Research Article
Genetic Polymorphism of Glutathione S-transferase T1, M1 and Asthma, A Meta-analysis of the Literature
Shiraz University, Shiraz 71454, Iran
Maryam Ansari-Lari
Shiraz University, Shiraz 71454, Iran
Combinations of genetic and environmental factors are important in the developing of asthma. Evidences for genetic contribution to asthma come from family aggregation and twin studies, also from genome wide searches and candidate gene approach (Blumenthal et al., 2006; Haagerup et al., 2001; Hakonarson and Halapi, 2002; Langefeld et al., 2001; Mauia, 1997; McCunney, 2005).
Inflammation is a central feature of asthma, as part of the inflammation, Reactive Oxygen Species (ROS) is formed. Genetic variations in enzymes that detoxify ROS and other products generated by several inflammatory cells may contribute to the exposure to an environmental trigger (Caramori and Papi, 2004; McCunney, 2005; Strange et al., 2001).
The members of the Glutathione S-Transferase (GST) (E.C. 2.5.1.18) super-family are potentially important in the protection of cells from ROS (McCunney, 2005; Strange et al., 2001). Members of the GST families have sequence similarity and shared catalytic properties for reaction of glutathione with reactive substrates. GSTs are well known as phase II xenobiotic detoxifying enzymes. One of the more recently recognized roles of GSTs is in oxidative defenses. Because antioxidants play a role in the pathobiology of a variety of disease and variants in the GST super-family are common, members of this super-family may be determinants of respiratory health (Strange et al., 2001).
Based on sequence homology and immunological cross-react ivity, human cytosolic GSTs have been grouped into eight distinct gene families designated GST Alpha, Mu, Pi, Sigma, Omega, Theta, Zeta and Kapa. Polymorphism has been described in many genes in these families though to date, most alteration has focused on allelism in the Pi, Mu and Theta families. Glutathione S-transferase M1 (GSTM1; a member of class μ) has two functional alleles (named GSTM1-A and GSTM1-B) and a non-functional null allele (named GSTM1-0). Homozygosity for the null-allele (null genotype) expresses no GSTM1. Glutathione S-transferase T1 (GSTT1; a member of class θ) has a functional (GSTT1-1) and a non-functional null-allele (GSTT1-0). Homozygous for non-functional allele of GSTT1 (null genotype), cause an absence of GSTT1 enzyme activity (Strange et al., 2001).
Published articles in the literature have confirming or refusing the association between genetic polymorphism of GSTM1 and/or GSTT1 and asthma risk (Ebrahimi et al., 2004; Freidin et al., 2002; Fryer et al., 2000; Gilliland et al., 2002a; Holla et al., 2006; Ivaschenko et al., 2002; Kabesch et al., 2004; Lee et al., 2005; Li et al., 2006; Makarova et al., 2004; Piirila et al., 2001; Saadat et al., 2004a; Tamer et al., 2004; Vavilin et al., 2005; Zhang et al., 2004). To clarify an association between genotypes and asthma risks, sample size is considered to be a crucial factor in the design of studies. Several studies evaluating GSTT1 and/or GSTM1 polymorphism as risk factors for asthma had small sample size (Gilliland et al., 2002a; Ivaschenko et al., 2002; Saadat et al., 2004a). In order to clarify the effect of GSTM1 and GSTT1 genotypes on the risk of developing asthma, we carried out a meta-analysis using published data from 2000 up to the May 2006, to obtain more precise estimates of risk.
Identification of studies: Studies published between 2000 and May 2006 with information of GSTT1 and/or GSTM1 genetic polymorphism and the risk of asthma were identified using electronic databases, MEDLINE (National Library of Medicine, Washington, DC, USA), EBSCO host Research Databases, ProQuest and CAB Abstract. Search terms were GSTT1 or glutathione S-transferase T1, GSTM1 or glutathione S-transferase M1 and asthma. Additional articles were also checked using the references cited in these publications.
Articles selected for analysis were studies with case-control or cross sectional designs and their primary references, which had no obvious overlap of cases with other studies. Study of Gilliland et al. (2002a) that included different ethnical groups were considered separately in our analysis. Five studies (Freidin et al., 2002; Li et al., 2006; Makarova et al., 2004; Vavilin et al., 2005; Zhang et al., 2002) were excluded, because in one of them (Li et al., 2006) there was overlapping data with the other paper of the investigators (Gilliland et al., 2002a) and in the other two studies (Makarova et al., 2004; Vavilin et al., 2005), there were no data for genotypes in control subjects and finally Freidin et al. (2002) and Zhang et al. (2004) published their article in Russian and Chinese, respectively. The application of these criteria yielded 14 studies eligible for meta-analysis (Ebrahimi et al., 2004; Fryer et al., 2000; Gilliland et al., 2002a; Holla et al., 2005; Ivaschenko et al., 2002; Kabesch et al., 2004; Lee et al., 2005; Piirila et al., 2001; Saadat et al., 2004a; Tamer et al., 2004). In all of the studies, GSTM1 and GSTT1 polymorphism were determined by PCR assays; many studies reported quality control measurements (Fryer et al., 2000; Gilliland et al., 2002a; Holla et al., 2006; Kabesch et al., 2004; Lee et al., 2005; Piirila et al., 2001; Saadat et al., 2004a; Tamer et al., 2004).
Quality score assessment: The quality of studies was also independently assessed by the same two reviewers who used quality assessment scores that were modified from previous meta-analysis of molecular association studies. These scores were based on both traditional epidemiologic considerations and genetic issues (Thakkinstian et al., 2005). Total scores ranged from 0 (worst) to 12 (best).
Statistical analysis: The odds ratio (OR) of asthma associated with GSTM1 and GSTT1 genetic polymorphism was re-calculated for each study and their corresponding 95% Confidence Intervals (CI) were estimated. The results might not be exactly the same as those of some studies as different criteria were used in the statistical analysis. The low-risk genotypes (presence of GSTM1 and presence of GSTT1) were used as the baselines for calculation ORs.
To take into account the possibility of heterogeneity across the studies, a statistical test for heterogeneity was performed based on the Q statistic, in which a P-value greater than 0.05 suggested a lack of heterogeneity (DerSimonian and Laird, 1986). We carried out meta-analysis using a fixed-effects model and a random-effects model. The fixed-effects model assumes no significant heterogeneity between the results of the individual studies being pooled, whereas, the random-effects model allows for such heterogeneity. The fixed-effects and random-effects models were used by Mantel-Haenszel method (Mantel and Haenszel, 1959) and DerSimonian and Laird method (DerSimonian and Laird, 1986), respectively.
The analyses were also conducted on the subgroups of studies based on the score of study quality (after excluding two studies having lowest quality scores), smoking behavior of subjects and age of subjects (childhood and adulthood). Because GSTT1 and GSTM1 genotypes may show additive effects in the development of the asthma, further analysis combining the GSTM1 and GSTT1 genotypes was carried out.
Association between polymorphism of GSTM1 and asthma risk: We identified 14 eligible studies, including 8010 subjects (2292 cases and 5718 controls) in relation to GSTM1 polymorphism and risk of asthma, which are summarized in Table 1. From these, 4, 5 and 5 studies were carried out in Asian, European and American countries, respectively. Case-control and cross sectional designs were 7 and 7 studies, respectively (Table 1). The number of subjects in these studies varied considerably (range 114 to 3054 individuals).
The frequencies of GSTM1 null genotype varied in the control participants, from 23.5 to 54.5% (Table 2). The distribution of GSTM1 polymorphism among control individuals is in agreements with other reports (La Torre et al., 2005; Ye and Sang, 2005).
Test for heterogeneity between studies showed a significant heterogeneity (Q statistic = 45.09, df = 13, p<0.0001). The overall OR of the asthma risk associated with GSTM1 null genotype was 1.21 (95% CI: 1.08-1.35). Exclusion of two studies with lowest quality scores (Ebrahimi et al., 2004; Ivaschenko et al., 2002 (Table 1) resulted in a dramatic decrease in Q statistic (45.09 to 27.68). Therefore, in order to reduce the heterogeneity, in further analysis these two studies were excluded.
Table 3 also summarizes the results of the stratified meta-analysis. Subgroup analysis, regarding age of subjects and smoking status of subjects were carried out.
Stratifying the meta-analysis by age of subjects, the pooled ORs for GSTM1 null genotype were 1.56 (95% CI: 1.25-1.94) in adults and 1.10 (95% CI: 0.97-1.25) in childhood. Although there was no identifiable evidence of heterogeneity in the analysis of GSTM1 and asthma in childhood, there was a borderline heterogeneity between studies in adulthood (Table 3).
Considering that cigarette smoking is an obvious risk factor for asthma and the GST genes are involved in the metabolism of various compounds present in cigarette smoke, further analysis regarding smoking status of subjects were carried out in adulthood subjects.
Table 1: | Studies used in the meta-analysis |
Note: Both means subjects were smokers and non-smokers |
Table 2: | Genetic polymorphism of GSTM1 and risk of asthma |
Table 3: | Summary of meta-analysis of GSTM1 polymorphism and the risk of asthma |
Unfortunately, only few studies reported the smoking status of their subjects and mentioned that their subjects were non-smokers (Fryer et al., 2000; Saadat et al., 2004a) other studies reported that some of their subjects were smokers and others were non-smokers (Holla et al., 2006; Ivaschenko et al., 2002; Piirila et al., 2001; Tamer et al., 2004). Unfortunately, there were no raw data about smoking behavior of the subjects in the studies using mixed of smoker and non-smoker subjects (Holla et al., 2006; Ivaschenko et al., 2002; Piirila et al., 2001; Tamer et al., 2004). After grouping according to smoking status, the GSTM1 null genotype was associated with an increased risk of asthma in either non-smokers (OR=1.95, 95% CI: 1.21-3.13) or mixed subjects of smokers and non-smokers (OR=1.47, 95% CI: 1.14-1.88).
Association between polymorphism of GSTT1 and asthma risk: We identified 8 eligible studies, including 4965 subjects (3765 controls and 1200 patients) in relation to GSTT1 polymorphism and risk of asthma, which are summarized in Table 4. From these, 2 and 6 studies were carried out in Asian and European countries, respectively. The numbers of subjects in these studies varied considerably (range 170 to 3054 individuals).
The frequencies of GSTT1 null genotype varied in the control participants, from 8.2 to 41.1% (Table 4). The distribution of GSTT1 polymorphism among control individuals is in agreements with other reports (La Torre et al., 2005; Saadat 2006; Ye and Sang, 2005).
Test for heterogeneity showed a significant heterogeneity between studies (Q statistic = 43.56, df = 7, p>0.001). Using the random effects model the overall OR of the asthma risk associated with GSTT1 null genotype was 1.36 (95% CI: 1.11-1.63). Exclusion of two studies with lowest quality scores (Ebrahimi et al., 2004; Ivaschenko et al., 2002; Table 1) resulted in a dramatic decrease in Q statistic (43.36 to 11.27). Therefore, in order to reduce the heterogeneity, in further analysis these two studies were excluded.
Table 5 also summarizes the results of the stratified meta-analysis. Subgroup analysis based on age and smoking status of subjects was carried out.
Table 4: | Genetic polymorphism of GSTT1 and risk of asthma |
Table 5: | Summary of meta-analysis of GSTT1 polymorphism and the risk of asthma |
In all the subgroup analyses, there was no identifiable evidence of heterogeneity in the analyses of GSTT1 and asthma risk. It should be mentioned that using reminder 6 studies, the overall OR of the asthma risk associated with GSTT1 null genotype was 1.06 (95% CI: 0.86-1.31). The GSTT1 null genotype was associated with asthma risk in non-smoker adults (OR = 2.06, 95% CI: 1.21-3.71). Other associations were not significant.
Combination of GSTM1 and GSTT1 genotypes and risk of asthma: GSTs have overlapping substrate specificities. Therefore, deficiency of an individual GST isoenzyme may be compensated by other isoforms (Hayes and Pulford, 1995; Strange et al., 2001). Therefore, simultaneous determination of all GST genotypes appears to be a prerequisite for reliable interpretation of the role of the GST family in asthma development. To investigate whether profile of GST genotypes are associated with the risk of the asthma, further analysis combining the GSTT1 and GSTM1 genotypes were also carried out. The reference group consisted of individuals with two putative low-risk genotypes, i.e. the presence of GSTM1 and GSTT1 functional alleles. Four studies have examined the relation between asthma risk and combination of GSTM1 and GSTT1 polymorphism (Holla et al., 2006; Ivaschenko et al., 2002; Saadat et al., 2004a; Tamer et al., 2004). These studies included 1210 subjects (601 cases and 609 controls) (Table 6). It should be mentioned that the observed frequencies for genotype combinations in control groups did not show significant difference with the expected frequencies according to the Hardy-Weinberg equilibrium (the values of χ2 in Russia, Iran, Turkey and Czech were equal to 0.22, 1.21, 2.11 and 1.86, respectively (for all comparisons df = 2, p>0.30).
Table 6: | Combination genotypes of GSTM1 and GSTT1 and asthma risk |
Note: The trend for none, one and two putative high-risk genotypes was significant using 4 studies χ2 = 37.53, df = 1, p<0.00001 and using 3 studies after excluding study from Russia χ2 = 12.07, df = 1, p = 0.00051, a Reference genotype. b After combination these two genotypes using 4 studies Q statistic = 9.92, df = 3; OR=1.29, 95% CI: 0.99-1.67 and using 3 studies after excluding study from Russia Q statistic=5.77, df=2; OR=1.18, 95% CI: 0.89-1.56. c Using 4 studies Q statistic = 29.616, df = 3; OR = 3.14, 95% CI: 2.12-4.66 and using 3 studies after excluding study from Russia Q statistic = 14.07, df = 2; OR = 2.15, 95% CI: 1.39-3.33.d There is no data about M1+/T1- and M1-/T1+ genotypes |
Table 6 displays the risk of asthma associated with each combination of genotypes as well as the trend in risk associated with zero, one and two putative high-risk genotypes. Subjects with null genotypes for both GSTM1 and GSTT1 were at a significant higher risk for developing asthma (OR = 3.14, 95% CI: 2.12-4.66; Q statistic = 29.616, df = 3) compared with subjects who had both active genes. The trend in risk associated with zero, one and two putative high-risk genotypes was significant (χ2 = 37.53, df = 1, p<0.00001). If the study of Ivaschenko et al. (2002) was excluded from the analysis the heterogeneity decrease and subjects with null genotypes for both GSTM1 and GSTT1 were at a significant higher risk for developing asthma (OR = 2.15, 95% CI: 1.39-3.33; Q statistic=14.07, df = 2) compared with subjects who had both active genes. The trend in risk associated with zero, one and two putative high-risk genotypes was significant (χ2 = 12.07, df = 1, p = 0.0005). Therefore there is an additive effect for GSTT1 and GSTM1 genotypes.
The overall goal of meta-analysis is to combine the results of previous studies to arrive at summary conclusions about a body of research. It is most useful in summarizing prior research when individual studies are small and they are individually too small to yield a valid conclusion. In the present study, 14 studies were found eligible for meta-analysis (Ebrahimi et al., 2004; Fryer et al., 2000; Gilliland et al., 2002a; Holla et al., 2005; Ivaschenko et al., 2002; Kabesch et al., 2004; Lee et al., 2005; Piirila et al., 2001; Saadat et al., 2004a; Tamer et al., 2004). There was considerable heterogeneity between studies in this meta-analysis in our initial results (Table 3, 5 and 6). Heterogeneity can be an important drawback of meta-analysis when a pooled estimate is the main objective. Other wise, heterogeneity can be also a major goal in meta-analysis: finding the variables associated with variation across the primary studies can help future research on a topic. In the present meta-analysis, two studies (Ebrahimi et al., 2004; Ivaschenko et al., 2002) were the major sources of heterogeneity. It might be due to uncontrolled confounding and inherent bias of study design. One study (Ivaschenko et al., 2002) used subjects with highly heterogeneous with respect to the age of subjects (adulthood and childhood). Selection bias may be a major source of heterogeneity; therefore such bias was reduced by removing the studies. Although there is some evidence of heterogeneity across studies, which will produce an overestimate of the true association, studies that contribute to heterogeneity, do not significantly alter the estimate of the overall OR of the asthma risk associated with GSTM1 null genotype (All studies: Q statistic = 45.09, df = 13; OR = 1.21, 95%CI: 1.08-1.35; After removing the two studies: Q statistic = 27.68, df = 11; OR=1.20, 95%CI: 1.08-1.35).
Numerous studies have reported an association between Environmental Tobacco Smoke (ETS) exposure or cigarette smoking and asthma risk (Cook and Stranchan, 1997; Cook et al., 1998). Unfortunately, only two studies (Fryer et al., 2000; Saadat et al., 2004a) mentioned that their subjects were adults and non-smokers. In other articles, authors did not report the GST polymorphism in smoker and non-smoker subjects (Holla et al., 2006; Ivaschenko et al., 2002; Piirila et al., 2001; Tamer et al., 2004). Overall our present meta-analysis revealed that the null genotype of GSTM1 was associated with risk of asthma in adults (OR = 1.56, 95% CI: 1.25-1.94) especially in non-smoker ones (OR = 1.95, 95% CI: 1.21-3.13). Also the null genotype of GSTT1 was associated with increased risk of asthma in non-smoker adults (OR = 2.06, 95% CI: 1.21-3.71). It might be suggested that chronic smoking carries such a high dose of toxins into the body that overloads the capacity of either GSTM1 or GSTT1 detoxification system. Therefore, the GSTM1 and GSTT1 lack their protective roles against development of asthma in adults with positive history of smoking. Interestingly, there are same story between risk of developing cataract and polymorphism of GSTM1 in non-smoker females (Saadat et al., 2004b). However, this result is not consistent with a report investigated association between in utero exposure to maternal smoking and prevalence of many asthmatic phenotypes with respect to the GSTM1 genotype of children. The authors reported that among children with null genotype of GSTM1, in utero exposure was associated with increased prevalence of early onset asthma, asthma with recurrent symptoms, persistent asthma, lifetime history of wheezing, wheezing with exercise, wheezing requiring medication and emergency room visits in the past year (Gilliland et al., 2002b).
It should be noted that the GSTM1 and GSTT1 are involved in detoxification of a variety of compounds, some that overlap between enzymes and some that are highly specific (Hayes and Pulford 1995; Strange et al., 2001). We found four studies that reported the combination genotypes of GSTM1 and GSTT1 in control subjects and asthma patients (Holla et al., 2005; Ivaschenko et al., 2002; Saadat et al., 2004a; Tamer et al., 2004). Statistical analysis showed that there is significant trend in the risk associated with zero, one and two putative high-risk genotypes (Table 6). Those who had null genotypes of GSTM1 and GSTT1 had an increased risk compared with those who had both active genes (OR = 3.14, 95% CI: 2.12-4.66). Because both GSTM1 and GSTT1 showed substrate overlapping (Hayes and Pulford 1995; Strange et al., 2001) and detoxify several compounds involved in development of asthma (Caramori and Papi, 2004; Mauia, 1997; McCunney, 2005), the observed additive effect might be interpreted. It should be mentioned that such additive effect of GSTM1 and GSTT1 was reported for developing colorectal (Saadat and Saadat, 2001) and gastric cancers (Saadat, 2006) and senile cataract (Saadat et al., 2004b).
The mRNA level of GSTT1 in lung tissue of women was affected positively by the serum level of cotinine (a metabolite of nicotine). Also the GSTT1 mRNA level was decreased as a function of age (Spivack et al., 2003). There are some indications of ethnic differences in the effects of GSTM1 genotype on FVC and FEV1 growth (Carroll et al., 2005; Gilliland et al., 2007). Taken together, future research in this field should take great care in the interaction between risk factors and combination genotypes of GSTs (such as GSTT1, GSTM1, GSTP1, etc) with respect to age and ethnicity of subjects. Also additive effect of involved genes should be investigated. Therefore, stratification of subjects according to their age, sex, ethnicity, smoking behavior and GST genotypes (including combination of genotypes) in future studies is necessary.
Finally in this meta-analysis, only published studies were used and in some analysis some studies were excluded because the raw data was not available. Therefore, publication bias is an issue.
This study supported by Shiraz University.