Advances in the genetics of endometriosis

Endometriosis is a gynecological disease characterized by implantation of endometrial tissue outside of the uterus. Early familial aggregation and twin studies noted a higher risk of endometriosis among relatives. Studies on the roles of the environment, genetics and aberrant regulation in the endometrium and endometriotic lesions of women with endometriosis suggest that endometriosis arises from the interplay between genetic variants and environmental factors. Elucidating the hereditary component has proven difficult because multiple genes seem to produce a susceptibility to developing endometriosis. Molecular techniques, including linkage and genome-wide analysis, have identified candidate genes located near known loci related to development and regulation of the female reproductive tract. As new candidate genes are discovered and hereditary pathways identified using technologies such as genome-wide analysis, the possibility of prevention and treatment becomes more tangible for millions of women affected by endometriosis. Here, we discuss the advances of genetic research in endometriosis and describe technologies that have contributed to the current understanding of the genetic variability in endometriosis, variability that includes regulatory polymorphisms in key genes.

with endometriosis [11]. Simpson et al. [2] published the first formal genetic studies on women with histologically proven endometriosis in 1980. In this study, he found that 6.9% of firstdegree relatives also had endometriosis; by contrast, less than 1% of their husbands' female relatives had endometriosis [2]. In a followup study by Malinak et al. [12], the clinical characteristics of patients with histologically confirmed pelvic endometriosis who had affected relatives were compared with patients who had endometriosis without affected relatives. The primary difference was the finding that women who had relatives with endometriosis had more severe disease, defined as stages III to IV. The result further supported the role of genetics in the pathophysiology of endo metriosis and established a 7% risk of developing endometriosis for firstdegree relatives. Lamb et al. [3] later estimated an overall risk of 1.9% for seconddegree relatives. In subsequent familial aggregation studies in the UK, Coxhead and Thomas [13] compared 64 women with laparoscopically confirmed endometriosis and 128 controls and found that 9.4% of the patients had first degree relatives with endometriosis, yet only 1.6% in the control group had relatives with endometriosis. They found a sixfold increased risk for firstdegree relatives of women with laparoscopically confirmed diagnosis of endometriosis. Moen and Magnus [14] conducted a large Norwegian study composed of 522 cases, which supported the findings of earlier studies looking at familial aggregation; 3.9% of mothers and 4.8% of sisters of affected individuals had endometriosis compared with only 0.6% of sisters in the control group. Interestingly, this Norwegian study [14] also concluded that symptom severity was increased among women who had relatives with endometriosis, a conclusion originally reported in a smaller study by Malinak et al. [12].
Icelandic people, who have been geographically and genetically isolated, have been an interesting population in which to study familial aggregation. A computerized database includes all 283,000 presently living Icelanders in addition to most of their ancestors since the island was settled in the late 9th century. Stefansson et al. [15] used this database to study women diagnosed with endo metri osis over a 12year period (1981 to 1993). They calculated the contribution of genetic factors by comparing average kinship coefficients. The calculated kinship coefficient in 750 women with endometriosis was 1.82 × 10 4 , which was significantly higher (P < 0.001) than that of matched controls (1.45 × 10 4 ).
Twin studies have been used to discern the separate contributions of genes and environment. A small Nor wegian twin trial reported that six of eight monozygotic twin pairs were concordant for endometriosis [16]. Hadfield et al. [17] described concordance in 9 out of 16 monozygotic pairs for stage III to IV endometriosis in a larger British population of twin pairs. Of the seven discordant pairs, there were five pairs in which one twin had stage I to II disease and the other had stage III to IV disease. Interestingly, the women in the two discordant monozygotic twin pairs were all infertile. The observation suggests that genes associated with the development of endometriosis may be linked to genes that cause infertility.

Linkage analysis studies
Linkage analysis has been an important technique for discovering the chromosomal location of diseaserelated genes. Many monogenetic Mendelian disorders, includ ing mutations in the familial breast and ovarian cancer genes BRCA1 and BRCA2, have been successfully mapped using linkage analysis [18,19]. This technique has been used for detection of genetic markers of endo metriosis. A large linkage study by Treloar et al. [20] combined Australian and UK families and genotyped 4,985 women, including 2,709 with endometriosis. The collaborative International Endogene Study [20] brought together two independent groups: the UKbased Oxford Endometriosis Gene Study and the Australian Genes behind Endometriosis Study. The study [20] used linkage analysis techniques to find two loci of significant linkage on chromosome 10q26 and another region of suggestive linkage on chromosome 20p13. Chromosome 10q26 had previously been implicated in a candidate gene study [21] that reported aberrant endometrial EMX2 expression in women with endometriosis. EMX2 encodes a trans crip tion factor required for reproductivetract development [22], but variants of EMX2 have not been confirmed to contribute to the development of endometriosis [23]. Although promising, linkage analysis has yielded few significant genetic markers [24].

Genetic association studies
Candidate genes are chosen on the basis of their patho physiological relevance and variants of candidate genes are assessed in samples of endometriosis cases and controls. Table 1 summarizes the most commonly studied candidate genes. Generally, candidate genes can be divided into several classes: genes involved in regulating xenobiotic metabolism, those involved in steroid action and receptors, and those involved in inflammatory or angiogenic responses (Table 1) [25].
Human endometrium, whether in its normal place (eutopic) or ectopic, is a hormoneresponsive tissue, which also produces steroids, including estrogen, that have been shown to promote endometriosis [26]. Risk factors of endometriosis include prolonged exposure to endogenous estrogen, such as that resulting from early menarche or late menopause, and exogenous estrogens. Dysregulation of progesterone signaling in the eutopic and ectopic endometrium leads to impaired physiological functions, including infertility and endometriosis in women [2729].
Several association studies on single nucleotide polymorphisms (SNPs) of steroid receptor genes and steroid metabolizing genes and endometriosis have been performed. Results on the association of estrogen recep tor polymorphisms and endometriosis are inconsistent. Some studies showed an association of a polymorphism (PvuII) in the estrogen receptor 1 (ESR1) and endo metriosis [30,31], whereas other studies did not support this association [32,33]. These inconsistencies are probably caused by ethnic differences and small sample size of association studies. Huber et al. [34] analyzed the genotype frequencies of ten estrogenmetabolizing SNPs in patients with endometriosis compared with controls. Of these, the vIV A→C variant in the hydroxysteroid (17β) dehydrogenase 1 (HSD17B1) gene was significantly associated with an increased risk of endometriosis in both univariate and multivariate analyses.
Insensitivity to progesterone has been suggested to affect both implantation and inflammation in endometri osis [26]. The PROGINS polymorphism identified in the progesterone receptor (PR) gene contributes to growth dysregulation in hormonesensitive tissues and has been associated with breast and ovarian cancer [35]. The receptor change caused by the PROGINS mutation is hypothesized to affect ligand and hormonebinding properties, leading to inadequate control of estrogen receptordriven proliferation and estrogen overactivation. The PROGINS polymorphism is found more frequently in women diagnosed with endometriosis [3638], but data using genetic variants in the PR gene to predict endo metriosis are inconclusive in various ethnic popu lations [35,39].
Polymorphisms in cytokine genes, including tumor necrosis factor (TNF)α, interleukin (IL)-1β, IL-6 and IL-10, and their receptors (IL-1RN and IL-2Rβ), have been extensively studied because of the involvement of cytokines in the pathogenesis of endometriosis [40]; however, genetic variants of cytokine genes have not always shown an association with endometriosis. For example, several studies have reported that polymor phisms in the promoter region of the TNF-α gene do not influence endometriosis risk in Korean, Taiwanese or Caucasian women [4143]. Similarly, inconsistent find ings on the association of genetic variants and endometriosis have been shown for IL-6 and intercellular adhesion molecule 1 (ICAM1) genes [44]. Association studies in English on the most commonly studied candidate genes were identified by performing a PubMed literature search up to 23 June 2010. If a published study identified one or more positive associations, we identified this study as positive, otherwise negative. Variants of the ESR2, peroxisome proliferatoractivated receptor γ2 (PPAR-γ2), nuclear factor κB1 (NFKB1), E-cadherin, matrix metalloproteinase 1 (MMP1), MMP9, cyclin dependent kinase inhibitor p27 (CDKN1B), neurokinin-1 (TAC1),nitric oxide synthase 3 (NOS3), fibroblast growth factor 1 (FGF1), FGF2 and catechol-O-methyltransferase (COMT) genes have also been investigated as candidate genes potentially associated with endometriosis.
Exposures to endocrinedisrupting chemicals, such as dioxin, have been implicated in the pathogenesis of endo metriosis in rhesus macaques [45] and in women [46,47]. Glutathione Stransferases (GSTs) are enzymes involved in the detoxification of a broad range of toxic compounds and carcinogens, including dioxin [48]. Two of the most studied SNPs in detoxification enzymes are in the GST genes GSTM1 and GSTT1, which are located on chromo some 1p13.3 and 22q11.2, respectively. A metaanalysis by Guo [49] involving 14 studies on GSTM1 and 9 studies on GSTT1 concluded that there was no increased risk for women with the GSTM1 null genotype and a slightly increased risk for those with a GSTT1 polymorphism. Table 1 summarizes the variable results of the endo metri osis and gene polymorphism association studies. These discrepancies may be due to methodological problems, including relatively small numbers of cases and controls and selection of control populations [44]. Another explanation may be the use of different geno typing techniques. In addition, there is difficulty control ling environmental factors and problems replicating results among different ethnic populations. Large, multi center prospective cohort studies with DNA genotyping arrays and genegene and geneenvironment studies of endometriosis are needed.

Genome-wide association studies
Highresolution genomic tools enabling genomewide association studies (GWASs) [50] offer the prospect of making real progress in the discovery of genes contri buting to risk of multifactorial diseases such as endo metriosis. GWASs have been performed successfully in the discovery of new genomic regions that are associated with multifactorial diseases [51,52]. The methodology examines variation across the genome using compu tational models to compare the genotypes of people with and without disease to identify SNPs associated with disease. Sufficient power using this approach necessitates thousands of subjects, but this can be achieved by collaboration of large consortia. The advantage of this method is that hundreds of thousands of SNPs can be quickly examined for potential associations [53]. Another advantage of GWASs is that they are unbiased and do not depend on biologically plausible candidate genes or specific polymorphisms [54]. Disadvantages are that very large sample sizes are necessary to derive meaningful and reproducible conclusions valuable for characterizing poly genic diseases such as endometriosis.
Several large populationbased prospective cohorts in the United States, Europe, Australia and Japan have gathered endometriosis data and represent potential sources for GWASs. These include the American Cali fornia Teacher Study [55], the Nurses' Health Study II [56,57], the National Health and Nutrition Examination Survey [58], the French E3N Cohort study [59], the Australian Genes behind Endometriosis Study [60], the Oxford Endometriosis Gene Study [61], the Icelandic Endometriosis Cohort Study [15] and the BioBank Japan project [62]. From these different cohorts, extensive timedependent covariate data have been collected [56]. The cohorts differ in prevalence of current infertility, surgical diagnosis, age distribution, selection of controls and ethnicity. The first large endometriosis GWAS was published by Uno et al. [63] using the endometriosis cohort from the Biobank Japan project (1,907 Japanese endometriosis cases and 5,292 controls). They identified an association of endometriosis with the SNP rs10965235 located in the cyclindependent kinase inhibitor 2B anti sense RNA (CDKN2BAS) gene on chromosome 9p21 (with an odds ratio of 1.44) and with rs16826658m in the linkage disequilibrium block that includes the signaling gene WNT4 on chromosome 1p36 [63]. WNT4 signaling has a role in the coordination of female ovarian follicle development and of the development of the fallopian tubes and uterus from the Müllerian (paramesonephric) ducts [64]. CDKN2BAS regulates the expression of CDKN2A, a cellcycle kinase inhibitor that controls endometrial proliferation and has been attributed with a role in endometriosis [65].
Interestingly, this GWAS [63] has confirmed loci such as 9p21 that are shared with conditions previously thought to be unrelated to endometriosis, including coro nary artery disease. Recent studies have revealed that genetic variants in 9p21 are associated with multi factorial diseases, including coronary artery disease, type 2 diabetes, malignant melanoma, basal cell carcinoma, nevi (moles) and glioma [6669]. The clinical relevance of identified genetic variants for predicting risk of endo metriosis is unclear because the overall increase in risk caused by the known variants is small [70]. However, GWASs represent a new frontier in the investigation of genegene and geneenvironment interactions. Predictive power may improve as more susceptibility loci are identified and new computational tools are applied.

Conclusions
We have seen progress in the understanding of endometriosis in the past decade through linkage analysis and association studies. However, the pathophysiology and genetics of endometriosis are still unclear. The long term goals in discovering the genetic basis of endo metriosis are early detection and improved treatment of endometriosisassociated symptoms, including infertility and pelvic pain. GWASs may bring insight into gene environment interaction and identify candidate genes that are associated with endometriosis. This promising new technology has the potential for finding modest associations for polygenic diseases such as endometriosis.
However, GWASs are prone to error and bias arising from the predominantly casecontrol study design and the difficulty of interpreting statistics that result from multiple hypothesis testing. They need to be further validated through replication studies that find similar associations among independent samples with large, appropriately powered populations. The credibility and validity of GWASs will be further increased when consistent statistical results are combined with evidence of biological functionality for the variants found. Thereafter, data from GWASs could be used to assess individual risk for endometriosis and may be useful in counseling women for early treatment in order to avoid sequelae such as infertility. In addition, GWAS dis coveries could be implemented in the development of novel therapies for the treatment of endometriosis. Geneticsbased diagnostic tests for endometriosis and genetically targeted therapies hold the promise to diagnose and treat women at a younger age and allow physicians to better advise women regarding prognosis.