Genetics of Endometriosis

Endometriosis is a common disease defined as the growth of endometrial tissue outside the uterine cavity that often results in a vast array of problems, including dyspareunia, dysmenorrhea, pelvic pain, and infertility. The true incidence of endometriosis is unknown, but reported estimates normally range from 10% to 15% for all women of reproductive age and from 25% to 35% for women who are infertile.¹ The histopathology of endometriosis is variable and dependent on the site of growth. Ovaries, uterosacral ligaments, retrovaginal septum, and pelvic peritoneum are principal sites. Because of the heterogeneity of this disease, development of a reliable classification system to aid in diagnosis of disease stage and therapy has been difficult. Currently, the disease is classified into stages ranging from mild to moderate to severe based on the size, number, and depth of endometriotic lesions.² However, these parameters do not always correlate with severity of symptoms (e.g., pain or infertility or both). Treatment involves surgical resection or use of various hormonal-based regimens (e.g., progestogens, Danocrine, gonadotropin-releasing hormone [GNRH] agonists).

Retrograde menstruation is believed to be a common mechanism involving backward flow of menstrual blood and uterine tissue that is refluxed through the Fallopian tubes into the peritoneal space. There have been many reports in support of this mechanism.^3,4,5 However, although most women are believed to exhibit retrograde menstruation, only some have endometriosis.⁶ Therefore, the development of endometriosis in only a select percentage of women implies increased susceptibility to disease.

Endometriosis has long been suspected of familial tendencies. Following the questionnaire survey of Ranney⁷ reported in 1971, Simpson and coworkers⁸ conducted the first formal genetic study. A total of 123 probands with histologically verified endometriosis were identified. Nine (5.9%) of 153 female siblings older than 18 years had endometriosis; 10 (8.1%) of the 123 mothers were affected. Only 1% of the patients' husbands' first-degree relatives (control subjects) had endometriosis. Of two sets of monozygotic (MZ) twins, one was concordantly affected. Women with an affected sibling or parent were more likely to have severe than mild or moderate endometriosis.⁹ Severe endometriosis was present in 11 (61%) of 18 probands having an affected first-degree relative. Severe endometriosis was present in only 25 (23%) of 105 having no affected first-degree relative.

Studies by other investigators have confirmed these results. Lamb and coworkers¹⁰ used questionnaires received from 491 members of the Endometriosis Association within the United States. Positive family history was reported by 18% of respondents. Of the 66 women who were evaluated in greater detail, 43 returned a detailed questionnaire completed as well by a friend (control subject). Endometriosis was present in 6.2% in mothers of proband and 3.8% in sisters; endometriosis was reported in fewer than 1% in first-degree relatives of friends. The frequency in second-degree relatives was 0.4% in grandmothers and 3.1% in aunts. In this study, most (93%) affected relatives were in the maternal lineage. A pitfall of this study is that no attempts were made to confirm the diagnosis in relatives believed to be affected; however, members of the Endometriosis Association can be assumed to be knowledgeable concerning the disorder.

In Norway, Moen and Magnus¹¹ conducted a study similar to that of Simpson and colleagues.⁸ Among 522 informative cases, 3.9% of mothers and 4.8% of sisters had endometriosis; only 0.6% of sisters of women not having endometriosis (control subjects undergoing laparoscopy for other reasons) were affected. In this study, either endometriosis or adenomyosis constituted the basis for diagnosis. Mothers were far more likely to have adenomyosis than affected sisters. As in the U.S. cases of Simpson and coworkers⁸ and Malinak and coworkers,⁹ familial cases in Norway proved more likely to show severe endometriosis than were nonfamilial cases. In another report from the same Norwegian center, eight MZ twins were observed among 515 endometriosis cases.¹² Six of the eight sets were concordant, and in three cases, mothers also were affected.

In the United Kingdom, Coxhead and Thomas¹³ reported a sixfold increase frequency of endometriosis among first-degree relatives. Of 64 patients confirmed to have endometriosis, 6 (9.4%) had an affected first-degree relative. Among the 128 control women, only 2 (1.6%) reported to have a first-degree relative affected with endometriosis. More recently, dos Reis and coworkers¹⁴ reported 7 (8.6%) of first-degree relatives of 81 probands to be affected compared with no relatives of 43 control subjects.

Investigations have been initiated to solicit familial cases for linkage studies by performing sibling pair analysis using DNA polymorphic variants. Recurrence data cannot be derived because of the selection biases, but these studies nonetheless confirmed that familial aggregates are not infrequent.^15,16,17 The OXGENE (Ox ford Endometriosis Gene) group recorded receipt of samples from 19 mother-daughter pairs and 56 sibling pairs.¹⁵ In 18 families, 3 or more relatives in more than 1 generation were observed. All but 2 of 16 MZ twin pairs have been concordant for endometriosis.¹⁸ A similar study in Iceland also yielded 15 familial aggregates.¹⁹ Of interest is that Kennedy and colleagues^20,21 recommend magnetic resonance imaging to diagnose endometriosis. Endometriosis was found on magnetic resonance imaging in 5 (14%) of 14 first-degree relatives and 1 (8%) of 12 other relatives.

Higher concordance for MZ than DZ twins has been observed.^12,15,22 In addition, endometriosis detected in the context as a cause for surgical menopause is more highly correlated in MZ twins than in DZ twins (r = 0.52 versus r = 0.19).²² Finally, catamenial pneumothorax was reported in two sisters having pelvic endometriosis.²³ Pneumothorax occurred on the right, the usual site.

Endometriosis is clearly heritable, but the precise mechanism or mechanisms remain unclear. Magnitude of the increased risk (5% to 8% of first-degree relatives) is more reminiscent of polygenic/multifactorial tendencies than a single mutant gene. However, this recurrence risk is higher than the 2% to 5% expected risk for polygenic inheritance. The frequency of affected relatives might be even higher if one could directly measure a gene product or products. Although Mendelian mechanisms cannot be excluded, polygenic inheritance seems more likely if one assumes all endometriosis is a single disorder. If it is, the increased severity in familial cases is also consistent with predictions based on a polygenic model. Such a model predicts that the greater the severity, the greater the underlying genetic liability and, hence, the greater the proportion of affected relatives. Endometriosis was more severe in familial cases, which also lessens the likelihood that the presence of an affected family member led to the identification of an affected relative merely because of a higher index of clinical suspicion.

The other formal explanation, and perhaps the most likely (i.e., not all endometriosis is the same disorder), is genetic heterogeneity. One or more forms of endometriosis might be Mendelian despite the larger proportion being nongenetic or polygenic. This has proved to be the likely explanation for peptic ulcer and other adult-onset disorders. No studies have shown HLA associations with endometriosis,^24,25,26 which is more suggestive, however, of polygenic/multifactorial inheritance than genetic heterogeneity.

As noted, genome-wide genetic analysis is underway by several groups seeking to localize the various genes paramount to the etiology of endometriosis. Kennedy and colleagues^15,17 are using sibling-pair analysis with polymorphic DNA markers. Identification of multiple nonlinked genes would be consistent with either polygenic inheritance or genetic heterogeneity. Several regions of exclusion have been identified by Kennedy and colleagues,¹⁵ but no linkage has been observed. This lack of success could reflect inaccuracies in diagnosis. In Icelandic women, a “suggestive locus” was found on 9q that is not in the region of the galactose-l-phosphate uridyl transferase (GALT) gene.¹⁹

Initial steps are underway to determine the molecular basis of endometriosis. One vexing problem has been difficulty in studying pure endometriosis tissue, necessary to avoid analysis of tissue samples in which normal endometrium or connective tissues are admixed with endometriotic tissue. Analysis of the normal tissue should prove genetically normal and may possibly explain the failure of Dangel and colleagues²⁷ to find cytogenetic abnormalities in any of 45 endometrial implants. Similarly, Vercellini and coworkers²⁸ failed to find ras oncogene- and p53 tumor suppressor gene-specific mutations in ectopic and eutopic endometrial tissue biopsy specimens from 10 women with severe endometriosis. Techniques are now being used that offer greater confidence for the genetic information truly being reflective of endometriosis-derived tissue. Our group has used touch preparations to study endometriotic tissues, in which histology can be confirmed in a mirror-image micrograph. Using fluorescence in situ hybridization with chromosome-specific probes, monosomy 16 and 17 and trisomy 11 cells were found.²⁹ We also found chromosome 17 monosomic cells consistently in each of 8 endometriotic samples.³⁰ Various chromosome abnormalities, including tetrasomy 17, also were observed in a human endometriosis-derived established cell line by Bouquet de Joliniere and colleagues.³¹ This French team has used comparative genome hybridization to find 1q+ , 4q-, 11p-, 13q-; loss of 9, 12, and 18; and amplification of 6p.³² Together, these studies provide evidence that acquired chromosome-specific alterations may be involved in endometriosis, possibly reflecting clonal expansion of chromosomally abnormal cells.

A polymorphism in the GALT gene—an adenine-to-guanine transition in exon 10, substituting aspartate for asparagine (N314D)—was found to be associated with endometriosis by Cramer and coworkers³³ but not by Morland and coworkers³⁴ and Hadfield and coworkers.³⁵

In endometriosis, we have hypothesized that the chromosomal pattern of aberrations combined with the invasive nature of the disorder parallels between endometriosis and neoplasia.³⁶ Pathogenesis of neoplasia is now accepted as involving clonal expansion of a progenitor cell, giving rise to selective advantage. By this scenario, two “hits” (mutational events) are necessary; both may be somatic mutations or one may be germline and the other somatic. Relevant to this hypothesis is that monoclonal cell expansion in endometriosis has been observed.^37,38 Some investigators now speculate that endometrioid ovarian cancers probably arise by the malignant transformation of endometriotic implants.³⁹ Although the frequency of malignant transformation is less than 1%,^40,41 there is clear association of endometriosis with certain “endometrioid” malignancies.^42,43,44 Cancers arising from endometriosis typically are divided into two main groups based on site: ovarian and extraovarian. Ovarian cancers account for more than 75% of the cases, consisting predominantly of endometrioid (70%) and clear-cell (14%) carcinoma. Of the extraovarian tumors, approximately 66% are endometrioid carcinomas localized to the rectovaginal septum (most common), uterus, Fallopian tubes, rectum, or bladder. Therefore, in the search for the genes involved in endometriosis, initial screening of chromosomal or gene-specific alterations or both identified in these cancers is warranted in endometriotic cells.

Jiang and colleagues⁴⁵ have reported using microdissected endometriotic glands and stroma from archival tissue to investigate whether loss of heterozygosity (LOH) in chromosomal regions associated with ovarian carcinomas occurs in late-stage endometriosis. They showed monoclonal origin in endometriotic cysts and provided evidence of LOH involving chromosomes 9p, 11q, and 22q compared with normal genomic DNA. In a subsequent study, the same authors show common genetic alterations in nine of 11 cases in which ovarian carcinoma had arisen within or adjacent to endometriosis.⁴⁶ They observed alterations in chromosome arms 5q, 6q, 9p, 11q, and 22q to be common in 25% to 30% of cases involving endometriosis with associated carcinoma. In one endometriosis sample, mutation in the p53 gene at codon 220 (Tyr to Cys) was observed. Similarly, we have reported preliminary findings of chromosome 17 alterations, namely LOH involving 17q in 4 (30%) of 12 cases⁴⁷ and monosomy 17 in all 8 cases.³⁰ In both of these studies, endometriotic DNA or cells, respectively, were analyzed and compared with matched “normal” tissue from patients confirmed to have stage III or IV disease. Therefore, genetic alterations associated with ovarian endometrioid cancers may be involved in the development of endometriosis.

Further support of the hypothesis that ovarian endometrioid cancers arise through malignant transformation of endometriotic lesions stems from recent evaluation of DNA alterations in the subtypes of epithelial ovarian cancers. Obata and coworkers⁴⁸ reported mutations in the tumor suppressor PTEN gene on chromosome 10q23 in endometrioid but not serous or mucinous epithelial ovarian tumors. They analyzed more than 81 ovarian tumors, including 34 endometrioid, 29 serous, 10 mucinous, and 8 clear-cell tumors for LOH on 10q23 and mutations in PTEN. Although LOH was common among the endometrioid (43%) and serous (28%) tumors, somatic mutations involving PTEN in the remaining allele was observed only in the endometrioid (21%) tumors. These results show that the developmental pathways of the 3 epithelial ovarian cancer subtypes (serous, mucinous, and endometrioid) appear to be different and that somatic mutations in PTEN may represent early events in the transformation of benign endometriotic cells to malignancy. The PTEN protein is believed to function as a tyrosine phosphatase and play a role in signal transduction. Interestingly, other studies have shown PTEN mutations to be common in endometrial cancers in which microsatellite instability is detected.^49,50

Immunocytochemical studies have also shown high protein levels of various proto-oncogene expression in endometriotic tissue compared with normal endometrium, including c-myc, c-fms, c-erbB-1/-2, and ras.^51,52 Amplification on the short arm of chromosome 6 (6p) has also been seen.³² However, none of several potential candidate genes on 6p were overexpressed. These results suggest that altered proto-oncogene expression may be involved in disregulated growth and differentiation of endometriotic cells.

The bcl-2 gene is a member of the Bcl-2 gene family, and overexpression of this gene leads to a decreased rate of cell death.⁵³ This gene is considered to play a role in the normal endometrium cycle by regulating cellular homeostasis and apoptosis; increased expression is detected in the proliferative endometrial phase but not in the secretory phase.⁵⁴ Although several studies have examined the expression of bcl-2 in endometrial carcinomas and hyperplasia, the results have been controversial.⁵⁵ In a recent study, Watanabe and coworkers⁵⁶ reported bcl-2 overexpression in ectopic endometrial lesions by immunohistochemical staining, indicating that endometriotic cells fail to undergo apoptosis. Dmowski and colleagues⁵⁷ have also reported decreased apoptosis of endometrial cells in endometriosis.

That only some women have endometriosis implies that there is increased susceptibility to development of disease among certain women. Individual susceptibility is not only influenced by genetic background but also by the interaction of genes with environmental factors. Dioxin (2,3,7,8-tetrachlorodibenzo-p-dioxin; TCDD) and dioxin-like compounds (DLCs) (e.g., polychlorinated biphenyls [PCBs]) have been implicated as factors involved in the development of endometriosis.⁵⁸ The major route of human exposure to dioxin and DLCs is through diet. These compounds are hydrophobic and resistant to metabolic degradation, with an estimated half-life of 7 years in humans. Consequently, they accumulate to high levels in lipids and membranes, being released slowly into the blood. In addition to well-described toxic effects, dioxin has adverse effects on the female reproductive system. Dioxin administration severely reduced the reproductive success and estrous cycling in rodents and nonhuman primates^59,60 and antagonized estrogen-induced increases in uterine weights and endometrial development in mice.⁶¹ Using rhesus monkeys, Rier and coworkers⁶² were first to show the direct relationship between dioxin and endometriosis. Autopsy evaluation of animals exposed daily for 5 years to 5 or 25 parts-per-trillion (ppt) of dioxin in their food showed widespread peritoneal endometriosis in 71% and 86% of animals, respectively. Among the control group, endometriosis was observed in only two (33%) animals; this is similar to the spontaneous endometriosis rate in colonies of rhesus monkeys. In humans, there is increasing epidemiologic evidence of the effects of dioxin. Gerhard and Runnebaum⁶³ reported higher levels of PCBs in 28 patients with endometriosis compared with 441 unaffected women in a population of infertile women. Koninckx and colleagues⁶⁴ indicated that the incidence of endometriosis in Belgium is approximately 60% to 80% in infertile women and pain, one of the highest reported incidences in the world. In 1989, the World Health Organization⁶⁵ reported dioxin concentrations in breast milk of Belgium women to be among the highest in the world. Increased concentration of dioxin in breast milk may reflect the high dioxin accumulation in women, possibly explaining the high incidence of endometriosis in these women. More recently, Mayani and coworkers⁶⁶ measured the concentration of dioxin in the peripheral blood of 44 women with endometriosis compared with 35 control subjects. Dioxin concentrations of 0.6 to 1.2 ppt were detected in 18% of affected cases compared with 0.4 ppt in one (3%) control case. Together, these studies suggest that dioxin and DLCs may play a role in the development of endometriosis in humans.

The phase II conjugating enzymes usually function to inactivate carcinogenic and procarcinogenic compounds. Among the phase II enzymes, two have been studied extensively, GSTs and NAT2, and are considered important cancer susceptibility genes. Both the GSTM1 and GSTT1 genes are polymorphic with null alleles that can be detected by polymerase chain reaction-based methods.^67,68 In the study of 84 cases of epithelial ovarian cancer, Sarhanis and coworkers⁶⁹ reported an influence of the GSTM1 and GSTT1 null alleles on increased p53 expression. The authors proposed that the GSTM1 and GSTT1 are critical in the detoxification of the products of oxidative stress produced during the repair of the ovarian epithelium. Thus, homozygous null alleles in both genes may function synergistically, causing inefficient detoxification of intermediates produced during stress that increase damage to various genes in the host cell, including p53, resulting in persistent expression of mutant protein. A recent study has shown a possible role of the homozygous GSTM1 null allele with the development of endometriosis in 86% (n = 50) of cases compared with 45.8% (n = 72) of control unaffected women (p < .0001).⁷⁰ Our preliminary results of an association of M1/M2 NAT2 slow-acetylator genotype provide further evidence that heritable allelic differences in drug-metabolizing enzymes may play an important role in the development of endometriosis.⁷¹