Clinical significance of MLH1/MSH2 for stage II/III sporadic colorectal cancer.

BACKGROUND: The development of colorectal cancer (CRC) is a complicated multistep process that involves an accumulation of mutations in tumor suppressor genes and oncogenes. In the process of DNA replication, base mismatch often occurs due to various factors leading to abnormal expression of mismatch repair genes (MMR), among which MLH1 and MSH2 are the most important. Recently, numerous studies indicated that MLH1/MSH2 phenotype is associated with CRC. We wanted to elucidate the role of MLH1/MSH2 in the prediction and prognosis of CRC through long-term clinical observation. AIM: To evaluate the prognostic and predictive significance of MLH1/MSH2 in patients with stage II-III CRC using immunohistochemical analysis and GeneScan.
METHODS: Specimens from 681 patients with CRC (395 stage II and 286 stage III, 387 males and 294 females) who underwent curative surgical resection from 2013 to 2016 were tested. Immunohistochemistry was used to analyze MMR status and the microsatellite status of 133 patients was determined by GeneScan analysis.
RESULTS: Five hundred and fifty (80.76%) patients were MLH1/MSH2 positive and 131 (19.24%) were negative by immunohistochemistry. MLH1/MSH2-positive tumors were significantly more frequent in the colon than in the rectum, and had poor differentiation and less mucin production (P < 0.05). Patients of different groups did not differ in terms of age, gender, tumor size, tumor stage, lymphocytic infiltration, or circumscribed margin. MLH1/MSH2-negative patients had a more favorable OS than MLH1/MSH2-positive patients (P < 0.001). Univariate and multivariate analyses demonstrated MLH1/MSH2 expression as an independent prognostic and predictive factor for stage II/III CRC. MLH1/MSH2 expression was a strong prognostic factor in all patients [P < 0.001, hazard ratio (HR) = 4.064, 95%CI: 2.241-7.369]. Adjuvant chemotherapy had a greater correlation with survival advantage in MLH1/MSH2-negative patients with stage III disease (P < 0.001, HR = 7.660, 95%CI: 2.974-15.883). However, patients with stage II disease or MLH1/MSH2-positive patients with stage III disease did not benefit from adjuvant chemotherapy. GeneScan analysis demonstrated that among 133 patients, 105 (78.95%) were microsatellite stable, and 28 (21.05%) had microsatellite instability (MSI), including 18 (13.53%) with high MSI and 10 (7.52%) with low MSI. This is consistent with the immunohistochemical results.
CONCLUSION: MLH1/MSH2 phenotype constitutes a pathologically and clinically distinct subtype of sporadic CRC. MLH1/MSH2 is an independent prognostic and predictive factor for outcome of stage II-III CRC. ©The Author(s) 2019. Published by Baishideng Publishing Group Inc. All rights reserved.

Core tip: Mutation or methylation of mismatch repair gene leads to microsatellite instability (MSI), which is one of the most important mechanisms for the development of colorectal cancer (CRC). The purpose of this study was to collect data on MLH1/MSH2 phenotype and MSI status in stage II-III CRC patients and to assess their predictive and prognostic value. This is the first large study in China to evaluate the role of MLH1/MSH2 in CRC and its relationship with adjuvant chemotherapy.

INTRODUCTION

Colorectal cancer (CRC) is one of the most common malignancies of the digestive tract. In 2017, there were nearly 135430 newly diagnosed CRC cases with 50260 associated deaths in the United States[1]. CRC is mainly associated with at least three distinct genetic pathways: Microsatellite instability (MSI), chromosomal instability (CIN), and CpG Island methylator phenotype (CIMP)[2]. Most hereditary nonpolyposis CRC and 15%[3] of sporadic CRC are characterized by MSI[4,5]. In contrast, 85% of CRC develop from the CIN pathway and are characterized by aneuploidy, allelic losses, amplifications, and translocations[6]. Meanwhile, many sporadic MSI CRC are also CIMP positive. These three pathways are not mutually exclusive, and most tumors are characterized by multiple pathways. The mismatch repair (MMR) gene[7] is a housekeeping gene that is highly conserved. MMR maintains correct DNA replication and high fidelity by repairing DNA base mismatches, which allows for genomic stability and reduces spontaneous mutations[8]. MSI is characterized by the deletion of DNA methylation or MMR caused by genetic mutation, which leads to widespread alterations in the length of short repeated sequences[9,10]. In China, the incidence of CRC has increased significantly in the last 10 years. While the mortality rate of male patients has increased annually, the mortality of female patients has tended to be relatively stable. CRC is the fifth leading cause of morbidity in men and fourth in women. The number of new cases of CRC in 2015 in China was 376300, including 215700 men and 160600 women. There were 191000 deaths due to CRC, including 111100 men and 80000 women[11].

Most cases of MSI appear to result from MMR deficiency. At least six of the genes involved in MMR have now been identified, including MutL homolog 1 (MLH1), MutS homolog 2 (MSH2), MutS homolog 3 (MSH3), postmeiotic segregation increased 1 (PMS1), postmeiotic segregation increased 2 (PMS2), and MutS homolog 6 (MSH6)[12]. MLH1 is located on chromosome 3p21–23 and is connected with the creation of MLH3, PMS2, and PMS1. MSH2 is located on chromosome 2p21 and has been shown to form MSH3 and MSH6[13]. Mutations in either MLH1 or MSH2 account for the majority of known germline mutations in CRC, and >90% of MMR deficiencies are deletions of MLH1 or MSH2 that rarely appear in other genes, which are the major causes leading to the mutated phenotype[14,15]. Mutations either in MLH1 or MSH2 or both of them are considered as MLH1/MSH2 negative, and no mutations in either of them are considered as MLH1/MSH2 positive. MLH1 recruits and allocates other proteins to the mismatch repair system and MSH2 can recognize any errors in DNA replication and replace the incorrect sequence using the parental strand sequence as the correct one[16].

The purpose of the current study was to evaluate the prognostic significance of MLH1/MSH2 status determined by immunohistochemical analysis in a large cohort of patients with stage II-III CRC. In particular, we sought to detect the relationship between MLH1/MSH2 and overall survival (OS).

MATERIALS AND METHODS

Patients

Initially, we enrolled 836 consecutive patients who underwent curative-intent surgical resection between January 2013 and December 2016 at the Nanjing Hospital of Chinese Medicine Affiliated to Nanjing University of Chinese Medicine (Nanjing, China). One hundred and fifty-five patients were excluded due to loss to follow-up or because their specimens were not available for immunohistochemical analysis. There were at least 681 patients who had all data available at the time of follow-up (387 males and 294 females with a median age of 63 years; range 22–87 years). Diagnosis of CRC was confirmed according to World Health Organization (WHO) criteria and the TNM stage classification[17]. Clinical data were obtained from hospital medical records and included details pertaining to patient gender and age; tumor differentiation, location, and size; and mucin, surgical margin, TNM stage, lymph node (LN) metastasis, and histopathological grade. Follow-up of all cases started from the postoperative period to December 2018. Of the 681 patients with CRC, 300 underwent surgery only and 381 surgery plus adjuvant chemotherapy. One hundred and twenty-two (30.9%) of the 395 patients with stage II CRC and 259 (90.6%) of 286 with stage III CRC received postoperative adjuvant chemotherapy. Patients with stage II CRC were subdivided into high- and low-risk categories according to poor prognostic features. The National Comprehensive Cancer Network (NCCN) guidelines[18] consider the following high-risk factors for recurrence: Poorly differentiated histology [exclusive of those cancers that have MSI (MSI-H)] or undifferentiated tumors; pathological T4 (pT4) disease; perineural invasion; bowel obstruction; indeterminate or positive margins or localized perforating tumors; and inadequate LN sampling (<12 LNs).

Treatment

Combining the NCCN guidelines with patients’ personal wishes, for stage II CRC, all patients with no high-risk factors, 90% (36/40) of those with high-risk factors combined with MLH1/MSH2 negativity, and 6.56% (12/183) of those with high-risk factors combined with MLH1/MSH2 positivity received only regular follow-up. The remaining stage II patients received the CAPEOX regimen (oxaliplatin 130 mg/m2 for 1 d, intravenous drip 2 h; capecitabine 1 g/m2/d for 14 d, oral administration, every 21 d for 6 cycles). For stage IIIA[19] (T1-2N1M0 or T1N2M0) CRC, 69.91%(79/113) of patients received the CAPEOX regimen, and for stage IIIB-IIIC (T3-4N1M0 or T1-4N2M0), 80.92%(140/173) of patients with high-risk factors for recurrence received the FOLFOX6 regimen (oxaliplatin 85 mg/m2 for 1 d, intravenous drip 2 h; calcium folinate 400 mg/m2 for 1 d, intravenous drip; 5-fluorouracil (5-FU) 400 mg/m2 for 1 d, intravenous injection; 5-FU 1.2 g/m2 for 46 h, intravenous drip, every 14 d for 12 cycles), and the remaining ones received only regular follow-up. No patients received radiotherapy, neoadjuvant chemotherapy, or immunotherapy before surgery. Written informed consent was obtained from each patient. Ethical approval to perform this research was issued by the Human Research Ethics Committee of the Nanjing Hospital of Chinese Medicine affiliated to Nanjing University of Chinese Medicine.

Histopathological and immunohistochemical analyses

According to histopathological and immunohistochemical detection, tumor types were classified into adenocarcinoma with or without mucin production. Three hundred and ninety-five patients had stage II CRC and 286 had stage III on the basis of the WHO criteria.

The tumor tissue specimens were immersed in 4% paraformaldehyde for 4 h, and transferred to 70% ethanol. Individual lobes of tumor tissue biopsy material were placed in processing cassettes, dehydrated through a graded series of ethanol, and embedded in paraffin wax blocks. Before immunostaining, 5-μm-thick colorectal tissue sections were dewaxed in xylene, rehydrated through decreasing concentrations of ethanol, washed in phosphate-buffered saline, and stained with hematoxylin and eosin. After staining, sections were dehydrated through increasing concentrations of ethanol and xylene. Staining was carried out using the BenchMark XT system (Roche, Shanghai, China).

The specimens were observed under a light microscope, with 10 fields with no fewer than 100 cells per field observed by a double-blind method. At least 5% of the tumor cells were stained positive, otherwise they were considered negative. The final result was reviewed by more than two senior pathologists. Complete loss of MLH1 or MSH2 expression was classified as MLH1 or MSH2 negative, which formed the MLH1/MSH2-negative group. Normal expression of MLH1 and MSH2 was classified as MLH1 positive or MSH2 positive, which formed the MLH1/MSH2-positive group[20].

Microsatellite analysis

One hundred and thirty-three fresh CRC tissues and matched tumor-adjacent normal tissues were collected, frozen in liquid nitrogen, and stored at -80 °C. DNA was extracted by a standard phenol–chloroform procedure. Before DNA extraction, frozen sections were cut from each tumor sample, which were stained with hematoxylin and eosin to verify by microscopic examination the presence of adequate neoplastic material (60%–70% of tumor cells). The primers, location, and sequence of MSI are listed in Table 1.

Table 1

Name of primers, location, and the sequence of microsatellite instability

SNP	Location	Primer sequence
BAT26	2p16	p1 TGACTACTTTTGACTTCAGCC
		p2 AACCATTCAACATTTTTAACCC
D2S123	2p21-2p16	p1 AAACAGGATGCCTGCCTTTA
		p2 GGACTTTCCACCTATGGGAC
D5S346	5q21-5q22	p1 ACTCACTCTAGTGATAAATCGGG
		p2 CAGATAAGACAGTATTACTAGTT
BAT25	4q12	p1 TCGCCTCCAAGAATGTAAGT
		p2 TCTGCATTTTAACTATGGCTC
D17S250	17q11.2-17q12	p1 GGAAGAATCAAATAGACAAT
		p2 GCTGGCCATATATATATTTAAACC

SNP: Single nucleotide polymorphism.

In all 133 cases, MSI was evaluated at five microsatellite loci (BAT26, BAT25, D2S123, D5S346, and D17S250) using fluorescence-based polymerase chain reaction (PCR). Analysis of the PCR products was performed with an automated laser-activated fluorescent DNA sequencer using the Applied Biosystems 3130XL and analyzed with GenScan 3.1 software. MSI-H was defined as ≥2 mutation sites; low MSI (MSI-L) was defined as only one mutation site; and microsatellite stability (MSS) was defined as no mutations.

Statistical analysis

The relationship between MLH1/MSH2 expression and clinicopathological factors was analyzed by the χ2 test. Clinical factors that were analyzed included age, gender, tumor stage, differentiation, lymphocytic infiltration, tumor size, mucin, and tumor margin. Survival was estimated by the Kaplan–Meier method. Univariate and multivariate analyses were carried out using Cox’s proportional hazards regression models. P < 0.05 was defined as significant. The statistical analyses were performed using SPSS version 20.0.

RESULTS

MLH1/MSH2 expression and clinicopathological features

Among the 681 patients, 131 (19.24%) were MLH1/MSH2-negative and 550 (80.76%) were MLH1/MSH2-positive. The relationship between clinicopathological characteristics and MLH1/MSH2 expression is shown in Table 2 and Figure 1. MLH1/MSH2-negative CRC occurred more frequently in the right than in the left colon (27.88% vs 17.86%, P = 0.029). It also occurred more frequently in the colon than in the rectum (22.82% vs 15.80%, P = 0.025) and in poorly differentiated than well–moderately differentiated CRC (33.33% vs 18.31%, P = 0.017). In addition, MLH1/MSH2-negative CRC was characterized by LN metastasis and mucinous tumor. MLH1/MSH2-negative tumors were more likely to contain mucin (P = 0.024). MLH1/MSH2 expression was not associated with age, gender, tumor stage, tumor size, lymphocytic infiltration, or circumscribed margin (P > 0.05).

Table 2

Relationship between clinicopathological features and MutL homolog1/MutS homolog2 expression in 681 patients, n (%)

Observation	Positivea	Negativeb	Total	χ2	P value
Gender
Male	310 (80.10)	77 (19.90)	387	0.252	0.625
Female	240 (81.63)	54 (18.37)	294
Age (yr)
≤50	63 (88.73)	8 (11.27)	71	3.240	0.072
>50	487 (79.84)	123 (20.16)	610
Location in the colon
Left	138 (82.14)	30 (17.86)	168	4.746	0.029
Right	119 (72.12)	46 (27.88)	165
Location of tumor Colon	257 (77.18)	76 (22.82)	333	5.395	0.025
Rectum	293 (84.20)	55 (15.80)	348
Grade of differentiation
Poor	28 (66.67)	14 (33.33)	42	5.725	0.017
Well-moderate	522 (81.69)	117 (18.31)	639
Tumor stage (TNM)
II	324 (82.03)	71 (17.97)	395	0.964	0.327
III	226 (79.02)	60 (20.98)	286
Tumor size (cm)
≤4	210 (78.95)	56 (21.05)	266	0.927	0. 370
>4	340 (81.93)	75 (18.07)	415
Lymphocytic infiltration
None or little	336 (82.76)	70 (17.24)	406	1.195	0.332
Marked or moderate	214 (75.76)	61 (27.24)	275
Mucin
Positive	108 (73.97)	38 (26.03)	146	5.517	0.024
Negative	442 (82.62)	93 (17.38)	535
Circumscribed margin
Negative	482 (79.80)	122 (20.20)	604	3.184	0.090
Positive	68 (88.31)	9 (11.69)	77

MLH1/MSH2 positive;

MLH1/MSH2 negative. MLH1: MutL homolog 1; MSH2: MutS homolog 2.

Figure 1

Expression of MLH1 and MSH2 in colorectal cancer tissues detected by immunohistochemistry (DAB staining, ×100). A: MLH1 positivity; B: MLH1 negativity; C: MSH2 positivity; D: MSH2 negativity.

Survival analysis

With a median follow-up period of 56 mo (range 8.0–72.0 mo), 36 of 395 (9.11%) patients with stage II CRC died and 42 (10.63%) had recurrence or metastasis during the study. For stage III CRC, 90 of 286 (31.47%) patients died and 107 (37.41%) had recurrence or metastasis. Causes of death included cancer recurrence (n = 46), metastasis to other organs (n = 51), pulmonary infection (n = 15), heart disease (n = 11), second primary cancer (n = 17), multiple organ failure (n = 7), and unspecified reasons (n = 8). Some patients had more than one cause of death.

Patients with MLH1/MSH2-negative stage II or III CRC showed a favorable trend for OS (68.62 ± 0.83 vs 62.11 ± 1.07 mo, P < 0.001). Stratified analyses showed that patients with MLH1/MSH2-negative stage II CRC had longer OS than those with MLH1/MSH2-positive CRC (70.67 ± 0.65 vs 66.02 ± 1.01 mo, P = 0.011). Patients with MLH1/MSH2-negative stage III CRC also had a longer OS than those with MLH1/MSH2-positive (66.05 ± 1.62 vs 63.40 ± 1.15 mo, P = 0.023). In patients who received adjuvant chemotherapy, those with MLH1/MSH2-negative CRC had an OS of 64.02 ± 1.61 mo compared with 62.11 ± 1.07 mo in those with MLH1/MSH2-positive CRC (P = 0.015). The 5-year survival rate for patients with MLH1/MSH2-negative CRC was 86.9%, compared with 59.1% for patients with MLH1/MSH2-positive CRC. The data for OS are listed in Figure 2.

Figure 2

Kaplan–Meier survival curves for patients with colorectal cancer, stratified according to expression of MLH1/MSH2. A: Overall survival of total colorectal cancer (CRC) patients classified as MLH1/MSH2-positive and MLH1/MSH2-negative subgroups; B: Overall survival of stage II CRC patients classified as MLH1/MSH2-positive and MLH1/MSH2-negative subgroups; C: Overall survival of stage III CRC patients classified as MLH1/MSH2-positive and MLH1/MSH2-negative subgroups; D: Overall survival of all adjuvant chemotherapy patients classified as MLH1/MSH2-positive and MLH1/MSH2-negative subgroups.

Univariate and multivariate analyses

In univariate analysis, patients with MLH1/MSH2-positive CRC had a significantly worse OS than those with MLH1/MSH2-negative CRC [P < 0.001, hazard ratio (HR) = 3.799, 95%CI: 2.205–6.546]. Several other factors were also associated with disease survival, such as age, tumor differentiation, tumor stage, lymphocytic infiltration, mucin, and circumscribed margin (P < 0.05).

In multivariate survival analysis incorporating status of MLH1/MSH2, gender, age, tumor location, tumor differentiation, tumor stage, tumor size, lymphocytic infiltration, mucin, and tumor margin, the status of MLH1/MSH2 was an independent prognostic factor for OS (P < 0.001, HR = 4.064, 95%CI: 1.241–7.369). Besides that, age, tumor location, tumor stage, lymphocytic infiltration, mucin, and circumscribed margin were also independent prognostic factors for OS (P < 0.05). In the subgroup analysis of stage II CRC, patients with MLH1/MSH2-negative tumor demonstrated a better OS than those with MLH1/MSH2-positive tumor (multivariate P < 0.011, HR = 5.583, 95%CI: 1.478–21.092). Patients with stage III disease had similar results but MLH1/MSH2 status was less significant than in patients with stage II disease (multivariate P = 0.023, HR 2.289, 95%CI: 1.270–4.125) (Table 3).

Table 3

Prognostic factors for survival in univariate and multivariate analyses

Observation	Univariate			Multivariate
	P	HR	95%CI	P	HR	95%CI
Model 1a
Gender
Male vs Female	0.932	0.984	0.684-1.417	0.991	0.998	0.685-1.454
Age/yr
≤50 vs >50	< 0.001	0.288	0.189-0.438	0.025	0.508	0.213-0.995
Location in the colon
Left vs Right	0.730	1.086	0.678-1.740	0.440	1.211	0.745-1.967
Location of tumor
Rectum vs Colon	0.232	1.311	0.840-2.047	0.017	1.795	1.111-2.902
Differentiation
Poor vs Well-moderate	0.002	0.426	0.472-0.734	0.923	0.972	0.542-1.741
Tumor stage
II vs III	0.034	0.651	0.437-0.968	0.041	0.601	0.321-0.932
Tumor size
<4 vs ≥4 cm	0.646	0.921	0.647-1.311	0.421	0.861	0.598-1.240
Lymphocytic infiltration
Positive vs Negative	< 0.001	2.282	1.092-5.756	0.022	3.665	1.207-7.128
Mucin
Positive vs Negative	0.001	2.361	1.647-3.383	< 0.001	2.512	1.714-4.682
Circumscribed margin
Positive vs Negative	< 0.001	3.908	2.654-5.755	0.011	2.474	1.433-4.270
MLH1/ MSH2
Positive vs Negative	< 0.001	3.799	2.205-6.546	< 0.001	4.064	2.241-7.369
Model 2b
Gender
Male vs Female	0.761	0.896	0.441-1.821	0.207	0.622	0.298-1.300
Age/yr
≤50 vs >50	< 0.001	0.150	0.075-0.302	< 0.001	0.131	0.063-0.271
MLH1/ MSH2
Positive vs Negative	0.014	4.833	1.382-16.899	0.011	5.583	1.478-21.092
Therapeutic regimen
Operation vs Operation + Chemotherapy	0.176	0.821	0.520-1.233	0.063	0.901	0.899-2.312
Model 3c
Gender
Male vs Female	0.964	0.990	0.647-1.517	0.748	0.932	0.607-1.432
Age/yr
≤50 vs >50	0.002	0.424	0.247-0.728	0.004	0.446	0.258-0.769
MLH1/ MSH2
Positive vs Negative	0.041	1.625	1.042-2.803	0.023	2.289	1.270-4.125
Therapeutic regimen
Operation vs Operation + Chemotherapy	0.028	2.891	1.209-6.372	< 0.001	4.002	1.929-9.425

All 681 patients;

395 Patients with stage II CRC;

286 Patients with stage III CRC. HR: hazard ratio; CI: confidence interval; CRC: Colorectal cancer; MLH1: MutL homolog 1; MSH2: MutS homolog 2.

We observed no significant benefit from adjuvant chemotherapy for patients with MLH1/MSH2-negative (multivariate P = 0.147, HR = 1.563, 95%CI: 0.481–4.441) or MLH1/MSH2-positive (multivariate P = 0.070, HR = 1.267, 95%CI: 0.212–5.052) stage II CRC. However, a better survival was observed for patients with MLH1/MSH2-negative stage III CRC who received adjuvant chemotherapy (multivariate P < 0.001, HR = 7.660, 95%CI: 2.974–15.883). But a nonsignificant trend for survival benefit from adjuvant chemotherapy was observed in patients with MLH1/MSH2-positive stage III disease (multivariate P = 0.052, HR = 2.817, 95%CI 0.223-6.671) (Table 4). All findings are consistent for the OS end point.

Table 4

Predictive factors for survival in univariate and multivariate analyses

Observation	Univariate			Multivariate
	P	HR	95%CI	P	HR	95%CI
Model 1a
MLH1/MSH2 negative
Operation vs Operation + Chemotherapy	0.098	1.021	0.342-2.741	0.147	1.563	0.481-4.441
MLH1/MSH2 positive
Operation vs Operation + Chemotherapy	0.081	1.899	0.127-4.114	0.070	1.267	0.212-5.052
Model 2b
MLH1/MSH2 negative
Operation vs Operation + Chemotherapy	0.001	4.393	2.068-12.316	< 0.001	7.660	2.974-15.883
MLH1/MSH2 positive
Operation vs Operation + Chemotherapy	0.063	2.015	0.648-5.997	0.052	2.817	0.223-6.671

395 Patients with stage II CRC;

286 Patients with stage III CRC. HR: hazard ratio; CI: confidence interval; MLH1: MutL homolog 1; MSH2: MutS homolog 2.

MSI

Microsatellite analysis was performed in 133 CRC patients (71 stage II and 62 stage III) using GenScan, and 105 had MSS (78.95%) and 28 (21.05%) had MSI, including 18 (13.53%) cases of MSI-H and 10 (7.52%) cases of MSI-L. All patients were detected by immunohistochemical analysis, which confirmed that patients with MSI-H carcinomas included 17 (94.4%) who were MLH1/MSH2-negative and one (5.6%) who was MLH1/MSH2-positive (P < 0.001). Patients with MSS and MSI-L who had MLH1 or MSH2 positivity were classified as MLH1/MSH2-positive. According to our data analysis, MSI-H was more frequent in patients aged > 50 years (P = 0.048), in the right colon (P < 0.001), in tumors with poor differentiation (P = 0.028), and in tumors with mucin (P = 0.037). The clinicopathological features of MSI are consistent with previous immunohistochemical results of MLH1/MSH2 expression. The analysis of MSI is shown in Figure 3.

Figure 3

Analysis of microsatellite instability in colorectal cancer tissues and corresponding normal mucosa using the five markers of the international workshop of Bethesda and fluorescence-based multiplex polymerase chain reaction. Fragment pattern of a high-frequency microsatellite instability tumor showing instability at all five loci examined is shown. A: Low microsatellite instability (MSI-L) of tumor tissue; B: MSI-L of corresponding normal tissue; C: High microsatellite instability (MSI-H) of tumor tissue; D: MSI-H of corresponding normal tissue; E: Microsatellite stability (MSS) of tumor tissue; F: MSS of corresponding normal tissue.

DISCUSSION

Defective DNA MMR is most often associated with loss of MLH1 and MSH2 gene functions and results in MSI mutation. MLH1 and MSH2 promoter hypermethylation is an important DNA MMR pathway in sporadic proximal CRC[21]. Abnormal methylation, causing alteration of MLH1/MSH2, can form transcriptional target genes for silencing[22]. Several studies[23-25] have revealed that MLH1 and MSH2 play a critical role, and mutations in either gene result in complete loss of function, with tumor formation preferentially in the proximal colon, and this hypermethylation is significantly more common in sporadic than in hereditary MSI-positive tumors. Thibodeau et al[26,27] reported that in most colorectal carcinomas with MSI-H phenotype , 91% of cases are confirmed as MLH1-negative. In another study, Herman et al[28] reported that hypermethylation of MLH1 in sporadic CRC with MSI-H was as high as 84%. By contrast, Vasen et al[29] demonstrated that MSH2 mutation is associated with a higher risk of developing cancer than MLH1 mutation. MSH2 generally forms a connection with MSH6 or MSH3, so it can control most of the hypermethylation that occurs with different bound proteins. And MLH1 forms with PMS2, and goes through one pathway to combine with other proteins[30,31]. Gene mutation is the key reason for the decrease of MLH1 and MSH2 expression and both of them are the most dominant parts of the MMR system, so detection for those two genes are important for discovering the pathogenesis of sporadic CRC.

In this study, we used immunohistochemistry to detect MLH1 and MSH2 expression in all postoperative patients because it is more accurate, rapid, and cost-effective for assessment of MMR status than other methods[32,33]. MLH1/MSH2 status can verify MMR expression. In contrast, using GenScan to analyze MSI status can more accurately and directly demonstrate the difference between normal and abnormal loci, although the test is expensive and not easy to analyze in each postoperative patient[34]. So, immunohistochemical staining for MMR is now performed as part of routine processing in the department of pathology in almost all hospitals after surgery[35-37]. In our study, immunohistochemical analysis for MMR found that MLH1/MSH2 negativity was more frequent in the right colon, in tumors with poor differentiation, and in tumors with mucin production. With regard to MSI detection, MSI-H occurred mostly in patients aged > 50 years, in the right colon, in tumors with poor differentiation, and in tumors with mucin production. We confirmed that these two clinical assays have more consistency and accuracy than other detections.

The result of our study clearly showed that MLH1/MSH2-negative tumors were mostly in the right colon with poor differentiation and contained mucin. This conclusion is consistent with most of the published research. Benatti et al[38] reported that MSI-H occurred in 256 (20.3%) of 1263 patients, more frequently in tumors which were in the less advanced stage, right sided, poorly differentiated with mucinous phenotype, and had infiltrative growth than MSS[39]. Numerous studies have a similar conclusion that most MSI-H tumors are mucinous adenocarcinoma, located in the right colon, and poorly differentiated (P < 0.05)[40,41]. Though in the early studies, CRC was divided into colonic and rectal by anatomical site, they had some differences in specific treatments even if they have been treated as the same disease. And in recent years, with the deepening of understanding of this disease and the increase of evidence-based medical proof, CRC in different parts was considered to have distinct clinical pathological expression and prognosis. So the location of tumor is instructive for prognosis and treatment[42,43]. More and more studies now tend to divide CRC into right site and left site according to the colonic splenic flexure[44]. It is based on the right site originating from the midgut of embryo, while the left site originates from the hindgut of embryo. Anatomically, the right site was supplied by the superior mesenteric artery, and the left site was supplied by the inferior mesenteric artery[45,46]. Therefore, the right-sided tumor has poorer differentiation, worse pathological stage, and earlier metastasis than the left-sided tumor[47]. Since the rectal blood is supplied from the internal iliac artery, and rectal cancer is different from the colonic cancer in clinical treatment, in our study we divided CRC into the right colon, left colon, and rectum parts. As found in other studies[48,49], the frequency of mismatch repair deficiency (dMMR) in right-sided tumors with poor differentiation was significantly higher than that in tumors of the left colon and rectum in our study, indicating that MSI is mainly involved in the development of right colon cancer. Thus, the occurrence of CRC in different parts is not the same at the genetic level.

In terms of prognostic value of MMR phenotype in CRC, abundant studies[50-52] have acquired positive results. Gryfe et al[53] analyzed 607 patients with CRC and divided them into TNM stages I–IV. All patients with MSI-H had a survival advantage compared with MSS patients (P < 0.001, HR 0.42, 95%CI 0.27–0.67). The incidences of distant metastasis (P = 0.02, HR 0.49) and regional LN metastasis (P < 0.001, HR 0.33) in MSI-H patients were lower than those in patients who had MSS tumors. Roth et al[54] investigated the effect of MMR at different stages. In 1404 patients with stage II or III CRC, the prognostic advantage conferred by MSI was more evident in stage II than in stage III (P = 0.04). In a large meta-analysis[55] of 12782 patients, there was a clear correlation between MSI-H tumors and improved OS. The data demonstrated that the OS of patients with MSI-H was significantly better than that of MSI-L and MSS patients (P < 0.001), with an overall odds ratio of 0.6 (95%CI 0.53–0.69). In addition, disease-free survival (DFS) was also significantly different (P < 0.001). In our study, patients with MLH1/MSH2-negative stage II-III CRC had a better clinical outcome than those with MLH1/MSH2-positive CRC. Moreover, in multivariate analysis, the survival advantage for MLH1/MSH2-negative patients was independent from several other clinical and pathological parameters. These conclusions are consistent with almost all related studies because of the difference in histology, anatomy, and accompanying degree of differentiation, histopathology in different locations of the intestine.

Adjuvant chemotherapy is considered as the gold standard for treatment of patients with stage III CRC. However, there was controversy in a previous study as to whether patients with stage II CRC should take adjuvant chemotherapy after surgery. The 2013 NCCN guidelines[18] suggested to test for MMR proteins for all patients < 50 years of age or with stage II disease. Stage II MSI-H patients may have a good prognosis and do not benefit from 5-FU adjuvant therapy[56]. Our study followed this guideline and respected patients’ wishes to formulate their treatment schedule. We found that stage II-III CRC had different results on adjuvant chemotherapy. In stage II CRC, OS was not strongly associated with adjuvant chemotherapy either in MLH1/MSH2-negative (multivariate P = 0.147, HR = 1.563, 95%CI: 0.481–4.441) or MLH1/MSH2-positive patients (multivariate P = 0.070, HR = 1.267, 95%CI: 0.212–5.052). However, in stage III CRC, we found that OS of the MLH1/MSH2-negative patients (multivariate P < 0.001, HR = 7.660, 95%CI: 2.974–15.883) was more strongly associated with adjuvant chemotherapy than that of MLH1/MSH2-positive patients (multivariate P = 0.052, HR = 2.817, 95%CI: 0.223–6.671). This is similar to the study by Elsaleh et al[57] which revealed that patients with stage III colon cancer with MSI had improved survival when treated with 5-FU-based chemotherapy (fluorouracil and levamisole) compared with no chemotherapy (HR = 0.07, 95%CI: 0.01–0.53 vs HR = 1.06, 95%CI: 0.65–1.72). In the study of Sinicrope et al[58] (NCCTG N0147) of 2720 stage III colon cancer patients for 5-year disease-free survival, it was found that MMR proficient (pMMR) patients had statistically shorter survival time than dMMR patients (P < 0.0001). But in patients with stage III CRC, the predictive function of MMR for adjuvant chemotherapy remains controversial because there are some studies supporting the opposite conclusion. Ribic et al[59] showed no benefit from adjuvant chemotherapy in stage II-III patients with MSI. Sargent et al[60] enrolled 457 stage II-III CRC patients divided into 5-FU-based therapy (n = 229) and postsurgical treatment groups (n = 228). Patients with MSI who received 5-FU had no improvement in DFS (P = 0.85, HR 1.10; 95%CI 0.42-2.91) compared with the postsurgical treatment group. Jover et al[61] confirmed that patients with dMMR colon cancer do not benefit from adjuvant 5-FU/leucovorin. Many studies that analyzed the relationship between MMR and prognosis enrolled patients with all TNM stages of disease rather than stageII- III patients. Some studies included fewer patients with MSI-H, which made it difficult or inaccurate to evaluate 5-FU-based chemotherapy regimens. Those are the main reasons leading to the inconsistency of the final results. We encouraged to make risk-stratification for patients in evaluating the effect of adjuvant chemotherapy for patients with dMMR colon cancer.

Tumor immunotherapy has greatly advanced in recent years, especially, PD-1/PD-L1 blocking therapy has shown encouraging effects and become a major pillar of immunotherapy. Patients with metastatic colorectal cancer also gain new hopes from immunotherapy, particularly in dMMR patients for whom immune checkpoint inhibitor antibody can achieve a 40% objective response rate (ORR) and up to 78% clinical benefit rate[62]. The clinical trial KEYNOTE-028[63] enrolled advanced colorectal adenocarcinoma patients who failed standard therapy and had PD-L1 expression in ≥ 1% of cells in tumor nests. The primary endpoints were ORR, safety, and tolerability. Patients received pembrolizumab 10 mg/kg every 2 wk and lasted more than 2 years or until confirmed unacceptable toxicity or progression. The results showed that in the 156 advanced CRC patients, 23 were PD-L1 positive, 1 gained complete remission (CR), and 1 experienced a partial response (ORR, 4%; 95%CI: 0–22%) who was confirmed as MSI-H. This trial revealed that PD-L1 expression cannot screen out the dominant population of anti-PD-1 immunotherapy. Based on the KEYNOTE-028 and several other clinical trial results, Le et al[64] designed a single-arm, phase II clinical study (NCT01876511) which aimed to explore the predictive value of MMR status in the treatment of PD-1. The investigator divided 41 patients with progressive metastatic carcinoma with or without dMMR into three groups, namely, 11 with dMMR CRC, 21 with pMMR CRC, and 9 with dMMR non-CRC (4 ampullary /cholangiocarcinoma, 2 endometrial cancer, 2 small intestine cancer, and 1 gastric cancer) and every patient was administered pembrolizumab 10 mg per kilogram of body weight every 14 d. The primary endpoints were ORR at 20 wk and progression-free survival (PFS). The results showed that the ORR for the three groups were 40%, 0, and 71%, respectively, and the PFS rates were 78%, 11%, and 67%, respectively, at 20 wk. Interestingly, the investigators used whole-exome sequencing to check somatic mutations and found that the mutation rate was higher in dMMR than in pMMR (P = 0.007). Moreover, the study demonstrated that high somatic mutation loads were associated with prolonged PMS (P = 0.02) and dMMR patients received clinical benefit of immune checkpoint blockade with pembrolizumab.

The data of these innovative single-arm clinical studies[65-67] led to accelerated approval of the United States FDA for pembrolizumab in patients with dMMR/MSI-H solid tumors (including CRC) who failed previous treatment in May 2017[68]. Subsequently, pembrolizumab and nivolumab were recommended for second-line or later treatment of dMMR/MSI-H CRC in the 2017 NCCN guidelines[69], but were approved in refractory or metastatic CRC, so there is just little evidence or trial using the new immunotherapeutic drugs, alone or in combination with chemotherapy, in patients with stage III CRC. Although checkpoint inhibitors (CPIs) have achieved remarkable efficacy in CRC, they still face the dilemma of limited effective drugs and limited access to dMMR patients. The proportion of dMMR with advanced CRC was less than 5%, so how to make the majority of pMMR patients benefit from immunotherapy in the future is an important problem to be solved. We may need to face the following questions: (A) How to screen out patients who may be effective in immunotherapy in a large pMMR population and then expand the indications for CPIs? (B) How to more optimize the treatment strategy to overcome the primary resistance of pMMR population to immunotherapy and improve the response to immunotherapy? (C) How to control the immune-related events more effectively? (D) Is there any other new targeted immune checkpoints?

In conclusion, this study demonstrated the function of MLH1/MSH2 expression in sporadic CRC, including its effect on prognostic and predictive factors, but we have few details about the correlation between MMR and tumorigenesis, loss of heterozygosity, and immunotherapy[70,71]. Further studies should clarify the cause and mechanisms of hypermethylation in MLH1/MSH2 and antineoplastic immunity. Data from the current study may be helpful to understand the roles of MLH1/MSH2 in the development and progression of CRC. They also suggest a new therapeutic strategy by regulating MMR expression to slow down the malignant progression of CRC and to improve the prognosis of CRC patients.

ARTICLE HIGHLIGHTS

Research background

Colorectal cancer (CRC) can arise through three distinct mutational pathways: Microsatellite instability, chromosomal instability, and CpG island methylator phenotype. We tested the hypothesis that CRC arising from the microsatellite-instability pathway through MLH1/MSH2-negative expression can lead a more favorable overall survival (OS) than MLH1/MSH2-positive patients. We also made an in-depth observation of the correlation between adjuvant chemotherapy and MLH1/MSH2 expression in different stages of CRC.

Research motivation

A larger sample size with a longer follow-up period was included to assess the effect of MLH1/MSH2 status on the prediction and prognosis of stage II-III CRC and its association with adjuvant chemotherapy. It is important for clinical doctors to choose optimal treatment regimen, especially adjuvant chemotherapy, for patients.

Research objectives

To evaluate the predictive and prognostic effects of MLH1/MSH2 status in stage II-III CRC patients and its significance in guiding adjuvant chemotherapy.

Research methods

We analyzed 681 postoperative patients with CRC with a median follow-up period of 56 mo (range, 8.0–72.0 mo) between January 2013 and December 2016. The main outcome data included MLH1/MSH2-positive rate, MLH1/MSH2-negative rate, and long-term follow-up outcomes.

Research results

The outcomes showed that 550 patients were MLH1/MSH2-positive and 131 were MLH1/MSH2-negative. MLH1/MSH2-positive tumors were significantly more frequent in the colon than in the rectum, and with poor differentiation and less mucin production (P < 0.05). Patients did not differ in terms of age, gender, tumor size, tumor stage, lymphocytic infiltration, or circumscribed margin. MLH1/MSH2-negative patients had a more favorable OS than MLH1/MSH2-positive patients (P < 0.001). In both stages II and III, MLH1/MSH2 expression was a strong prognostic factor in all patients [P < 0.001, hazard ratio (HR) = 4.064, 95% confidence interval (CI): 2.241–7.369]. Adjuvant chemotherapy had a greater correlation with survival advantage in MLH1/MSH2-negative patients with stage III disease (P < 0.001, HR = 7.660, 95%CI: 2.974–15.883). Patients with stage II disease or MLH1/MSH2-positive stage III patients did not benefit from adjuvant chemotherapy.

Research conclusions

MLH1/MSH2 phenotype constitutes a pathologically and clinically distinct subtype of sporadic CRC. MLH1/MSH2 is an independent prognostic and predictive factor for outcome of stage II-III CRC.

Research perspectives

Our study demonstrated mismatch repair (MMR) is an important prognostic and predictive biomarker for stage II-III CRC, but we did not enroll the patients who used any PD-1/PD-L1 blocking therapy and had no data for survival improvement with different MMR statuses. In addition to detecting MMR status and tumor mutational burden, are there any indicators that are more sensitive to immunotherapy? The currently found immunologic drugs are only effective for high microsatellite instability (MSI-H)/dMMR population, but are ineffective for most patients with microsatellite stability (MSS). Would any new effective immune drugs be found for MSS patients? Is it possible to subdivide MSI-H (through the number of mutation sites) for enriching the dominant population in future? These issues will be the focal points and difficulty in our later research.

ACKNOWLEDGEMENTS

We thank Ya-Nan Shi, Xiao Chen, Xin-Yang Wu, Jing-Yu Shen, and Zong-Lan Zhang for data collection and Youping Deng’s team for excellent collaboration. We also thank Chris K Deng for his extensively revising the article.