Fig. 7.1
Four histopathological subtypes of HGSOC. MT mesenchymal transition, IR immune reactive, SP solid and proliferative, PG papilloglandular. Reused from [4] with permission
In 2014, two interesting data were presented at the American Society of Clinical Oncology meeting analyzing relationship between the HGSOC gene expression subtypes and the bevacizumab (Bev) sensitivity. Gourley et al. performed mRNA expression microarray analysis of 283 HGSOC cases registered for the ICON7 trial [5], which demonstrated improved progression-free survival by addition of Bev to paclitaxel-carboplatin regimen in ovarian cancer [6]. Surprisingly, they found that addition of Bev was significantly associated with poor prognosis in tumors with elevated expression of immune-related genes. Winterhoff et al. also performed mRNA expression microarray analysis of 359 ovarian cancer cases registered for the ICON7 trial [7]. They divided the cases into four gene expression subtypes like the TCGA cases and found that addition of Bev improved survival in proliferative and mesenchymal subtypes. These studies suggest gene expression profile could serve a biomarker to select HGSOC cases Bev should be used for.
7.1.3 Molecular Mechanisms of Chemoresistance in HGSOC
Patch et al. analyzed whole-genome sequencing, transcriptome, DNA methylation, copy number alterations, and microRNA expression of 92 cases of HGSOC by focusing on the mechanism of chemoresistance [8]. They found gene breakage frequently occurs in tumor suppressor genes including RB1, NF1, RAD51B, and PTEN. CCNE1 amplification was common in primary resistant and refractory cases. On the other hand, homologous recombination pathway-deficient cases, having extensive copy number alterations and increased single nucleotide variants, were sensitive to chemotherapy. Mechanism of acquired resistance included gene breakage of tumor suppressor genes, reversion mutation of BRCA1/2 mutated cases, and upregulation of BRCA1 gene expression by demethylation of the methylated BRCA1 promoter region in a primary tumor. Additionally, gene fusion of ABCB1 with SLC25A40 promoter caused upregulation of ABCB1 expression, which can cause increased excretion of chemotherapeutic agents.
7.1.4 PARP Inhibitors
Among the different subtypes of ovarian cancers, the highest rates of germline BRCA1 and BRCA2 mutations (8–18%) occur in HGSOC [9, 10]. Germline BRCA mutation-associated ovarian cancers have a relatively distinct clinical behavior characterized by an earlier age at diagnosis, improved survival, higher response rates to platinum, and sensitivity to poly(ADP-ribose) polymerase (PARP) inhibitors [11]. Single-strand DNA breaks, repaired by the mechanism PARP is involved in, are not repaired until DNA replication and generation of double-strand DNA breaks in the presence of a PARP inhibitor. Therefore, following the concept of “synthetic lethality,” PARP inhibitors cause apoptosis of homologous recombination (HR)-deficient tumor cells. The clinical proof of this concept in ovarian cancer was first shown in a phase I study of the PARP inhibitor, olaparib [12], and subsequently supported in a phase II trial of olaparib in recurrent ovarian cancer patients with germline BRCA1 or BRCA2 mutation [13].
Ovarian cancers with somatic mutations of BRCA1/2 are observed in about one-fifth of germline mutation cases [11]. Ovarian cancer patients with somatic BRCA1 or BRCA2 mutations probably benefit from treatment with PARP inhibitors like those with germline BRCA mutations. The European Medicines Agency approves olaparib as maintenance therapy in platinum-sensitive ovarian cancer including all patients with a germline or somatic BRCA mutation. The number of patients with somatic BRCA mutations analyzed so far is relatively low. Further clinical trials will clarify the clinical significance of somatic BRCA mutations as biomarkers of PARP inhibitors.
It became increasingly apparent that a proportion of ovarian cancers without BRCA mutation also share clinical features of BRCA mutation-associated cases, including platinum sensitivity. This concept, termed as “BRCAness,” now indicates a situation where HR DNA repair deficiency is present, but no BRCA1 or BRCA2 mutation is detected. Importantly, PARP inhibitors show activity in clinical trials for ovarian cancer without BRCA mutations [14, 15]. Several groups have studied to find markers of the “BRCAness” by using gene expression profiling, proteomics, or genomic instability scores [11]. Additionally, as the accumulation of RAD51 at the DNA lesion is a marker of HR proficiency, its absence following DNA damage can be a functional biomarker of HR deficiency [16]. Actually, detection of RAD51 foci predicted response to chemotherapy and PARP inhibition [17–19]. The quantification of phosphorylated gamma-H2AX has been used to measure the amount of DNA damage [20]. More recently, SNP array-based signatures of chromosome instability have been reported as biomarkers of HR deficiency and sensitivity to PARP inhibitors [21].
7.2 Ovarian Clear Cell Carcinoma (OCCC)
7.2.1 Gene Expression of OCCC Closely Related to Oxidative Stress
Ovarian clear cell carcinoma (OCCC) is a chemoresistant subtype. Although it is rare in Western countries, it is increasing in Japan. It is often associated with an endometriotic cyst. We analyzed the content of endometriotic cysts and revealed it contained huge amount of iron, oxidative stress marker LPO, and 8-0HdG, a marker of DNA damage caused by oxidative stress. When the content of endometriotic cyst or iron was added to immortalized ovarian surface epithelial cells, intracellular reactive oxygen species (ROS) was elevated. Then the content of endometriotic cyst or iron increased DNA mutations. Therefore, we hypothesized iron-mediated reactive oxygen species may cause DNA mutations and carcinogenesis [22].
Next we conducted a gene expression microarray analysis and identified clear cell-specific genes, termed “OCCC signature.” These genes contained HNF1B, SOD2, HIF1A, IL6, and STAT3. They enriched gene ontology terms related to oxidative stress and glucose metabolism. Many of these genes had HNF1-binding motif in their promoter regions, suggesting that many are downstream genes of HNF1B. Interestingly, the OCCC signature genes were upregulated in immortalized ovarian surface epithelial cells by adding the content of endometriotic cyst or iron [23]. A methylation DNA microarray analysis revealed OCCC is distinct from other subtypes in terms of the methylation profile. ER pathway genes were hypermethylated and downregulated, while HNF-1 pathway genes were hypomethylated and upregulated [24]. Therefore, the OCCC-specific gene expression seems to be stabilized via the epigenetic mechanism.
7.2.2 Roles of HNF1B in OCCC
We investigated the roles of HNF1B in metabolism of OCCC cells. We found HNF1B increases glucose uptake by increasing GLUT1 expression, a major glucose transporter [25]. We conducted a metabolome analysis and found that the upregulated HNF1B expression enhances anaerobic glucose metabolism, that is, Warburg effect, which is known to cause resistance to oxidative stress. We further analyzed the relationship between HNF1B and oxidative stress in clear cell carcinoma. Knockdown of HNF1B decreased the amount of glutathione, a redox substance. This was due to the decreased intracellular cystine, a substrate for the biosynthesis of glutathione, via the decreased expression of rBAT, a cystine transporter. Then, we found HNF1B knockdown increased intracellular ROS and cytotoxicity by iron-induced oxidative stress. Furthermore, in hypoxia, suppression of HNF1B increased sensitivity to cisplatin. Collectively, HNFB in clear cell carcinoma causes resistance to oxidative stress and platinum [26].
It is known that a germline mutation of HNF1B causes hereditary diabetes mellitus and renal hypoplasia. We found clear cell carcinoma is very similar to kidney cancer through the expression of HNF1B and its target genes. As sorafenib is effective for kidney cancer, we treated OCCC on nude mice by sorafenib and observed a prominent effect [27]. Then we treated two chemoresistant ovarian clear cell carcinoma patients by sorafenib and observed antitumor effect [28]. Therefore, the similarity of ovarian clear cell carcinoma with kidney cancer implies the efficacy of sorafenib for ovarian clear cell carcinoma.
7.2.3 Genetic Analyses of OCCC
By an analysis of exome sequences of eight OCCC tumors, Jones et al. identified four genes that were mutated in at least two tumors; PIK3CA, KRAS, PPP2R1A, and ARID1A. Out of 42 OCCCs, 57% had mutations in ARID1A [29]. In another study, ARID1A mutations were seen in 55 of 119 OCCCs (46%), 10 of 33 endometrioid carcinomas (30%), and none of the 76 HGSOCs [30]. Out of 97 OCCCs, mutations of PIK3CA, TP53, KRAS, PTEN, CTNNB1, and BRAF occurred in 33%, 15%, 7%, 5%, 3%, and 1% of samples, respectively [31]. Consistently, by an exome sequencing analysis of 39 OCCCs, we recently found ARID1A was the top mutated gene and PIK3CA was the second one (paper in preparation). The integrated analysis of gene mutations and copy number variations revealed KRAS-PI3K pathway, SWI/SNF complex, and MYC-RB pathway were the most frequently altered pathways.
7.2.4 Analyses of Precursor Lesions of OCCC
ARID1A protein expression was analyzed in endometriosis-associated OCCCs (n = 28) and clear cell adenofibroma-associated OCCCs (n = 14). Among the precursor lesions adjacent to the 23 ARID1A-deficient carcinomas, 86% of the nonatypical endometriosis (12 of 14) and 100% of the atypical endometriosis (14 of 14), benign (3 of 3), and borderline (6 of 6) clear cell adenofibroma components were ARID1A deficient. In contrast, in the 19 patients with ARID1A-intact carcinomas, all of the adjacent precursor lesions were ARID1A positive [32]. In an analysis of 23 endometriosis-associated OCCCs, PIK3CA gene mutations were detected in 10/23 (43%) carcinomas. The identical mutations were detected in the adjacent endometriotic epithelium in nine of ten (90%) cases [33]. Using whole-genome sequencing of seven endometriosis-associated OCCCs, ARID1A and PIK3CA mutations were found in concurrent endometriosis regardless of any cytological atypia when present in the OCCC [34]. Collectively, these data indicate ARID1A and PIK3CA mutations are early event in the carcinogenic process of OCCC, which mutations are usually found in endometriotic lesions adjacent to OCCCs.
Recently, it was reported that Pik3ca and Arid1a mutations in the ovaries generate clear cell carcinoma in mice [35]. This tumor highly expressed Hnf1b. We hypothesize that iron-induced oxidative stress in endometriotic lesions may cause DNA damage, causing mutations of PIK3CA and ARID1A, which may lead to carcinogenesis of ovarian clear cell carcinoma. HNF1B plays important roles in Warburg effect and resistance to oxidative stress. This may be important for the progression of OCCC in the stressful condition of endometriotic cysts and for the development of platinum resistance. Further epigenetic changes, gene mutations, and copy number alterations may cause stabilization of OCCC-specific gene expression and biological features including chemoresistance.
7.3 Ovarian Endometrioid Carcinoma (OEC)
7.3.1 Genetic Analysis of OEC
Wu et al. analyzed gene mutations in OEC samples with different grades (grade 1; n = 20, grade 2; n = 26, grade 3; n = 26) and found mutations in CTNNB1 (13, 5, 0%), APC (5, 0, 0%), KRAS (10, 12, 0%), PTEN (20, 8, 0%), PIK3CA (20, 8, 0%), and TP53 (15, 46, 65%), respectively. Therefore, high-grade OECs are likely to harbor TP53 mutations, while low-grade OECs frequently harbor mutations of Wnt/β-catenin pathway and/or KRAS/PI3K pathway genes. Additionally, inactivation of the Pten and Apc in murine ovaries resulted in the formation of adenocarcinomas morphologically and biologically similar to human OECs [36]. More recently, ARID1A mutations were reported in 10 of 33 EOCs (30%) [30]. Consistently, another group reported mutations of CTNNB1 (53%), PIK3CA (40%), ARID1A (30%), PTEN (17%), KRAS (33%), PPP2R1A (17%), and TP53 (7%) in low-grade (grade 1 and 2) OECs (n = 30) [37].
7.3.2 Genetic Analysis of Synchronous Endometrial and Ovarian Carcinoma
Five to ten percentage of women with OECs present with concurrent endometrial carcinoma. Based on both targeted and exome sequencing of 18 synchronous endometrial and ovarian tumors, most (17/18) cases showed evidence of clonality. Importantly, 10 of 11 cases that fulfilled clinicopathological criteria that would lead to classification as independent endometrial and ovarian primary carcinomas showed evidence of clonality [38]. Therefore, the genome-wide analysis demonstrated that most synchronous endometrial and ovarian carcinoma tumors develop from a clonal origin.
7.4 Mucinous Ovarian Tumors
7.4.1 Genetic Analysis of Mucinous Ovarian Tumors
Ryland et al. performed genetic analysis of a total of 82 mucinous ovarian tumors, which included exome sequencing of 24 tumors and a validation cohort of benign 58 tumors for specific gene regions. Benign, borderline, and carcinoma samples harbored mutations in BRAF (0, 10, 23%), TP53 (9, 14, 52%), and RNF43 (0, 7, 20%), respectively, in which mutations were associated with progression of the disease. Other recurrent, but not associated with progression, mutations were found in KRAS (54%), CDKN2A (16%), ARID1A (8%), ELF3 (6%), GNAS (6%), ERBB3 (5%), and KLF5 (5%) [39]. In another study, RNF43 mutations were observed with a frequency of 2/22 (9%) in mucinous ovarian borderline tumors and 6/29 (21%) in mucinous ovarian carcinomas [40].
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
