Introduction

Colorectal cancer (CRC) is one of the most malignant types of cancers. In the USA and Europe, CRC is the second most frequent cancer that leads to death, which ranks only below lung cancer [1]. It is expected that there will be approximately 8 % of new CRC cases in the USA in 2014 [2]. Surgical resection can be performed to remove the tumor if neither lymph node nor distant metastasis was present, and the recurrence rate after surgery remains high [3, 4]. Thus, further study on the mechanism of initiation, progression, and identification of prognostic marker and potential drug target is still needed and will help select the patients with high chance of CRC recurrence and provide better prognosis and individualized treatment. Epigenetic modification of gene expression, such as DNA methylation, plays an important role in carcinogenesis. Aberrant methylation of CpG dinucleotides, such as DNA hypermethylation and DNA hypomethylation, is a commonly observed epigenetic modification in human cancer [5–7]. Therefore, analysis of gene methylation as a tool for diagnosis of tumors and its use as a prognostic marker have been widely used for various types of cancer including CRC.

O6-methylguanine-DNA methyltransferase (MGMT) is a DNA repair protein with the ability to remove mutagenic and cytotoxic adducts from O6-guanine in DNA [8]. MGMT transfers the alkyl group from the O6-guanine in DNA to an active cysteine within its own sequence. One MGMT molecule is inactivated for each lesion repaired [9]. The active site of MGMT molecule cannot be recovered after such an enzymatic reaction [10]. The role that MGMT losses in colorectal tumorigenesis is complex and not well characterized. Low level of MGMT methylation has been detected in the normal colorectal mucosa taking from the margin of the resected CRC as well as individuals without CRC [11, 12]. MGMT methylation has also been detected more frequently in colorectal adenomas than in normal tissues. These findings suggest that MGMT methylation may play a role in preceding the development of CRC [11, 13]. Therefore, MGMT methylation status might be a valuable biomarker for early detection and prediction of prognosis in CRC. Some studies have demonstrated that MGMT methylation is a weak prognostic marker or has no survival advantage, but the studies fail to show any significance of MGMT methylation in CRC due to limited power. We conducted this systematic meta-analysis with a total of 28 inconsistent observational studies to assess the value of MGMT methylation as an early detection and prognosis biomarker in CRC (Table 1).

Table 1 Basic characteristics of the included studies
Full size table

Material and methods

Articles search strategy and selection criteria

The following electronic databases were searched for relevant articles without any language restrictions: Web of Science (1945~2014), the Cochrane Library Database (Issue 12, 2014), PubMed (1966~2014), EMBASE (1980~2014), CINAHL (1982~2014), and the Chinese Biomedical Database (CBM) (1982~2014). We also checked Google scholar for additional articles.

We searched articles using following terms: MGMT promoter methylation and CRC, adenoma, and normal tissue. We included studies that met the following criteria: (1) the association between MGMT methylation and CRC and/or adenoma and (2) the association of MGMT methylation and prognosis in CRC patients. The exclusion criteria included the following: (1) letters, reviews, case reports, conference abstracts, editorials, and expert opinion; (2) all publications regarding in vitro/ex vivo studies, cell lines, and human xenografts were also excluded; and (3) if the study utilized the same population or overlapping database.

Data extraction

Two reviews independently extracted data using a standardized form. The information extracted from each article included the following: first author, year of publication, countries, number of patients, methods to detect MGMT methylation, stages, prognosis, hazard ratios with corresponding 95 % confidence intervals for patients with methylated MGMT, and the total number of participants.

Statistics analysis

Hazard ratio (HR) with a 95 % confidence interval was calculated for the association between MGMT methylation and prognosis. Odds ratios (ORs) with 95 % confidence intervals were calculated by using a fixed or random effect model depending on heterogeneity (a fixed effect model for I 2 <50 %, a random effect model for I 2 >50 %) [14]. The I 2 index was proposed to quantify the degree of heterogeneity in a meta-analysis [15]. Subgroup analysis was performed to compare MGMT methylation between normal tissues and CRC tissues, normal tissues and adenoma, early and late stages of CRC, male and female, and adenoma and CRC. All p values were two sided. Funnel plots were designed for publication bias. All analyses were performed with Review Manager 5.2 [16].

Results

Identification of relevant studies

One hundred and twenty publications were identified by the search method as described above. Ninety-eight of those were excluded due to laboratory studies, non-original articles (review), or studies irrelevant to the current analysis. Eventually, there were 28 studies included in the final meta-analysis [11–13, 17–41] as shown in Fig. 1. The frequency of MGMT methylation ranged from 22.7 to 77.8 % (average 41.2 %) in cancer tissues. The frequency of MGMT methylation in cancer tissues ranged from 25.0 to 77.8 % (average 42.7 %) and from 22.7 to 60.7 % (average 39.5 %) in Caucasian and Asian, respectively.

Fig. 1
figure 1

Schematic flow diagram for selection of included studies

Full size image

The correlation of MGMT hypermethylation with clinicopathological features

The inactivation of MGMT through hypermethylation in CRC and adenoma

MGMT hypermethylation is significantly higher in CRC than in normal colorectal mucosa. The pooled OR from 13 studies including 1085 CRC and 899 normal colorectal mucosa is shown in Fig. 2 (OR = 6.04, 95 % confidence interval (CI) = 4.69–7.77, p < 0.00001), which indicates that MGMT inactivation through hypermethylation plays an important role in the pathogenesis of CRC. MGMT hypermethylation also occurs in colorectal adenoma, but significantly less compared to that in CRC. The pooled OR from nine studies including 709 CRC and 450 colorectal adenoma is shown in Fig. 3 (OR = 1.30, 95 % CI = 1.00–1.69, p = 0.05). In addition, MGMT hypermethylation is also significantly higher in colorectal adenoma than in normal colorectal mucosa. The pooled OR from seven studies including 373 colorectal adenoma and 511 normal colorectal mucosa is shown in Fig. 4 (OR = 4.81, 95 % CI = 3.28–7.04, p < 0.00001).

Fig. 2
figure 2

Forest plot for MGMT methylation in CRC and normal tissue

Full size image
Fig. 3
figure 3

Forest plot for MGMT methylation in CRC and adenoma

Full size image
Fig. 4
figure 4

Forest plot for MGMT methylation in adenoma and normal tissue

Full size image

Relationship between the frequency of MGMT hypermethylation and sex status

MGMT hypermethylation rate in female CRC patients is significantly higher than that in male patients. The pooled OR from ten studies including 1824 male and 1634 female CRC is shown in Fig. 5 (OR = 0.86, 95 % CI = 0.74–0.99, p = 0.03), indicating that MGMT hypermethylation is strongly associated with sex status in CRC patients.

Fig. 5
figure 5

Forest plot for MGMT methylation in female and male CRC patients

Full size image

The role of MGMT hypermethylation in CRC development

We analyzed 2320 CRC patients pooled from ten studies to assess whether or not the aberrant MGMT hypermethylation in CRC was associated with the advanced stage. As shown in Fig. 6, aberrant MGMT hypermethylation is not significantly high in advanced CRC (III and IV) than that in early staged CRC (I and II), OR = 0.99, 95 % CI = 0.84–1.18, p = 0.94. These results suggest that epigenetic silencing of MGMT gene expression by promoter hypermethylation may play an important role in the early stage of CRC progression and development.

Fig. 6
figure 6

Forest plot for MGMT methylation in late and early stages of CRC

Full size image

The correlation of MGMT hypermethylation with MSI in CRC

We also determined 2506 CRC patients pooled from five studies to assess whether or not the aberrant MGMT hypermethylation in CRC was associated with microsatellite instability (MSI). As shown in Fig. 7, aberrant MGMT hypermethylation is significantly higher in MSI-positive CRC than that in MSI-negative CRC, OR = 1.43, 95 % CI = 1.14–1.80, p = 0.002. These results suggest that epigenetic silencing of MGMT gene expression by promoter hypermethylation may play an important role in increasing levels of MSI.

Fig. 7
figure 7

Forest plot for MGMT methylation and microsatellite instability (MSI) in CRC patients

Full size image

MGMT hypermethylation as a prognostic factor for CRC

Only three studies estimated the relationship between OS and MGMT hypermethylation in CRC. The pooled HR for OS shows that MGMT hypermethylation is not associated with worse survival in CRC patients as shown in Fig. 8 (HR = 1.05, 95 % CI = 0.91–1.21, p = 0.51).

Fig. 8
figure 8

Forest plot for MGMT methylation and overall survival of CRC patients

Full size image

Sensitivity analyses and publication bias

Finally, we performed a sensitivity analysis to assess the result stability, in which one study is removed at a time. The stability of the analysis is acceptable, since the pooled ORs and HRs are not significantly become different. The results also show no publication biases in the meta-analysis of MGMT methylation and clinicopathological features, and the funnel plots are largely symmetric (Fig. 9).

Fig. 9
figure 9

Funnel plot for publication bias. a MGMT methylation in CRC and normal tissue. b MGMT methylation in CRC and adenoma. c MGMT methylation in adenoma and normal tissue. d MGMT methylation in different genders of CRC. e MGMT methylation in late and early stages of CRC. f MGMT methylation in different levels of microsatellite instability (MSI) of CRC. g The association of MGMT methylation and overall survival of CRC patients. X axis: value of odds ratio (OR). Y axis: standard errors (SE) multiply log scale of OR

Full size image

Discussion

MGMT is a DNA repair enzyme which can remove alkyl group from O6 position of guanine. Alkylguanine adducts result in mispairing with thymine by leading to G:C to A:T transitions during DNA replication [9]. MGMT is genetically or epigenetically altered in different kinds of primary or advanced carcinomas. MGMT is critical to protect normal cells from exogenous carcinogens. Its inactivation by the promoter hypermethylation plays an important role in tumorigenesis in several types of tumors including CRC [42–46].

To date, there have been some studies describing the methylation status of MGMT in CRC; however, the roles of methylation of MGMT in CRC and clinical significance have not been thoroughly investigated. We conducted the meta-analysis to determine the correlation between MGMT hypermethylation and clinicopathological characteristics in CRC. Analysis of the pooled data showed that (1) CRC had a higher proportion of hypermethylation rate than colorectal mucosa; MGMT hypermethylation also occurs in colorectal adenoma, but its proportion of rate is significantly less compared to CRC. In addition, MGMT hypermethylation also had significantly higher proportion of hypermethylation in colorectal adenoma than in normal colorectal mucosa. (2) MGMT hypermethylation is correlated with sex status and significantly higher in female than in male. (3) Aberrant MGMT hypermethylation is not significantly higher in advanced CRC than that in early staged CRC. These results suggest that epigenetic silencing of MGMT gene expression by promoter hypermethylation may play an important role in the early stage of CRC progression and development. (4) Aberrant MGMT hypermethylation is significantly higher in MSI-positive CRC than that in MSI-negative CRC, which indicates that epigenetic silencing of MGMT gene expression by the promoter hypermethylation may play an important role in increasing levels of MSI. (5) The pooled HR for OS shows that MGMT hypermethylation is not associated with worse survival in CRC patients. The results from the current study demonstrate that the hypermethylation rate of MGMT gene promoter in CRC is significantly higher than that in the normal colorectal mucosa, as well as colorectal adenoma, indicating that MGMT promoter hypermethylation is common in CRC. Since changes in MGMT promoter hypermethylation are reversible, drug treatment through demethylation may be useful to delay carcinogenesis and progression and to improve prognosis. This approach may bring new direction and hope for cancer treatment through gene-targeted therapy. Therefore, MGMT has a role as a potential drug target of CRC. The finding that MGMT hypermethylation is correlated with sex status and significantly higher in female than in male is also interesting. Female sex hormones have been implicated in the etiology of proximal CRC, and they may participate in different tumorigenic pathways that are associated with distinct DNA methylation-based molecular signatures and specific DNA methylation alterations [47].

Epigenetic alteration, particularly aberrant DNA methylation, is one of the best-characterized epigenetic modifications contributing to tumor initiation and progression [6, 7]. A cell only has limited resources to repair abnormal adducts since the active site cannot be regenerated after MGMT methylation. The DNA mutation repairing ability depends on the rate of MGMT synthesis. Therefore, the preciseness of MGMT protein expression indicates the capability of MGMT DNA mutation repairing during protection against tumorigenesis. MGMT methylation or loss of MGMT has been associated with unfavorable prognosis in brain tumor and B cell lymphoma [48–50]. However, in CRC, inconsistent results have shown the association between MGMT methylation and CRC. This is likely due to the previous studies that are limited by low statistical power. Another reason is that previous studies use one or the other MGMT promoter regions to correlate methylation data with the loss of MGMT protein [17, 20, 51, 52]. Therefore, MGMT can be considered as a tumor suppressor, and its inactivation could contribute to tumor progression and poor prognosis. For the first time, we used a meta-analysis to assess MGMT methylation on CRC progression. Our result indicates that MGMT methylation is not significantly associated with the prognosis of CRC.

This study has several potential limitations. The different methods used to measure MGMT methylation may also affect the individual results, and the different methods were not able to measure the same CpG sites within MGMT promoter in individual study. There are other factors that may also affect MGMT methylation. For example, MGMT germline polymorphism was reported to be associated with somatic MGMT promoter methylation and gene silencing in CRC patients [53]. Age is also a factor that may affect MGMT methylation status in CRC patients. However, we were not able to do meta-analysis due to limited available data. Therefore, cautions should be taken when these analysis results are interpreted among the general populations.

In conclusion, our meta-analysis shows that MGMT may play an important role in CRC initiation, and MGMT hypermethylation is associated with high levels of MSI. MGMT hypermethylation may play an important role in the early stage of CRC progression and development. Further large-scale studies, especially multicenter and well-matched cohort research, will provide more insights into the role of MGMT in the prognosis and clinical implementation of CRC patients.