Introduction

Lung cancer is the leading cause of cancer-related death in the worldwide (lberg and Samet 2003; Kuper et al. 2002). Human lung cancer is associated with exposure to carcinogens such as polycyclic aromatic hydrocarbons (PAH) and asbestos; these mainly come from tobacco smoke, diet, and occupational exposure (Peto et al. 1996; Beckett 1993; Cote et al. 2009). However, not all of those who have been exposed to the risk factors will develop lung cancer, suggesting that other causes, including genetic susceptibility, might contribute to the variation in individual lung cancer risk (Schabath et al. 2002; Hayashi et al. 1991a, b; Agarwal 2001; Kiyohara et al. 2004). The exact mechanism of lung cancer is still under investigation. This genetic susceptibility may result from inherited polymorphisms in the genes involved in carcinogen metabolism. To our knowledge, many studies have reported that the variations of several drug-metabolising enzymes, such as cytochrome P450, NAD(P)H quinone reductase 1, myeloperoxidase, glutathione S-transferase, and arylamine N-acetyltransferses, are associated with the sensitivity of lung cancer (Agundez 2008; Carlsten et al. 2008; Raimondi et al. 2006; Kiyohara et al. 2005; Le Marchand et al. 2003; Uematsu et al. 1991). Cytochrome P450 2E1 (CYP2E1), a member of the cytochrome P450 superfamily, is a natural ethanol-inducible enzyme that is involved in the metabolic oxidation of low-molecular weight carcinogens such as N-nitrosoamines, benzene and vinyl chloride and aniline. CYP2E1 gene is located on 10q24.3–qter. It is 18,754 bp long consisting of nine exons and eight introns, which encodes a 493 amino acid protein. The variant type of this polymorphic site can enhance the transcription and increase the level of CYP2E1 enzymatic activity in vitro (Liu et al. 2009; Hayashi et al. 1991a). Genetic mutations in the CYP2E1 gene are considered to be associated with increased CYP2E1 activity and may be linked to the carcinogenic process. CYP2E1 is an ethanol-inducible enzyme that metabolically activates various carcinogens, such as benzene, vinyl chloride and N-dimethylnitrosamines (Yamazaki et al. 1992; Bellec et al. 1996). N-nitrosamines are present in tobacco smoke, and activation of nitrosamines has been linked to the development of various cancers (Hoffmann and Hecht 1985; Hecht and Hoffmann 1988). Several CYP2E1 polymorphisms had been identified by restriction fragment length polymorphism analysis (Hayashi et al. 1991a; Uematsu et al. 1991). The most extensively studied single nucleotide polymorphisms of CYP2E1 are RsaI polymorphism in the 50-flanking region and the DraI polymorphism in intron 6.

However, many epidemiologic studies have reported to evaluate the association between CYP2E1 RsaI (rs2031920), DraI (rs6413432) polymorphisms and lung cancer risk in diverse populations (Li et al. 2000, 2004, 2005, 2008, 2012; Klinchid et al. 2009; Eom et al. 2009; Chen et al. 2002; Zienolddiny et al. 2008; Minegishi et al. 2007; Lee et al. 2006; Oyama et al. 2003; Liu et al. 2010; Liang et al. 2004; Gu et al. 2004; Su et al. 2011; Qu et al. 1998; Quiñones et al. 2001; Wang et al. 1999, 2003, 2006; Persson et al. 1999; Le Marchand et al. 1998; Wu et al. 1997, 1998; El-Zein et al. 1997a, b; Oyama et al. 1997; London et al. 1996; Watanabe et al. 1995; Sugimura et al. 1995; Hamada et al. 1995; Kato et al. 1994; Huang et al. 2000; Persson et al. 1993; Kato et al. 1992; Hirvonen et al. 1992, 1993; Uematsu et al. 1991; Shi et al. 2002; Ye et al. 2006; Zou et al. 2004). The results were inconsistent or even contradictory. The reason for this disagreement may be related to gene–gene, gene–environment interactions in lung cancer carcinogenesis. Therefore, we performed a comprehensive meta-analysis by including the most recent and relevant articles to identify statistical evidence of the association between CYP2E1 RsaI (rs2031920), DraI (rs6413432) polymorphisms and the risk of lung cancer that have been investigated.

Materials and methods

Identification and eligibility of relevant studies

A bibliographical search was performed in PubMed, CNKI, and EMBASE database to identify studies that evaluated CYP2E1 polymorphisms and lung cancer up to May 10, 2014. The search terms used were: (polymorphism or mutation or variant) and (CYP2E1 or “cytochrome P-450 2E1’’ or ‘‘cytochrome P450 2E1’’) and lung. The search was not limited to language. Additional studies were identified by hand searching references in original articles and review articles. Authors were contacted directly regarding crucial data not reported in original articles. In addition, studies were identified by a manual search of the reference lists of reviews and retrieved studies. We included all the case–control studies and cohort studies that investigated the association between CYP2E1 RsaI and DraI polymorphisms and lung cancer risk with genotyping data. All eligible studies were retrieved, and their bibliographies were checked for other relevant publications.

Inclusion criteria

The included studies needed to have met the following criteria: (1) only the case–control studies or cohort studies were considered, (2) evaluated the CYP2E1 RsaI and DraI polymorphisms and lung cancer risk, and (3) the genotype distributions of the polymorphisms in cases and controls were described in detail and the results were expressed as odds ratio (OR) and corresponding 95 % confidence interval (95 % CI). Major reasons for exclusion of studies were as follows: (1) not for lung cancer research, (2) only case population, and (3) duplicate of previous publication (when the same patient population was used in several publications, only the most recent, largest or complete study was included following careful examination).

Data extraction

Information was carefully extracted from all eligible studies independently by two investigators according to the inclusion criteria listed above. The following data were collected from each study: first author’s name, year of publication, country of origin, ethnicity, source of controls, genotyping method, and numbers of cases and controls in the CYP2E1 RsaI and DraI genotypes whenever possible. Ethnicity was categorized as ‘‘Caucasian,’’ “African,” (including African Americans) and ‘‘Asian.’’ When one study did not state as to which ethnic groups were included or if it was impossible to separate participants according to phenotype, the sample was termed as ‘‘mixed population.’’ We did not define any minimum number of patients to include in this meta-analysis. For articles that reported different ethnic groups and different countries or locations, we considered them as different study samples for each category cited above.

Statistical analysis

Crude ORs together with their corresponding 95 % CIs were used to assess the strength of association between the CYP2E1 RsaI and DraI polymorphisms and lung cancer risk. The pooled ORs were performed for dominant model (RsaI: C2/C2 + C1/C2 vs. C1/C1 and DraI: CD + DD vs. CC); recessive model (RsaI: C2/C2 vs. C1/C2 + C1/C1 and DraI: DD vs. CD + CC); homozygous model (RsaI: C2/C2 vs. C1/C1 and DraI: DD vs. CC), heterozygous model (RsaI: C1/C2 vs. C1/C1 and DraI: CD vs. CC), and additive model (RsaI: C2 vs. C1 and DraI: D vs. C), respectively. Heterogeneity assumption was checked by a Chi square-based Q test (heterogeneity was considered statistically significant if P < 0.10) (Davey and Egger 1997) and quantified using the I 2 value, a value that describes the percentage of variation across studies that are due to heterogeneity rather than chance, where I 2 0 % indicates no observed heterogeneity, with 25 % regarded as low, 50 % as moderate, and 75 % as high (Higgins et al. 2003). If results were not heterogeneous, the pooled ORs were calculated by the fixed-effect model (we used the Q-statistic, which represents the magnitude of heterogeneity between-studies) (Mantel and Haenszel 1959). Otherwise, a random-effect model was used (when the heterogeneity between-studies were significant) (DerSimonian and Laird 1986). In addition to the comparison among all subjects, we also performed stratification analyses by ethnicity, source of controls, smoking status, and histological type. Moreover, the extent to which the combined risk estimate might be affected by individual studies was assessed by consecutively omitting every study from the meta-analysis (leave-one-out sensitivity analysis). This approach would also capture the effect of the oldest or first positive study (first study effect). In addition, we also ranked studies according to sample size, and then repeated this meta-analysis. Sample size was classified according to a minimum of 200 participants and those with fewer than 200 participants. The cite criteria were previously described (Klug et al. 2009). Last, sensitivity analysis was also performed, excluding studies whose allele frequencies in controls exhibited significant deviation from the Hardy–Weinberg equilibrium (HWE), given that the deviation may denote bias. Deviation of HWE may reflect methodological problems such as genotyping errors, population stratification or selection bias. HWE was calculated by using the goodness-of-fit test, and deviation was considered when P < 0.05. Begg’s funnel plots (Begg and Mazumdar 1994) and Egger’s linear regression test (Egger et al. 1997) were used to assess publication bias. A meta-regression analysis was carried out to identify the major sources of between-studies variation in the results, using the log of the ORs from each study as dependent variables, and ethnicity, sample size, HWE, and source of controls as the possible sources of heterogeneity. All of the calculations were performed using STATA version 10.0 (STATA Corporation, College Station, TX, USA).

Results

Literature search and meta-analysis databases

Relevant publications were retrieved and preliminarily screened. As shown in Fig. 1, 318 publications were identified, among which 132 irrelevant papers were excluded. Thus, 186 publications were eligible. Among these publications, 144 articles were excluded because they were review articles, case reports, and other polymorphisms of CYP2E1. In addition, of these published articles, four articles (El-Zein et al. 1997a; Hirvonen et al. 1992; Sugimura et al. 1995; Oyama et al. 2003) were excluded because of their populations overlapped with another included five articles (El-Zein et al. 1997b, Hirvonen et al. 1993; Hamada et al. 1995; Oyama et al. 1997). As summarized in Table 1, 38 publications with 50 case–control studies publications were selected in the final meta-analysis, including 5,074 cases and 6,828 controls for CYP2E1 RsaI (from 34 studies) and 2,093 cases and 2,508 controls for DraI (from 16 studies). Table 1 lists all essential information such as the publication year, first author, Country, ethnicity, source of controls, and Genotyping method for CYP2E1 RsaI and DraI, respectively. Genotype frequencies for lung cancer cases and controls are listed in Tables 2 and 3. And six studies (Klinchid et al. 2009; Li et al. 2008; El-Zein et al. 1997b; Zienolddiny et al. 2008; Eom et al. 2009; Gu et al. 2004) were analyzed only in dominant model because their provided the limited genotyping information for CYP2E1 RsaI and DraI polymorphisms. All of the cases were pathologically confirmed.

Fig. 1
figure 1

Study flow chart explaining the selection of the 38 eligible case–control studies included in the meta-analysis

Full size image
Table 1 Main characteristics of all studies included in the meta-analysis
Full size table
Table 2 Genotype distribution of CYP2E1 RsaI polymorphism used in the meta-analysis
Full size table
Table 3 Genotype distribution of CYP2E1 DraI (rs6413432) polymorphism used in the meta-analysis
Full size table

Quantitative synthesis

Table 4 lists the main results of the meta-analysis of CYP2E1 RsaI polymorphism and lung cancer risk. Significantly decreased lung cancer risk was observed (dominant model: OR 0.80, 95 % CI 0.71–0.90, P value of heterogeneity test [P h] = 0.015, I 2 = 37.8 %; heterozygous model: OR 0.80, 95 % CI 0.70–0.90, P h = 0.050, I 2 = 32.3 %; additive model: OR 0.82, 95 % CI 0.72–0.94, P h < 0.001, I 2 = 58.9 %) when all the eligible studies were pooled into the meta-analysis. However, we did not observe an association between lung cancer risk and the risk of lung cancer among recessive and homozygous models. In the subgroup analysis by ethnicity, significantly decreased lung cancer risk was observed among Asians (dominant model: OR 0.81; 95 % CI 0.71–0.93; P h = 0.003, I 2 = 50.3 %, Fig. 2; heterozygous model: OR 0.81, 95 % CI 0.69–0.95, P h = 0.012, I 2 = 46.7 %) and Caucasians (heterozygous model: OR 0.56, 95 % CI 0.32–0.98, P h = 0.441, I 2 = 0.0 %; additive model: OR 0.55, 95 % CI 0.32–0.94, P h = 0.347, I 2 = 0.0 %). There were only two studies of Africans and no significant association was found among any genetic model (Table 4). In the subgroup analysis by pathological type, significant association was found among lung adenocarcinoma (AC) (heterozygous model: OR 0.84, 95 % CI 0.71–1.00, P h = 0.129, I 2 = 36.1 %). However, no significant association was found in lung squamous cell carcinomas (SC) or non-small cell lung cancer (NSCLC). In the subgroup analysis by source of controls, significant association was observed among the population-based studies (dominant model: OR 0.69, 95 % CI 0.54–0.88, P h = 0.304, I 2 = 17.4 %; recessive model: OR 0.39, 95 % CI 0.16–0.91, P h = 0.327, I 2 = 10.6 %; additive model: OR 0.67, 95 % CI 0.53–0.84, P h = 0.677, I 2 = 0.0 %; homozygous model: OR 0.34, 95 % CI 0.14–0.80, P h = 0.445, I 2 = 0.0 %; heterozygous model: OR 0.70, 95 % CI 0.54–0.91, P h = 0.194, I 2 = 39.1 %) and hospital-based studies (dominant model: OR 0.80, 95 % CI 0.69–0.93, P h = 0.008, I 2 = 45.2 %; additive model: OR 0.84, 95 % CI 0.70–1.00, P h < 0.001, I 2 = 64.9 %; heterozygous model: OR 0.80, 95 % CI 0.68–0.95, P h = 0.025, I 2 = 40.9 %). In the subgroup analysis by smoking status, there was significant association among smokers (dominant model: OR 0.72, 95 % CI 0.55–0.94, P h = 0.374, I 2 = 7.2 %) and non-smokers (dominant model: OR 0.74, 95 % CI 0.61–0.91, P h = 0.311, I 2 = 14.7 %).

Table 4 Results of meta-analysis for CYP2E1 RsaI (rs2031920) polymorphism and the risk of lung cancer
Full size table
Fig. 2
figure 2

Forest plot of CYP2E1 RsaI polymorphism and lung cancer risk among Asians (dominant model)

Full size image

Table 5 shows the summary ORs of CYP2E1 DraI on the basis of 2,093 cases and 2,508 controls. Overall, there was no significant association between CYP2E1 DraI polymorphism and the risk of lung cancer when all the eligible studies were pooled into the meta-analysis. In the subgroup analysis by ethnicity, significantly decreased lung cancer risk was found among Asians (dominant model: OR 0.79, 95 % CI 0.66–0.95, P h = 0.319, I 2 = 14.2 %; additive model: OR 0.82, 95 % CI 0.69–0.97, P h = 0.579, I 2 = 0.0 %). In the subgroup analyses by source of controls and histological type, no significant association was observed among population-based studies, hospital-based studies, lung NSCLC, lung AC, and lung SC. In the subgroup analysis by smoking status, significant association was observed among smokers (dominant model: OR 0.49, 95 % CI 0.35–0.69, P h = 0.149, I 2 = 43.8 %).

Table 5 Results of meta-analysis for CYP2E1 DraI (rs6413432) polymorphism and the risk of lung cancer
Full size table

Test of heterogeneity and sensitivity

There was significant heterogeneity among these studies for dominant model (RsaI: P h = 0.015), recessive model (RsaI: P h = 0.041 and DraI: P h = 0.075), homozygote model (RsaI: P h < 0.001), heterozygote model (RsaI: P h = 0.050 and DraI: P h = 0.033), and additive model (RsaI: P h < 0.001 and DraI: P h = 0.079). Then, we assessed the source of heterogeneity by ethnicity and source of controls. The results of meta-regression indicated that ethnicity (dominant model: P = 0.713 for RsaI and P = 0.094 for DraI; recessive model: P = 0.161 for RsaI and P = 0.140 for DraI; additive model: P = 0.314 for RsaI and P = 0.062 for DraI; homozygote model: P = 0.161 for RsaI; heterozygote model: P = 0.637 for RsaI) and source of controls (dominant model: P = 0.752 for RsaI and P = 0.248 for DraI; recessive model: P = 0.691 for RsaI and and P = 0.115 for DraI; additive model: P = 0.982 for RsaI and P = 0.578 for DraI; homozygote model: P = 0.637 for RsaI; heterozygote model: P = 0.989 for RsaI) did not contribute to substantial heterogeneity among the meta-analysis. Although there were five studies (Zou et al. 2004; Minegishi et al. 2007; Qu et al. 1998; Wang et al. 1999; Uematsu et al. 1991) deviated from HWE for this meta-analysis, the corresponding pooled ORs were not materially altered by excluding these studies in overall and subgroup analyses. The sample size for cases and controls in all eligible studies ranged from 96 to 819, the corresponding pooled ORs were not qualitatively altered with or without the study of small sample in the overall analysis and all subgroup analyses. However, when the study of Su et al. (2011) was excluded, the results were changed in Asians for DraI (dominant model: OR 0.84, 95 % CI 0.70–1.01; additive model: OR 0.85, 95 % CI 0.72–1.02). In addition, when the study of Persson et al. (1993) was excluded, the results were changed in Caucasians for RsaI (additive model: OR 0.68, 95 % CI 0.34–1.39; dominant model: OR 0.68, 95 % CI 0.33–1.39).

Publication bias

Both Begg’s funnel plot and Egger’s test were performed to access the publication bias of this meta-analysis. Begg’s funnel plots did not reveal any evidence of obvious asymmetry in any genetic model in the overall meta-analysis (Fig. 3). The Egger’s test results also suggested no evidence of publication bias in the meta-analysis of RsaI (P = 0.325 for dominant model, P = 0.147 for recessive model, P = 0.065 for additive model, P = 0.101 for homozygote model, and P = 0.119 for heterozygote model) and DraI (P = 0.247 for dominant model, P = 0.607 for recessive model, P = 0.237 for additive model, P = 0.605 for homozygote model, and P = 0.353 for heterozygote model), respectively.

Fig. 3
figure 3

Begg’s funnel plot of publication bias between the CYP2E1 polymorphisms and lung cancer risk in dominant model (a DraI, b RsaI)

Full size image

Discussion

A number of epidemiologic studies have reported the association of CYP2E1 with lung cancer risk. However, the results remained controversial. Some original studies thought that CYP2E1 RsaI (rs2031920), DraI (rs6413432) polymorphisms were associated with lung cancer risk, but others had different opinions. Available data on the effect of CYP2E1 polymorphisms in lung cancer are scarce, especially in comparison with the bulk of studies on other genes involved in carcinogen activation/detoxification. In order to resolve this conflict, a meta-analysis was conducted to explore the association between CYP2E1 RsaI (rs2031920), DraI (rs6413432) polymorphisms and lung cancer risk.

When all the eligible studies were pooled into the meta-analysis of polymorphism, there was no significant association between CYP2E1 DraI polymorphism and the risk of lung cancer. In further stratified and sensitivity analyses, no significant association was observed in any subgroup analysis, but not smoking status. Persson et al. (1999) and Zienolddiny et al. (2008) found that CYP2E1 DraI polymorphism was not associated with lung cancer risk in Caucasians. Persson et al. (1999), Wang et al. (1999), Qu et al. (1998), Liang et al. (2004), Li et al. (2008), and Klinchid et al. (2009) found that CYP2E1 DraI polymorphism was not associated with lung cancer risk in Asians. The results of our meta-analysis supported the negative association between CYP2E1 DraI polymorphism and lung cancer risk. In the subgroup analysis by smoking status, significant association was observed among smokers (dominant model: OR 0.49, 95 % CI 0.35–0.69). However, at any case, the association between CYP2E1 DraI polymorphism and lung cancer risk among smokers essentially remains an open field, as the number of studies (n = 4) is considerably smaller than that needed for the achievement of robust conclusions (Higgins and Green 2008).

When all the eligible studies were pooled into the meta-analysis of CYP2E1 RsaI polymorphism, significantly decreased lung cancer risk in the total population (dominant model: OR 0.80, 95 %CI 0.71–0.90; heterozygote model: OR 0.80, 95 % CI 0.70–0.90; additive model: OR 0.82, 95 % CI 0.72–0.94). In further stratified and sensitivity analyses by ethnicity, significantly decreased lung cancer risk was only observed among Asians. Wang et al. (1999), Li et al. (2000), Shi et al. (2002), Wang et al. (2003), Sunaga et al. (2002), and Gu et al. (2004) found that CYP2E1 RsaI polymorphism contributed to the development of lung cancer in Asians. The results of our meta-analysis supported the positive association between CYP2E1 RsaI polymorphism and lung cancer risk. We did not observe significantly decreased lung cancer risk among Caucasians and Africans, the reason may be because only two small studies are included among Africans and three small studies are included among Caucasians in the meta-analysis. Hence, at any case, the association between CYP2E1 RsaI polymorphism and lung cancer risk among Caucasians and Africans essentially remains an open field, as the number of studies (n = 2 for Africans and n = 3 for Caucasians) is considerably smaller than that needed for the achievement of robust conclusions (Higgins and Green 2008). In the subgroup analysis by source of controls, significant association was observed among the population-based studies and hospital-based studies. The hospital-based studies have some biases because such controls may be a sample of ill-defined reference population, particularly when the genotypes under investigation were associated with the disease conditions. Hence, using a proper and representative population-based study was much important in the studies. In the subgroup analysis by smoking status, there was significant association among smokers and non-smokers. The results indicated that there could be an interaction between cigarette smoking and CYP2E1 RsaI polymorphism. It is possible that some non-smokers with the variant CYP2E1 RsaI polymorphism were susceptible to be exposed to low levels of tobacco smoke, and it is also likely that these nonsmokers may have been exposed to passive smoking. These hypotheses also need to be tested in future studies. However, only small number of studies examined the association between the CYP2E1 RsaI polymorphism and lung cancer risk in smokers or nonsmokers, hence, our meta-analysis should be interpreted with caution. It is well known that the development of squamous and small cell carcinoma is strongly associated with smoking, whereas that of adenocarcinoma is less associated compared with those two subtypes, indicating that carcinogenic processes are different among the histological subtypes of lung cancer (Sato et al. 1994). Therefore, stratified analyses were performed by histological type. In the subgroup analysis by pathological type, significantly decreased lung AC risk was observed for CYP2E1 RsaI polymorphism.

Lung cancer is a multi-factorial disease that results from complex interactions between many genetic and environmental factors. This means that there will not be single gene or single environmental factor that has large effects on lung cancer susceptibility. For lung cancer, although different results in published meta-analyses were partly explained by the different ethnic populations included in the analyses, large studies with detailed genetic and environmental exposure information are needed to evaluate reliably any moderate genetic effects. In order to control these environmental factors, some statistical methods, such as the logistic regression models, multilevel models and artificial neural networks (ANNs) could be applied in future analyses.

We noticed that two meta-analysis had been reported on the lung cancer risk with CYP2E1 polymorphisms. Wang et al. (2010) included 26 case–control studies (4,436 cases and 6,385 controls) for CYP2E1 RraI and 13 case–control studies (1,666 cases and 2,093 controls) for CYP2E1 DraI. Their meta-analysis had observed a decreased lung cancer risk among subjects carrying c1/c2 and c1/c2 + c2/c2 genotypes in the Asians and on the basis of population control in stratified analysis. Their meta-analysis also found a protective effect of the CYP2E1 DraI CC and CD + CC polymorphisms for lung cancer (OR 0.58, 95 % CI 0.41–0.81 and OR 0.84, 95 % CI 0.73–0.96, respectively). Zhan et al. (2010) included 21 case–control studies (3,984 cases and 5,496 controls) for CYP2E1 RraI. Their meta-analysis suggests that CYP2E1 RraI polymorphism was a decreased risk factor for the developing lung cancer among Asians and lung SC. However, the results of the present meta-analysis are not in accordance with those reported the previous two meta-analyses (Wang et al. 2010; Zhan et al. 2010). Our meta-analysis included more studies than previous two meta-analyses, there are 5,074 cases and 6,828 controls for CYP2E1 RsaI (from 34 studies) and 2,093 cases and 2,508 controls for DraI (from 16 studies). Our meta-analysis indicates that CYP2E1 RsaI polymorphism is associated with lung cancer risk among Asians, CYP2E1 RsaI polymorphism may be associated with lung adenocarcinoma risk, and CYP2E1 RsaI and DraI polymorphisms may be associated with decreased lung cancer risk in smokers.

There are several limitations in this meta-analysis. First, the controls were not uniformly defined. Although all the controls were healthy populations, most of them were common populations, some controls were population-based; other controls were hospital-based. Hence, non-differential misclassification bias is possible. Second, in the subgroup analysis may have had insufficient statistical power to check an association. Third, we were also unable to examine the interactions among gene–environment, lacking of the original data of the included studies limited our further evaluation of potential interactions, which may be an important component of the association between CYP2E1 RsaI (rs2031920) and DraI (rs6413432) polymorphisms and environment and lung cancer risk. Fourth, it was much difficult to get the all articles published in various languages. We only included the studies published in English and Chinese. Last, our results were based on unadjusted published estimates. Because of data limitations, we were unable to adjust them such as age, smoking, alcohol consumption, etc.

In summary, this meta-analysis indicates that CYP2E1 RsaI polymorphism is associated with lung cancer risk among Asians, CYP2E1 RsaI polymorphism may be associated with lung adenocarcinoma risk, and CYP2E1 RsaI and DraI polymorphisms may be associated with decreased lung cancer risk in smokers. However, and a study with a larger sample size is needed to further evaluate gene–environment interaction on CYP2E1 polymorphisms and lung cancer risk.