Secondary Logo

Journal Logo

Original Articles

Interobserver Agreement for Mismatch Repair Protein Immunohistochemistry in Endometrial and Nonserous, Nonmucinous Ovarian Carcinomas

Sari, Aysegul MD*; Pollett, Aaron MD*; Eiriksson, Lua R. MD; Lumsden-Johanson, Brenda MSc; Van de Laar, Emily MSc; Kazerouni, Hamid MD§; Salehi, Amir MD§; Sur, Monalisa MD§; Lytwyn, Alice MD§; Ferguson, Sarah E. MD

Author Information
The American Journal of Surgical Pathology: May 2019 - Volume 43 - Issue 5 - p 591-600
doi: 10.1097/PAS.0000000000001220


Lynch syndrome (LS) is a hereditary cancer syndrome caused by germline mutations in the mismatch repair (MMR) genes. It is associated with an increased risk of colorectal cancer (CRC) and endometrial cancer (EC) as well as adenocarcinomas of other organs including ovary, brain, stomach, pancreas, and urinary tract. CRC and EC are the most common malignancies. Individuals with LS have a lifetime risk of 40% to 80% for CRC, 30% to 60% for EC, and 10% to 12% for ovarian cancer (OC).1–3 Identification of patients with LS is important as LS patients and their relatives benefit from surveillance programs and risk-reducing measures.3

MMR-immunohistochemistry (IHC) is a sensitive triage test to identify LS in patients with CRC, EC, and OC.2,4 Universal screening of CRC with MMR-IHC is widely recommended5,6 and has recently become a common practice for CRCs in many centers. Given that the lifetime risk for EC is similar to that of CRCs and that gynecologic cancers are often the first malignancies diagnosed in patients with LS,7–9 MMR-IHC has been recommended to identify LS in women with endometrial and OC.2,10

Given the potential impact of MMR-IHC results on patient management, it is important to know whether the test, when performed on EC and OC, has acceptable reproducibility. Technical factors, pathologist experience, as well as various staining variabilities in MMR-deficient CRC may cause differences in the interpretation of MMR-IHC.4 Reproducibility of MMR-IHC in CRC has been reported in 2 studies as moderate to substantial.11,12 Interobserver reliability of MMR-IHC in gynecologic malignancies has not been reported to our knowledge. The aim of this study was to assess the multi-institutional interobserver variability in the interpretation of MMR-IHC on EC and OC and to compare the reproducibility of MMR-IHC reporting in a cancer center without previous experience to a reference center experienced in MMR-IHC interpretation.


Study Cohort and Setting

This study was approved by the Research Ethics Board and is a part of a larger prospective study evaluating enhanced genetic referral program to identify women with LS who present with newly diagnosed nonserous/mucinous OC or EC. Written informed consent was obtained from all patients. We followed the Guidelines for Reporting Reliability and Agreement Studies.13 The study sample consisted of 73 consecutive ECs (n=48) and nonserous, nonmucinous epithelial OCs (n=25) from a single cancer center (Table 1). The prospective study aimed to perform MMR-IHC staining on final hysterectomy specimens unless the patient was not a surgical candidate or was treated with progestin therapy for fertility preservation. In our cohort, all endometrial samples were hysterectomy specimens. Tissue sections were stained with MMR-IHC for MLH1, MSH2, MSH6, and PMS2. Six pathologists from 2 cancer centers took part in the study. Four pathologists with a subspecialty interest in gynecologic pathology were from the same cancer center and had no previous experience with MMR-IHC on gynecologic tumors. Two pathologists were from a referral cancer center. One was a consultant pathologist (A.P.) working with referral cases for MMR-IHC and microsatellite instability (MSI) analysis and had >10 years of experience interpreting MMR-IHC in CRC and gynecologic cancers. This pathologist had also previously participated in a study assessing the performance characteristics of MMR-IHC in EC in which MMR-IHC identified all cancers with MMR germline mutations.2 The other pathologist was a trainee in gastrointestinal pathology and had previously been exposed to MMR-IHC interpretation on CRC.

Distribution of Ovarian and Endometrial Cancers According to MMR-IHC Status


All cases were fixed in 10% formalin and paraffin-embedded (FFPE). One representative tumor block was chosen for IHC staining. IHC analysis was performed on 4-μm-thick FFPE tissue sections using monoclonal mouse antibodies against: MLH1 (clone ES05) 15-minute incubations for primary antibody, polymer, and mouse linker, as well as a 5-minute protein blocking step before antibody application; MSH2 (clone FE11) 30-minute incubation times for primary antibody and polymer and the same protein blocking step mentioned previously; MSH6 (clone EP49) 10-minute incubation time for antibody and polymer and the same protein blocking step; and PMS2 (clone EP51) 30-minute incubation times for primary antibody and polymer with no protein blocking step. All stains were performed on the Dako Autostainer Link 48 AS480 instrument using the EnVision FLEX+ Mouse High pH kit (pH9.0). Antigen retrieval was performed for 20 minutes at 97°C and cooled to 75°C using the Dako PT Link pretreatment module.

Evaluation of MMR-IHC

Before the study, a review of 9 selected teaching cases that were not part of the study sample was led by the experienced pathologist. Examples of intact (normal) and deficient (abnormal) protein expression patterns were examined in various tissue specimens (see Figs., Supplemental Digital Content 1–9,,,,,,,,,, showing various MMR protein expression patterns in tumor tissue). Detailed assessment of each case by the experienced pathologist was provided (see Fig., Supplemental Digital Content 10,, detailed description of staining pattern for each teaching case).

To help guide the observers in MMR staining interpretation, a decision tool was developed (Fig. 1). This tool was intended to resemble the diagnostic reasoning that pathologists use when assessing histologic findings. The content information on MMR-IHC criteria was obtained from both the literature4 and experience of the consultant pathologist (A.P.).

The decision tool.

Each pathologist reviewed all the study slides independently and were blinded to each other’s interpretation. The staining was interpreted by each pathologist as intact (normal) (Fig. 2), deficient (abnormal) (Fig. 3), equivocal (indefinite) or not assessable (suspected technical failure) for each protein. Nontumoral background tissue was used as an internal control for all cases. An overall final interpretation of the patient MMR status (MMR intact, MMR deficient: MLH1-PMS2 deficient; PMS2 deficient; MLH1 deficient, MSH2-MSH6 deficient, MSH2 deficient, MSH6 deficient, or equivocal) was also reported by each pathologist. A consensus interpretation was obtained for each patient and was defined as majority opinion for each case. At the end of the study all the pathologists reviewed the discordant cases together at a multiheaded microscope.

Endometrioid adenocarcinoma of the endometrium, MLH1 stain. The tumor is MMR intact; the tumor and internal control tissue show strong nuclear staining for all MMR-IHC stains.
Ovarian clear cell carcinoma MSH2 staining. The tumor shows abnormal staining with loss of staining in tumor nuclei. The background stroma (internal control) shows strong nuclear staining.

Data Analysis

Estimated sample size was 62 cases, based on 6 observers and expected substantial agreement (κ=0.8; 95% CI, 0.7-0.9). Interobserver reliability was assessed using criteria described by Landis and Koch.14 The strength of the agreement between observers was evaluated by calculating the κ value.


The consensus interpretation defined 51 (70%) cases as MMR intact (Fig. 2) and 22 (30%) cases as MMR deficient (Fig. 3, Table 1). The κ coefficients for pair-wised interobserver agreement for MMR status were all “almost perfect,” ranging from 0.85 to 1.00 (Table 2). The overall κ coefficient was 0.930 (95% CI, 0.874-0.987) (almost perfect). All observers agreed in 67 (92%) cases (Table 3).

Test Agreement Between Intact Versus Deficient Versus Equivocal Among 6 Raters: Pair-wised κ Coefficient
Distribution of MMR-IHC Interpretation for Study Cases

The most experienced pathologist was in full agreement with the consensus interpretation. Four of the less experienced pathologists had at least 1 discrepant interpretation compared with the consensus interpretation. There were 6 discordant cases. Three tumors considered MMR deficient and 2 as MMR intact on consensus interpretation were called equivocal by at least one of the less experienced observers. In 1 discordant case, the individual pathologist interpretation was MMR intact while the consensus was MMR deficient (Table 4 and Figs. 4–9, discordant cases). The review of the discordant cases at the end of the study agreed with the consensus opinions and no cases were considered equivocal by consensus opinion. There were no cases interpreted as not assessable (technical failure).

Pathologists’ MMR Diagnosis in Discordant Cases
Ovarian clear cell carcinoma MSH6 staining. The tumor is MMR intact. MSH6 shows focal regional weaker staining areas at low power.
A region of Figure 4 at high power shows admixed variable loss and strong staining in tumor cells, but there is no clear regional area of isolated loss of staining in the tumor.
Ovarian clear cell carcinoma, PMS2 staining. The tumor is PMS2 deficient with a heterogeneous staining pattern.
A region of Figure 6 at higher power shows there are sharply demarcated area of tumor with loss of staining.
Endometrial endometrioid adenocarcinoma, MSH2 staining. The tumor is MSH2/MSH6 deficient (inset). Both stains show regional areas of faint staining in the internal control tissue.
Endometrial undifferentiated carcinoma, MLH1 staining. The tumor is MLH1/PMS2 deficient. There is loss of staining in tumor with strong staining in the internal control tissue. The dense staining of intratumoral lymphocytes and undifferentiated nature of the tumor may be interpreted as tumor staining.

The rate of agreement was 96% for MMR-intact cases and 82% for MMR-deficient cases. Discrepant evaluations were made in 4% (3/73) of MLH1 stains (all MMR deficient by consensus), 4% (3/73) of PMS2 stains (all MMR deficient by consensus), 1.3% (1/73) of MSH2 stains (MMR deficient by consensus), and 5.4% (4/73) of MSH6 stains (2 MMR intact, 2 MMR deficient by consensus). Discordant interpretations did not appear related to tumor site or morphology.

Several reasons were identified for discordant interpretations. Three cases had MSH6 staining in which scattered tumor cells showed absent or weak staining adjacent to tumor cells with strong nuclear staining (Figs. 4, 5). Taking the entire tumor into consideration, the clear majority of the tumor showed nuclear staining. These cases were considered intact by consensus diagnosis, but some pathologists were concerned by the scattered tumor cells with absent nuclear staining and had considered the cases equivocal.

Two discordant cases, a MLH1-PMS2-deficient and a PMS2-deficient case by consensus, showed heterogeneous, or subclonal staining (Figs. 6, 7), characterized by abrupt loss of staining (internal control tissue positive) with sharply demarcated borders from the adjacent staining areas. These had been classified as “equivocal” and “MMR intact,” respectively.

The other discordant cases in our study included 1 MSH2/MSH6-deficient case with areas of reduced staining in the internal control tissue which was called equivocal by 1 pathologist (Fig. 8) and 1 MLH1/PMS2-deficient case which had prominent intraepithelial lymphocytes (called equivocal by 1 pathologist; Fig. 9).


MMR-IHC is a widely used method as an initial step to detect MMR-deficient cases. The staining in routine practice is generally classified as intact (normal), deficient (abnormal), equivocal (indefinite), or not assessable (uninterpretable, usually due to technical factors). Cases with equivocal MMR-IHC can have further investigation such as MSI analysis to determine a tumor’s MMR status. Inaccurate interpretation of MMR-IHC, especially interpreting an abnormal staining as intact, might miss a proportion of patients with LS and with the emerging role of immune modulating therapy, may miss an opportunity for individuals to be treated with immune modulating therapy (such as checkpoint inhibitors).15

In spite of the impact of MMR-IHC on patient management and prognosis, there is limited data on the reproducibility of MMR-IHC, which up to now has primarily focused on CRC.11,12 To the best of our knowledge this is the first study assessing interobserver variability in the interpretation of MMR-IHC on EC and OC. In addition, it is the first to compare the reproducibility of MMR-IHC reporting in a cancer center without previous experience to a reference center experienced in MMR-IHC interpretation.

Two previous studies on MMR-IHC in CRC reported pair-wised κ coefficients ranging from 0.35 to 0.95 (fair to almost perfect agreement) among 6 observers11 and from 0.49 to 0.79 (moderate to substantial agreement) among 7 observers.12 Both studies included experienced and nonexperienced pathologists in the interpretation of MMR-IHC. Overbeek et al12 suggested that MMR-IHC interpretation be restricted to experienced pathologists. Klarskov et al11 did not demonstrate a difference between experienced and nonexperienced pathologists, but stated that some contradictory evaluations could be related to lack of experience in the evaluation of weak staining.

In contrast, our study showed consistent “almost perfect” agreement regardless of the pathologist’s level of experience. This could be explained by a number of factors such as staining quality in EC/OC versus CRC, or improvements in the MMR-IHC antibodies. It is also possible that the decision tool that we developed helped guide the inexperienced pathologists and contributed to improved interpretative uniformity.

Five of the 6 pathologists did not miss a MMR-deficient tumor in our study. Only a single case (1/73 patients, 1.4%) had an unequivocal disagreement. One of the less experienced pathologist misclassified this case as intact when the tumor was deficient. Four of the less experienced pathologists had at least 1 discrepant interpretation compared with the consensus interpretation. All but 1 discrepant case were called equivocal and would have further investigation to determine the patient’s MMR status and therefore would unlikely be misclassified after final evaluation. The rate of agreement among MMR-deficient cases were slightly lower than the rate of agreement MMR-intact tumors. This is the same finding as in the CRC interobserver studies and is not an unexpected result as it is well known that MMR-deficient tumors occasionally have unusual staining patterns.4 Klarskov et al11 reported that the predominant cause of MMR-IHC discordancy among observers was due to weak IHC staining that had not been clearly defined at study start. This “weak” staining, in which there is a true difference in the intensity of the tumor staining and the adjacent internal control tissue has been shown to be associated with MSI-high tumors and underlying genetic alterations.3,11,15,16 This pattern was not seen in our study cohort. In the interobserver study by Overbeek et al12 details regarding the staining patterns of cases with discrepancy were not reported.

We identified several interpretive issues in our study: (1) The presence of scattered absent or weak staining cells adjacent to cells with strong nuclear staining was misinterpreted as equivocal. This pattern of staining has been described in the literature, predominately with MLH1 and MSH6, and has not been shown to be associated with any underlying genetic alteration11,17,18; (2) Heterogeneous staining was not identified as abnormal. This is a rare staining pattern characterized by 2 population of tumor cells with retained and abrupt/subclonal complete loss of MMR protein expression.11,16–21 Heterogeneous staining was shown to occur up to 2.6% to 7.2% of ECs,20,21 around 8% of proximal CRCs,16 and 0.7% of CRCs in below 60 years of age.19 Pathologists need to recognize this pattern and report these cases as MMR deficient, as they usually represents an MSI-high tumor and have been associated with germline mutations.17,19 Among EC, MLH1 and PMS2 are the most common stains showing heterogeneous expression and all cases were hypermethylated18,20,21; (3) Regional areas of reduced staining in the internal control tissue in a typical deficient case, resulted in interpreting a case as equivocal. This reduced staining is a technical artefact and likely a result of suboptimal fixation. Reduced/weak staining resulting from technical issues have been reported11; and (4) Prominent intratumoral lymphocytes were misinterpreted as intact tumoral cell staining.

Potential pitfalls in interpretation of MMR-IHC, including misinterpretation of intratumoral lymphocytes, aberrant staining patterns including cytoplasmic positivity, variable staining, and heterogeneous staining patterns, have been extensively reported.3,4,11,17,19,22–25

Some authors have suggested that MMR-IHC interpretation guidelines addressing different staining patterns and/or educational training sessions for young pathologists are required before reading MMR-IHC.11,21 Before the start of our study, we had a teaching session at the multiheaded microscope, and all were encouraged to ask questions and raise any pertinent issues. We also created a MMR-IHC decision tool that the pathologists could use during their sign-out. The use of a decision tool, a teaching session before the introduction of MMR-IHC, and the support of an experienced pathologist may have contributed to the high level of interobserver agreement in our study.

In reviewing the decision tool with the pathologists at the end of the interobserver study, 2 potential changes were suggested to improve the decision tool. The first modification was that the cut point for Normal staining could be changed from 90% to any tumoral cells with intact staining where the internal control tissue shows similar (or weaker) staining throughout. This change would make the interpretation of cases showing patchy faint/weak nuclear staining easier and avoid classifying these tumors as equivocal. In addition, we suggest adding another category of staining leading to Equivocal: “weak” tumor staining, that is, tumor with intensity weaker than the adjacent internal control tissue. This could help inexperienced pathologists to be aware of this pattern and also help to distinguish this staining from the patchy weaker staining where both the tumor and reference tissue show similar weak staining (Fig. 10).

Revised schematic tool.


This study showed a high interobserver agreement for MMR-IHC even with inexperienced pathologists and shows that implementation into a center without previous experience interpreting MMR-IHC for OC and EC can be done with high reliability. A MMR-IHC staining decision tool which addresses the uncommon but diagnostically important staining patterns and support from experienced expert pathologist for unusual cases may be useful in guiding the interpretation of MMR-IHC by less experienced pathologists.


1. Chui MH, Gilks CB, Cooper K, et al. Identifying Lynch syndrome in patients with ovarian carcinoma: the significance of tumor subtype. Adv Anat Pathol. 2013;20:378–386.
2. Ferguson SE, Aronson M, Pollett A, et al. Performance characteristics of screening strategies for Lynch syndrome in unselected women with newly diagnosed endometrial cancer who have undergone universal germline mutation testing. Cancer. 2014;120:3932–3939.
3. Halvarsson B, Lindblom A, Rambech E, et al. Microsatellite instability analysis and/or immunostaining for the diagnosis of hereditary nonpolyposis colorectal cancer? Virchows Arch. 2004;444:135–141.
4. Shia J, Ellis NA, Klimstra DS. The utility of immunohistochemical detection of DNA mismatch repair gene proteins. Virchows Arch. 2004;445:431–441.
5. Evaluation of Genomic Applications in Practice and Prevention (EGAPP) Working Group. Recommendations from the EGAPP Working Group: genetic testing strategies in newly diagnosed individuals with colorectal cancer aimed at reducing morbidity and mortality from Lynch syndrome in relatives. Genet Med. 2009;11:35–41.
6. Boland CR, Goel A. Microsatellite instability in colorectal cancer. Gastroenterology. 2010;138:2073–2087; e2073.
7. Clarke BA, Cooper K. Identifying Lynch syndrome in patients with endometrial carcinoma: shortcomings of morphologic and clinical schemas. Adv Anat Pathol. 2012;19:231–238.
8. Mills AM, Sloan EA, Thomas M, et al. Clinicopathologic comparison of Lynch syndrome-associated and “Lynch-like” endometrial carcinomas identified on universal screening using mismatch repair protein immunohistochemistry. Am J Surg Pathol. 2016;40:155–165.
9. Rabban JT, Calkins SM, Karnezis AN, et al. Association of tumor morphology with mismatch-repair protein status in older endometrial cancer patients: implications for universal versus selective screening strategies for Lynch syndrome. Am J Surg Pathol. 2014;38:793–800.
10. Chui MH, Ryan P, Radigan J, et al. The histomorphology of Lynch syndrome-associated ovarian carcinomas: toward a subtype-specific screening strategy. Am J Surg Pathol. 2014;38:1173–1181.
11. Klarskov L, Ladelund S, Holck S, et al. Interobserver variability in the evaluation of mismatch repair protein immunostaining. Hum Pathol. 2010;41:1387–1396.
12. Overbeek LI, Ligtenberg MJ, Willems RW, et al. Interpretation of immunohistochemistry for mismatch repair proteins is only reliable in a specialized setting. Am J Surg Pathol. 2008;32:1246–1251.
13. Kottner J, Audige L, Brorson S, et al. Guidelines for Reporting Reliability and Agreement Studies (GRRAS) were proposed. J Clin Epidemiol. 2011;64:96–106.
14. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1977;33:159–174.
15. Stenger M. Stenger M. “Pembrolizumab in MSI-H or dMMR Solid Tumors: ‘First Tissue/Site-Agnostic Approval by FDA’.” The ASCO Post, HSP News Service, February 10, 2018. Available at:
16. Chapusot C, Martin L, Bouvier AM, et al. Microsatellite instability and intratumoural heterogeneity in 100 right-sided sporadic colon carcinomas. Br J Cancer. 2002;87:400–404.
17. Joost P, Veurink N, Holck S, et al. Heterogeneous mismatch-repair status in colorectal cancer. Diagn Pathol. 2014;9:126.
18. Pai RK, Plesec TP, Abdul-Karim FW, et al. Abrupt loss of MLH1 and PMS2 expression in endometrial carcinoma: molecular and morphologic analysis of 6 cases. Am J Surg Pathol. 2015;39:993–999.
19. Watson N, Grieu F, Morris M, et al. Heterogeneous staining for mismatch repair proteins during population-based prescreening for hereditary nonpolyposis colorectal cancer. J Mol Diagn. 2007;9:472–478.
20. Stelloo E, Jansen AML, Osse EM, et al. Practical guidance for mismatch repair-deficiency testing in endometrial cancer. Ann Oncol. 2017;28:96–102.
21. Watkins JC, Nucci MR, Ritterhouse LL, et al. Unusual mismatch repair immunohistochemical patterns in endometrial carcinoma. Am J Surg Pathol. 2016;40:909–916.
22. Muller A, Giuffre G, Edmonston TB, et al. Challenges and pitfalls in HNPCC screening by microsatellite analysis and immunohistochemistry. J Mol Diagn. 2004;6:308–315.
23. Mangold E, Pagenstecher C, Friedl W, et al. Tumours from MSH2 mutation carriers show loss of MSH2 expression but many tumours from MLH1 mutation carriers exhibit weak positive MLH1 staining. J Pathol. 2005;207:385–395.
24. Shia J. Immunohistochemistry versus microsatellite instability testing for screening colorectal cancer patients at risk for hereditary nonpolyposis colorectal cancer syndrome. Part I. The utility of immunohistochemistry. J Mol Diagn. 2008;10:293–300.
25. Shia J, Klimstra DS, Nafa K, et al. Value of immunohistochemical detection of DNA mismatch repair proteins in predicting germline mutation in hereditary colorectal neoplasms. Am J Surg Pathol. 2005;29:96–104.

Lynch syndrome; endometrial cancer; ovarian cancer; mismatch repair proteins; immunohistochemistry; observer agreement

Supplemental Digital Content

Copyright © 2019 The Author(s). Published by Wolters Kluwer Health, Inc