Association between GSTM1 and GSTT1 Allelic Variants and Head and Neck Squamous Cell Cancinoma

Backgrounds GSTM1 and GSTT1 are involved in the detoxification of carcinogens such as smoking by-products, and polymorphisms in these two genes with a result of loss of enzyme activity may increase risk of carcinogenesis. Although many epidemiological studies have investigated the association between GSTM1 or GSTT1 null genotype and head and neck squamous cell carcinoma (HNSCC), the results remain conflicting. To elucidate the overall association of GSTM1, GSTT1 and HNSCC, we included all available studies and performed this meta-analysis. Methodology/Principal Findings A dataset including 42 articles for GSTM1, 32 articles for GSTT1, and 15 articles for GSTM1 and GSTT1 in combination were identified by a search in PubMed. Associations beween HNSCC and polymorphisms of GSTM1 and GSTT1 alone and in combination were analysed by software RevMan 5.1. Stratification analysis on ethnicity and smoking status, sensitivity analysis, heterogeneity among studies and their publication bias were also tested. Association was found in overall analysis between HNSCC and GSTM1 and GSTT1 null genotype. Stratified by ethnicity, we found increased risks of HNSCC in carriers with GSTM1 null genotype in Asian, GSTT1 null genotype in South American, and dual null genotype in European and Asian. When stratified by smoking, a more significant association of GSTM1 null genotype with HNSCC risk was observed in smokers. Conclusions/Significance This meta-analysis presented additional evidence of the association between GSTM1 and GSTT1 polymorphisms and HNSCC risk.


Introduction
Head and neck neoplasms are the sixth leading cause of death by cancer [1]. The most common histological type is the squamous cell carcinoma, accounting for about 90% of all cases [2,3]. Being a multifactorial disease, the etiology of head and neck squamous cell carcinoma (HNSCC) is still a much debated question.
Smoking of cigarettes, consumption of alcohol and genetic causes are some of the foci of former etiological studies.
Enzymes of the glutathione S-transferase (GST) family are present in eukaryotes and in prokaryotes, which are composed of many cytosolic, mitochondrial, and microsomal proteins. They catalyze various reactions and participate in the phase II biotransformation of xenobiotics. GSTs contribute to the detoxification of by-products of smoking and alcohol and other exogenous chemical carcinogens which may induce HNSCC, so they have been considered as potential candidates for HNSCC susceptibility. Classesi and m of the GST superfamily have been paid lots of attention, which are encoded by GSTT1 and GSTM1 genes. The GSTM1 and GSTT1 gene have been localized to chromosome 1p13.3 and 22q11.2. Both of the genes are polymorphic and frequent homozygous deletions of the genes presenting null genotype are associated with loss of the corresponding enzyme activity. Therefore, carriers with null genotype will increase the risk of the development of HNSCC due to the decreased ability to detoxify carcinogens theoretically.
In 2003, a meta-analysis conducted by Hashibe et al. indicated modest associations of GSTM1 and GSTT1 genotypes with head and neck cancer risk [4]. However, more than twenty independent studies from various populations have further examined the relationships between these two genes and HNSCC risk, and still reported conflicting results. Some studies in HNSCC have indicated that the null genotype of GSTM1 or GSTT1 is a risk factor of HNSCC development [5][6][7]. However, such an association was not observed in some other groups [8][9][10]. Therefore, it is necessary to reevaluate the association of GSTM1 or GSTT1 null genotype with the risk of HNSCC by pooling the new published studies using meta-analysis. The present study included all eligible published case-control studies to establish a relatively comprehensive picture of the relationship between these two genes and HNSCC.

Selection criteria and identification of eligible studies
Candidate studies were identified through computer-aided literature searches in PubMed for relevant articles in English and Chinese (1995 to May 2012). To identify all articles that studied the association of GSTT1 and GSTM1 polymorphisms with HNSCC, we conducted the search using the following keywords and subject terms: 'GSTT1' or 'GSTM1', and 'squamous'. We also searched the references cited in the articles and included published works. Abstracts, case-only articles, editorials, review articles and repeated literatures were excluded. Of the articles with the overlapping data, we only included the publication with the most extensive information. The inclusion criteria in the current meta-analysis were as follows: (a) they are unrelated studies; (b) identification of squamous cell carcinoma was histologically confirmed; and (c) they have original data of genotype frequency and provided sufficient information to calculate the odds ratio (OR) or P-value.

Data extraction
Two reviewers (Zhang Y and Ni Y) independently examined the studies for inclusion in the meta-analysis and collected data on the genotype of GSTT1 and GSTM1. We extracted the following information from each study: first author, year of publication, country, ethnicity, numbers of case and control, smoking status and genotyping information. Disagreements between two reviewers were discussed and resolved with consensus. When essential information was not found in articles, we made effort to get the data from the authors (Figure 1).

Statistical analysis
The meta-analysis for GSTM1 or GSTT1 null genotype or dual null genotype compared HNSCC vs. controls. Odds ratio (OR) and its 95% confidence interval (CI) were assessed for each study. The Cochran's Q-statistic was used to test heterogeneity, and the heterogeneity was considered statistically significant when P,0.1 [11]. The Mantel-Haenszel method was used to calculate the OR for the included data in a fixed effects model in the absence of between-study heterogeneity, while random effects model was used for those with heterogeneity. P-value,0.05 was considered statistically significant, and 0.05#P-value,0.10 was indicated suggestive. In addition, we also performed stratification analyses on ethnicity, smoking and combined analyses of GSTM1 and GSTT1 on HNSCC risk. The sensitivity analysis was carried out to test the stability of the pooled effect after excluding individual studies. Begg's funnel plot was used to evaluate publication bias.   All above statistical analysis was carried out using the software packages Review Manager (RevMan) 5.1.

Heterogeneity result
Cochran's Q tests indicated heterogeneity exist in different studies in the analysis except studies of dual genes in South American (P = 0.51, I 2 = 0%) and GSTM1 in non-smokers (P = 0.65, I 2 = 0%). The random or fixed effect model was selected for comparisons with or without heterogeneity, respectively.

Meta-analysis results
A total of 7584 HNSCC cases and 8576 controls for GSTM1, 6255 cases and 7138 controls for GSTT1, 2657 cases and 3092 controls for dual genes were investigated.
For GSTM1 polymorphism, the overall meta-analysis showed a suggestively increased risk in null genotype as compared to wild genotype (OR = 1.145, 95% CI: 1.00-1.29, P = 0.05) (Figure 2). In sensitivity analysis by temporarily excluding individual studies, no single study substantially affected the pooled OR, indicating that the results of these meta-analyses are stable. Analysis after stratification by ethnicity indicated GSTM1 null genotype tended to be associated with HNSCC in Asian (OR = 1.48, 95% CI: 1.24-1.75, P,0.01), while no significant association was found in European or South American (Table 2).
We further performed stratification analysis by smoking status. As shown in Table 3, significant association of GSTM1 deletion with risk of HNSCC was observed in smoking group (OR = 1.51, 95% CI: 1.05-2.17, P = 0.03) but not in non-smoking group (OR = 1.14, 95% CI: 0.90-1.43, P = 0.28). However, we did not found any significant associations for GSTT1 in either smokers (OR = 1.01, 95% CI: 0.64-1.60, P = 0.96) or non-smokers   (Table 3), which may be due to the limited number of study with smoking information.

Publication bias
Funnel plots were performed to assess the publication bias, and these shapes did not suggest any obvious evidence of asymmetry in the analyses of GSTM1, gene-gene interaction, and GSTT1 analysis stratified by smoking status. When one study [32] for GSTT1 analysis and two studies [46,49] for GSTM1 analysis stratified by smoking status were omitted, funnel plots illustrated symmetric shape.

Discussion
Genetic factors play an important role in the etiology of tumors. For HNSCC, genes encoding xenobiotic-metabolizing enzymes (XMEs) are some of the most likely candidates that could affect individual's susceptibility to the disease, due to their involvement of the metabolic activation and detoxification of the environmental carcinogens [61]. Conjugation is one of the most common pathways of xenobiotic metabolism and is considered phase II metabolism which is catalyzed by multiple enzyme superfamilies including Glutathione S-transferases (GSTs). GSTs mediate the reactions of glutathione with electrophiles, resulting in the elimination of potentially carcinogenic chemicals [62]. GSTM1 and GSTT1 genes belonging to GSTs have been studied extensively due to their important detoxification function and high-frequency polymorphisms. GSTM1 and GSTT1 homozygous deletions (null genotype) may lead to deficient enzyme activity [63]. In the present study, the overall frequency of GSTM1 and GSTT1 null genotype in controls were 47.65% and 23.77% respectively in accordance with other studies [64][65][66][67]. After stratification for ethnicity, the frequency of GSTM1 and GSTT1 null genotype in controls in European, Asian and South American were 53.14%, 40.06%, 49.29% and 21.86%, 23.67%, 26.34% respectively, which indicated ethnic differences.
The importance of GSTM1 and GSTT1 polymorphisms effects on HNSCC has been a concern in recent years, but the data of existing studies are contradictory. An increase in the risk of HNSCCC was observed in cases with null genotypes of GSTM1 or GSTT1 in some studies [5][6][7][8]. However the risk was not found in other studies. For example, Boccia [9] and Biselli [10] did not find the association between GSTM1 or GSTT1 and HNSCC. Although the confused effect of these polymorphisms may be a result of various reasons such as demographic features of subjects and different life styles, comparatively small sample size in individual study might lead to lower statistical power and bias. The present meta-analyses of 42 studies including 7584 cases and 8651 controls for analysis of GSTM1, 32 studies including 6255 cases and 7138 controls for analysis of GSTT1, and 15 studies including 2657 cases and 3092 controls provide more comprehensive information on the relationships between two genes and HNSCC.
This meta-analysis showed that both GSTM1 and GSTT1 null genotype confers susceptibility to HNSCC in the overall analysis. GSTM1 can deals with large hydrophobic electrophiles including polycyclic aromatic hydrocarbons derived epoxides (PAH) [68,69], while GSTT1 targets a more restricted kind of compounds, like monohalomethane and ethylene oxide [70]. Different GST isoforms exhibit overlapping substrate specificity, combinations of GSTM1 and GSTT1 null genotype may theoretically confer a higher risk to HNSCC. Comparing to homozygous deletion of GSTM1 and GSTT1 alone, deletion of two genes in combination significantly increases the risk of HNSCC as showed in our combined analysis, indicating a synergenic role of GSTT1 and GSTM1 in cancergenesis.
Analyses after stratification by ethnicity revealed ethnicityspecific associations between two genes and HNSCC. Our findings indicate that GSTM1 may be an important factor in Asians in the development of HNSCC, which is similar to the results reported by Hashibe et al. [4]. However, GSTT1 but not GSTM1 may be important in South Americans, while GSTM1 and GSTT1 in combination play a vital role in Europeans and Asians. This result may be attributed to the different habits of smoking, alcohol consumption, intake of food and different genetic backgrounds in different ethnic groups.
Both GSTT1 and GSTM1 can prevent the accumulation of tobacco smoke carcinogens, and compared with non-smokers, mutations of these two genes theoretically increase the risk of HNSCC in smokers. To investigate potential gene-environment interaction, we stratified the data by smoking status. A significant association was observed in smokers with GSTM1, whereas no difference was observed between smokers and non-smokers for GSTT1. Previous studies showed that GSTT1 and GSTM1 are involved in the detoxification of carcinogens such as smoking byproducts, and polymorphisms in these two genes with a result of loss of enzyme activity may increase risk of carcinogenesis and have different role in detoxification. [68][69][70]. Although we found higher risk of GSTM1 null genotype in smokers (OR = 1.51) than non-smokers (OR = 1.14), further individual large study are required to evaluate the interaction of GSTM1 and smoking on HNSCC risk.
Although our result of this meta-analysis is constructive, its limitations and some potential bias should be addressed. First, despite that a well-designed search strategy was used to identify eligible studies, it was possible that some relevant studies were not included. This study only focused on full-text papers published in English and Chinese in PubMed, so some eligible studies in other languages or in other databases might be missed. Second, adjustments over age, gender and other environmental factors such as alcohol drinking might help better detect the association between GSTM1, GSTT1 and HNSCC. If available detailed individual data are enough for an adjusted estimate in the future, a more precise analysis should be conducted. Third, ethnicity was determined roughly by subject's country due to inadequate available data, and this classification can help us have a regional concept of these genes functions. Fourth, the controls in the included studies were recruited in different ways and not uniformly defined, which may have distorted the meta-analysis. Finally, because all the studies were designed with retrospective studies, we cannot clearly determine the causal relationship between the risk factor and HNSCC. Given the limitations and biases above, the conclusions or interpretations made from the results of this metaanalysis should be explained with caution.

Conclusions
The results of this meta-analysis suggest that GSTM1 and GSTT1 null genotypes may be associated with an increased risk of HNSCC. Further large well-designed studies are warranted to confirm these findings.

Author Contributions
Conceived and designed the experiments: JM LW. Analyzed the data: YZ YN HZ. Wrote the paper: JM YP.