MMR protein immunohistochemistry and microsatellite instability in gastric cancers

Microsatellite instability-high (MSI-H) is emerging as a new therapeutic target for cancer immunotherapy. In May 2017, the United States Food and Drug Administration (FDA) granted accelerated approval of pembrolizumab for treatment of patients with unresectable or metastatic MSI-H or mismatch repair deficient (dMMR) solid tumours.1Cho J. Lee J. Bang H. et al.Programmed cell death-ligand 1 expression predicts survival in patients with gastric carcinoma with microsatellite instability.Oncotarget. 2017; 8: 13320-13328Crossref PubMed Scopus (57) Google Scholar MSI-H is one major molecular subtype of gastric cancer (GC)2Cancer Genome Atlas Research NetworkComprehensive molecular characterization of gastric adenocarcinoma.Nature. 2014; 513: 202-209Crossref PubMed Scopus (3912) Google Scholar, 3Cristescu R. Lee J. Nebozhyn M. et al.Molecular analysis of gastric cancer identifies subtypes associated with distinct clinical outcomes.Nat Med. 2015; 21: 449-456Crossref PubMed Scopus (1194) Google Scholar and is associated with high expression of programmed cell death-ligand 1 (PD-L1).1Cho J. Lee J. Bang H. et al.Programmed cell death-ligand 1 expression predicts survival in patients with gastric carcinoma with microsatellite instability.Oncotarget. 2017; 8: 13320-13328Crossref PubMed Scopus (57) Google Scholar, 4Ma C. Patel K. Singhi A.D. et al.Programmed death-ligand 1 expression is common in gastric cancer associated with Epstein-Barr virus or microsatellite instability.Am J Surg Pathol. 2016; 40: 1496-1506Crossref PubMed Scopus (120) Google Scholar Given the relatively high incidence of MSI-H GC in patients with stage III–IV GC,5An J.Y. Kim H. Cheong J.H. et al.Microsatellite instability in sporadic gastric cancer: its prognostic role and guidance for 5-FU based chemotherapy after R0 resection.Int J Cancer. 2012; 131: 505-511Crossref PubMed Scopus (110) Google Scholar the effective therapeutic use of anti-PD-1/PD-L1 inhibitor in MSI-H GC could contribute to remarkable improvement in survival of patients with GC. The most accurate method for detecting MSI to date is polymerase chain reaction (PCR), which requires a long test time, is expensive, and is difficult to interpret. So, faster, cheaper, and easily accessible diagnostic algorithms are needed. To investigate how accurately MSI-H cancers can be detected by MMR protein IHC and the distributions of protein losses in dMMR GC, we performed 4 MMR protein IHC and MSI pentaplex test in 580 GC from Asian/Korean patients. The correlations of MSI, dMMR, and clinicopathological characteristics are summarised in Table 1. The mean age of all 580 patients was 55.6 years (range 24–86), and 385 (66.4%) patients were male. Epstein–Barr encoding region (EBER) in situ hybridisation was performed in 569 cases and EBV was positive in 25 cases (4.4%). In Lauren classification, there were 271 (46.7%) diffuse type GCs and 246 (42.4%) intestinal type GCs. There were 11 (1.9%) stage I cancers, 119 (20.5%) stage II cancers, 235 (40.5%) stage III cancers, and 215 (37.1%) stage IV cancers. The mean follow-up duration was 43.6 ± 35.0 months. Of 580 patients, 304 (52.4%) died during the follow up period, and the 5-year survival rate was 36.7%.Table 1The correlation of mismatch repair protein expression, microsatellite instability, and clinicopathological characteristics in 580 gastric cancer patientsMLH1/PMS2p valueMSH2/MSH6p valueMMR proteinp valueTotalMSIp valuePreservedN=523LossN=57PreservedN=571LossN=9ProficientN=519DeficientN=61MSSN=520MSI-HN=60Age ≤60338 (64.6)19 (33.3)<0.001352 (61.6)5 (55.6)0.709336 (64.7)21 (34.4)<0.001357337 (64.8)20 (33.3)<0.001 >60185 (35.4)38 (66.7)219 (38.4)4 (44.4)183 (35.3)40 (65.6)223183 (35.2)40 (66.7)Sex Male353 (67.5)32 (56.1)0.085380 (66.5)5 (55.6)0.488350 (67.4)35 (57.4)0.116385350 (67.3)35 (58.3)0.164 Female170 (32.5)25 (43.9)191 (33.5)4 (44.4)169 (32.6)26 (42.6)195170 (32.7)25 (41.7)LaurenIntestinal212 (40.5)34 (59.6)0.009240 (42.0)6 (66.7)0.455209 (40.3)37 (60.7)0.005246209 (40.2)37 (61.7)0.003 Mixed32 (6.1)6 (10.5)38 (6.7)0 (0.0)32 (6.2)6 (9.8)3832 (6.2)6 (10.0) Diffuse255 (48.8)16 (28.1)268 (46.9)3 (33.3)254 (48.9)17 (27.9)271255 (49.0)16 (26.7) Indeterminate24 (4.6)1 (1.8)25 (4.4)0 (0.0)24 (4.6)1 (1.6)2524 (4.6)1 (1.7)Histology Tubular WD9 (1.7)0 (0.0)<0.0019 (1.6)0 (0.0)0.9429 (1.7)0 (0.0)<0.00199 (1.7)0 (0.0)<0.001 Tubular MD151 (28.9)25 (43.9)171 (29.9)5 (55.6)148 (28.5)28 (45.9)176148 (28.5)28 (46.7) Tubular PD114 (21.8)24 (42.1)136 (23.8)2 (22.2)114 (22.0)24 (39.3)138114 (21.9)24 (40.0) SRCC179 (34.2)4 (7.0)181 (31.7)2 (22.2)178 (34.3)5 (8.2)183179 (34.4)4 (6.7) Mucinous19 (3.6)1 (1.8)20 (3.5)0 (0.0)19 (3.7)1 (1.6)2019 (3.7)1 (1.7) Papillary32 (6.1)2 (3.5)34 (6.0)0 (0.0)32 (6.2)2 (3.3)3432 (6.2)2 (3.3) Adenosquamous0 (0.0)1 (1.8)1 (0.2)0 (0.0)0 (0.0)1 (1.6)10 (0.0)1 (1.7) Undifferentiated5 (1.0)0 (0.0)5 (0.9)0 (0.0)5 (1.0)0 (0.0)55 (1.0)0 (0.0) Hepatoid1 (0.2)0 (0.0)1 (0.2)0 (0.0)1 (0.2)0 (0.0)11 (0.2)0 (0.0) Others13 (2.5)0 (0.0)13 (2.3)0 (0.0)13 (2.5)0 (0.0)1313 (2.5)0 (0.0)TNM I11 (2.1)0 (0.0)0.004aLinear-to-linear association.10 (1.8)1 (11.1)0.185aLinear-to-linear association.10 (1.9)1 (1.6)0.001aLinear-to-linear association.1110 (1.9)1 (1.7)0.003aLinear-to-linear association. II94 (18.0)25 (43.9)116 (20.3)3 (33.3)93 (17.9)26 (42.6)11994 (18.1)25 (41.7) III219 (41.9)16 (28.1)233 (40.8)2 (22.2)218 (42.0)17 (27.9)235218 (41.9)17 (28.3) IV199 (38.0)16 (28.1)212 (37.1)3 (33.3)198 (38.2)17 (27.9)215198 (38.1)17 (28.3)MSI MSS519 (99.2)1 (1.8)<0.001520 (91.1)0 (0.0)<0.001519 (100)1 (1.6)<0.001520 MSI-H4 (0.8)56 (98.2)51 (8.9)9 (100)0 (0.0)60 (98.4)60MD, moderately differentiated; MMR, mismatch repair gene; MSI, microsatellite instability; MSI-H, microsatellite instability-high; MSS, microsatellite stable; PD, poorly differentiated; SRCC, signet ring cell carcinoma; WD, well differentiated.a Linear-to-linear association. Open table in a new tab MD, moderately differentiated; MMR, mismatch repair gene; MSI, microsatellite instability; MSI-H, microsatellite instability-high; MSS, microsatellite stable; PD, poorly differentiated; SRCC, signet ring cell carcinoma; WD, well differentiated. In all cases, IHC was performed in representative whole blocks using MLH1 (M1; Ventana, USA) with a BenchMark XT autostainer (Ventana), and MSH2 (G219-1129, 1:500; Cell Marque, USA), PMS2 (MRQ-28, 1:20; Cell Marque), and MSH6 (44/MSH6, 1:500; BD Biosciences, USA) with a Bond-Max autoimmunostainer (Leica Biosystems, Australia). Expression was reported as MMR proficient (pMMR; strong to weak nuclear staining with positive internal controls) or MMR deficient [dMMR; unequivocal loss of nuclear staining or focal (<20%) weak equivocal nuclear staining in the viable tumour cells in the presence of internal positive controls]. For MSI PCR testing, multiplex PCR was performed with five quasi-monomorphic mononucleotide repeat markers in all cases, as previously described.3Cristescu R. Lee J. Nebozhyn M. et al.Molecular analysis of gastric cancer identifies subtypes associated with distinct clinical outcomes.Nat Med. 2015; 21: 449-456Crossref PubMed Scopus (1194) Google Scholar, 6Min B.H. Tae C.H. Ahn S.M. et al.Epstein-Barr virus infection serves as an independent predictor of survival in patients with lymphoepithelioma-like gastric carcinoma.Gastric Cancer. 2016; 19: 852-859Crossref PubMed Scopus (29) Google Scholar Samples with allelic size variation in fewer than two microsatellites were classified as microsatellite-stable (MSS) and allelic size variations in two or more microsatellite markers were considered MSI-H. For cases with discrepant results between IHC and PCR tests, both experiments were repeated three times and the consistent results were used for final analyses. For statistical analyses, we used age, sex, histological type by Lauren classification, World Health Organization (WHO) tumour classification, TNM stage (AJCC 8th edition), and overall survival (OS) of patients as clinical variables with the SPSS 24.0 statistical software program (IBM, USA). MSI/dMMR and clinicopathological variables were compared using Pearson's chi square test and were further analysed with linear-by-linear association. The Kaplan–Meier method was used to estimate OS. p values less than 0.05 were considered statistically significant. MSI-H was found in 60 cases (10.3%) and dMMR was observed in 61 cases (10.5%) out of 580 GC (Fig. 1). All EBV-positive GC was pMMR and MSS. Both dMMR and MSI-H were significantly correlated with old age (p<0.001), low AJCC stage (p<0.003), intestinal-type by Lauren classification (p<0.005), and longer OS (p<0.001) (Table 1, Fig. 2). In multivariate analyses, dMMR and MSI-H were independent favourable prognostic factors (p = 0.001) in addition to AJCC stage (Supplementary Table 1, Appendix A).Fig. 2The Kaplan–Meier curves of overall survival. (A) MSI and (B) dMMR GC were significantly associated with longer overall survival. dMMR, mismatch repair deficient; GC, gastric cancer; MSI, microsatellite instability; MSS, microsatellite stability; pMMR, mismatch repair proficient.View Large Image Figure ViewerDownload Hi-res image Download (PPT) In MMR IHC, 61 dMMR cases consisted of 52 MLH1/PMS2 losses (85.2%), four MSH2/MSH6 losses (6.6%), and five MLH1/PMS2/MSH2/MSH6 losses (8.2%). Of 52 cases with MLH1/PMS2 losses, MSI-H was found in 51 cases (98.1%). All four cases with MSH2/MSH6 losses and five cases with MLH1/PMS2/MSH2/MSH6 losses were confirmed as MSI-H. In total, 60 of 61 dMMR GC with either MLH1/PMS2 or MSH2/MSH6 losses were confirmed to be MSI-H. The positive predictive value of dMMR for MSI-H was 98.4% (60/61). Among 519 MMR-proficient cases, PCR showed MSS in all cases and negative predictive value was 100% (519/519). The sensitivity and specificity of MMR protein IHC for MSI testing were 100% (60/60) and 99.8% (519/520), respectively. We interpreted MMR IHC as unequivocal loss or focal (<20%) weak equivocal nuclear staining in viable tumour nuclei and we found only one (0.2%) case with discrepant results between MSI and MMR IHC. This patient was a 40-year-old female patient and showed 90% loss of MLH1 and 100% loss of PMS2 in IHC and multiplex PCR showed MSS (Supplementary Fig. 1, Appendix A). The patient's GC tumour was stage II poorly differentiated tubular adenocarcinoma and she was still alive without disease after 48.8 months of follow up. However, when we used focal weak equivocal nuclear staining in more than 20% of tumour cells for interpretation of dMMR, we found five discrepant cases and the specificity (99.0%) was decreased. The clinicopathological characteristics of the remaining four patients with focal MMR losses are described in Supplementary Table 2 (Appendix A). Those patients with GC and focal MMR losses were confirmed as MSS and three of them showed distant metastasis after surgery and died of disease. Immunotherapy is a promising cancer therapeutic modality that has recently emerged. MSI-H is a biomarker for anti-PD-1/PD-L1 inhibitors in GC. Clinical trials using pembrolizumab have also been underway in GC.7Muro K. Chung H.C. Shankaran V. et al.Pembrolizumab for patients with PD-L1-positive advanced gastric cancer (KEYNOTE-012): a multicentre, open-label, phase 1b trial.Lancet Oncol. 2016; 17: 717-726Abstract Full Text Full Text PDF PubMed Scopus (773) Google Scholar Therefore, diagnosing MSI is very important to establish a precise therapeutic strategy in patients with relapsed or refractory metastatic GC. Depending on the selection of MSI markers, the frequencies of MSI-H and dMMR GC vary and there has been no standardised MSI test algorithm that can replace PCR in GC. Defects in the DNA MMR proteins result in a phenotype called MSI-H. The prevalence of MSI-H GC has ranged from 0% to 44.5%.8Lee J. Kim K.-M. Biomarkers for gastric cancer: molecular classification revisited.Precis Future Med. 2017; 1: 59-68Google Scholar The difference in overall prevalence of MSI is related to patient cohort, methods used to detect MSI, and use of tissue microarrays for dMMR screening.9Mathiak M. Warneke V.S. Behrens H.M. et al.Clinicopathologic characteristics of microsatellite instable gastric carcinomas revisited: urgent need for standardization.Appl Immunohistochem Mol Morphol. 2017; 25: 12-24Crossref PubMed Scopus (74) Google Scholar In this study of 580 GC cases with their representative whole block, the prevalence of MSI-H was 10.3%, and dMMR was found in 10.5% of cases, which is very similar to recent studies using both MMR IHC and the same PCR testing.10Bae Y.S. Kim H. Noh S.H. et al.Usefulness of immunohistochemistry for microsatellite instability screening in gastric cancer.Gut Liver. 2015; 9: 629-635Crossref PubMed Scopus (33) Google Scholar Intriguingly, we found five MSI-H GC cases with concomitant losses of MLH1/PMS2/MSH2/MSH6 proteins and none of 580 cases showed a single MMR protein loss. For MSI-H in GC, direct comparison with dMMR is not well studied. In a previous large Korean cohort study in 464 GC cases with direct comparison of MSI and MLH1/MSH2 IHC, MLH1 loss was found in 88.2% and MSH2 in 7.4% and they also found co-losses of MLH1 and MSH2 in 4.4% of cases.10Bae Y.S. Kim H. Noh S.H. et al.Usefulness of immunohistochemistry for microsatellite instability screening in gastric cancer.Gut Liver. 2015; 9: 629-635Crossref PubMed Scopus (33) Google Scholar In MSI-GC, it is reported that about ∼90% of cases are associated with MLH1/PMS2 losses. However, depending on microsatellite markers for MSI test, the frequencies of MSH2/MSH6 loss varied from 0% to 37%.8Lee J. Kim K.-M. Biomarkers for gastric cancer: molecular classification revisited.Precis Future Med. 2017; 1: 59-68Google Scholar In the present study with four MMR proteins, we found MLH1/PMS2 losses in 85.2%, MSH2/MSH6 losses in 8.3% and co-losses of MLH1/PMS2/MSH2/MSH6 in 6.7% of MSI-H GC tumours, and the overall prevalence is similar to the previous observations.10Bae Y.S. Kim H. Noh S.H. et al.Usefulness of immunohistochemistry for microsatellite instability screening in gastric cancer.Gut Liver. 2015; 9: 629-635Crossref PubMed Scopus (33) Google Scholar In MMR protein IHC for MSI testing in GC, the sensitivity was 100% and the specificity was 99.8%. This high concordance would be due to strict interpretation criteria to define dMMR (complete or >80% loss of expression). However, when we interpret focal losses of MMR as dMMR, the specificity was decreased although sensitivity was the same. In a previous study using two MMR proteins (MLH1 and MSH2), discordant results were found in 4.7% of cases.10Bae Y.S. Kim H. Noh S.H. et al.Usefulness of immunohistochemistry for microsatellite instability screening in gastric cancer.Gut Liver. 2015; 9: 629-635Crossref PubMed Scopus (33) Google Scholar Use of two more additional antibodies (PMS2 and MSH6) might have worked to increase the sensitivity and specificity because MMR IHC staining varied due to the sensitivities of primary antibodies and age of the paraffin blocks. So, for screening of MSI-H in GC, use of four antibodies is recommended to increase the accuracy to predict MSI. In the present study, we found focal (<20%) losses of MMR in four MSS GC patients and three of them showed distant metastasis shortly after surgery and died of disease. Although we failed to find heterogeneous MSI-H in the present study, focal losses of MMR are interesting findings and warrant further studies exploring the mechanisms underlying focal MMR protein inactivation. Recently, we11Kim S.T. Cristescu R. Bass A.J. et al.Comprehensive molecular characterization of clinical responses to PD-1 inhibition in metastatic gastric cancer.Nat Med. 2018; 24: 1449-1458Crossref PubMed Scopus (731) Google Scholar and Mathiak et al.9Mathiak M. Warneke V.S. Behrens H.M. et al.Clinicopathologic characteristics of microsatellite instable gastric carcinomas revisited: urgent need for standardization.Appl Immunohistochem Mol Morphol. 2017; 25: 12-24Crossref PubMed Scopus (74) Google Scholar found focal losses of MMR proteins and intratumoural heterogeneity in MSI-H patients with advanced stage GC. Given recently reported intratumoural heterogeneity in MSI-H GC and subsequent non-responsiveness for immune checkpoint inhibitors,11Kim S.T. Cristescu R. Bass A.J. et al.Comprehensive molecular characterization of clinical responses to PD-1 inhibition in metastatic gastric cancer.Nat Med. 2018; 24: 1449-1458Crossref PubMed Scopus (731) Google Scholar strict criteria for dMMR are recommended and further clinical correlation studies are recommended to support our hypothesis. This research was supported by grant (NRF-2017R1A2B4012436) from the National Research Foundation of Korea (NRF). The authors state that there are no conflicts of interest to disclose.