Clinicopathological features and prognostic value of mismatch repair protein deficiency in gastric cancer

Abstract

The microsatellite instability (MSI) tumor is one of the four molecular subtypes in gastric cancer (GC). MSI tumors are sensitive to immune checkpoint blockade therapy. However the prevalence and characteristics of MSI in GCs remains unclear. We aimed to clarify relationships between MSI and clinicopathological features along with patients’ survival rates. Data was collected from a cohort of 567 consecutive GC patients who received radical gastrectomy in Fudan University Shanghai Cancer Center. Expression of four DNA mismatch repair proteins (MMRPs)-MLH1, PSM2, MSH2, MSH6 was assessed using immunohistochemistry staining. Absence of any of the four MMRPs was defined as deficiency mismatch repair (dMMR). Tumors with preserved expression of all MMRPs were considered MMR-proficient (pMMR). Chi-squared test or Fisher’s exact probability test was used to detect correlation between MMR status and clinicopathological parameters. Kaplan-Meier method and Log-rank test were used for survival analysis. Fifty-seven cases (57/567, 10.1%) were confirmed as dMMR. The dMMR status was in significant correlation with older age (p<0.001), female gender (p=0.016), distal tumor location in stomach (p=0.002), intestinal Lauren classification (p<0.001), less lymph node metastasis (p=0.040), and less nerve invasion (p=0.016). The dMMR tumors often exhibited unique nested, trabecular or solid growth pattern with an expanding margin and many infiltrating lymphocytes. Patients with dMMR phenotype had improved disease-free survival (p=0.024) and overall survival rates (p=0.025) compared to those with pMMR status. Cox regression analysis manifested dMMR status was an independent factor of better prognosis. In summary, GC with dMMR subtype had distinct clinicopathological features.

Introduction

Gastric carcinoma (GC) is a major public health problem and the leading cause of cancer-related mortality in China [1]. Along with the wide application of gastroscopy, substantial early stage GC with a five-year survival rate of 80-90% was detected [2]. In contrast, advanced GC exhibits a poor prognosis, as it tends to metastasize and recur after radical gastrectomy and chemotherapy. As a clinically heterogeneous disease, patients with the same histological characteristics respond differently to chemotherapy. More specific classifications are demanded for treatment guidance.

Recently, The Cancer Genome Atlas (TCGA) Research Network project analyzed the genomic and proteomic data and presented a brand new molecular classification that divided four subtypes of GC: tumors positive for Epstein-Barr virus, microsatellite instability (MSI) tumors, genomic stable tumors, and tumors with chromosomal instability [3]. Among them, MSI tumors are an important subtype, constitute of 21.7% GC patients.

Maintenance of mismatch repair (MMR) protein function is essential for the fidelity of genetics. Microsatellite instability leads to dysfunction of mismatch repair with subsequent replication errors, and accumulation of mutations throughout the genome. Some of those mutated genes are involved in carcinogenesis [4]. For example, it has been reported that MSI gastric carcinoma frequently shows activation of EGFR-MAPK and PI3K pathways [3,5]. Moreover, currently accumulating clinical trials have demonstrated that MSI status predicted benefit of immune checkpoint blockade therapy in colorectal cancers (CRC) [6,7]. Therefore, as one important subtype of GC, it’s urgent to achieve a more understanding of MSI. However, compared with other MSI related tumors-CRC [8,9], data concerning the clinical role of MSI in GC remains scare and discordant. Quite a few studies reported that MSI was associated with a better prognosis [10,11], while An J et al. reported that MSI status itself was not a prognostic indicator in GC [12].

Previous literature has reported that the prevalence of MSI in GC ranged from 7% to 25% [11,13,14]. Geographic disparities and different detecting methods can lead to the variance. Therefore we sought to evaluate the prevalence of mismatch repair-deficiency (dMMR) phenotype in a large cohort of GC patients. We detected the four DNA mismatch repair proteins (MMRPs)-MLH1, PSM2, MSH2, MSH6 by immunohistochemistry staining, which is the method similar to the CRC MMR standard evaluation.

The aim of the current study was to evaluate the clinical role of MMR status in GC and provide information for further immunotherapy. We tried to elucidate the correlations between MMR status and clinicopathological characteristics as well as patients’ outcomes.

Methods

Patients

A total of 567 patients who underwent radical surgical resection for primary gastric carcinoma at Fudan University Shanghai Cancer Center from 2010 to 2012 were evaluated retrospectively. This study was approved by the Clinical Research Ethics Committee of Fudan University Shanghai Cancer Center. Written informed consent was obtained from all the participants involved in this study.

All the cases were histopathologically confirmed as gastric adenocarcinoma according the WHO classifications of Tumors of Digestive System 2010 version. Gastric adenocarcinoma stage was determined using the TNM classification system [15]. Clinical data, including patients’ age, gender, tumor location, and tumor size were collected from the medical record system. Patient follow-up was performed every 3 months during the first year after surgery and every 3-6 months thereafter until April 30, 2017. Disease-free survival (DFS) was calculated from the date of surgery to the date of recurrence or the last follow-up. Overall survival (OS) was calculated from the date of surgery to the date of death or last follow-up.

Evaluation of pathological features and Tissue handling

Hematoxylin and eosin (H&E) slides were collected from the archives of the Pathology Department of Fudan University Shanghai Cancer Center. One experienced pathologist specialized at gastroenterology reviewed the H&E-stained slides to evaluate the following pathological features including histologic stage, depth of invasion, and the status of lymphovascular and nerve invasion. This pathologist also selected the representational H&E slide. Two cores of 1 mm in diameter were drilled from the selected paraffin-embedded block with the most obvious lesions in each case. Then tissue cores were transplanted to the tissue micro arrays (TMAs) blocks.

Immunohistochemistry

Immunohistochemistry (IHC) was a widely used method for CRC dMMR screening due to its relatively high specificity and sensitivity [16,17]. So IHC including four-antibody MMRPs was applied in our MMR status detection. It was performed using the 4-μm TMAs sections of paraffin-embedded specimens on an automated stainer (Ventana, Tucson, AZ, USA). Primary antibodies included anti-MLH1 (Clone G168-728), anti-MSH2 (Clone G219-1129), anti-MSH6 (Clone 44), and anti-PMS2 (Clone EPR3947), all purchased from Roche (Basel, Switzerland). Omission of the primary antibody and substitution with non-specific immunoglobulins were used as negative controls.

The stained slides were evaluated by two independent experienced pathologists specialized in gastroenterology. Normal colonic mucosa (adjacent to the carcinoma) and lymphocytes were used as positive internal controls. Representative images of IHC staining are shown in Figures 1 and 2. Complete loss of nuclear staining in tumor cells was considered loss of expression. The absence of expression of any of the four MMRPs was defined as dMMR. Tumors with preserved expression of all MMRPs were considered MMR-proficient (pMMR).

Figure 1.

Representative images of MLH1 and PMS2 immunostaining (40×). Negative (A) and positive (B) staining of MLH1 in different gastric cancer tissues. Negative (C) and positive (D) staining of PMS2 in different gastric cancer tissues.

Figure 2.

Representative images of MSH2 and MSH6 immunostaining (40×). Negative (A) and positive (B) staining of MSH2 in different gastric cancer tissues. Negative (C) and positive (D) staining of MSH6 in different gastric cancer tissues.

Statistical analysis

Associations between clinicopathological features and MMR status were explored utilizing the Chi-square test or Fisher’s exact probability test. DFS and OS curves were calculated using the Kaplan-Meier method and analyzed using the Log-rank test. Features proved to have significant impact in survival were further included in the Cox regression analysis. All statistical analyses were performed using SPSS 20.0 (IBM, SPSS, Chicago, IL, USA). A value of P<0.05 indicated a significant difference.

Results

Patient characteristics and MMR status

As shown in Table 1, the study was composed of 567 GC cases, including 388 males and 179 females. The mean and median age of the patients was 59.4 and 59 years (range 21-84 years). The median tumor size was 4 cm (range 2-15 cm). A total of 57 (57/567, 10.1%) cases were defined as dMMR by IHC. Table 2 displays detailed expression modes of those dMMR cases. Among them, 39 cases (39/57, 68.4%) lost the expression of MLH1 and PMS2. 5 cases (5/57, 8.8%) negatively expressed MSH2 and MSH6. Concurrent loss of MLH1, PMS2, and MSH6 was observed in 6 cases (6/57, 10.5%). 2 cases (2/57, 3.5%) showed isolated absence of MSH6 staining. 5 tumors (5/57, 8.8%) exhibited the negative staining of the four MMR proteins.

Table 1.

Relationship between clinicopathological factors and MMR status

Clinicopathological feature		pMMR%	dMMR%	X2	p-value
All cases	567	510 (89.9)	57 (10.1)
Age	<60	273 (53.5)	15 (26.3)	15.192	<0.001*
	≥60	237 (46.5)	42 (73.7)
Gender	Male	357 (70.0)	31 (54.4)	5.786	0.016*
	Female	153 (30.0)	26 (45.6)
Location (Stomach)	Upper	138 (27.1)	10 (17.5)	6.346	0.042*
	Middle	189 (37.1)	17 (29.8)
	Lower	183 (35.9)	30 (52.6)
Size (cm)	<5	336 (65.9)	31 (54.4)	2.968	0.085
	≥5	174 (34.1)	26 (45.6)
T stage	T1	80 (15.7)	4 (7.0)	4.469	0.215
	T2	58 (11.4)	10 (17.5)
	T3	283 (55.5)	34 (59.6)
	T4	89 (17.5)	9 (15.8)
N stage	0	158 (31.0)	23 (40.4)	10.052	0.040*
	1	102 (20.0)	16 (28.1)
	2	112 (22.0)	6 (10.5)
	3a	88 (17.3)	11 (19.3)
	3b	50 (9.8)	1 (1.8)
M stage	0	498 (97.6)	57 (100.0)	1.370	0.242
	1	12 (2.4)	0 (0.0)
TNM stage	1	102 (20.0)	9 (15.8)	5.090	0.165
	2	118 (23.1)	20 (35.1)
	3	278 (54.5)	28 (49.1)
	4	12 (2.4)	0 (0.0)
Tumor grade	Low	58 (11.4)	8 (14.0)	0.353	0.552
	High	452 (88.6)	49 (86.0)
Lauren classification	Intestinal	206 (40.4)	29 (50.9)	22.655	<0.001*
	Diffuse	84 (16.5)	6 (10.5)
	Mixed	206 (40.4)	14 (24.6)
	Unclassified	14 (2.7)	8 (14.0)
LVI	-	253 (49.6)	34 (59.6)	2.068	0.150
	+	257 (50.4)	23 (40.4)
Nerve invasion	-	232 (45.5)	38 (66.7)	9.218	0.002*
	+	278 (54.5)	19 (33.3)

LVI, Lymphovascular invasion.

^* p<0.05.

Table 2.

The detail expression patterns of the 57 dMMR cases

Markers	MLH1	PMS2	MSH2	MSH6	N (%)
Expression status	-	-	+	+	39 (68.4)
	-	-	+	-	6 (10.5)
	-	-	-	-	5 (8.8)
	+	+	-	-	5 (8.8)
	+	+	+	-	2 (3.5)

Correlation between MMR status and clinicopathological features

We then evaluated the correlation between MMR status and clinicopathological factors (Table 1), dMMR phenotype was significantly associated with older age (p<0.001), female gender (p=0.016), distal tumor location (p=0.002), intestinal Lauren classification (p<0.001), less advanced N station (p=0.040), and less nerve invasion (p=0.016). Other features such as tumor size, tumor differentiation grade, TNM stage, and the status of lymphovascular invasion didn’t show any relationships with the MMR expression phenotype.

Moreover, besides the correlation with intestinal classification, dMMR tumors frequently exhibited nested or solid growth pattern with a pushing border in the deep tumor front. There was lack of desmoplastic reaction in dMMR tumors compared with pMMR tumors. Instead a plenty of infiltrating lymphocytes were often seen in dMMR cases.

Correlation between MMR status and clinical prognosis

The median DFS and OS time for patients was 55 and 57 months. The Kaplan-Meier analysis revealed that patients with dMMR phenotype had longer DFS (p=0.024) and OS (p=0.025) compared those with pMMR status (Figure 3A and 3B). Cox regression analysis also manifested that the dMMR status was an independent factor of improved DFS and OS (Tables 3 and 4). Patient’s age and TNM stage were also proved as independent factors of DFS and OS.

Figure 3.

Kaplan-Meier plots of disease-free survival and overall survival according MMR status in the entire cohort (A, B) and in stage II GC patients (C, D).

Table 3.

Univariate and Multivariate analysis of disease-free survival in GC

Variables	Categories	Univariate analysis		P value	Multivariate analysis		P value
		HR	95% CI		HR	95% CI
Patient age	≥60/<60 y	1.49	1.11-2.00	0.007	0.67	1.11-2.90	0.008*
TNM stage	I/II/III/IV	3.24	2.46-4.26	<0.001	2.98	2.21-4.03	<0.001*
Tumor size	≥5/5 cm	1.40	1.04-1.88	0.027
Lymphatic invasion	+/-	2.28	1.68-3.09	<0.001
Nerve invasion	+/-	2.23	1.63-3.06	<0.001
MMR status	dMMR/pMMR	0.49	0.26-0.93	0.028	0.50	0.27-0.97	0.037*

HR, hazard ratio, 95% CI, 95% confidence interval.

^* p<0.05.

Table 4.

Univariate and Multivariate analysis of overall survival in GC

Variables	Categories	Univariate analysis		P value	Multivariate analysis		P value
		HR	95% CI		HR	95% CI
Patients’age	≥60/<60 y	1.66	1.23-2.24	0.001	1.69	1.24-2.29	0.001*
TNM stage	I/II/III/IV	3.17	2.41-4.18	<0.001	2.98	2.20-4.04	<0.001*
Tumor size	≥5/5 cm	1.44	1.07-1.95	0.017
Lymphatic invasion	+/-	2.22	1.63-3.03	<0.001
Nerve invasion	+/-	2.11	1.54-2.91	<0.001
MMR status	dMMR/pMMR	0.47	0.24-0.93	0.029	0.47	0.24-0.93	0.031*

HR, hazard ratio, 95% CI, 95% confidence interval.

^* p<0.05.

We further performed strata survival stratification analysis according the TNM stage. Because the 12 stage IV cases were all pMMR, the survival curves of different MMR status couldn’t be evaluated in stage IV patients. In stage II GC patients, the DFS rate was significantly higher in dMMR cohorts than the pMMR cohorts (Figure 3C), but the difference between OS and MMR status didn’t reach statistical significance (Figure 3D). As for stage I and III GC patients, the survival rates didn’t show any difference according the varied MMR status (Figure 4).

Figure 4.

Kaplan-Meier plots of disease-free survival and overall survival according MMR status in stage I (A, B) and stage III GC patients (C, D).

Discussion

Recently, immunotherapy has achieved great clinical success in combating cancers. Objective responses of programmed death-1 (PD-1) blockade were achieved in in CRC patients with MSI status [6,18]. This encouraging effect raises the interest of researches in GC with MSI. As an important subset of gastric cancer, the biological features of MSI calls for more research focus. Our study first investigated the MMR status in a large group of Chinese patients. The prevalence of dMMR in our cohort was 10.1%, dMMR GC had distinct clinopathlogical characteristics, and was significantly associated with improved DFS and OS.

Our research manifested patients with dMMR subtype had favorable prognosis. dMMR tumors were associated with older age, female gender, intestinal histology, distal location, and reduced lymph node metastasis, along with the distinct morphology appearance of solid and medullary growth pattern, which was in concordance with previous reports [14,19]. Moreover, we also found that dMMR tumors tended to show less nerve invasion. The favorable prognostic impact along with those less aggressive clinicopathological characteristics indicates that MMR status is a valuable biomarker. We recommend routine testing for MMR status in every GC case, similar to CRC.

Although several studies reported that MSI status itself was not a prognostic indicator in GC [12,13], accumulating evidences demonstrated dMMR GC exhibited better survival [10,20,21]. Our research showed that patients with dMMR subtype had favorable prognosis. The reason of better outcomes of MSI GC has not been clearly clarified. In our study, abundant infiltrating lymphocytes were observed in dMMR tumors, we consider there occur due to an active host-tumor interaction process as a result of increased tumor antigens in dMMR cancers. It’s been reported that MSI-high tumors harbor more neoantigens that might be recognized by the immune system [4]. Some studies also implied that dMMR tumors were highly infiltrated with T-cells including cytotoxic T-lymphocytes [22], which may actively inhibit tumor growth. Moreover, it’s been confirmed in the studies of CRCs that some of targeted checkpoints including PD-1, programmed death-ligand 1 (PD-L1) were up-regulated in MSI subtype tumors [23]. The result that the blockade of PD-1 was selectively efficacious in MSI CRC patients indicated that MSI was a promising factor for immunotherapy [6,18]. Similarly, researchers also found that MSI GC overexpressed PD-L1 at tumor invasive front [24]. Recently a Phase 1b clinical trial showed that anti-PD-1 monoclonal antibody pembrolizumab provides sustained antitumor activity in PD-L1 positive advanced gastric cancer [25]. In addition, Hause R et al. revealed that dMMR cancers were sensitive to PD-1 blockade, regardless of the tissue of origin [26]. Those studies confirmed the important role of MMR status in predicting the benefit of immunotherapy.

Some research has also suggested that dMMR status can be used to predict the efficacy of chemotherapy in gastric cancer. Kim S et al. evaluated the impact of chemotherapy in stage II and III gastric cancer and demonstrated that chemotherapy attenuated the benefits of MSI gastric cancers [27]. Similarly, Sohn BH et al. reported that patients with the MSI subtype showed only moderate benefit from adjuvant chemotherapy [28]. Our research manifested that patients with dMMR subtype had favorable prognosis in Kaplan-Meier analysis. The subgroup analysis indicated dMMR cases linked to better DFS rate in stage II patients, while stage III dMMR GC patients didn’t exhibit any survival advantages. However, the Cox regression analysis manifested that the dMMR status was an independent factor of improved DFS and OS in the whole GC group. Since patients in our study received different chemotherapeutic regimens and targeted drugs, we supposed it might be the confounding factor of prognosis analysis. The application of chemotherapy might worsen the improved survival advantage result from dMMR. Further investigation can focus on the response of specific chemotherapy or targeted immunotherapy in different MMR status GC patients.

This study has some limitations. The expression status from two tissue cores of 1 mm in diameter wouldn’t always be consistent with the real condition. The prevalence of dMMR detected by TMA in our GC cohort was 10.1%. Our result of dMMR rates might be underestimated. First, as we all know that GC is a carcinoma with high heterogeneity, so the staining of MMRPs would be different in distinct tissue areas. Future study can raise the number of cores to elevate the accuracy of detection. Second, although the sensitivity of IHC was near 90%, a proportion of MSI tumors might be missed by the four-antibody IHC method since some uncommon MMRPs, such as MSH3 and PMS1 could also lead to MSI [4,29].

Conclusion

In summary, our study found that the prevalence of dMMR in GC patients was 10.1%. Gastric cancer with microsatellite instability had unique morphology, distinct clinicopathological features and significant prognosis impact. dMMR-GC is an important subtype of GC in the emerging development of immune check point blockade therapy.

Disclosure of conflict of interest

None.