The Impact of Mismatch Repair Status on Prognosis of Patients With Gastric Cancer: A Multicenter Analysis

Background The clinical role of deficient DNA mismatch repair (dMMR)/microsatellite instability-high (MSI-H) in gastric cancer (GC) is still controversial. We aimed to analyze the relationship between dMMR/MSI-H and clinicopathological features along with survival. Methods Patients who were diagnosed with GC at the three big cancer centers in China from 2015 to 2020 were evaluated retrospectively. MMR/MSI status was assessed using immunohistochemistry/PCR. Clinical and pathological data were collected from the medical record system. Results A total of 196 patients with dMMR/MSI-H status were enrolled for analysis. The prevalence of MSI-H/dMMR in GC was 6.6%. Another 694 proficient MMR (pMMR) GC patients were enrolled for comparison. Compared with pMMR patients, dMMR/MSI-H patients were associated with older age, female predominance, distal location in the stomach, earlier TNM stage, intestinal subtype, better differentiation, and more negative HER2 status. The median overall survival (OS) of the dMMR/MSI-H group was better than that of the pMMR/microsatellite stability (MSS) group (not reached vs. 53.9 months, p = 0.014). Adjuvant chemotherapy had no impact in both disease-free survival (DFS) and OS of dMMR/MSI-H patients (p = 0.135 and 0.818, respectively). dMMR/MSI-H patients had poorer response and progression-free survival (PFS) of first-line chemotherapy, though they were statistically significant (p = 0.361 and 0.124, respectively). Conclusions dMMR/MSI-H GC patients have specific clinicopathological characteristics and better prognosis than pMMR patients.


INTRODUCTION
Gastric cancer (GC) is one of the most common causes of cancerrelated mortality worldwide. As a heterogeneous disease, patients with the same TNM stage and histological characteristics may respond differently to treatment and have different survival. Hence, specific classification was presented for guidance of clinical decision making. The Cancer Genome Atlas (TCGA) Research Network has identified four distinct molecular subtypes of GC through molecular evaluation of 295 GC patients: Epstein-Barr virus (EBV) positive, microsatellite instability (MSI), chromosomal instability (CIN), and genomically stable (GS) (1). As a result of dysfunction of mismatch repair (MMR), MSI leads to increased rate of replication error and hypermutational status, which results in increased probability of mutations in oncogenes or tumor suppressors. MSI status is commonly assessed using PCR. MMR proteins, including FmutL homologue 1 (MLH1), mutS homologue 2 (MSH2), mutS homologue 6 (MSH6), and PMS1 homologue 2 (PMS2), were determined by immunohistochemical (IHC) analysis. The concordance between MSI-high (MSI-H) status and deficiency of MMR protein function (dMMR) was 97.6%-99% (2,3).
According to the data of TCGA, 21.7% patients of GC were identified as MSI. However, it did not distinguish between MSI-H and MSI-low (MSI-L). The prevalence of MSI-H/dMMR status in GC ranged from 8% to 25% in previous reports (4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14). MSI-H/dMMR status was a predictive marker of response to immunotherapy (15). Besides, MSI-H/dMMR status has been reported to be a prognostic and predictive factor in the adjuvant setting. In colorectal cancer, MSI-H/dMMR tumor shows better prognosis, and may be associated with lack of benefit from adjuvant chemotherapy in stage II disease (16,17). Similar results were reported in GC patients (14). Recently, an individual patient data meta-analysis from four large randomized clinical trials performed in patients with resectable GC (MAGIC, CLASSIC, ARTIST, and ITACA-S) showed that patients with MSI-H/dMMR GC did not benefit from adjuvant chemotherapy after radical surgery (14). However, in this pooled analysis of four clinical trials, the number of MSI GC patients was still relatively low (N = 121), which made the statistical power limited. Moreover, due to the low prevalence of MSI-H/dMMR in GC, the clinical and pathological features, response to chemotherapy, and overall survival (OS) are still controversial.
With this in mind, we conducted this retrospective study, enrolling MSI-H/dMMR GC patients from three big cancer centers in China. By this, we tried to expand the sample size and explore the clinicopathological characteristics and predictive and prognostic values of MSI-H/dMMR status for GC.

Patients
Patients who were diagnosed as GC at the three big cancer centers in China (Harbin Medical University Cancer Hospital, Fudan University Zhongshan Hospital, and Sun Yat-sen University Cancer Center) from 2015 to 2020 were evaluated retrospectively. These three hospitals all have large patient volume in China. The studies involving human participants were reviewed and approved by the ethics committee of each hospital. All the patients were diagnosed as GC by H&E staining and histological analysis. The stage was determined using the American Joint Committee on Cancer (AJCC) TNM 8th stage system. Clinical data, including sex, age, family history, and primary tumor location, were collected from the medical record system. Patients with dMMR/MSI-H status were enrolled for analysis (a total of 196 cases, including 72 from Harbin, 71 from Shanghai, and 53 from Guangzhou). Besides, 694 cases with proficient MMR (pMMR) status diagnosed at Sun Yat-sen University Cancer Center in the same period were enrolled as comparison.

Mismatch Repair/Microsatellite Instability Assessment
For MMR protein IHC analysis, 4-mm formalin-fixed, paraffinembedded sections were prepared from the tissue blocks and stained for the MLH1, MSH2, MSH6, and PMS2 proteins. Primary antibodies included anti-MLH1 (M1, Ventana, USA), anti-MSH-2 (RED2, ZSGB-bio, China), anti-MSH6 (SP93, Ventana, USA), and anti-PMS2 (EP51, Dako, Denmark). Loss of MMR protein expression was designated when none of the neoplastic epithelial cells had nuclear staining, while positive internal control nuclei (lymphocytes and stromal cells) were present in the immediate vicinity of the tumor infiltrate. Normal expression was defined as the presence of nuclear staining of tumor cells irrespective of the proportion or intensity. A case was classified as dMMR if tumor cell nuclei were negative for one or the four MMR proteins in the presence of positively stained lymphocytes or fibroblasts as internal control. pMMR was defined if tumor cell nuclei, irrespective of the number or intensity, were positive for all MMR proteins tested.
The MSI status was evaluated using PCR. DNA was extracted from paired normal/tumor tissues that were formalin-fixed, paraffin embedded. Then PCR amplification was performed for two mononucleotide repeat markers (BAT25 and BAT26) and three dinucleotide markers (D5S346, D2S123, and D17S250) (11). MSI-H was defined as two or more markers mentioned above with instability. Otherwise, it was defined as microsatellite stability (MSS).

Statistical Analysis
The patients' clinicopathological features were summarized with descriptive statistics. Categorical variables were compared using chi-square test, and comparisons of continuous variables were performed using Student's t-test. Five-year cause-specific survival (CSS) was calculated from the date of diagnosis to the date of cancer-specific death. Survival among different variables was compared using the Kaplan-Meier estimates and the log-rank test. Statistical analysis was carried out by the IBM SPSS Statistics 22.0.0 package software (SPSS Inc.) and the Intercooled Stata 13.0 (Stata Corporation, College Station, TX). All the p-values were two-sided, and statistical significance was set at p < 0.05.

Mismatch Repair Expression Mode
The detail of MMR protein expression mode is shown in Table 2. Most common defective expression was seen in MLH1 and PMS2 (153, 78.1%). Twenty-two cases (11.2%) only negatively expressed PMS2, and eight cases (4.1%) had concurrent loss of MSH2 and MSH6. We also found two cases that were pMMR but turned out to be MSI-H using PCR.

Survival of Patients With Deficient Mismatch Repair/Microsatellite Instability-High Gastric Cancer
The median OS of dMMR/MSI-H group was significantly better than that of the pMMR/MSS group (not reached vs. 53.9 months, p = 0.014, Figure 1). We examined the outcomes stratified by TNM stages. We found that the OS was not remarkably different in each TNM stage group ( Figure 2). In stage IV, the OS of dMMR/MSI-H patients was numerically better than that of pMMR/MSS patients (56.5 vs. 25.6 months, p = 0.052), but it was not statistically significant. The multivariate analysis showed that TNM stage was the only prognostic factor associated with OS ( Table 3).

Adjuvant Chemotherapy in Deficient Mismatch Repair/Microsatellite Instability-High Gastric Cancer
One hundred nineteen (60.7%) dMMR/MSI-H cases were diagnosed at stage II or III, and 117 of them received radical surgery. Eighty-six patients with detail record of adjuvant therapy and follow-up were enrolled for analysis. Seventy-one of them received adjuvant chemotherapy after surgery, and the remaining 15 only received surgery. Compared with the DFS of pMMR GC, the DFS of dMMR/MSI-H was longer (46.9 vs. 37.1 months), though the statistical significance was still not reached (p = 0.486, Figure 3). The disease-free survival (DFS) was 46.9 and 21.9 months for patients with and without adjuvant chemotherapy, respectively, but the difference was not significant (p = 0.135, Figure 3 and Table 4).

Chemotherapy and Immunotherapy in Advanced Deficient Mismatch Repair/ Microsatellite Instability-High Gastric Cancer
The response and progression-free survival (PFS) of first-line chemotherapy (without combination of immunotherapy) are shown in  Figure 5).

DISCUSSION
This study investigated the clinical and pathological features of dMMR/MSI-H GC and the role of the MMR/MSI status as prognostic and predictive biomarkers of GC. To our knowledge, this is the largest cohort of dMMR/MSI-H GC reported. The prevalence of dMMR/MSI-H in our cohort was 6.6% (4.7%-8.3% in three cancer centers), which was similar to the results reported in previous eastern studies (8%-10%) (4-9), but lower than that in western studies (20% and more) (10,12,13). The possible explanation is the difference in gastric carcinogenesis background between eastern and western patients. Besides, MMR/MSI status was associated with several clinical and pathological features such as age, sex, primary site, histology, and Lauren classification. As a result, different clinical and pathological characteristics in eastern and western GC could be responsible of the incidence of MMR/MSI reported in previous studies. In our study, dMMR/MSI-H was associated with older age, female patients, distal location, intestinal subtype, and better differentiation.
Like colorectal cancer, dMMR/MSI-H GC was more often seen in early stage. Whether MMR/MSI status was an independent prognostic factor was still controversial. Some researchers reported it was not a prognostic indicator in GC (4,18), but others demonstrated that dMMR GC exhibited favorable OS (13,19,20). Zhang et al. reported that dMMR status was an independent factor for better prognosis (8). Our study did show that GC patients with dMMR/MSI-H subtype had better OS. However, the prognostic impact of MMR/MSI status was lost on multivariate analysis. As we have mentioned above, dMMR/MSI-H subtype came along with those less aggressive clinical and pathological characteristics such as intestinal histotype and better differentiation. Moreover, dMMR/MSI-H was commonly seen in early TNM stage; therefore, the prognostic value of dMMR/MSI-H may be confounded by other clinical factors, especially the TNM stage.
Although the prognostic value of MMR/MSI status was still controversial, accumulating evidences had identified MSI status     respectively) (26). Our findings were partly in contrast to previous observations. The PFS and ORR of dMMR/MSI-H patients in our study were worse than those in pMMR patients, though the statistical significance was not reached.
First, the regimens of chemotherapy were various in our study, which might explain the conflicting results. Second, most recurrent GC patients with dMMR/MSI-H in our study received adjuvant chemotherapy before. According to the replication result in expressions of neoantigens, which result in high mutation burden and act as a potential target for immunecells (27). Attraction of immune cells into tumor environment leads to immune stimulation. It has been reported that high density of intratumoral CD8+ and FoxP3+ tumor-infiltrating lymphocytes were associated with good prognosis (22). Hence, it is reasonable to administrate immune checkpoint inhibitors to enhance the effect of immune stimulation in dMMR/MSI-H tumors (15). In KEYNOTE-059 trial, the response rate of pembrolizumab in GC patients with MSI-H was 57.1%, while it was 9.0% in MSS GC (28). Furthermore, in KEYNOTE-061 trial, anti-PD-1 monotherapy showed better response rate than chemotherapy (paclitaxel) alone in MSI-H GC patients (29). Our study also showed that immunotherapy with or without chemotherapy had better response rate in dMMR/MSI-H patients, though the statistical significance was not reached due to the small sample size. This result suggested that dMMR/MSI-H was a reliable biomarker in predicting the effect of immunotherapy.
Considering the low prevalence of dMMR/MSI-H in GC, this cohort, which enrolled 196 dMMR/MSI-H cases, may be the largest one to date. However, our study has several limitations. First of all, it is a retrospective study, and selection bias inevitably exists. For example, most patients after radical surgery also received standard adjuvant chemotherapy, so it is hard to evaluate the prognostic and predictive roles of adjuvant chemotherapy. Second, the retrospective design and various regimens of chemotherapy used in adjuvant or first-line made it difficult to conduct more subgroup analysis for the effect of chemotherapy or immunotherapy. Third, some important biomarkers associated with immunotherapy, such as tumor mutation burden (TMB) and PD-L1 expression, were not available in this study.

CONCLUSION
In summary, dMMR/MSI-H GC patients have specific clinical and pathological characteristics, such as older age, female predominance, distal location in the stomach, earlier TNM stage, intestinal subtype, better differentiation, and more negative HER2 status. Although dMMR/MSI-H is a predictive factor of immunotherapy in advanced stage, it was not an independent prognostic factor in GC. Moreover, the predictive and prognostic value of chemotherapy for dMMR/MSI-H GC in adjuvant or first-line setting is not clear, which should be further investigated in prospective clinical trials.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary Material.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by the ethics committee of Sun Yat-sen University Cancer Center (B2020-335-01). Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.