Background

Ovarian cancer has the highest mortality among gynecological cancers and is associated with 4.2 % of all cancer-related deaths in women [1]. The four major histological subtypes include serous adenocarcinoma (SAC), clear cell carcinoma (CCC), endometrioid adenocarcinoma, and mucinous adenocarcinoma. Although the underlying reason remains unknown, CCC prevalence varies with race, with an estimated prevalence of 1–12 % in Europe and North America [2] and 15–25 % in Japan. Thus, CCC is the second most common histological subtype of epithelial ovarian cancer (EOC) in Japan [2, 3]. Over the past decade, advances in surgical and chemotherapeutic management have improved progression-free survival (PFS) and overall survival (OS) rates.

Platinum and taxane agents are typically included in standard intravenous regimens administered to women requiring first-line chemotherapy for ovarian cancer [4], and high response rates (60–80 %) have been shown with them. Although these chemotherapies have improved PFS and OS in ovarian cancer, some histological subtypes have shown low response rates. Moreover, standard chemotherapy using paclitaxel and carboplatin has exhibited an approximately 70 % response rate in the treatment of SAC, the most common ovarian carcinoma subtype, but only 18–56 % for CCC [3, 5, 6].

CCC tumors tend to present at earlier stages, with 47–81 % diagnosed at stage I or II, and showing a similar prognosis to SACs [7]. However, advanced CCC (i.e., FIGO stage III or IV) has a poorer prognosis, and available treatments are less effective owing to its resistance toward chemotherapeutic agents. Accordingly, in comparison with advanced SAC, the clinical prognosis for advanced CCC is remarkably inferior, primarily because of its chemoresistant behavior [8–10].

As an alternative to platinum drugs, irinotecan has been shown to be a promising candidate for the treatment of CCC in retrospective studies [11, 12] and a randomized phase II trial [13]. However, the combination therapy of irinotecan plus cisplatin (CPT-P) failed to show efficacy. A recent randomized phase III trial of paclitaxel plus carboplatin (TC) therapy versus irinotecan plus cisplatin (CPT-P) was conducted by the Japanese Gynecologic Oncology Group (JGOG 3017). This trial was the first CCC-specific international clinical trial. With a 44.3-month median follow-up, the 2-year PFS was 73.0 % in the CPT-P arm and 77.6 % in the TC arm. The 2-year OS was 85.5 % in the CPT-P arm and 87.4 % in the TC arm. That is, there were no significant changes in PFS and OS at 2 years between the two groups.

Based on these findings, CCC is considered a highly malignant and chemoresistant type of ovarian cancer, and conventional chemotherapy is not regarded as an effective treatment. In this review, we focus on potential therapeutic molecular targets and discuss prospective treatments.

Review

Molecular mechanisms of platinum resistance in ovarian CCC

Several mechanisms involved in platinum resistance have been proposed, including pre-, on-, post-, and off-target mechanisms as well as the speed of cell proliferation [14]. Our consideration of platinum resistance in CCC is shown in Fig. 1 and summarized in Table 1 [15–53].

Fig. 1
figure 1

Mechanisms of platinum resistance in CCC. At present, five mechanisms of platinum resistance have been characterized in CCC, including pre-, on-, post-, and off-target mechanisms, and slow cell proliferation. ABCC3 ATP-binding cassette, subfamily C, member 3, ERCC1 excision repair cross-complementing rodent repair deficiency complementation group 1, EGFR epidermal growth factor receptor, HER2 human epidermal growth factor receptor 2

Full size image
Table 1 Summary of potential targets for platinum resistance in CCC
Full size table

Pre-target mechanisms include at least two mechanisms using which cancer cells elude the cytotoxic potential of cisplatin before binding to cytoplasmic targets and DNA: (1) a reduced intracellular accumulation of cisplatin and (2) an increased sequestration of cisplatin by glutathione, metallothioneins, and other cytoplasmic scavengers with nucleophilic properties [14].

Previous studies have shown that the expression of the ABCC3 gene is significantly greater in CCC than in SAC. Increased expression of ABCC3 may, at least in part, be associated with the chemoresistant phenotype of CCC [19]. Moreover, AnxA4 overexpression reportedly stimulates efflux of platinum drugs and induces platinum resistance [20, 54, 55]. Increased sequestration of platinum agents in CCC has also been reported, with significantly increased glutathione concentrations in cell lines after cisplatin exposure [56]. Furthermore, a study of gene expression showed that glutathione peroxidase 3, glutaredoxin, and superoxide dismutase 2 were highly expressed in CCC tumors and that elevated levels of these proteins may render the tumors more resistant to chemotherapy [57].

The on-target mechanism involves repair of adducts at an increased pace and/or the ability to tolerate unrepaired DNA lesions, reflecting activity of a particular class of DNA polymerases [14]. The majority of cisplatin lesions are removed from DNA by the nucleotide excision repair (NER) system [14, 58]. One study revealed higher mRNA expression of ERCC1 and XPB genes in CCC. These genes are involved in the NER pathway of EOC and are more prevalent in CCC than in other histological subtypes of EOC [35]. This phenomenon may be related to de novo drug resistance against chemotherapeutic agents in CCC.

Post-target resistance to cisplatin may follow several alterations, including defects in signal transduction pathways and issues with cell death machinery [14]. In this mechanism, galectin-3 is associated with CCC platinum resistance, and suppression of galectin-3 reportedly leads to cis-diamminedichloroplatinum-induced apoptosis via decreases in p27 protein expression [42].

Regarding the off-target mechanism, accumulating evidence suggests that the cisplatin-resistant phenotype can also be maintained by alterations in signaling pathways that are not directly engaged by cisplatin, and yet, these compensate for cisplatin-induced lethal signals [14]. Epidermal growth factor receptor (EGFR) and human epidermal growth factor receptor 2 (HER2) are cell-surface receptor tyrosine kinases that are capable of activating both mitogen-activated protein kinase and phosphatidylinositol 3-kinase (PI3K)/AKT signaling pathways. This leads to phosphorylation of BAD and Bcl-2 and results in the inhibition of chemotherapy-induced apoptosis. An immunohistochemical study reported that the expression of EGFR was detected in 61 % of CCC tumors [46]. Furthermore, HER2 was reportedly overexpressed in CCC in comparison with other major histological subtypes of EOC. In ovarian cancer, the HER2 protein was overexpressed as a consequence of gene amplification in 20–25 % of cases and predicted poor prognosis [59, 60]. Moreover, HER2 was overexpressed in 42.9 % of CCC cases, as investigated using immunohistochemistry (IHC) [61].

Slow cell proliferation is also associated with platinum resistance because cytotoxic drugs primarily target proliferating cells [5], and doubling times for CCC cells were significantly longer than those for SAC cells (61.4 vs. 29.8 h) [5].

Determining the mechanisms underlying platinum resistance in CCC is important, because no novel drugs have yet proven effective for CCC treatment. Our review revealed that annexin A4 (AnxA4) is one of the most well-investigated platinum resistance factors in CCC [20, 22–25]. A recent study showed that AnxA4 knockout improved platinum resistance of CCC in vitro and in vivo [56], and the functional site of AnxA4 that is responsible for conferring platinum resistance has been identified. If an AnxA4 blockade drug was developed, its use in combination with platinum drugs could have therapeutic activity against CCC.

Characteristics of ovarian CCC

The molecular features of CCC are summarized in Table 2 [22, 24, 46, 52, 53, 61–72]. One major distinguishing characteristic is its higher incidence among Asian populations, particularly among Japanese women [2, 3, 56]. The reason for this is unknown, although CCC has been associated with endometriosis and endometriosis-associated ovarian cancers in 22–70 % of younger female patients [73]. Previous studies showed that ovarian endometrioma increases the risk for ovarian cancer, and 0.72 % of all cases of ovarian endometrioma later develop neoplasms [74].

Table 2 Characteristics of ovarian clear cell carcinomas and potential molecular targets
Full size table

CCC tends to present at significantly earlier stages than other ovarian cancers, possibly owing to slow tumor growth and frequent presentation of tumors as large pelvic masses [75], and the proportion of stage I/II tumors ranges from 59 to 71 % [3].

Unlike high-grade serous EOCs, CCCs usually display a wild-type p53 and have a lower frequency of BRCA1 and BRCA2 mutations [2, 76, 77]. Significant differences were reported in the distribution of mutations among histological subtypes, and TP53 mutations were reportedly present in 67 and 21 % of cases with serous and nonserous cancers, respectively [76]. Similar studies report much lower frequencies of p53 mutations (approximately 15 %) in CCC than in other EOC types [2]. Similarly, Alsop et al. reported that the frequency of BRCA1 and BRCA2 mutations of CCC was 6.3 and 0 %, respectively [77]. However, other recent studies have revealed that several genes/proteins are mutated and/or overexpressed in CCC, and that these proteins may serve as therapeutic targets for CCC (Table 2) [22, 24, 46, 52, 53, 61–72].

Novel therapeutic modalities for CCC

Paclitaxel plus carboplatin combination therapy is currently the primary treatment strategy in postoperative chemotherapy. However, advanced-stage patients eventually relapse after adjuvant therapy and have a high risk of recurrence [3, 5, 6]. The mechanism underlying resistance to standard chemotherapy has been studied but remains unknown. Nonetheless, novel drugs that target specific molecular pathways are being developed to improve the outcomes of chemotherapy-resistant ovarian cancer. Moreover, studies of chemotherapy-resistant ovarian cancer therapy indicate that effective treatments for CCC are available. However, clinical data relating directly to the treatment of CCC subtypes are limited.

Clinical trials have shown that in combination with chemotherapy, targeting vascular endothelial growth factor (VEGF), VEGF receptor (VEGFR), mammalian target of rapamycin (mTOR), EGFR, and poly (ADP-ribose) polymerase elicits positive results for EOC (Table 3) [52, 61, 71, 78–83]. However, limited data are available pertaining to novel therapies for either EOC or CCC, and further studies of novel treatment strategies are required to focus on the clinical features of CCC.

Table 3 Examples of targeted molecular cancer therapeutics for epithelial ovarian cancer
Full size table

Targeting VEGF

VEGF is an important therapeutic target in several solid tumors, including ovarian cancers, and the monoclonal antibody bevacizumab has been shown to bind to VEGF, inhibit receptor binding, and prevent the growth of tumor vasculature. Accordingly, the International Collaborative Ovarian Neoplasm Group (ICON7) and the Gynecologic Oncology Group (GOG) trials (GOG218) of bevacizumab addition to standard chemotherapy in newly diagnosed advanced ovarian cancer both reported significant improvements in PFS. The ICON7 trial specifically reported increased OS in a predefined group of patients with a high risk of disease progression [84, 85]. In addition, efficacy and safety of bevacizumab has been reported both in patients with platinum-sensitive and in those with platinum-resistant recurrent ovarian cancers [84, 85]. However, in these studies, histological subgroup analyses were not performed, and the clinical utility of VEGF as a therapeutic target for CCC has not been evaluated. Mabuchi et al. demonstrated the efficacy of bevacizumab in in vitro and in vivo CCC models, which showed that VEGF is frequently expressed and may be a promising therapeutic target for the management of CCC [66]. However, a clinical trial has not been performed. Sunitinib is another possible therapeutic option for renal cell carcinoma treatment and acts as an oral, small-molecule, multitargeted receptor tyrosine kinase inhibitor (targeting VEGFR, platelet-derived growth factor receptor, and c-Kit). However, only a few small clinical studies have reported the efficacy of sunitinib for ovarian CCC [86].

Targeting the PI3K/AKT/mTOR signaling pathway

PI3Ks are lipid kinases that regulate signaling pathways that are vital for neoplasia, including cell proliferation, adhesion, survival, and motility. The frequency of PIK3CA mutations in CCC has been estimated to be 40 %, and some studies suggest that the PI3K/AKT/mTOR pathway is a target with therapeutic potential [53]. Moreover, immunohistochemical analyses have shown that mTOR is frequently activated in CCC (86.6 %) and that mTOR inhibition by RAD001 may be an effective treatment for CCC [63]. Furthermore, Rahman et al. reported that PIK3CA mutations were associated with more favorable prognoses but did not predict the sensitivity of ovarian CCC cells to PI3K/AKT/mTOR inhibitors [87]. Since more than 80 % of ovarian CCC shows activation of the AKT/mTOR pathway, it is of great interest to explore the potential of mTOR inhibitors [88]. A very important GOG clinical trial is currently being conducted. The GOG-268 trial is an open-label phase II trial for newly diagnosed stage III and IV ovarian CCC to examine the activity of one of the mTOR inhibitors, temsirolimus. The primary endpoint of this trial is PFS at 12 months, and secondary endpoints include adverse events, duration of PFS and OS, and tumor response. IHC expression of components of the mTOR signaling pathway will be explored. Temsirolimus will be administered in combination with paclitaxel and carboplatin for six cycles. For the maintenance phase, temsirolimus will be administered on days 1, 8, and 15 every 3 weeks for 11 cycles. This clinical trial closed, and clinicians are awaiting the results with high expectations. Although targeting the PI3K/AKT/mTOR signaling pathway is promising, some problems remain, and there is no evidence of effective clinical management of CCCs, warranting further studies and clinical trials to prove the efficacy of PI3K/AKT/mTOR inhibition.

Targeting AnxA4

AnxA4 is reportedly involved in exocytosis and regulation of epithelial Cl− secretion [20], and its overexpression in CCC has been shown to induce platinum resistance [22]. Accordingly, IHC analyses of AnxA4 in CCC samples showed strong staining in 30 of 43 samples but moderate staining in the remaining 13 samples [22].

Nonetheless, enhanced AnxA4 expression was recently shown to increase chemoresistance to carboplatin by contributing to extracellular efflux of the drug [22]. Recently, we demonstrated molecular mechanisms underlying AnxA4-induced promotion of platinum drug efflux [54]. Exposure of an AnxA4-overexpressing endometrial carcinoma cell line to platinum drugs caused relocalization of AnxA4 from the cytoplasm to the membrane fraction, and colocalization of P-type ATPase ATP7A (a copper and platinum transporter) to cell membranes. This colocalization promoted platinum drug efflux via ATP7A and induced the platinum resistance [54].

Several studies have shown that AnxA4 induces drug resistance [20], and suppression of AnxA4 expression improved platinum sensitivity of CCC in vitro and in vivo [55]. In addition, Morimoto et al. showed that annexin repeat domains and calcium-binding sites of repeated annexin sequences are required for resistance to platinum-based drugs [55]. The structure of AnxA4 and the mechanism of platinum resistance induced by AnxA4 is shown in Fig. 2. Taken together, these reports suggest the potential of AnxA4 targets for the treatment of ovarian CCC. However, no drugs have been shown to suppress AnxA4 expression. Nonetheless, in a study of the related annexin A2 (AnxA2) using chick chorioallantoic membrane assays, neutralizing antibodies significantly inhibited OV-90 cell motility and invasion in vitro and in vivo, suggesting the potential of AnxA2-neutralizing antibodies as therapeutic targets for AnxA2-overexpressing cancers [89]. Similarly, AnxA4 blockade using neutralizing antibodies might limit the platinum resistance of CCC and is currently under investigation by our research group. Potentially, drugs that inhibit the function of AnxA4 in combination with platinum drugs may offer promising therapies for the treatment of CCC.

Fig. 2
figure 2

The structure of annexin A4 (AnxA4) and related mechanisms of platinum resistance. a No treatment. AnxA4 is localized in the cytoplasm before exposure to platinum drugs. b After platinum drug exposure. Exposure to platinum drugs leads to AnxA4 relocalization from the cytoplasm to the cell membrane. AnxA4 has four annexin repeats that are packed into an α-helical disk within the C-terminal polypeptide core, which contains Ca2+-binding sites. These Ca2+-binding sites are involved in platinum resistance. c After platinum drug exposure. AnxA4 is attached to the membrane surface through bound Ca2+ ions and contributes to the efflux of platinum drugs via platinum transporters

Full size image

Targeting the ARID1A gene

The AT-rich interactive domain 1A (SWI-like) gene (ARID1A) encodes BAF250A, which is a member of the SWI/SNF ATP-dependent chromatin remodeling complex. ARID1A plays an indispensable role in controlling gene expression and in tissue development and cellular differentiation [70]. Moreover, CCCs reportedly have the highest frequency of ARID1A mutations (43–57 %) [62].

Yamamoto et al. reported two critical associations between the AT-rich interactive domain-containing protein 1A (ARID1A) and CCC. In particular, deficiencies of ARID1A immunoreactions were evident at the stage of precursor lesions that lacked atypical cytology, indicating that the loss of ARID1A protein may occur as an early event in tumorigenesis. Moreover, loss of ARID1A protein expression is often coincident (not mutually exclusive) with PIK3CA mutation [90]. However, a previous study demonstrated that inactivation of ARID1A alone is insufficient for tumor initiation, suggesting that additional genetic alterations are required to drive tumorigenesis [91]. Chandler et al. also showed that the coexistent ARID1A-PIK3CA mutations promote ovarian clear cell tumorigenesis [92]. Taken together, these studies suggest that ARID1A is related to CCC tumorigenesis, although the precise mechanisms remain unknown, and no drugs have yet been shown to target ARID1A. A recent study by Bitler et al. is an important study regarding the targeting of ARID1A for ovarian CCC [93]. The study showed that ARID1A is mutated in over 50 % of ovarian CCCs and pharmacological inhibition of enhancer of zeste homolog 2 (EZH2) represents a novel treatment strategy for cancers involving ARID1A mutations. The study showed that EZH2 inhibitor selectively suppressed the growth of ARID1A mutated cells in vitro and in vivo via upregulating the expression of PIK3IP1, which negatively regulates PI3K/AKT signals. Further studies are expected to elucidate the detailed mechanisms of the cellular dysfunction caused by ARID1A mutations in CCCs and to expose the potential of ARID1A as a therapeutic target.

Targeting hepatocyte nuclear factor-1β

Hepatocyte nuclear factor-1β (HNF-1β) is a transcription activator that regulates the promoters and enhancers of genes expressed in the liver, digestive tract, pancreas, and kidneys [64]. Recent studies have reported specific expression of HNF-1β in endometriosis and CCC and suggest that early differentiation into the clear cell lineage occurs in endometriosis [64, 74]. However, the role of HNF-1β expression in ovarian clear cell tumors and endometriosis remains uncertain. Nonetheless, RNA interference has been used to decrease HNF-1β expression and reportedly led to apoptotic cell death in CCC cell lines, indicating that HNF-1β expression may be tightly linked to CCC and that it could be essential for its survival [74]. In addition, HNF-1β is reportedly expressed in almost all CCC cases [64].

Accordingly, Kajihara et al. concluded that the HNF-1β-dependent pathway may provide novel insights into the regulation of glycogen synthesis, detoxification, and resistance to anticancer agents [94]. In support of these conclusions, HNF-1β directly regulates multiple cancer-related genes, including those for dipeptidyl peptidase IV, osteopontin, tissue factor pathway inhibitor 2, AnxA4, and angiotensin-converting enzyme 2 [74, 94, 95]. Genes that are upregulated in CCC are likely direct targets of HNF-1β. However, drugs that target HNF-1β have not been developed, warranting further studies of the mechanisms by which HNF-1β regulates various genes and its association with CCC.

Targeting ZNF217

The ZNF217 gene on human 20q13.2 encodes a transcription factor that is overexpressed in 30 % of breast tumors and in several cell lines [96]. Several studies show that overexpression of ZNF217 in several cancers is associated with poor prognosis [96, 97]. Among these, Littlepage et al. reported that ZNF217 overexpression promotes metastasis and resistance to chemotherapy and inhibits signaling events in vivo [96]. These authors also showed that triciribine inhibits the growth of ZNF217-overexpressing cells in vitro and in vivo, indicating that it is a potential target for the treatment of ZNF217-overexpressing cancers.

In a previous study, ZNF217 overexpression was reported in 20.0 % of CCC cases [65]. Moreover, Rahman et al. showed that ZNF217 gene overexpression is significantly correlated with lymph node metastasis in ovarian CCC. In comparison with small interfering RNA-treated cells without ZNF217 overexpression, profound inhibition of cell migration and invasion was observed in cells overexpressing ZNF217 [98], suggesting that ZNF217 is a potential therapeutic target for CCC.

Conclusion

As discussed above, the loss of ARID1A expression and/or PI3K activation is crucial for CCC tumorigenesis. Moreover, synergic effects of the loss of ARID1A expression and PI3K/AKT pathway activation and ZNF217 overexpression may be related to ovarian CCC development [99], warranting further studies of these associations and assessments of their potential as co-therapeutic targets for CCC.

CCC is highly resistant to current platinum-based treatment. However, if an AnxA4 blockade drug was developed, its use in combination with platinum drugs may have therapeutic activity against CCC.