Genetic polymorphisms of glutathione S‐transferase T1 (GSTT1) and susceptibility to gastric cancer: a meta‐analysis

Abstract

The association between glutathione S‐transferase T1 (GSTT1) polymorphism and gastric cancer risk has been both confirmed and refuted in a number of published studies. Most of these studies were based on small sample sizes. We carried out a meta‐analysis of the research published up to August 2005 to obtain more precise estimates of gastric cancer risk associated with GSTT1 polymorphism. In the present study, 16 case‐control studies (with a total of 6717 subjects) were eligible for meta‐analysis. There was no evidence of heterogeneity between the studies. The GSTT1 null genotype conferred a 1.06‐fold increased risk of gastric cancer, which was not significant (95% confidence interval [CI]: 0.94–1.19). However, in the analysis of ethnic groups, we observed distinct differences associated with GSTT1 status. Restricting analyses to ethnic groups, the pooled odd ratios for the GSTT1 genotype were 1.27 in Caucasians (95% CI: 1.03–1.57) and 0.98 in Asians (95% CI: 0.86–1.13). Glutathione S‐transferase M1 (GSTM1) and GSTT1 are involved in detoxification of a variety of compounds, some that overlap between enzymes and some that are highly specific. To investigate whether the profile of glutathione S‐transferase genotypes was associated with risk of gastric cancer, further analyses combining the GSTT1 and GSTM1 genotypes were also carried out. There was a significant trend in risk associated with zero, one and two putative high‐risk genotypes (χ² = 9.326, d.f. = 1, P = 0.0023). Those who had null genotypes of GSTM1 and GSTT1 had an increased gastric cancer risk compared with those who had both active genes (odds ratio = 2.08, 95% CI: 1.42–3.10). (Cancer Sci 2006; 97: 505–509)

Gastric cancer represents the second most frequent cancer in the world and the fourth most frequent in Europe. Nutritional, infectious and genetic factors have been shown to play a role in the complex multifactorial and multistage process of gastric carcinogenesis.^{( 1} , ² , ^{3 )} Also, several lines of evidence have indicated that cigarette smoking is a risk factor for developing stomach cancer^{( 4} , ^{5 )} and that the carcinogens in tobacco smoke, such as benzo[α]pyrene,^{( 6 )} and drinking water^{( 7 )} are involved in gastric carcinogenesis.

Interindividual differences in the cellular mechanisms of activation and detoxification of carcinogenic chemicals could confer different degrees of susceptibility to cancer.^{( 8 )} Several enzymes are involved in the detoxification of xenobiotic compounds. The glutathione S‐transferase (GST) family of genes has a critical function in protection against electrophiles and the products of oxidative stress.^{( 9} , ^{10 )} GST are involved in the metabolism of many xenobiotics, including an array of environmental carcinogens, chemotherapeutic agents and endogenously derived reactive oxygen species.^{( 9 )} Based on sequence homology and immunological cross‐reactivity, human cytosolic GST have been grouped into seven subfamilies designated GST α, , π, σ, ω, θ and ζ.^{( 11} , ^{12 )} GSTT1 (GSTT1), a member of class θ, has a functional and a non‐functional allele (GSTT1‐0). Homozygosity for the non‐functional allele of GSTT1 (null‐genotype) causes an absence of GSTT1 enzyme activity.^{( 13 )} Individuals with GSTT1 null‐genotype may be at increased risk for genotoxic damage from environmental or occupational 1,3‐butadene exposure.^{( 14 )} A study of the GSTT1 gene may provide insights into the nature of common environmental or dietary exposures that produce chromosomal damage. People with the GSTT1 null genotype show reduced ability to detoxify metabolites of ethylene oxide.^{( 15 )}

Very recently, La Torre et al. reported the results of a meta‐analysis concerning GSTM1 polymorphism and risk of gastric cancer.^{( 16 )} As GSTT1 is expressed at relatively high levels in many cell types of the human gastrointestinal tract,^{( 17} , ^{18 )} it has been suggested that genetic polymorphism of GSTT1 might play an important role in protection against carcinogens. Studies have been published both confirming and refuting the association between GSTT1 polymorphism and gastric cancer risk.^{( 16} , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁰ , ³¹ , ³² , ³³ , ³⁴ , ³⁵ , ^{36 )} In clarifying an association between genotype and cancer risk, sample size is considered to be a crucial factor in the design of case‐control studies. However, several case‐control studies evaluating GSTT1 polymorphism as risk factor for gastric cancer had small sample sizes.^{( 21} , ²² , ²³ , ²⁸ , ^{30 )} In order to clarify the effect of GSTT1 genotype on the risk of developing gastric cancer, we carried out a meta‐analysis using published data from 1996 to August 2005 to obtain more precise estimates of risk.

Materials and Methods

Identification of studies

Studies published between January 1996 and August 2005 containing information on GSTT1 genetic polymorphism and the risk of gastric cancer were identified using the electronic database Medline (National Library of Medicine, Washington, DC, USA). Search terms were ‘GSTT1’ or ‘glutathione S‐transferase T1’ and ‘gastric cancer’ or ‘stomach cancer’. Additional studies were also checked using the references cited in these publications.

Publications selected for analysis were studies with case‐control design and primary references that had no obvious overlap of cancer cases with other studies. Three studies were excluded: one showed overlap with another study,^{( 21} , ^{35 )} one was published in Chinese,^{( 34 )} and a third had no raw data was available.^{( 36 )} The study of Colombo et al.^{( 29 )}, which included different ethnic groups, was considered separately in our analysis. The application of these criteria yielded 16 case‐control studies eligible for meta‐analysis.^{( 19} , ²⁰ , ²¹ , ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁰ , ³¹ , ³² , ^{33 )} In all of the studies, GSTT1 polymorphism was determined by polymerase chain reaction assays; many studies reported quality control measurements.^{( 21} , ²² , ²⁴ , ²⁸ , ²⁹ , ³² , ^{33 )}

Statistical analysis

The odds ratio (OR) of gastric cancer associated with GSTT1 genetic polymorphism was recalculated for each study, and their corresponding 95% confidence intervals (CI) were estimated. Our results might not be exactly the same as those within each study as different criteria were used in the statistical analyses. A low‐risk genotype (presence of GSTT1) was used as the baseline for calculating OR.

Meta‐analysis is not a simple average of the effects in all studies. Rather, it is a more precise study. To take into account the possibility of heterogeneity across the studies, a statistical test for heterogeneity was carried out based on the Q statistic, in which a P‐value greater than 0.05 suggested a lack of heterogeneity.^{( 37 )} We carried out meta‐analysis using both a fixed‐effects and a random‐effects model. The fixed‐effects model assumed no significant heterogeneity between the results of the individual studies being pooled, whereas the random‐effects model allowed for such heterogeneity. The fixed‐effects and random‐effects models used the Mantel–Haenszel^{( 38 )} and DerSimonian and Laird^{( 37 )} methods, respectively. Analyses were also conducted on subgroups of studies based on ethnic origin (Caucasians and Asians), sample size, smoking behavior (smokers and non‐smokers) and control sources. Because GSTT1 and GSTM1 genotypes may show additive effects in the development of gastric cancer, further analysis combining the GSTM1 and GSTT1 genotypes was carried out.

Results

We identified 16 eligible studies, including a total of 6717 subjects, that related to GSTT1 polymorphism and risk of gastric cancer, which are summarized in Table 1. Of these, 10 were carried out in Asian countries, three were in European countries and three were in American countries. Hospital‐based controls were used in six studies (Table 1). The numbers in the case‐control studies varied considerably (range 28–1014 individuals).

Table 1.

Genetic polymorphism of glutathione S‐transferase T1 and risk of gastric cancer

Investigators	Reference	Country	Ethnicity	Source of controls	Controls		Cases		OR	95% CI
					T1+	T1−	T1+	T1−
Deakin et al. 1996	19	UK	Caucasian	Hospital‐based	415	94	93	21	1.00	0.57–1.73
Kato et al. 1996	20	Japan	Asian	Hospital‐based	70	56	73	66	1.13	0.68–1.89
Setiawan et al. 2000	21	China	Asian	Population‐based	228	190	37	44	1.43	0.86–2.36
Saadat and Saadat 2001	22	Iran	Caucasian	Population‐based	90	41	27	15	1.22	0.55–2.69
Cai et al. 2001	23	China	Asian	Population‐based	47	47	54	41	0.76	0.41–1.40
Lan et al. 2001	24	Poland	Caucasian	Population‐based	352	66	233	60	1.37	0.92–2.06
Gao et al. 2002	25	China	Asian	Population‐based	104	119	82	71	0.76	0.49–1.17
Wu et al. 2002	26	Taiwan	Asian	Hospital‐based	130	148	181	175	0.85	0.61–1.18
Choi et al. 2003	27	South Korea	Asian	Population‐based	83	94	37	43	1.03	0.58–1.80
Tamer et al. 2004	28	Turkey	Caucasian	Hospital‐based	151	53	49	21	1.22	0.64–2.31
Colombo et al. 2004	29	Brazil	Caucasian	Population‐based	114	21	73	14	1.04	0.47–2.31
Colombo et al. 2004	29	Brazil	Negroid	Population‐based	8	7	10	3	0.34	0.05–2.26
La Torres et al. 2004	30	Colombia	Caucasian	Hospital‐based	82	14	38	8	1.23	0.43–3.48
Palli et al. 2005	31	Italy	Caucasian	Population‐based	455	91	134	41	1.53	0.99–2.37
Nan et al. 2005	32	China	Asian	Hospital‐based	367	247	229	171	1.11	0.85–1.44
Mu et al. 2005	33	China	Asian	Population‐based	201	192	103	93	0.95	0.66–1.35

CI, confidence interval; OR, odds ratio.

The frequency of GSTT1 null genotype varied in the control participants, from 14.5 to 53.4%. The distribution of GSTT1 polymorphism among control individuals was in agreements with other reports.^{( 39 )}

A test for heterogeneity between studies showed no evidence of heterogeneity (Q statistic = 14.94, d.f. = 15, P > 0.05). Therefore, the fixed‐effects model and random‐effects model results were generally the same in terms of their OR and 95% CI values (Table 2). The overall OR of gastric cancer risk associated with the GSTT1 null genotype was 1.06 (95% CI: 0.94–1.19).

Table 2.

Summary of meta‐analysis of case‐control studies of glutathione S‐transferase T1 polymorphism and risk of gastric cancer

Category	Q statistic	d.f.	Fixed‐effects		Random‐effects
			OR	95% CI	OR	95% CI
All studies	14.94	15	1.06	0.94–1.19	1.06	0.94–1.18
Asians	6.75	7	0.98	0.86–1.13	0.98	0.86–1.13
Caucasians	2.06	6	1.27	1.03–1.57	1.27	1.04–1.57
Hospital‐based studies	2.38	5	1.03	0.87–1.22	1.03	0.88–1.22
Population‐based studies	12.40	9	1.08	0.92–1.26	1.08	0.93–1.26
Non‐smokers	4.88	3	0.99	0.67–1.45	0.99	0.69–1.44
Smokers	3.23	3	1.27	0.94–1.72	1.27	0.95–1.71
Studies with fewer than 100 cases and 100 controls	5.02	7	1.09	0.86–1.38	1.09	0.87–1.37
Studies with at least 100 cases and 100 controls	9.84	7	1.05	0.92–1.19	1.05	0.92–1.19

There was no heterogeneity between studies: P > 0.05. CI, confidence interval; OR, odds ratio.

Table 2 also summaries the results of the stratified meta‐analysis. Subgroup analyses were carried out for ethnicity, sample size, smoking status and control source. In all of the subgroup analyses, there was no identifiable evidence of heterogeneity for GSTT1 and gastric cancer risk.

The frequency of the GSTT1 null genotype in control participants of Asian and Caucasians populations varied from 40 to 53.4% and from 14.6 to 31.3%, respectively. We observed distinct differences in GSTT1 status in the analysis of ethnic groups. Restricting analyses to ethnic groups, the pooled OR for GSTT1 polymorphism were 1.27 (95% CI: 1.03–1.57) in Caucasians, and 0.98 (95% CI: 0.86–1.13) in Asians.

Considering that cigarette smoking is an obvious risk factor for stomach cancer, and that GST genes are involved in the metabolism of various carcinogens present in cigarette smoke, further analysis regarding smoking status of subjects was carried out. Unfortunately, only few studies reported the smoking status of their subjects.^{( 21} , ²⁴ , ²⁵ , ^{28 )} After grouping according to smoking status, the GSTT1 null genotype was not associated with an increased risk of gastric cancer in either smokers (OR = 1.27, 95% CI: 0.94–1.72) or non‐smokers (OR = 0.99, 95% CI: 0.67–1.45).

Stratifying the meta‐analysis by sample size, the pooled OR for GSTT1 genotype were 1.05 (95% CI: 0.92–1.19) in studies with at least 100 cases and 100 controls, and 1.09 (95% CI: 0.86–1.37) in studies with fewer than 100 cases and 100 controls. As it is conceivable that the GSTT1 gene might confer susceptibility to non‐cancer disease, its genotype frequency might have differed between the population‐based and hospital‐based controls (non‐cancer diseases). For example, it was reported that there is a significant association between GSTT1 polymorphism and multifactorial diseases such as asthma^{( 40 )} and cardiovascular disease.^{( 41 )} Therefore, stratification by source of control was considered in the meta‐analysis. There was no difference between population‐based and hospital‐based control studies for association between GSTT1 genotype and risk of gastric cancer (Table 2).

Because the different GST classes have overlapping substrate specificities,^{( 9} , ^{10 )} deficiency of an individual GST isoenzyme may be compensated by other isoforms.^{( 42 )} Therefore, simultaneous determination of all GST genotypes appears to be a prerequisite for reliable interpretation of the role of the GST family in gastric cancer development. To investigate whether the profile of GST genotypes is associated with risk of gastric cancer, further analyses combining the GSTT1 and GSTM1 genotypes were also carried out. The reference group consisted of individuals with two putative low‐risk genotypes, that is, the presence of functional GSTM1 and GSTT1 alleles. Four studies have examined the relationship between gastric cancer risk and combination of GSTM1 and GSTT1 polymorphisms.^{( 22} , ²³ , ²⁸ , ^{31 )} These studies included 1357 subjects (382 cases and 975 controls). It should be mentioned that the observed frequencies for genotype combinations in control groups did not show significant differences in expected frequencies according to the Hardy–Weinberg equilibrium; the values of χ² in Iran, China, Italy and Turkey were equal to 0.99, 3.47, 0.76 and 1.58, respectively (for all comparisons d.f. = 2, P > 0.05). Table 3 displays the risk of gastric cancer associated with each combination of genotype, 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 gastric cancer (OR = 2.08, 95% CI: 1.42–3.10) compared with subjects who had both active genes. It should be mentioned that the GSTM1 null genotype alone was associated with an increased gastric cancer risk (OR = 1.43, 95% CI: 1.12–1.83; Q statistic = 7.23, d.f. = 3, P > 0.05). Therefore there is an additive effect for GSTT1 and GSTM1 genotypes. The trend in risk associated with zero, one and two putative high‐risk genotypes was significant (χ² = 9.326, d.f. = 1, P = 0.0023).

Table 3.

Combination genotypes of glutathione S‐transferase M1 and glutathione S‐transferase T1, and gastric cancer risk

Country	Reference	Category	Combinations of genotypes				Total
			M1+/T1+ †	M1+/T1− ‡	M1−/T1+ ‡	M1−/T1− §
Iran	22	Controls	51	27	39	14	131
		Cases	11	5	16	10	42
Italy	31	Controls	222	49	233	42	546
		Cases	72	13	62	28	175
Turkey	28	Controls	82	34	69	19	204
		Cases	21	9	28	12	70
China	23	Controls	30	21	17	26	94
		Cases	21	14	33	27	95

The trend for none, one and two putative high‐risk genotypes was significant: χ² = 9.326, d.f. = 1, P = 0.0023. ^†Reference genotype. ^‡After combination of these two genotypes: odds ratio (OR) = 1.10; 95% confidence interval (CI): 0.83–1.45. ^§OR = 2.08, 95% CI: 1.42–3.10.

Discussion

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 when they are individually too small to yield a valid conclusion. In the present study, 16 case‐control studies were found eligible for meta‐analysis.^{( 19} , ²⁰ , ²¹ , ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁰ , ³¹ , ³² , ^{33 )} There was no evidence of heterogeneity between studies in this systematic‐review and meta‐analysis (Table 1). In the present meta‐analysis with a total of 6717 subjects, the GSTT1 null genotype conferred a 1.06‐fold increased risk of gastric cancer, which was not statistically significant (95% CI: 0.94–1.19).

It is now widely accepted that differences in the distribution of various ethnicities between cases and controls may be a source of confounding when pooling studies.^{( 43 )} In the pooled analysis, the frequency of null genotype of GSTT1 showed distinct differences in Caucasians and Asians (Table 1), as reported previously.^{( 16} , ^{39 )} Stratification by ethnicity was considered in the meta‐analysis (Table 2). The results revealed that the pooled OR associated with GSTT1 polymorphism was significant in Caucasians (OR = 1.27, 95% CI: 1.03–1.57), while the above‐mentioned association was not significant in Asian populations (OR = 0.98, 95% CI: 0.86–1.13). Differences between ethnic populations were observed in published reports of meta‐analysis studies concerning GST genotypes at risk of acute leukemia^{( 39 )} and gastric cancer.^{( 16 )}

Based on a recently published meta‐analysis study, there is a significant association between GSTM1 polymorphism and risk of gastric cancer.^{( 16 )} It should be noted that GSTM1 and GSTT1 are involved in detoxification of a variety of compounds, some that overlap between enzymes and some that are highly specific.^{( 9} , ¹⁰ , ^{41 )} We found four studies that reported the combination genotypes of GSTM1 and GSTT1 in control subjects and gastric cancer patients.^{( 22} , ²³ , ²⁸ , ^{31 )} Statistical analysis showed that there was a significant trend in the risk associated with zero, one and two putative high‐risk genotypes (χ² = 9.326, d.f. = 1, P = 0.0023). Those who had null genotypes of GSTM1 and GSTT1 had an increased risk compared with those who had both active genes (OR = 2.08, 95% CI: 1.42–3.10) (Table 3). Because both the GSTM1 and GSTT1 genes are expressed in the stomach,^{( 17} , ^{18 )} the effects of their overlapping substrate specificities^{( 10} , ^{41 )} and detoxification of carcinogens involved in the development of gastric cancer^{( 14} , ¹⁵ , ¹⁶ , ^{44 )} might be additive.

Previous studies have shown that Helicobacter pylori infection leads to an increased production of reactive oxygen species within the gastric mucosa, which is thought to play a major role in the development of gastric cancer.^{( 3} , ^{45 )} Even though the prevalence of infection may be very high (70–90% in developing countries, 25–50% in developed countries), most infected individuals are asymptomatic and only a few patients develop peptic ulcer or gastric cancer.^{( 45} , ^{46 )} Unfortunately, almost all of the published studies that we used in the present meta‐analysis did not mention the status of H. pylori infection of their subjects. It should be mentioned that the GSTT1 null genotype showed a significant difference between ethnic groups. At present it is very difficult to say whether H. pylori infection has any influence on the relationship between GSTT1 polymorphism and gastric cancer risk, especially when we know that H. pylori has distinct genotypes with different pathogenecity.^{( 47 )} However, the ethnic differences in GSTT1 genotypes and H. pylori genotypes, and their interactions, might be involved in the development of gastric cancer. Unfortunately, at present time, there is no data describing this.

In this meta‐analysis, only published studies were used, and some studies were excluded because the raw data was not available. Therefore, publication bias is an issue. Future research in this field should take great care in the interaction between risk factors (life‐style conditions, such as smoking behavior, H. pylori genotypes, and alcohol and drug consumption) and combination genotypes of GST (such as GSTT1, GSTM1 and GSTP1). Also additive effects should be investigated for the genes involved.

Acknowlegments

We would like to thank Dr Mayram Ansari‐Lari for her invaluable discussion. This study was supported by Shiraz University.