Tumor Biology

, Volume 34, Issue 5, pp 3227–3237

The polymorphisms in the MGMT gene and the risk of cancer: a meta-analysis

Authors

  • Liang Du
    • West China Medical School/West China HospitalSichuan University
    • The Periodical Press of West China HospitalSichuan University
  • Haichuan Wang
    • West China Medical School/West China HospitalSichuan University
  • Tianyuan Xiong
    • West China Medical School/West China HospitalSichuan University
  • Yaxian Ma
    • West China Medical School/West China HospitalSichuan University
  • Jiqiao Yang
    • West China Medical School/West China HospitalSichuan University
  • Jichong Huang
    • West China Medical School/West China HospitalSichuan University
  • Dong Zeng
    • West China Medical School/West China HospitalSichuan University
  • Xiaoze Wang
    • West China Medical School/West China HospitalSichuan University
  • He Huang
    • West China Medical School/West China HospitalSichuan University
    • West China Medical School/West China HospitalSichuan University
Research Article

DOI: 10.1007/s13277-013-0893-x

Cite this article as:
Du, L., Wang, H., Xiong, T. et al. Tumor Biol. (2013) 34: 3227. doi:10.1007/s13277-013-0893-x
  • 384 Views

Abstract

Polymorphisms in the MGMT gene have been implicated in susceptibility to cancer, but the published studies have reported inconclusive results. The objective of the current study was to investigate the genetic risk of polymorphisms in the MGMT gene for cancer. A meta-analysis was carried out to analyze the association between polymorphisms in the MGMT gene and cancer risk. Five polymorphisms (Leu84Phe, Leu53Leu, Ile143Val, Lys178Arg, and −485C/A) with 98 case–control studies from 49 articles were analyzed. The results indicated that individuals who carried the Phe/Phe homozygote genotype of Leu84Phe had a 31 % increased risk of cancer compared with the Leu allele (Leu + Leu/Phe) carriers (odds ratio [OR] = 1.32, 95 % confidence interval [CI] = 1.15–1.52, P < 0.0001 for Phe/Phe vs. Phe/Leu + Leu/Leu). However, there was no significant association between the risk of cancer and the other four polymorphisms (Leu53Leu, Ile143Val, Lys178Arg, and −485C/A). In further stratified analyses for the Leu84Phe and Ile143Val polymorphisms, the increased risk of cancer remained in subgroups of Caucasians, patients with esophageal cancer for the Leu84Phe polymorphism, and patients with lung cancer for the Ile143Val polymorphism. Results from the current meta-analysis suggested that Leu84Phe and Ile143Val in the MGMT gene are risk factors for cancer. In the future, more studies should be performed to validate our results.

Keywords

MGMTPolymorphismsCancerMeta-analysis

Introduction

Cancers result from multiple interactions between genetic and environmental factors [1, 2]. Multiple somatic mutations in proto-oncogenes and tumor suppressor genes are suggested as risk factors. The host factors that increase the likelihood of somatic mutations can increase cancer risk [35]. Many functional polymorphisms of carcinogen-metabolizing enzymes have been found to be associated with increased risk of cancers. Recently, polymorphisms of DNA repair genes have been extensively examined as host risk factors for cancers, and the DNA repair genes have been suggested as cancer candidate genes. Among these DNA repair genes, the O6-methylguanine-DNA methyltransferase (MGMT) gene is one of the most widely studied [4, 6, 7].

The MGMT is a well-known protein involved in the cellular defense against alkylating agents in the DNA direct reversal repair pathway in mammals [6, 8]. Its physiological role is the protection of cells against the potentially adverse effects of DNA alkylation damage at the O6-position of guanine generated by endogenous and exogenous alkylating species (including cigarette smoking, environmental contaminants, and diet) [7, 9, 10]. In addition, inactivity of the MGMT gene by its promoter methylation may participate in epigenetic regulation in gene expression, and the phenomenon has been observed in a variety of human cancers [1113]. Thus, the MGMT gene has been reported to play important roles in the pathogenesis of cancers [1113]. The human MGMT gene is located on 10q26 and spans at least 15 kb [6, 9, 14]. Up to now, many polymorphisms have been identified in this gene, and several polymorphisms have been extensively studied for the associations with the risk of cancers, including the Leu84Phe (rs12917) and Leu53Leu (rs1803865) in exon 3, Ile143Val (rs2308321) and Lys178Arg (rs2308327) in exon 7, and −485C/A (rs1625649) polymorphism in the promoter [15]. Although there were many studies investigating the associations between the polymorphisms with the risk of cancers, the results were inconclusive. Zhong et al. performed a meta-analysis to identify the associations with the risk of the cancers and found that the Leu84Phe polymorphism was associated with increased risk of cancer [16]. However, the results should be updated. First, Zhong et al. only included the two most widely studied polymorphisms in the MGMT gene; some other polymorphisms should also be studied, such as the Leu53Leu and Lys178Arg polymorphisms. Second, Zhong et al. only identified the studies from the Medline database; some studies which were not included by that database might be missed, especially for the studies included in the EMBASE database. Third, Zhong et al. included studies published before April of 2009, and in the past few years, many studies concerning the topic of the MGMT gene variants with the risk of cancers have been identified, and these studies should also be included. Thus, we performed a meta-analysis to identify the associations between the polymorphisms in the MGMT gene with the risk of cancers. This is, to our knowledge, the most comprehensive meta-analysis concerning the associations between the MGMT gene variants and the risk of cancers.

Methods and materials

Study identification and selection

A comprehensive literature search was performed in the PubMed database, EMBASE database, CNKI database, and Wanfang database to identify studies investigating the associations between the MGMT gene variants and cancer risk (last search was performed on January 10, 2013). The search terms used were as follows: cancer or carcinoma in combination with O6-methylguanine-DNA methyltransferase or MGMT and in combination with polymorphism or variant or mutation. The languages were limited to English and Chinese. The inclusion criteria were as follows: (a) studies evaluating the association between the MGMT gene polymorphisms and cancer risk, (b) the design of the study should be a case–control design (cancer participants and noncancer participants), (c) sufficient data (genotype distributions of cases and controls) available to calculate an OR with its 95 % CI, and (d) genotype distributions in the control group should be consistent with Hardy–Weinberg equilibrium (HWE). The following exclusion criteria were used: (a) the studied populations were based on family or sibling pairs, (b) genotype frequencies or numbers were not presented in the original studies, and (c) reviews and abstracts. If more than one case–control studies were published by the same authors using the same case series or overlapped case series, studies with the largest sample sizes or the newest were included.

Data extraction

Two reviewers independently extracted the data and reached a consensus on all items. The following items were extracted from each study if available: first author’s surname, publication year, country of origin, ethnicity, sample size, cancer type, genotyping method, and genotype number in cases and controls.

Statistical analysis

For each case–control study, we first examined whether the genotype distribution in the control population was consistent with HWE by Pearson’s chi-squared test. Any polymorphism that had ever been studied in at least five case–control studies was included for data analysis. Finally, the following five polymorphisms were included: Leu84Phe, Leu53Leu, Ile143Val, Lys178Arg, and −485C/A. The strength of association between MGMT gene polymorphisms and cancer risks were assessed by ORs with corresponding 95 % CIs. The genetic models that were used for data analysis for each polymorphism were as follows: Phe/Phe vs. Phe/Leu + Leu/Leu for the Leu84Phe polymorphism, Val/Val + Val/Ile vs. Ile/Ile for the Ile143Val polymorphism, GG + GA vs. AA for the Lys178Arg polymorphism, AA + AC vs. CC for the −485C/A polymorphism, and TT + TC vs. CC for the Leu53Leu polymorphism. The heterogeneity was assessed by a chi-square-based Q statistic and was considered statistically significant at P < 0.10. When the P value was >0.10, the pooled OR was calculated by the fixed-effects model; otherwise, the random-effects model was used. The pooled OR was analyzed by a fixed-effects model (the Mantel–Haenszel method) or a random-effects model (the DerSimonian and Laird method) according to heterogeneity. The significance of the pooled OR was determined by the Z test, and P < 0.05 was considered as statistically significant. To analyze the caner type-specific effects and the ethnic-specific effects, subgroup analyses were performed by cancer types and ethnicities for the Leu84Phe, Ile143Val, and Lys178Arg polymorphisms, which were investigated in a sufficient number of studies (if a cancer type was investigated by <3 individual case–control studies, it was combined into the group of “other cancers”; if a given ethnic group was investigated by <3 individual case–control studies, it was categorized into the group of “others”). Publication bias was analyzed by Begg’s test and Egger’s test. For each polymorphism, other genetic models were also used: Phe/Leu + Phe/Phe vs. Leu/Leu, Phe/Phe vs. Leu/Leu, Phe/Leu vs. Leu/Leu, and Phe vs. Leu for the Leu84Phe polymorphism; Val/Val vs. Ile/Ile + Val/Ile, Val/Val vs. Ile/Ile, Val/Ile vs. Ile/Ile, and Val vs. Ile for the Ile143Val polymorphism; GG vs. AA + GA, GG vs. AA, GA vs. AA, and G vs.A for the Lys178Arg polymorphism, AA vs. CC + AC, AA vs. CC, AC vs. CC, and A vs. C for the −485C/A polymorphism; TT vs. CC + TC, TT vs. CC, TC vs. CC, and T vs. C for the Leu53Leu polymorphism. All statistical tests were performed using the Revman4.2 software and Stata10.0 software.

Results

Studies selection process and characteristics

The studies selection process is showed in Fig. 1. Briefly, a total of 2,310 results were identified after an initial search in the selected databases. After reading the titles and abstracts, 106 potential studies which investigated the association between genetic variants in the MGMT gene and the risk of cancers were included for full-text view. After reading the full texts, four studies were excluded for not being relevant to the association between the cancer risk and MGMT gene variants. Thus, 102 studies were left for the data extraction. In this step, 42 studies were excluded for not reporting usable data and 4 studies were excluded for not being case–control studies. Thus, a total of 145 case–control studies in 56 studies were extracted. The genotype in the control group for 9 case–control studies were not consistent with HWE and were excluded, 12 case–control studies were excluded for data overlapping or duplicating, and 26 case–control studies where the polymorphisms were analyzed in <5 case–control studies were excluded. Finally, a total of 5 polymorphisms with 98 case–control studies in 49 articles which met our inclusion criteria were identified [3, 4, 615, 1753]. Twenty-one articles were performed in Asians [6, 11, 18, 22, 23, 25, 3035, 38, 42, 43, 45, 46, 4952], 15 articles were performed in Caucasians [3, 7, 8, 12, 14, 15, 19, 36, 37, 3941, 44, 48, 53], 12 were performed in the mixed ethnic populations in USA [112], and 1 was performed in African [4]. The most studied cancers were lung cancer, esophageal cancer, and colorectal cancer. The characteristics of each case–control study are listed in Table 1. Genotype and allele distributions for each case–control study are shown in Supplement Table 1.
https://static-content.springer.com/image/art%3A10.1007%2Fs13277-013-0893-x/MediaObjects/13277_2013_893_Fig1_HTML.gif
Fig. 1

Flow diagram of included/excluded studies

Table 1

Characteristics of publications included in the meta-analysis

Author

Year

Country

Ethnicity

Cancer

Polymorphism

Agalliu, I [17]

2010

USA

Caucasian and African-American

Prostate

Leu84Phe, Ile143Val, Lys178Arg

Akbari, M [18]

2009

Iran

Asian

Esophageal

Leu84Phe, Lys178Arg

Betti, M [3]

2011

Italy

Caucasian

Mesothelioma

Leu84Phe

Bye, H [4]

2011

South Africa

African

Esophageal

Leu84Phe

Chae, MW [6]

2006

Korea

Asian

Lung

Leu84Phe, −485C/A, Leu53Leu

Crosbie, PAJ [8]

2008

UK

Caucasian

Lung

Lys178Arg

Doecke, J [19]

2008

Australia

Caucasian

Esophageal

Leu84Phe, Ile143Val

Dong, X [20]

2012

USA

Mixed

Pancreatic

Lys178Arg

Feng, XX [22]

2008

China

Asian

Esophageal

Leu84Phe

Felini, MJ [21]

2007

USA

Mixed

Gliomas

Leu84Phe, Ile143Val

Gu, S [23]

2008

China

Asian

Lung

Leu53Leu

Hall, J [7]

2007

Romania, Poland, Russia, Slovakia, Czech Republic

Caucasian

Upper aerodigestive tract

Leu84Phe

Han, J [9]

2006

USA

Caucasian-American

Endometrial

Leu84Phe, Ile143Val

Han, J [24]

2006

USA

Caucasian-American

Breast

Leu84Phe

Hazra, A [10]

2008

USA

Caucasian-American

Colorectal

Lys178Arg

Huang, J [11]

2007

China

Asian

Cervical

Leu84Phe, Ile143Val, Lys178Arg

Huang, SH [25]

2010

China

Asian

Oral

Leu84Phe

Huang, W [26]

2005

USA

Mixed

Heck and neck

Leu84Phe, Ile143Val

Huang, WY [12]

2005

Poland

Caucasian

Gastric

Leu84Phe, Ile143Val

Hung, RJ [27]

2008

USA

Mixed

Lung

Leu84Phe, Ile143Val

Jiao, L [28]

2006

USA

Mixed

Pancreatic

Ile143Val

Kaur, TB [29]

2000

USA

Caucasian, African-American

Lung

Ile143Val

Khatami, F [30]

2009

Iran

Asian

Colorectal

Ile143Val

Li, C [32]

2005

China

Asian

Bladder

Leu84Phe, Leu53Leu

Liu, RQ [34]

2002

China

Asian

Cancers

Leu84Phe

Liu, RQ [33]

2002

China

Asian

Lung

Leu84Phe

Liu, SH [35]

2006

China

Asian

Esophageal

Leu84Phe, Ile143Val

Loh, YH [37]

2011

UK

Caucasian

Cancers

Leu84Phe

Loh, YH [36]

2010

UK

Caucasian

Breast, prostate, colorectal

Ile143Val

Lu, Y [38]

2006

China

Asian

Gastric

Leu84Phe, Leu53Leu

Moreno, V [39]

2006

Germany

Caucasian

Colorectal

Ile143Val

O’Mar, TA [40]

2011

Australia, Poland

Caucasian

Endometrial

Leu84Phe, Ile143Val

Palli, D [41]

2010

Italy

Caucasian

Gastric

Leu84Phe

Ritchey, JD [42]

2005

China

Asian

Prostate

Leu84Phe, Ile143Val

Shah, MA [43]

2012

India

Asian

Esophageal

Leu84Phe, Ile143Val

Shen, J [13]

2005

USA

Mixed

Breast

Leu84Phe, Ile143Val, Lys178Arg

Shen, M [44]

2007

Australia

Caucasian

Non-Hodgkin’s lymphoma

Leu84Phe, Ile143Val, Lys178Arg

Shi, JY [45]

2011

China

Asian

Acute myeloid leukemia

Leu84Phe, Ile143Val, Lys178Arg, −485C/A

Stern, MC [46]

2007

Singapore

Asian

Colorectal

Leu84Phe

Kietthubthew, S [31]

2006

Thailand

Asian

Oral

Leu84Phe

Tranah, GJ [47]

2006

USA

Mixed

Colorectal

Leu84Phe, Ile143Val

Wang, L [48]

2006

USA

Caucasian

Lung

Leu84Phe, Ile143Val, −485C/A

Yang, F [49]

2009

China

Asian

Non-Hodgkin’s lymphoma

Leu84Phe, Lys178Arg

Yang, M [14]

2004

USA

Caucasian

Lung

Lys178Arg

Yu, HJ [50]

2009

China

Asian

Esophageal

Ile143Val, Lys178Arg

Yu, L [51]

2009

China

Asian

Esophageal

Ile143Val, −485C/A

Zhang, MD [52]

2008

China

Asian

Biliary tract

Leu84Phe

Zhang, Z [15]

2010

USA

Caucasian

Heck and neck

Leu84Phe, −485C/A, Leu53Leu

Zienolddiny, S [53]

2006

UK

Caucasian

Lung

Leu84Phe, Ile143Val, Leu53Leu

Quantitative synthesis

The Leu84Phe polymorphism

A total of 16,643 cancer cases and 26,720 controls in 41 case–control studies were included in the meta-analysis on the relationship between the Leu84Phe polymorphism and risk of cancer. Seventeen case–control studies were from Asians, 15 from Caucasians, 6 from a mixed US population, 1 from an African-American population, 1 from an African population, and 1 from a South African colored population. Overall, there was statistical evidence of an association between the Leu84Phe polymorphism and overall cancer risks (OR = 1.32, 95 % CI = 1.15–1.52, P < 0.0001 for Phe/Phe vs. Phe/Leu + Leu/Leu) (Fig. 2). No publication bias was detected by either the funnel plot (figure not shown) or the Egger’s test (t = −1.88, P = 0.068). In the subgroup analysis by cancer types (Supplement Fig. 1), results indicated that individuals with variant genotypes had a significantly increased risk of esophageal cancer (OR = 1.51, 95 % CI = 1.10–2.08, P = 0.01 for Phe/Phe vs. Phe/Leu + Leu/Leu). In the subgroup analysis by ethnicity (Supplement Fig. 2), significantly increased cancer risks were found among Caucasians (OR = 1.36, 95 % CI = 1.13–1.64, P = 0.001 for Phe/Phe vs. Phe/Leu + Leu/Leu) but not in Asians or mixed US populations. Other comparison results are listed in Supplement Table 2.
https://static-content.springer.com/image/art%3A10.1007%2Fs13277-013-0893-x/MediaObjects/13277_2013_893_Fig2_HTML.gif
Fig. 2

Meta-analysis for the association between cancer risk and the Leu84Phe polymorphism in the MGMT gene (Phe/Phe vs. Phe/Leu + Leu/Leu)

The Ile143Val polymorphism

A total of 14,477 cancer cases and 25,579 controls in 31 case–control studies were included in the meta-analysis on the relationship between the Ile143Val polymorphism and risk of cancer. Eight case–control studies were from Asians, 14 from Caucasians, 7 from a mixed US population, and 2 from an African-American population. Overall, there was no statistical evidence of an association between the Ile143Val polymorphism and overall cancer risks (OR = 1.07, 95 % CI = 0.99–1.16, P = 0.08 for Val/Ile + Val/Val vs. Ile/Ile) (Fig. 3). No publication bias was detected by either the funnel plot (figure not shown) or the Egger’s test (t = −0.46, P = 0.648). In the subgroup analysis by cancer types (Supplement Fig. 3), results indicated that individuals with variant alleles had a significantly increased risk of lung cancer (OR = 1.20, 95 % CI = 1.08–1.33, P = 0.0005 for Val/Ile + Val/Val vs. Ile/Ile). In the subgroup analysis by ethnicity (Supplement Fig. 4), significantly increased cancer risks were found among Caucasians (OR = 1.11, 95 % CI = 1.03–1.19, P = 0.004 for Val/Ile + Val/Val vs. Ile/Ile) but not in Asians. Other comparison results are listed in Supplement Table 2.
https://static-content.springer.com/image/art%3A10.1007%2Fs13277-013-0893-x/MediaObjects/13277_2013_893_Fig3_HTML.gif
Fig. 3

Meta-analysis for the association between cancer risk and the Ile143Val polymorphism in the MGMT gene (Val/Ile + Val/Val vs. Ile/Ile)

The Lys178Arg polymorphism

A total of 5,542 cancer cases and 6,979 controls in 13 case–control studies were included in the meta-analysis on the relationship between the Lys178Arg polymorphism and risk of cancer. Five case–control studies were from Asians, 4 from Caucasians, 3 from a mixed US population, and 2 from African-Americans. Overall, there was no statistical evidence of an association between the Lys178Arg polymorphism and overall cancer risks (OR = 0.96, 95 % CI = 0.82–1.12, P = 0.59 for Arg/Arg + Arg/Lys vs. Lys/Lys) (Fig. 4). No publication bias was detected by either the funnel plot (figure not shown) or the Egger’s test (t = −0.28, P = 0.788). In the subgroup analysis by ethnicity (Supplement Fig. 5), no significantly increased cancer risks were found either among Asians or Caucasians. Other comparison results are listed in Supplement Table 2.
https://static-content.springer.com/image/art%3A10.1007%2Fs13277-013-0893-x/MediaObjects/13277_2013_893_Fig4_HTML.gif
Fig. 4

Meta-analysis for the association between cancer risk and the Lys178Arg polymorphism in the MGMT gene (Arg/Arg + Arg/Lys vs. Lys/Lys)

The Leu53Leu polymorphism

A total of 2,119 cancer cases and 2,797 controls in 8 case–control studies were included in the meta-analysis on the relationship between the Leu53Leu(−16195C/T) polymorphism and risk of cancer. Overall, there was no statistical evidence of an association between the Leu53Leu polymorphism and overall cancer risks (OR = 0.97, 95 % CI = 0.84–1.11, P = 0.65 for TT + TC vs. CC) (Fig. 5). No publication bias was detected by either the funnel plot (figure not shown) or the Egger’s test (t = 0.34, P = 0.749). Other comparison results are listed in Supplement Table 2.
https://static-content.springer.com/image/art%3A10.1007%2Fs13277-013-0893-x/MediaObjects/13277_2013_893_Fig5_HTML.gif
Fig. 5

Meta-analysis for the association between cancer risk and the Leu53Leu polymorphism in the MGMT gene (TT + TC vs. CC)

The −485C/A polymorphism

A total of 2,504 cancer cases and 3,494 controls in 5 case–control studies were included in the meta-analysis on the relationship between the −485C/A polymorphism and risk of cancer. Overall, there was no statistical evidence of an association between the −485C/A polymorphism and overall cancer risks (OR = 1.09, 95 % CI = 0.98–1.21, P = 0.12 for AA + AC vs. CC) (Fig. 6). No publication bias was detected by either the funnel plot (figure not shown) or the Egger’s test (t = 1.43, P = 0.247). Other comparison results are listed in Supplement Table 2.
https://static-content.springer.com/image/art%3A10.1007%2Fs13277-013-0893-x/MediaObjects/13277_2013_893_Fig6_HTML.gif
Fig. 6

Meta-analysis for the association between cancer risk and the −485C/A polymorphism in the MGMT gene (AA + AC vs. CC)

Discussion

It is well known that genetic variants in genes in the pathway of the pathogenesis of cancer may alter protein function and individual’s susceptibility to cancer. The MGMT gene is an important DNA repair gene and plays a central role in the pathogenesis of cancer. Several polymorphisms in the MGMT gene might contribute to the development of cancer. The correlation of these polymorphisms and cancer risk had been studied, but the results were inconclusive. Therefore, we performed a meta-analysis to clarify the relationship between the polymorphisms and susceptibility to cancer.

A total of 5 polymorphisms in 98 case–control studies from 49 studies were included in the meta-analysis. The results indicated that the variant homozygote genotype Phe/Phe of the Leu84Phe polymorphism is associated with a significant increase in the overall risk of cancer, whereas the Ile143Val, Lys178Arg, Leu53Leu, and −485C/A polymorphisms did not appear to have a significant association on the overall cancer risk. The results from our stratification analyses suggested that there were effect modification of the cancer risk by cancer type and ethnicity for the Leu84Phe and Ile143Val polymorphism, whereas no effect modification of cancer risk was observed for the Lys178Arg polymorphism.

One important property of the gene variants is that different racial or ethnic populations have different frequencies of alleles. Thus, different genetic backgrounds may contribute differently to susceptibilities. Therefore, subgroup analyses were preformed according to ethnicity. In our meta-analysis, the populations were stratified into Asians, Caucasians, mixed populations in the USA, and other populations. We have to explain the group of mixed populations in the USA. In our studies, several studies were performed in American populations. As is known to all, the US population is mixed with Caucasian-American, African-American, and some other populations. Some studies independently investigated the Caucasian-Americans or African-Americans, and those studies were included in the Caucasians group or the African-American group. However, some studies did not provide the origin of the populations, some studies studied both Caucasian-Americans and African-Americans, and in some studies, the data were not independent, then these studies were stratified into the group of mixed populations in the USA. In the subgroup analysis by ethnicity, we found that the Leu84Phe and Ile143Val polymorphisms were significant associated with increased risks of cancer in Caucasians but not in Asians, while the Lys178Arg polymorphism was not associated with increased risk either in Asians or in Caucasians. As to the mixed populations in the USA, we did not find any significant association. These results suggested that, even for the same gene, genetic backgrounds might contribute differently to the susceptibility of cancer.

Because cancer origin can affect the results of meta-analysis, we performed subgroup analyses by cancer types for the Leu84Phe and Ile143Val polymorphisms. The results indicated that the Leu84Phe polymorphism is associated with increased risk of esophageal cancer but not of lung cancer, colorectal cancer, endometrial cancer, prostate cancer, or gastric cancer. In addition, our results indicated that the Ile143Val polymorphism was associated with increased risk of lung cancer but not of esophageal cancer, colorectal cancer, prostate cancer, or endometrial cancer. Since only three case–control studies were included for some cancer types, this might have limited the power to reveal a reliable association. In the future, more studies should be carried out to analyze these associations, especially for some common cancers.

It is worthy to mention some important issues in meta-analysis. The first important point is heterogeneity. Significant heterogeneity existed in overall comparisons in the dominant model of the Ile143Val polymorphism. After subgroup analysis by ethnicity and cancer types, heterogeneity was effectively decreased or removed in some subgroups. Possible explanations may be that differences in genetic backgrounds and environmental exposures existed among different ethnicities, and difference in etiology may exist in different types of cancers. Publication bias is another important issue in meta-analyses. In the current meta-analysis, publication bias was analyzed by using the funnel plots and the Egger test. No statistically significant publication bias was found for all these polymorphisms, suggesting the reliability of our results.

There were several limitations that should be considered when explaining the results of the studies. First, only published studies in English and Chinese which were included by the selected electronic databases were included for data analysis. It is possible that some potential studies which were included by other databases or published in other languages or unpublished could be missed. Second, due to lack of the original data, we could not assess the potential interactions between genes and genes or between genes and environmental factors. Third, as most of the included studies were performed in Asians, Caucasians, and mixed populations in the USA and only a few studies were performed in other populations, the results may be only applicable to only these ethnic groups.

In spite of these limitations, the current meta-analysis also has several advantages. First, it updated the newest data for the polymorphisms in the MGMT gene variants and cancer risk. Second, our results indicated that there are genetic-specific associations in different cancers and ethnicities. Third, the methodological issues for meta-analysis are well investigated.

To our knowledge, this study is the most comprehensive meta-analysis to date which has assessed the relationship between MGMT gene polymorphisms and cancer risk. In summary, this meta-analysis included the five most studied polymorphisms in the MGMT gene. We found that the Leu84Phe polymorphism is significantly associated with increased risk of overall caners. However, similar results were not found in the other four polymorphisms. As to the subgroup analysis for ethnicity, both Leu84Phe and Ile143Val were found to be associated with the increased risk of cancer in Caucasians. As to the subgroup analysis for cancer types, the Leu84Phe polymorphism was associated with increased risk of esophageal cancer, while the Ile143Val polymorphism was associated with increased risk of lung cancer. Future large-scale, well-designed studies are needed to validate these results in different populations, particularly in Africans.

Acknowledgments

This study was supported by grant 81101939 from the National Natural Science Foundation of China.

Conflict of interest

None.

Supplementary material

13277_2013_893_MOESM1_ESM.doc (611 kb)
ESM 1(DOC 611 kb)

Copyright information

© International Society of Oncology and BioMarkers (ISOBM) 2013