Introduction

Ovarian cancer is one of the most frequent cancers (age standardized rate 6.4/100,000 person) and has the highest mortality rate among all gynecological malignancies (1,2). Well-accepted prognostic parameters for ovarian cancer are International Federation of Gynecology and Obstetrics (FIGO) stage, age, tumor type and histological grade (2). Approximately one fifth of ovarian tumors are hereditary and Lynch syndrome (LS) is responsible for 10-15% of these cancers (2,3). In LS, DNA mismatch repair (MMR) genes (mostly MLH1, MSH2, MSH6, and PMS2) have hereditary mutations which lead to the development of microsatellite instability (MSI). In addition, deletions in EPCAM gene responsible for MSH2 gene promoter methylation might be found in LS. Microsatellites are short, repetitive DNA sequences unevenly distributed across the genome (3) on which insertion-deletion type mutations can emerge during DNA synthesis. These mutations, named MSI, are normally repaired by the MMR system. Due to mutations in repair genes, defects might not be repaired and subsequently, mutant copies might accumulate. Mutations developing on microsatellite foci that encode genes might lead to neoplastic alterations (2,4-7). MSI was first discovered in LS-related colon cancer, and subsequently, suggested to have a role in the pathogenesis of various genetic and sporadic cancers. MSI is also reported in approximately 10% of ovarian cancers (3,7,8).

MSI, which has a well-known prognostic and therapeutical significance in colon and endometrial cancers, might help clarify certain topics related with ovarian tumors, such as growth rate, precursor lesions, prognosis and response to therapy (3). MMR mutation and MSI are detected using methods such as immunohistochemisty (IHC) and polymerase chain reaction (PCR) (5,7,8). Although PCR and gene analysis give some idea on MSI status, these are troublesome, time-consuming and expensive methods (5,7,9,10). On the other hand, IHC analysis, which is routinely used in many pathology laboratories, is a simple, inexpensive, rapid and convenient method to investigate familial and sporadic MMR defects. In tumors with MMR pathway defects, it is possible to detect loss of one or more protein expressions.

In this study, we aimed to determine MSI status of invasive and borderline epithelial ovarian tumors using the IHC method and evaluate its correlation with several prognostic parameters and survival.

Methods

One hundred fifty-nine ovarian tumors diagnosed with morphological and immunohistochemical analysis in our pathology clinic between January 2012 and December 2016 were included in the study (ethic committee approval no: 2017/514/104/5, Kartal Dr. Lütfi Kirdar City Hospital). Slides, paraffin blocks and reports belonging to the cases were retrospectively analysed. Sections prepared from the paraffin blocks of selected slides were immunohistochemically stained, as explained below. The cases were reviewed by two pathologists using light microscopy. Demographic data were extracted from patient records. For staging, data from the archives of the medical oncology department were analysed according to the 2014 FIGO staging. For inclusion, the diagnosis of epithelial tumor should be made from surgical specimens, adequate tumor tissue should be present at blocks and slides of the case, positive internal control should be present, and besides, surgical specimen should be included if both biopsy and surgical specimens were available. Consultation cases, cases without paraffin blocks and cases with technically unsatisfactory IHC results were excluded. According to the Death Registration System, patients alive during the study period were recorded as “survived” cases.

For immunohistochemical examination, 4 µm thick sections prepared from formalin fixed-paraffin embedded tissues were used. Tissue sections were taken onto electrostatic loaded slides (isotherm) and placed into 70 °C incubator for at least 1 hour. All IHC procedures, including deparaffinization and antigen retrieval were performed in a full automated IHC stainer (Ventana BenchMark XT, Ventana Medical Systems, Tucson, AZ).

Instrument-ready, biotinylated, horseradish peroxidase multimer based, ready-to-use kit containing hydrogen peroxide substrate and 3.3’-diaminobenzidine tetrahydrochloride (DAB) chromogen (ultraView™ Universal DAB Detection Kit, Catalog number 760-500, Ventana Medical Systems, Tuscon, AZ) were used. Staining with MLH1 (Clone: ES05, Dako North America, Ready to use), MSH2 (Clone: FE11, Dako, North America, Ready to use), MSH6 (Clone: EP49, Dako, North America, Ready to use), and PMS2 (Clone: EP51, Dako, North America, Ready to use) antibodies, counterstaining with hematoxylin, dehydration and xylene clearing were performed and the slides were cover-slipped. Inflammatory and stromal cells on slides were considered positive internal control and nuclear staining for each antibody, positive staining. Absence of nuclear staining in tumor cells was interpreted as “loss of MMR proteins”.

Statistical Analysis

Statistical analysis with SPSS version 15.0 software was used to investigate the appropriateness of the variables to normal distribution by using histogram graphs and the Kolmogorov-Smirnov test. Mean, standard deviation and median values were used for descriptive analyses. Two x two grids were compared with Pearson’s chi-square and Fisher’s exact tests. ANOVA was used when variables with normal distribution (parametric) evaluated between more than two groups; the Mann-Whitney U test was used when variables showing non-normal distribution (nonparametric) were evaluated between binary groups; and the Kruskal-Wallis test, for more than two groups. The Spearman correlation coefficient was used to analyze the measurable data with each other. For overall comparison of survival function, log-rank test (Mantel-Cox) was used. A p value of less than 0.05 was considered statistically significant.

Results

Of the 159 study patients, 68% was under the age of 60 (18-79 year, mean age: 53.4). In 60% of the cases, tumor type was serous carcinoma and most of the cases (87%) were high grade. The most common borderline tumor was borderline mucinous carcinoma. Regarding stage, 43% of the patients had FIGO1 and 42% had FIGO3 disease (Table 1).

MMR loss was found in three of 159 cases of ovarian carcinoma (1 serious carcinoma, 1 clear cell carcinoma and 1 borderline endometrioid carcinoma). When the relationship between age, tumor type, FIGO stage, histological grade and MMR protein expression loss were analysed, a significant relationship was found only between tumor type and MMR loss (p<0.001). MMR protein expression loss was present in 1 of 2 (50%) borderline endometrioid tumors, one of nine (11%) clear cell carcinoma and 1 of 96 (1 %) serous carcinoma (p<0.001) (Table 2). In two patients, only MSH2 protein expression loss and in one patient, both MLH1 and PMS2 protein expression loss were detected (Table 3) (Figure 1 and 2).

Totally, in 159 ovarian cancers, the rate of patients without MMR loss and who were still alive (70.5%) was below the rate of the all patients who were alive (71.1%) while the rate of patients who had MMR loss and survived was 67% (Table 4). The patient who had MMR loss and was paraplegic died in September 2018 due to a non-tumoral reason. A statistically significant relationship was not found between MMR protein expression loss and mortality [c2 (1)=0.044, p>0.005].

Discussion

Most frequent hereditary cancer syndromes related with ovaries are hereditary breast and ovarian cancer syndrome (HBOC) and LS. HBOC with BRCA1/2 mutations constitute 90% of cases and LS, the remaining 10% (11). LS previously was called hereditary nonpolyposis colorectal cancer, is an autosomal dominant familial syndrome seen at an early age, associated with a genetic predisposition to colorectal, endometrial, gastric or ovarian cancers and rarely found in patients having small intestine, pancreas, and brain tumors. In LS, hereditary mutations are observed in MMR genes. According to the literature, the most frequent histological subtypes in LS-related ovary tumors are clear cell and endometrioid type ovarian cancers (12,13). In a study by Song et al. (14), the mean cumulative risks of ovarian cancers by age 80 years were estimated to be 64%, 24% and 3.7% in the presence of BRCA1, BRCA2 and MMR gene mutations.

In the study of Song et al. (14), the median age at diagnosis of hereditary ovarian cancer was 52 for BRCA1 mutation carriers, 57 for BRCA2 mutation carriers and 54 for MMR gene mutation carriers. Vierkoetter et al. (12) reported that patients with MMR protein expression loss were younger than patients without MMR loss (mean age, 47 and 58 years, respectively,) (p=0.014) In our study, three patients with MMR protein expression loss were 33, 48 and 60 years old, respectively (mean age: 48.6 year). Although the patients were younger in the group with MMR protein expression loss, we did not find any relationship between age and MMR loss. This may be due to small sample size.

Regarding tumor types, MSI was more frequently reported in several histological subtypes (10,13,15-17). In a study including 42 clear cell ovarian cancers, Cai et al. (18) reported that nine of the cases (21%) were related with MSI and high level of MSI was involved in the development of a subset of ovarian clear cell carcinomas. Gras et al. (13) found MSI in only endometrioid and clear cell ovarian cancers and its incidence was 12.5%. Lu et al. (16) found MSI more frequently, in poorly differentiated mucinous and clear cell ovarian cancers, however MSI level was not related with age, tumor differentiation or tumor type, but related with only stage (16). In a metaanalysis of 15 studies (n=159) that investigated the correlation between histological subtypes and MMR gene defect, Pal et al. (3) reported that, in ovarian tumors with MMR gene defect, non-serous histological subtypes were dominant and besides, mucinous and endometrioid type ovarian carcinomas were analogous to colon and endometrium cancers of LS (3). Ryan et al. (19), explained that, in LS-related cancers, dominant histological type was endometrioid carcinoma, however, MMR gene defect was also seen in high grade serous carcinomas. In our study, we found a relationship between tumor type and MMR protein expression loss (MMR loss was seen in a borderline endometrioid tumor, clear cell carcinoma and high grade mixed serous carcinoma). Even if we found a statistically significant relationship between MMR protein expression loss and tumor type, our patient group with MMR loss was small and this result should be supported with larger studies.

A retrospective study by the International LS Working Group on correlation between histological grade and MSI protein expression loss reported that LS-related ovarian cancers were mostly in early stage and low grade (20). Colle et al. (7) found that LS-related ovarian cancers were low-grade endometrioid and early stage tumors. Dellas et al. (21) reported that although MSI was more frequent in poorly differentiated cancers (p>0.05), a statistically significant relationship existed only between early stage and MSI (p=0.03). In our study, one of the patients with MMR loss had borderline endometrioid carcinoma, one, high-grade clear cell carcinoma and the other had high-grade mixed serous carcinoma. Statistical analysis revealed that histological grade was not correlated with MMR loss in our series. Since the stages in our patients with MMR loss were Stage IA, Stage IIB and Stage IIIC, respectively, a statistically significant relationship was not established between MMR loss and FIGO stage. Small sample size might be the reason for not finding any relationship between histological grade, stage and MMR loss; therefore, we think that further analysis might be clarifying.

In a study by Ryan at al. (19) examining 1047 patients with MMR gene mutation, 53 patients had LS-related ovarian cancer; 85% of these patients presented at stage 1/2 and 5 years survival was 80%. Mallorca Group prospectively investigated 1942 patients without previous cancer, who were MMR mutation carriers, with colonoscopy and gynecological examination. In this study, 19 of 314 cancers developed during follow-up were ovarian cancers and cumulative ovarian cancer risk at age 70 was 11% for MLH1 gene defect, 15% for MSH2 gene defect and 0% for MSH6 and PMS2 gene defects. Most of the patients with ovarian cancer were younger than 50 and 10-year survival was 89% (22). The same study group, in a recent study including LS patients who did not have cancer previously, had early cancers and still alive, found that 5-year survival was 83% and 10-year survival was 74% in ovarian cancers (23). In our study, one of the three patients with MMR loss was paraplegic and died within 5 years of follow up. Statistical analysis revealed no relationship between patient survival and MMR loss. Nevertheless, we believe that survival should also be further investigated in larger series.

In the study by Mallorca group, MSH6 and PMS2 loss were not found in ovarian cancers. On the contrary, Norquist et al. (24) found MMR gene mutation in eight of 1,915 patients with ovarian cancer and 88% of these patients had PMS2 or MSH6 gene defect. In 2/3 patients with MSH6 mutation, there were endometrioid and early-stage cancers. The tumor type in four patients with PMS2 mutations was high grade serous carcinoma and was advanced stage. In the study by Song et al. (14), MMR gene mutation was seen in 17 patients and 10 of them had MSH6, four had MSH2, two had MLH1 and one had PMS2 gene mutation. In our study, immunohistochemical analysis with MLH1, MSH2, MSH6, and PMS2 markers revealed that two of three patients with MMR loss had only MSH2 protein expression loss and, one patient had both MLH1 and PMS2 loss. Normally, as a result of MSH2 protein expression loss, due to heterodimeric structure, MSH6 protein expression loss is also expected. But in two patients with MSH2 loss, no staining loss was seen in MSH6. This might be because of a variable immunohistochemical staining pattern of MSH6 (5). According to Terui et al. (25), 24% of mutations identified in LS are missense substitutions and this mostly occurs on MSH6 gene. Mutant proteins that develop after missense mutations on genes, might be catalytically inactive but antigenically active (7). In our cases that show only MSH2 loss, MSH6 protein expression loss may not have been found because of missense mutation. In these patients, MSH2 loss might depend on EPCAM gene mutation, thus, we think that MSI status should be clarified with molecular analysis and, EPCAM gene analysis should be added.

Defect on DNA mismatch repair genes occur as a result of two mechanisms, one of which is germline mutations and the other is hypermethylation of CpG island on MLH1 gene. In normal cells, these islands are not methylated. Since BRAF mutation is seen in 70% of cancers that occur after MLH1 methylation and is not seen in LS, MMR defect might be sporadic (26,27). When immunohistochemical MLH1 and PMS2 protein expression loss are found in a case, since the tumor may be hereditary as well as sporadic, the patient should be examined for BRAF mutation and MLH1 methylation. Our patient who had both MHL1 and PMS2 loss was old and the tumor type was high-grade serous mixed carcinoma. We assume that in such cases, besides genetic analysis for LS, BRAF mutation should also be analysed for identification of sporadic cases.

In a study by Park and Kim (11) investigating hereditary risk in borderline ovarian tumors immunohistochemically with MLH1, MSH2, BRCA1, and BRCA2 (n=32), 3% of the cases showed MSH2 protein expression loss. According to this study, although a relationship between borderline tumors and hereditary syndromes was not known, family history should be investigated and necessary tests should be done in suspicious cases (11). In our study, MMR protein expression loss was present in almost 2-3% of borderline tumors (1/39 borderline tumors). We found MMR protein expression loss in 1 of 2 borderline endometrioid tumors and in which MSH2 protein expression loss was present. In the borderline cases in our study, the incidence of MSH2 protein loss was similar to that in the study by Park and Kim (11) The patient with MMR protein loss had an endometriosis history as well as tumor in family history. Altogether, the results suggest a hereditary tumor in that patient.

In a study evaluating 834 ovarian cancers with IHC and molecular MSI test, Lee et al. (8) reported poor overall concordance (68%) between the two methods. They suggested that disconcordance might be due to different genetic features of ovarian tumors and also, due to benign cases with MSI expression. Dellas et al. (21) found immunohistochemical MMR protein expression loss in 10/41 of patients and IHC could detect only a few (24%) MSI-positive cases. In our study, IHC detected MMR loss in three of 159 cases (1.8%). MSI is expected in 10-12% of ovarian cancers according to literature search, thus, IHC might have detected a small part of actual MSI cases (8).

Study Limitations

Our study has some limitations. Since some of ovarian cancers diagnosed between 2012 and 2016 were consultation cases, paraffin blocks were not available. Also, in some cases, due to technical problems, IHC did not reveal optimal results. Besides, detailed demographic and clinipathological data were not present in records of patients who received chemotherapy. These, altogether, negatively affected the included number of cases.

Conclusion

In our study, though MSI status of epithelial ovarian tumors was significantly related with histological subtypes, any relationship with other prognostic parameters and survival was not detected. We suggest that in patients with clear cell and endometrioid ovarian tumors under age 50 years, who have personal and/or family history of malignancy, or have concomitant secondary tumors; immunohistochemical MMR analysis might be helpful for monitoring the patients or patients’ relatives. We think that since we had a limited number of cases, further studies with larger series will be clarifying.

Authorship Contributions

Concept: F.İ.T., A.E.G. Design: F.İ.T., A.E.G. Data Collection or Processing: F.İ.T., A.E.G., S.S., S.H.K., N.Ö.B. Analysis or Interpretation: F.İ.T., A.E.G., S.S., S.H.K., N.Ö.B. Literature Search: F.İ.T., S.S. Writing: S.S.

Conflict of Interest: The authors declared no conflicts of interest with respect to the authorship and/or publication of this article.

Financial Disclosure: The authors received no financial support for the research and/or authorship of this article.