Among ovarian carcinomas, clear cell adenocarcinoma (CCA) has been recognized as a lethal histological subtype, mainly because of its highly chemoresistant nature 1–4. Despite progress in surgical techniques and modalities of chemotherapy for ovarian cancer, the mortality rate of patients with CCA remains largely unchanged. Identification of the molecular mechanisms that occur early in tumour development is important for the understanding of tumourigenesis, guidance of proper management and development of new therapeutic strategies.
Phosphatidylinositol-3 kinases (PI3K) are a large and complex family of lipid kinases and play important roles in cellular growth, proliferation, motility, differentiation and angiogenesis 5, 6. PI3Ks are activated in response to signals transmitted from growth factor receptor tyrosine kinases, and increased PI3K activity has been demonstrated in a number of human cancers 7, 8. The PIK3CA gene, encoding the catalytic subunit p110a of PI3K, is located on chromosome 3q26.3 and amplification of this locus has been shown to increase PI3K activity 9. Recently, somatic mutations in the PIK3CA gene have been reported in several human cancer types, including cancers of the colon, brain, stomach, liver, breast and ovary 10–15. Functional analysis revealed that several ‘hotspot’ mutants of p110a in PIK3CA, ie exon 9 (residues E542, E545 and Q546 of the helical domain) and exon 20 (residue H1047 of the kinase domain), elevate its lipid kinase activity and lead to the activation of the downstream Akt signalling pathway 10, 16.
Although little is known about the pathogenesis of CCA, amongst the most significant molecular genetic changes that have been described in CCA at present are the frequent somatic activating mutations of PIK3CA, which were detected in 46% of purified CCA cells from clinical samples or cell lines and in 25–32% of paraffin-embedded tumour samples 17. On the other hand, the frequencies of PIK3CA mutations are relatively low in the other histological subtypes of ovarian cancer 17.
In this study, we confined our analysis to the CCAs with synchronous endometriosis; the latter has been considered a putative precursor lesion of the former 18–23. This allowed us to delineate the early molecular genetic events in their pathogenesis. Specifically, we compared the mutational status of PIK3CA in the CCA with that of the adjacent epithelium of endometriosis. In ovarian carcinomas, > 80% of mutations in PIK3CA have been found in the helical (exon 9) and kinase domains (exon 20) 10, 12, 15, 17; therefore, we restricted our analyses to these two regions of the gene.
Materials and methods
A consecutive series of 79 primary ovarian CCAs were identified from the files of the Department of Clinical Laboratory, National Defence Medical College Hospital, Japan. These 79 cases with CCA had been surgically resected between 1986 and 2007 and the patients had not undergone chemotherapy or radiation therapy before surgery. All pathology specimens were formalin-fixed and paraffin-embedded, and haematoxylin and eosin (H&E)-stained sections were reviewed in our institution. Tumours were classified according to the criteria of the World Health Organization 1. The research protocol was approved by the ethics committee of the National Defence Medical College, Tokorozawa, Japan.
Criteria for endometriosis synchronous with CCA and for atypical endometriosis
Endometriosis synchronous with CCA was defined as: (a) endometriosis existing in histological continuity with, or adjacent to, the CCA; or (2) an endometriotic cyst in which CCA was observed, regardless of the presence of histological continuity between the carcinoma and the endometriotic epithelium 20, 21.
Histological criteria for ‘atypical endometriosis’ were described previously 19–21. Cytological atypia was determined as present if at least one of the following features was histologically evident in the endometriotic epithelium: large hyperchromatic or pale nuclei with moderate to marked pleomorphism; an increased nuclear : cytoplasmic ratio; or cellular crowding with stratification or tufting—but these findings were not so high as those seen in the carcinoma cells. When the epithelial cells lining an endometriotic cyst showed cytological features corresponding to the adjacent CCA cells, such as high cytological atypia and obvious ‘clear-cell’ cytoplasm, these epithelial cells were regarded as cancer cells growing along the cyst wall and were excluded from the atypical endometriotic lesions.
Consequently, endometrioses synchronous with CCA were identified in 23 (29%) of the 79 cases, and these 23 endometriosis-associated carcinomas were subjected to the following genetic analyses. Clinical staging of disease was done according to the International Federation of Gynecology and Obstetrics system. Of the 23 cases with endometriosis-associated carcinoma, 15 (65%) were stage I, two (9%) were stage II, four (17%) were stage III and two (9%) were stage IV.
Deparaffinized unstained tissue sections (8 µm) from carcinoma cell-rich areas were manually dissected under microscopic guidance. When cancerous tissue contained many non-neoplastic cells (ie inflammatory cells and stromal cells), laser microdissection was performed using a Leica LMD 6000 (Leica; Narishige Micromanipulator, Wetzlar, Germany) as described previously 24. If the carcinoma components were judged to be harbouring a PIK3CA mutation, their matched endometriotic lesions coexisting with CCA were also subjected to mutation analysis, and genomic DNA from corresponding normal tissue was isolated and subjected to genetic analysis to confirm the somatic nature of tumour mutations. For the examined endometriotic lesions, to obtain a specific cell population while minimizing possible contamination of stromal or cancer cells, epithelia of interest were carefully microdissected using the Leica LMD 6000 (Figure 1). The dissected tissues were incubated with 1 × PCR buffer containing 100 µg/ml proteinase K for 3 h at 56 °C. After heat inactivation at 95 °C for 3 min, the solution was directly used as template genomic DNA for the mutation analysis.
Exons 9 and 20 of the PIK3CA gene were amplified using PCR for genomic DNA. PCR was done using AmpliTaq Gold (Applied Biosystems, Foster City, CA, USA) and the primers for PCR and sequencing were as follows:
exon 9: forward primer, 5′-CCAGAGGGGAAAAATA TGACAA-3′
reverse primer, 9, 5′-ACCTGTGACTCCATAG AAA-3′
sequencing primer, 5′-AGACTAGCTAGAGAC AATGAAT-3′
exon 20: forward primer, 5′-TTGATGACATTGCATA CATTCG-3′
reverse primer, 5′-AATTGTGTGGAAGATCC AATCC-3′
sequencing primer, 5′-TGCATACATTCGAAA GACCCTA-3′
Primer pairs flanking PIK3CA exon 9 were selected to avoid the frequent cross-amplification of chromosome 22q (known PIK3CA pseudogene) observed in previous reports 25. The PCR conditions were as follows: 1 cycle of 95 °C for 11 min, 50 cycles of 95 °C for 30 s, 55 °C for 40 s, and 72 °C for 40 s, followed by 1 cycle of 72 °C for 5 min. The PCR products were subsequently subjected to direct sequencing PCR with BigDye terminator v 3.1/1.1 cycle sequencing reagents (Applied Biosystems). The samples were finally analysed on an ABI PRISM 3130 Genetic Analyser (Applied Biosystems) with DNA Sequencing Analysis Software v 5.2 (Applied Biosystems). When mutations were once detected by sequencing analysis, genetic analysis was repeatedly performed at least twice from a step of genomic DNA amplification.
To determine the mutational statuses, sequencing chromatograms for PIK3CA exons 9 and 20 were obtained from all 23 carcinomas studied. Direct sequencing identified mutations in 10 (43%) of the 23 endometriosis-associated CCAs. All of the mutations identified were in exon 20 and were the H1047R substitution; no mutations were found in exon 9 (Table 1). Matching DNA from normal tissue was available for all the mutation-positive carcinomas and showed no mutation. Therefore, in all instances the mutations were shown to be somatic.
Table 1. PIK3CA mutations in ovarian clear cell adenocarcinomas and adjacent endometriosis
Of the 10 cases of endometriosis-associated CCA containing PIK3CA mutation, six cases had both non-atypical and atypical endometrioses (case nos 5, 6, 12, 14, 17 and 18), two had non-atypical endometriosis only (case nos 8 and 20), and two had atypical endometriosis only (case nos 13 and 15) (Table 1). Analysis of the mutational status in the epithelium from these endometriotic lesions revealed that six (75%) of eight non-atypical endometriotic lesions and seven (88%) of eight atypical endometriotic lesions contained the H1047R mutation (Table 1). Representative sequence results are shown in Figure 2. In case no. 8, foci of endometriosis, which were distant from carcinoma components, were also identified in the adnexa ipsilateral to the carcinoma. In this distant endometriosis, PIK3CA mutations were not documented (Figure 2). Regarding the histomorphology or extent of cellular atypia, the mutation-harbouring endometriosis (adjacent to CCA) and the distant one lacking mutation appeared almost similar (see insets to Figure 2).
A hypothesis of multi-step tumourigenesis of ovarian CCA, with histologically benign-appearing precursor lesions (ie endometriosis) progressing to their atypical counterparts (ie atypical endometriosis) and ultimately to CCA, has been proposed and widely accepted 18–23. Given the high frequency of PIK3CA mutations in ovarian CCAs, we investigated the mutational statuses of the putative precursors of PIK3CA mutation-harbouring carcinomas as a means to study their pathogenetics. Consequently, the presence of identical mutations in the endometriotic epithelium, even in those that displayed no evidence of cytological atypia, strongly suggests that mutations of the PIK3CA gene occur before the development of the atypical precancerous lesions (ie atypical endometriosis) and that these non-atypical endometrioses are the true precursors for CCA and are already neoplastic in nature. These results also indicate that mutations of the PIK3CA gene are very early events during the development of endometriosis-associated ovarian CCA, as are KRAS and BRAF mutations occurring in the development of serous borderline tumours of the ovary 26. In two cases, endometriosis did not show the mutation by direct genomic DNA sequencing. Because the method employed can stably detect mutations when > 25% of the constituted cells were mutant, the presence of mutation in a minority of cells in these cases cannot be ruled out, even though careful and accurate enrichment of epithelial cells by laser microdissection was performed.
Endometriosis is a common gynaecological disease with an estimated prevalence of approximately 10% in women of reproductive age 27. On the other hand, the frequency of malignant transformation of endometriosis has been estimated as 0.7–1.6% over an average of 8 years 28–30, suggesting that only a small proportion of endometrioses have the potential to progress into carcinoma. Because mutations of the PIK3CA gene were found to be associated with the development of endometriosis-associated CCAs, detection of the mutations might be able to screen endometrioses with a potential risk of malignant transformation. The current mutation assay for endometriosis was confined to the cases with PIK3CA-mutation-positive carcinomas and not in the whole population. Therefore, it remains unclear as to the prevalence of PIK3CA mutations in cases where the cancer exhibits no mutation, or whether or not the detection of PIK3CA mutations in endometriosis would differentiate endometrioses with a high risk of malignant transformation from most endometrioses with a very low risk of progression; the latter are probably lacking mutations. In future analyses we should examine whether endometriosis found in the mutation-negative carcinoma cases and solitary endometriosis could exhibit PIK3CA mutation, because it has been thought that some of the solitary endometriosis were already in the neoplastic process 31. A previous retrospective study demonstrated, using multivariate analysis, that cystic endometriosis (also called ‘endometrioma’) with a diameter of ≥ 9 cm and postmenopausal status were independent predictive factors for subsequent carcinoma development 32. Therefore, development of novel molecular assays that can detect such mutations, possibly from the endometriotic cyst fluid, could play a significant role in the management of these patients.
To our surprise, all mutations found in this study were in exon 20, whereas no mutations were found in exon 9. This pattern is different from those in previous reports, where the mutations in ovarian CCA were typically found in both exons 9 and 20 17, 33. To exclude the possibility of pseudogene complication by PCR-based analysis, ie detection of the A1634C (E545A) pseudomutation by interferences from the homologue sequence of the PIK3CA pseudogene on chromosome 22q 25, we examined mutations in genomic DNA from histologically unremarkable tissues, such as fallopian tube or lymph nodes, in the same patients. Because such a pseudomutation was not detected in genomic DNA, either from the tumours and endometriotic epithelia or from these unremarkable tissues, we judged that the true PIK3CA gene was assayed and exon 9 mutations in the PIK3CA gene were absent in the present series. The frequency of PIK3CA mutations observed in our study (43%) was similar to, or even higher than, those reported in paraffin-embedded clinical samples of CCA examined by Kuo et al(25% and 32% in different cohorts) 17. These results might be affected to some extent by our specific patient cohort, such as CCAs with synchronous putative precursors, especially endometriosis-associated carcinomas, or by racial differences. Mutations in the kinase domain of exon 20 are close to the hinge region of the catalytic loop, and it has been suggested that they influence its activation 34. In contrast, mutations in the helical domain of exon 9 are clustered on an exposed surface patch of the protein and change its ability to interact with other regulatory proteins 34. Thus, changes affecting the kinase domain (exon 20) might exert changes in different biological or biochemical properties as compared with those occurring in the helical domain (exon 9).
Several investigators have assessed the genetic linkage between endometriosis and coexisting CCAs. Sato et al35 found allelic losses at the PTEN locus on 10q23.3 in 27% of CCAs, and also demonstrated that 43% of CCAs synchronous with endometriosis displayed the same allelic losses as those found in the coexisting endometriosis. Recently, with the use of whole-exome sequencing and transcriptome sequencing, two independent studies reported novel somatic mutations of the ARID1A (also known as BAF250a) gene in 43% and 56% of ovarian CCAs 36, 37. One of these studies also found ARID1A mutations in adjacent atypical endometriosis 37. Therefore, it is conceivable that PTEN and ARID1A losses will also be early molecular genetic events in the pathogenesis of endometriosis-associated CCAs.
Recently, a number of groups have developed PIK3CA-selective inhibitors, and some have demonstrated efficacy in vitro and in vivo38. Therefore, new therapeutic strategies aimed at interfering with the oncogenic properties of PIK3CA-mutated cancer alleles are hoped for, as well as targeted therapies, such as imatinib mesylate (anti-BCR/ABL and cKIT), gefitinib and erlotinib (anti-EGFR), showing a high degree of specificity for translocated/mutated oncogenes 39–41. Given the frequent occurrence of PIK3CA mutations in ovarian CCA, this could have a tremendous impact on eliminating the morbidity and mortality of this carcinoma.
This work was supported in part by a grant-in-aid for the promotion of defence medicine from the Ministry of Defence, Japan (to SY, KI and OM) and by a grant-in-aid for cancer research from the Ministry of Health, Labour, and Welfare, Japan (to HT). We are grateful to Kozue Suzuki, Department of Basic Pathology, National Defence Medical College, Saitama, Japan, for technical assistance.
SY carried out the experiments. SY, HT, KI and OM were responsible for the study design, data analysis, and data interpretation. SY and HT conceived the experiments, analysed the data, and wrote the manuscript. MT and ST collected the patient material. SY carried out the histopathological analysis. All authors had final approval of the submitted and published versions.