Epithelial ovarian cancer (EOC) remains the number one cause of death from gynecological malignancies. More than 50% of the patients relapse within a few years. Among the EOC, the most common morphological subtype is serous adenocarcinoma (SAC), with less common subtypes including clear cell (CCC), mucinous (MAC) and endometrioid (EAC), generally believed to originate from the ovarian surface epithelium, distal fallopian tube epithelium or peritoneal mesothelial cells. In Japan, however, CCC is not uncommon and is the second most frequent (22%) type of EOC (1). This tumor demonstrates a clinical behavior different from that of the most common histotype SAC. Patients with CCC have early stage with a large pelvic mass, less ascites, chemoresistance and higher incidence of endometriosis and thromboembolism (2). The survival rates of patients with advanced CCC are lower than those of patients with advanced SAC.

Although two histologic types, CCC and EAC, are the common histology in ovarian cancer patients who have associated endometriosis (also known as endometriosis-associated ovarian cancer; EAOC) (3), both tumor types have also distinct clinicopathological characteristics and molecular phenotypes (4,5). The exact mechanisms that turn a process of endometriosis into a cancer precursor are recent topics of intense research. Animal models that resemble the EAOC are of paramount importance in studying the carcinogenesis of these diseases. Several methods for modeling EAC in animals are available. A variety of molecular events, such as PTEN (phosphatase and tensin homolog) silencing and KRAS (v-Ki-ras2 Kirsten rat sarcoma viral oncogene homolog) mutations/activation, have been frequently identified in EAC (6). Despite advancements in the molecular analyses of EAOC and the development of animal models, the mechanisms that underlie the pathogenesis of CCC remain largely unknown. Representative CCC animal models have not been available to date. The current knowledge on the mechanisms involved in CCC carcinogenesis shows that ovarian hemorrhage or retrograde menstruation carries highly pro-oxidant factors, such as heme and iron, into the ovarian endometrioma or peritoneal cavity (7). The persistence of a redox active iron in the blood at the site of endometriosis development is important to understand both induction and promotion of the EAOC. Taken together, DNA damage or loss of heterozygosity (LOH) in an ectopic endometrium caused by iron-induced oxidative stress may be a critical factor in the carcinogenic process (8–11).

CCC carries worse prognoses than EAC tumors. EAC was predominantly positive for estrogen receptor (ER), but CCC specifically exhibits lower ER expression. Aberrant DNA methylation of CpG islands has been recognized in carcinogenesis as a common alteration associated with the loss of expression of a number of key regulatory genes. Many studies have found that the loss of ER expression is a result of the hypermethylation of the ER-α promoter. Although hypermethylation is one of the best known epigenetic events, other epigenetic events, such as histone modification (deacetylation and methylation of histones), are involved in the complex mechanism that regulates promoter transcription, and drive stable, clonally propagated changes in gene expression. The exact interplay of these factors in transcriptional repression activity in CCC is not yet well understood. This study reviews the current understanding of the role of the ER information in the pathogenesis of CCC.

The present article reviews the English language literature for biological, pathogenetic and pathophysiological studies on endometriosis and EAOC. We searched PubMed electronic database for a 20-year period (1990–2010), combining the keywords ‘genome-wide’, ‘microarray’, ‘proteomics’, ‘estrogen receptor’, ‘oxidative stress’, ‘ROS’, ‘carcinogenesis’, ‘iron’, ‘HNF-1β’, ‘mutation’ and ‘chromatin remodeling’ with ‘endometriosis’, ‘ovarian cancer’, ‘clear cell carcinoma’ or ‘endometriosis-associated ovarian cancer’. Several recent studies are discussed in the context of pathogenesis of CCC and DNA mutations. Additionally, references in each article were searched to identify potentially missed studies for a 20-year period.

Although the main interest is the regulation of ER expression obtained from human ovarian cancer samples, we have included the in vitro studies in the knowledgebase. We also included the animal models performed to support human data. Initially, 58 potentially relevant studies were identified by screening electronic databases. A total of 21 peer-reviewed journal articles were additionally identified from references in each article.

Estrogen plays a role in ovarian tumorigenesis. In vitro studies indicate that ovarian cancer cell growth is dependent on estrogen stimulation. In a case of breast cancer, ER-negative tumors fail to respond to endocrine therapy and have a poor overall prognosis when compared to ER-positive tumors. The ER-α mRNA level was a significantly positive prognostic factor for patient survival. The prognostic significance of ER expression by ovarian cancers has received little attention to date. One study reported a correlation between levels of ER expression and cancer disease stage, with levels declining with increased severity of disease, suggesting that loss of ER expression in ovarian cancer is a feature of malignant transformation and aggressiveness (12). Despite the poor prognosis of patients with ER-negative disease, there remains considerable heterogeneity in individual outcomes. Indeed, other studies demonstrated that since ER has emerged as a mitogenic factor, ER status is a prognostic factor for ovarian cancer with better survival for ER-negative tumors (12). Furthermore, a higher ER expression at the mRNA and protein levels was found to be associated with a longer progression-free survival and overall survival (13). The remaining investigators reported, however, that neither ER nor progesterone receptor (PR) independently correlated with survival in the overall study population. Therefore, there is a controversy as to whether ER expression is a prognostic factor for outcomes in ovarian cancer.

Among EAOC, EAC were predominantly positive for ER and PR (14), but CCC specifically exhibited lower ER and PR expression (14). The investigators put forward a model postulating that additional events, particularly deletion of ER expression, are required for CCC lesion progression. CCC pathogenesis may be a model to study the disease progression from estrogen-dependent to estrogen-independent, allowing design of new strategies targeting the hormone response, thereby modifying disease outcome. Therefore, loss of estrogen function may be a turning point in CCC development. There are basically the following hypotheses regarding the carcinogenesis or pathogenesis of CCC: initially, the heme and iron-mediated oxidative stress and persistent inflammation processes occur due to repeated hemorrhage in endometriosis. These compounds oxidatively modify DNA, proteins and lipids, and subsequently hypermethylation or ER depletion may be observed (Fig. 1). Suzuki et al reported that ER is inactivated mainly through aberrant DNA methylation (15). A dualistic model, which has been established on morphological and genomic basis, differentiates EAOC into two broad categories: estrogen-dependent ovarian cancers with an EAC morphology, and estrogen-independent carcinoma with the CCC morphology (4). The genetic pathways utilized by ER-negative tumors to proliferate in the absence of a mitogenic estrogen signal are poorly understood. Elucidation of these pathways is required for the development of improved therapies for ER-negative CCC patients.

Estrogen is supposed to actively participate in the early stages of endometrial tumorigenesis. ER also plays a role in the onset and development of tumors arising from or outside the reproductive system. Genomic aberrations of ER-dependent downstream targets provide cellular growth advantages when the cells are exposed to estrogen. There are a number of factors that interfere with ER expression and estrogen activity. Many factors are implicated as an important regulator of distinct aspects of ER action and expression, including gene expression, DNA methylation of the promoter region, histone deacetylation and heme and iron binding (16) (Fig. 1).

Epidemiologically and clinicopathologically, EAC and CCC are believed to develop from endometriosis (EAOC). Under certain conditions in endometriosis, the biological behavior change of the local microenvironment may promote the development of EAOC. Recent studies have demonstrated that a redox active iron overload and the accompanying chronic inflammatory processes will supply critical initiators inducing cell growth, survival, anti-apoptosis and finally the oxidative stress-response, leading to carcinogenesis. Among EAOC, CCC differ from EAC with respect to their clinical characteristics and carcinogenesis (10). A growing body of evidence has accumulated that, compared to EAC, CCC is relatively resistant to conventional taxane plus platinum-based chemotherapy (31). Therefore, EAC shows favorable prognosis, whereas CCC contributes to poor prognosis. There is a review focusing on the specific gene expression that may predict the response to chemotherapy and/or prognosis (7).

The genome-wide expression analyses identified differences in the expression of several genes and proteins between CCC and EAC (4,5,7–11,32). The literature data on the expression of ER in CCC differ from that in EAC (14). Compared to EAC, CCC exhibits very low ER expression. CCC characterized by ER negativity may be an important reason for an aggressive tumor with poor prognosis. Several studies have defined sets of genes with differential expression levels between ER-positive and ER-negative tumor types. For example, ER-negative breast cancer exhibits a greater proliferation signal, failure of chemotherapy and an aggressive tumor with poor prognosis. We put forward a model postulating that deletion of ER expression is required for CCC lesion progression and aggressiveness.

The present review demonstrated that many factors are implicated as critical regulators of distinct aspects of estrogenic action and ER expression, including gene overexpression and silencing, DNA methylation of the promoter region, histone deacetylation, heme and iron binding, PPARγ action and ubiquitin ligase activity. Consequently, loss of ER expression may be associated with a specific additional risk for developing CCC, but not EAC. Our preliminary microarray data on the HNF-1β siRNA experiments support the notion that the estrogenic action and ER expression are down-regulated in response to HNF-1β (Shigetomi et al, 2011, unpublished data). The up-regulation of HNF-1β expression led to subsequent modulation of its downstream targets that directly or indirectly suppress the estrogenic action. We have defined the gene set in CCC, with a view to identify the HNF-1β-dependent estrogenic action. Some of the genes predominantly identified in CCC were related to ER modulation, which may be involved in downstream targets of HNF-1β gene. This result is particularly interesting because CCC cells that show overexpression of HNF-1β gene may down-regulate the estrogenic action and ER expression.

More recent data showed that the specific genes that were somatically mutated in CCC are ARID1A, PIK3CA, PPP2R1A and KRAS (28,29). ARID1A mutation results in aberrant chromatin remodeling and may contribute to the development of CCC, possibly through a lack of tumor suppressor gene (28,29) and also suppression of estrogenic action. Additionally, PPP2R1A is another gene that exhibits acquired somatic mutations in the genome in CCC (29). PPP2R1A functions, when mutated, as an oncogene. PPP2R1A is a regulatory subunit A-α of PP2A (serine/threonine protein phosphatase type 2A). PP2A plays an essential role in cell cycle regulation and induction of G2 arrest by a mechanism of phosphorylation/dephosphorylation with a variety of protein kinases. Our preliminary data showed that PPP2R1A (29) and PP2A can be deregulated in CCC either by somatic mutation and transcriptional repression by HNF-1β, respectively (Shigetomi et al, 2010, unpublished data). Furthermore, the PP2A complex is also associated with ER and leads to its dephosphorylation and inhibiting transcription.

HNF-1β is differentially expressed among the EAOC. The nuclear immunostaining of HNF-1β protein is observed in the majority of CCC, but not in EAC, demonstrating that this aberration is histotype-specific. Recent biochemical studies based on genome-wide microarray analysis technology have noted specific expression of HNF-1β in CCC (33). Kato et al reported that hypomethylation of the HNF-1β CpG island participates in the HNF-1β up-regulation in CCC (34). Among the genes highly up-regulated in CCC, more than 80% genes were associated with the redox-related gene (32). Several important CCC-related genes overlap with those known to be regulated by HNF-1β (32).

Several investigators have described the HNF-1β-dependent pathophysiology of CCC and discussed its role in oxidative stress-induced carcinogenesis (7–11). A majority of CCC likely develops as the result of a stepwise accumulation of alterations in cellular regulatory pathways, such as DNA methylation, genomic DNA mutation, LOH, oncogene activation, tumor suppressor gene inactivation and aberrant chromatin remodeling. The gene expression profiling and signaling pathways of HNF-1β, so called ‘HNF signature’, has been assessed in CCC: the catalog of CCC-specificity may be a manifestation of several essential alterations in cell physiology, including oxidative stress, detoxification and metabolism (7,8). The ‘HNF signature’ shows evidence of biological functions, including glycogen storage (energy accumulation), anti-apoptosis (survival) and detoxification (stress-resistance).

Up-regulation of HNF-1β expression was evident in the CCC and contiguous atypical endometriosis and even in distant endometriotic lesions (35). A candidate early precursor to CCC, the ‘HNF signature’, is commonly found even in endometriosis in the absence of any morphological abnormality (35–38). The ‘HNF signature’ may be an early event of the clear cell directionality (4,7–10,32). Therefore, up-regulation of HNF-1β expression is not sufficient for the development of CCC.

Birrer proposed that the switch from a normal-stress-response phenotype to a stress-resistant phenotype may involve the mutation of the specific genes in CCC (39). Additional candidate gene abnormalities or molecular alterations may cause the progression towards the CCC. ARID1A mutations are evident in the CCC and contiguous atypical endometriosis, but not in distant endometriotic lesions (28). Therefore, ARID1A is considered to be an early event in the transformation of endometriosis into ovarian cancer (28).

In conclusion, this review examines the suppression of estrogenic action and ER expression in CCC. The basic findings of this comprehensive review suggest that the interactions between the iron-mediated oxidative stress and down-regulation of ER expression may have clear implications for developing CCC via the up-regulation of HNF-1β expression. These genes and proteins significantly elevated and repressed in CCC subjects may provide a foundation for further validation in larger patient cohorts. Future studies will address the mechanism of synergistic effects of the specific gene mutations and up-regulation of HNF-1β gene and its downstream target genes on the estrogenic action. Thus, targeting HNF-1β-mediated signaling may serve as a novel therapeutic strategy for the treatment of CCC. Certain inhibitors are currently being studied as new drugs able to restore ER-α protein expression in ER-α-negative cancer cells and to promote apoptosis and differentiation. Demethylating agents and HDAC inhibitors are candidates for becoming potent new drugs in cancer therapy.