often adherent to the ovarian fossa were mobilized. During laparoscopy, the degree of endometriosis was assessed according to the revised AFS classification system (1985) and the ovarian cyst diameter was measured with a probe introduced through a second-puncture incision. After careful examination of the pelvic cavity, the ovarian cyst was incised to a length of 2 to 3 cm, rinsed and flushed out. All the chocolate-colored fluid was aspirated and the internal wall of the cyst (Fig. 1-4) was inspected to ensure the absence of any intracystic polypoid projection. The ovarian endometrial cyst wall was biopsied with a biopsy punch forceps (26-175 DH; Storz; Tuttlingen; Germany) to confirm the diagnosis of endometriosis.
Patients received a SC injection of gosereline (Zoladex®) implant at weeks 0, 4, 8 and 12. The initial pituitary stimulation phase, associated with a rise in estrogens, was curtailed in our study by injecting the implant during the luteal phase.
Twelve weeks after the first-look laparoscopy, a second laparoscopy was performed. The degree of endometriosis was assessed by the same two observers (J.D. and M.N.) according to the revised AFS classification system. The size of the ovarian cyst was measured. Ovarian biopsy specimens were taken from the endometrial cyst wall in 48 cases in order to compare histological data before and after therapy. After biopsy, the internal wall of the ovarian cyst was vaporized with CO2 laser (Donnez et al., 1987, 1991). Since 1990 the Swiftlase® (Sharplan, Laser Industries, Tel Aviv, Israel) laser was used (Fig. 5-7). Patients were followed up for a maximum of 13 years.
Modifications of AFS Scores
The mean laparoscopic scores found before and at the end of treatment are compared in Table 1. The mean score decreased from 43.4 ± 2.1 to 33.2 ± 2.1. The difference was statistically (P < 0.001) significant. Laparoscopic scoring was done separately for implants and adhesions. In the case of implants, the score decreased from 27.1 ± 1.8 to 16.9 ± 1.7. The difference is statistically (P < 0.005) significant. The initial adhesion score did not decrease.
Modifications of Cyst Diameter
In all cases, a residual cyst was present. Complete disappearance of the cyst was never observed. The response of the ovarian disease was evaluated by measuring the modifications of the ovarian cyst diameter (Table 1). After drainage and a 12-week GnRH-a therapy, a significant (P < 0.001) decrease was observed from 47.3 mm ± 6.2 to 21.7 ± 3.8 mm. In 8 cases, the endometrioma was larger than 15 cm (mean 17.5 ± 2.8). The ovarian cortex was evaluated by echography as being less than 2 mm in thickness. Eight weeks after drainage and GnRH agonist, the size was 8.8 ± 4.2 cm.
Aspiration of the remaining chocolate-colored fluid and flushing out were then performed by vaginal puncture under echography. Four weeks later, the diameter was 4.6 ± 2.4 cm. It was obvious that the remaining ovarian cortex surrounding the endometrioma became significantly (p < 0.001) thicker (between 8 mm and 1 cm) when compared to the values obtained before drainage. In all cases, vaporization of the internal lining of the cyst was easily performed at week 12.
Pregnancy rate and recurrence rate
The cumulative pregnancy rate was evaluated from the first postoperative menstruation. Forty-two patients who did not menstruate (because they became pregnant from the first ovulation) were included in the « first cycle ». A pregnancy rate of more than 51 % was achieved. (Fig. 1). The majority of pregnancies occurred during the first 10 months following surgery.
In our series of 814 patients with ovarian endometriomas treated with the combined (GnRH-a laser and laser surgery) therapy, 607 patients were followed up for 2 to 13 years. A recurrence of ovarian endometriomas was suspected by vaginal echography and then confirmed by laparoscopy in 51 cases (8 %). Among these 51 cases, previous laparoscopy had revealed a unilateral endometrioma in 15 (30 %). The recurrence was observed in 8 cases in the heterolateral ovary, and in 7 cases in the previously treated ovary.
Ovarian Cyst Histology
The diagnosis of ovarian endometriosis was confirmed morphologically in all cases except two in which glandular epithelium and stroma were absent and replaced by fibrous tissue.
Serial sections were carried out in a series of 93 large biopsies in order to observe endometrial epithelium which is often very thin and flattened. Biopsies of the endometrial wall often reveal flattened columnar epithelium with a stroma of endometrial type, often surrounded by fibroreactive tissue with haemosiderin-like macrophages. There is no real plane of « cleavage » between the endometrial-type stroma and the ovarian cortex and it is not unusual to find ovocytes in the vicinity of the endometrial stroma (Fig. 8). In some areas, mucosal endometrium can be seen. In other areas, the epithelium is very thin, often detached from the subepithelial layers. The epithelial and stroma lining of an endometriotic cyst frequently becomes attenuated and the former may be reduced to a single layer of cuboïdal cells which is frequently devoid of specific features. In such circumstances, recognition of the cyst as endometriotic may only be possible if a rim of subjacent endometrial stroma persists. Commonly, the cyst lining of endometrial epithelium and stroma is totally lost and replaced by granulation tissue, dense fibrous and numerous pigmented macrophages. In this case, diagnosis of endometriosis can be difficult because similar findings can be seen in a old corpus luteum cyst.
When biopsies were obtained from the site of adhesions (inversion stigma according to Hughesdon’s hypothesis) and from the deepest area of the endometrioma, histology revealed the same data in both areas : areas of fibrosis covered by flattened epithelium surrounded by stroma with haemosiderin-laden macrophages were observed and small foci of full-thickness endometriomal tissue were found with the same incidence close to the inversion stigma as well as in the other parts. Flattened epithelium was in continuum with active glandular epithelium.
Extracystic endometriotic lesions were observed in the endometriotic stroma or deeper in the ovarian cortex in 33 % of cases. These endometriotic lesions were considered as invagination of the ovarian cortex by epithelium surrounded by endometrial-type stroma. The 3-dimensional architecture revealed this process of invagination. In some cases, invagination of the mesothelial cells covering the ovary (Fig. 9) was clearly demonstrated and metaplasia into endometrial tissue could be observed. A continuum between the invaginated mesothelium and the endometriotic tissue was observed in some serial sections.
Since 1992, different proposals (Nezhat et al., 1992; Donnez et al., 1993; Brosens, 1995) have been made in order to improve our knowledge of the physiopathology of endometriomas (chocolate cysts). The term « chocolate cyst » was applied by Sampson (1921, 1927) to describe the endometrial cyst of the ovary. According to Hughesdon (1957), the internal surface of a chocolate cyst is really the external surface of the ovary; the ovarian cortex is identifiable by the presence of primordial follicles. The endometrial cyst may be lined with free endometrial tissue similar histologically and functionally to eutopic endometrium (Brosens et al., 1994). It was thus suggested (Hughesdon, 1957; Brosens, 1995) that the endometrioma is a pseudocyst formed by an accumulation of menstrual debris from endometrial implants adherent to the peritoneal layer. The adhesions are the cause of the endometrioma by shedding of active superficial implants.
We have focused our study on the evaluation of « endometriomas » (larger than 3 cm) in order to analyze the histological data and the clinical implications. Ovarian endometriomas do not respond very well to medical therapy (Buttram et al., 1985; Nisolle et al., 1988; Donnez et al., 1990, 1994; Shaw, 1992). After drainage alone, quick recurrence of the cyst size proved that drainage alone is completely ineffective in the management of large endometriomas (Donnez et al., 1994) although some authors (Fayez and Voghel, 1991) have suggested its efficacy. However, three months after drainage of the endometrioma and GnRH-a therapy (which provokes amenorrhea), a significant reduction (from 47.3 ± 4.2 um to 21.7 ± 3.8) was observed. But, the reformation of chocolate-coloured fluid during an amenorrheic period suggests tha endometrial shedding is not responsible, by itself, for chocolate-colored fluid formation. In our opinion, its origin could be: (1) cyst wall exudation, (2) congested blood vessels of the cyst wall, (3) inflammation around persistent intracystic endometrial foci which are resistant to medical therapy.
The persistence of a chocolate cyst at laparoscopy after GnRH agonist therapy was confirmed by the histologic study. Indeed, in all cases, histologic examination of residual ovarian endometriotic lesions after hormonal therapy revealed glandular epithelium and stroma. In more than 40 % of cases, areas with persistent active endometriosis were observed after GnRH agonist. The precise reason why a number of endometrial cells of endometrial cysts do not respond to hormonal therapy is unknown but five hypethoses could be proposed:
The persistence of active endometriotic tissue after GnRH agonist suggests the need for the surgical removal of invasive endometriosis (Nehzat et al., 1992, 1994; Brosens et al., 1995; Buttram et al., 1985; Donnez et al., 1987, 1989, 1990, 1993). The cumulative pregnancy rate achieved after combined (GnRH-a and endoscopy) therapy was similar to that obtained after microsurgery (Gordts et al., 1984), and allows us to propose this form of therapy in the management of large endometriomas. Indeed, the reduction of both cyst size and internal wall thickness observed in our study after GnRH-a therapy facilitates the laparoscopic management of large endometrial cysts. In our series of large cysts (more than 15 cm), an echoguided puncture was performed after an 8-week GnRH agonist therapy and the size of the cyst decreased to a diameter of 4.6 ± 2.4 cm. The recurrence rate of ovarian endometriomas was low (8 %), similar to the rate observed after microsurgery (Gordts et al., 1984) or after a laparoscopic procedure (cystectomy) (Canis et al., 1992, 1993).
We recommend the vaporization of the internal wall of the endometrioma because the active implants are located on the internal surface of the cyst. The incidence of endometriotic lesions found deep in the stroma in our study (33 %) cannot be considered as an argument in favour of cystectomy because deep- infiltrating intra-ovarian endometriosis was found with the same incidence. Another argument in favour of vaporization is the frequent absence of a thickened capsule around the endometrial cyst which makes capsule removal difficult and is responsible for the simultaneous removal of numerous ovocytes. In large cysts, the remaining ovarian cortex surrounding the endometrioma is often thinner than 5 mm. There is a risk of removing the normal ovarian cortex containing ovocytes during the attempt of endometrioma wall removal.
We do not agree with a recent paper (Dubuisson and Chapron, 1995) which recommended cystectomy in every case because histological analysis of endometriotic lesions of the ovary showed that 4 % of cases were atypical (Czernosbilsky and Morris, 1979). This incidence was interpreted by Dubuisson and Chapron (1995) as a risk factor for the development of cancer. We encountered no cases of ovarian cancer in our series of 814 patients, although we observed atypical areas in 12 % of cases. As reported by Czernobilski and Morris (1979) and by our group (Nisolle et al., 1988), hyperplasia and atypia have to be considered as endometriotic tissue completely unresponsive to hormones and not as a potential for ovarian cancer. Although two recent papers and debates (Brosens, 1995; Nezhat et al, 1995) have tried to classify endometrioamas, considerable uncertaintly still exists. Indeed, in our series, we found 12 % of endometriomas not fixed to the broad ligament. The theory of Hughesdon cannot explain the formation of the endometrioma in these cases. In our opinion, the epithelium covering the ovary, which is the mesothelium able to form any type of tissue, can invaginate in the ovarian cortex. Indeed, invaginations of the mesothelial layer covering the ovarian tissue were descried by Motta et al (1980) in animals and foetal ovaries. Morevoer, in our serial sections of the ovary, it was frequent to see mesothelial inclusions. Under the influence of so far unknown growth factors, these inclusions could be transformed into intra-ovarian endometriosis by metaplasia. Our main argument is the demonstration in our study of the capacity of the tissue to invaginate secondarily from the endometrial wall itself. Our second argument is the presence of epithelial invaginations in continuum with endometrial tissue, proving the metaplasia theory.
Our theory is thus different from the theory of Hughesdon (1957) and Brosens (1993, 1995) who consider that the pathogenesis of the typical ovarian endometrioma has now been clarified as a process originating from a free superficial implant which is in contact with the ovarian surface and is sealed off by adhesions, the menstrual shedding and bleeding of this small implant resulting in a progressive invagination of the ovarian cortex and the formation of the pseudocyst.
Our theory contradicts Hughesdon’s and Brosen’s hypotheses. The endometrioma must indeed be considered as an invagination but not as the result of bleeding of a superficial implant. Indeed, metaplasia of the coelomic epithelium invaginated in the ovarian cortex was proved in our study and explains the endometrioma formation.
Table 1: Scores (revised AFS classification) and ovarian cyst diameter after drainage and GnRH-a. Incidence of « active » endometriosis and tubal metaplasia. Mitotic index and epithelial height in active endometriosis ovarian foci.
* Significantly different (p < 0.001) from other values.
Edited by Aldo Campana,