CYP1A1, GSTM1, and GSTT1 Polymorphisms, Smoking, and Lung Cancer Risk in a Pooled Analysis among Asian Populations

Lung cancer mortality has increased rapidly during recent years in Asian countries. Cigarette smoking is the strongest established risk factor for lung cancer, but genetically determined variations in metabolism of tobacco-derived carcinogens may affect individual susceptibility to lung cancer. Cigarette smoke contains a variety of carcinogens, such as polycyclic aromatic hydrocarbons, N-nitrosoamines, and aromatic heterocyclic amines (1). These carcinogens undergo biotransformation by several enzymatic pathways, including P450s (CYP), glutathione S-transferase (GST), and N-acetyltransferase.

CYP1A1 plays an important role in the metabolism of polycyclic aromatic hydrocarbons, including benzo(a)pyrene, as a phase I enzyme and two variants (i.e., Ile⁴⁶²Val and T⁶²³⁵C), which are potentially functional (2-4), have been evaluated as susceptibility factors for lung cancer by a number of investigators. An increased risk of lung cancer has been observed with the ⁶²³⁵C variant among smokers (5) and with ⁴⁶²Val among nonsmokers (6) in previous pooled analyses using the Genetic Susceptibility to Environmental Carcinogenesis (GSEC) database, whereas a separate meta-analysis did not find a significant association with lung cancer risk (7).

GSTM1 catalyzes reactive electrophilic intermediates derived from cigarette smoking, such as benzo(a)pyrene-7,8-diol-9,10-epoxides (BPDE), to less reactive and more easily excreted glutathione conjugates (8). Deletion of GSTM1 has most widely been evaluated for the association with lung cancer risk and a significant association was found in several studies. Although three meta-analyses concluded that the GSTM1-null genotype is associated with an increased lung cancer risk (9-11), a GSEC pooled analysis indicated that there is no strong evidence for increased risk of lung cancer among those with the GSTM1-null genotype (12). Another isoform of GST (GSTT1) is also involved in carcinogen detoxification and its deletion polymorphism has been suggested to be associated with lung cancer in several studies. In a recent GSEC pooled analysis, the association was not significant for either Asians or Caucasians and no interaction was observed between GSTT1-null genotype and smoking on lung cancer (13).

Pooled analyses based on the GSEC data suggest that the effects of these variants tend to differ according to ethnicity possibly because of differences in linkage disequilibrium and environmental exposures. Consequently, gene-environment or gene-gene interactions might differ by ethnic group. Thus, we focused on Asian populations and evaluated the potential role of four selected polymorphisms in the three aforementioned genes (CYP1A1 Ile⁴⁶²Val and T⁶²³⁵C, and null genotypes for GSTM1 and GSTT1) in the development of lung cancer and its specific cell types.

Subjects were recruited from the International Collaborative Study on GSEC. The design of this collaborative project is explained in detail elsewhere (14). We obtained the original data of 15 case-control studies on genetic polymorphisms in CYP1A1, GSTM1, or GSTT1 and risk of lung cancer conducted in Asian populations (15-30). Two studies were excluded due to a sample size of <10 subjects (29) or Caucasian ethnicity (Turkish; Table 1; ref. 30). The participation in GSEC was voluntary, and therefore, some relevant studies were not included in our analysis. The number of subjects included in this pooled analysis was 1,971 cases and 2,130 controls.

All statistical procedures were conducted using Statistical Analysis System version 9.1.3 (SAS Institute) unless otherwise indicated. We estimated the study-specific odds ratios (OR) of lung cancer for each polymorphism using unconditional logistic regression. Results might vary slightly from those reported for some of the published studies because of differences in the inclusion criteria of cases and controls and in the statistical analyses. Heterogeneity among the studies was evaluated by means of the Cochrane Q test and publication bias was assessed by Begg's and Egger's test using STATA version 9. In the pooled analysis, lung cancer risk was estimated with the ORs and 95% confidence intervals (95% CI) by unconditional logistic regression, adjusting for age, sex, and pack-year.

In addition to conducting analyses of all lung cancer, we calculated cell type–specific ORs for the three most prevalent histologic subtypes of lung cancer: adenocarcinoma (n = 905), squamous cell carcinoma (n = 542), and small cell carcinoma (n = 181). Subgroup analyses for other histologic subtypes were not conducted due to small numbers of cases.

Hardy-Weinberg equilibrium for each single nucleotide polymorphism of CYP1A1 was tested among controls with a Pearson χ² and linkage disequilibrium was assessed with D′ and r². Individual haplotypes for two CYP1A1 polymorphisms (Ile⁴⁶²Val and T⁶²³⁵C) were estimated by expectation-maximization method and the overall difference in haplotype frequency profiles between cases and controls was assessed using the likelihood ratio test. The subjects missing both polymorphisms were excluded in haplotype analysis. The program uses a weighting scheme based on expectation-maximization–derived haplotype frequency estimates. Thus, every haplotype is weighted by the probability of carrying each pair of haplotypes rather than assigning a most likely haplotype to an individual. Missing genotypes result in more low-probability haplotype pairs and each haplotype is weighted as such. An unconditional logistic regression model was used to estimate the effect of individual haplotypes by fitting an additive model, adjusting for sex, age, and pack-year.

Gene-smoking interactions (i.e., the modification of increasing pattern of lung cancer risk as the pack-year increases by different genotype) were evaluated by the significance of the coefficient of product term genotype*pack-year in the model. The test was equal to evaluate the difference of the slopes of two fitted lines stratified by categorized genotypes. Additionally, we tested the significance of the product term in the model without main effect term of genotype, which assumes that if there is no exposure to cigarette smoking, there is no difference in the risk of lung cancer between genotypes (27, 31). The assumption of no genotype effect when there is no smoking exposure was equal to common intercept assumption for two fitted lines by genotypes.

The distributions by age, sex, smoking status, and cell types of the 1,971 lung cancer cases and 2,130 controls are presented in Table 2. The mean age was 62.6 (±10.7 years) in cases and 58.4 (±13.2 years) in controls (P = 0.0001). The proportion of ever smokers was much greater in cases (75.1%) than in controls (61.7%; P = 0.0001). In terms of cell types, adenocarcinoma (50.2%) and squamous cell carcinoma (30.1%) were the most common.

Genotype frequencies of CYP1A1 Ile⁴⁶²Val and T⁶²³⁵C were consistent with Hardy-Weinberg equilibrium in the control group (P > 0.35) and the two polymorphisms were in moderate linkage disequilibrium (D′ = 0.86 and r² = 0.35). The variant allele frequencies of the three polymorphisms (CYP1A1 ⁴⁶²Val, 0.25; ⁶²³⁵C, 0.42; and GSTT1 null, 0.48) in the controls were higher compared with those of Caucasian or African populations (13, 32). The frequency of the GSTM1 null (0.56) was similar to that of Caucasians but higher compared with Africans (32). The CYP1A1 ⁶²³⁵C variant was associated with squamous cell lung cancer (TC versus TT: OR, 1.42; 95% CI, 0.96-2.09; CC versus TT: OR, 1.97; 95% CI, 1.26-3.07; P_trend = 0.003; Table 3). The CYP1A1 ⁴⁶²Val variant was moderately associated with adenocarcinoma (Val/Val versus Ile/Ile or Ile/Val: OR, 1.57; 95% CI, 0.96-2.59).

In haplotype analysis, ⁴⁶²Val-⁶²³⁵T and Ile-C haplotypes were associated with lung cancer risk with reference to the Ile-T haplotype (OR, 3.41; 95% CI, 1.78-6.53 and OR, 1.39; 95% CI, 1.12-1.71, respectively). An omnibus test showed that the distribution of the CYP1A1 haplotypes was significantly different between all lung cancer cases and controls (P = 0.0001). In subgroup analysis, the difference was also significant for adenocarcinoma (P = 0.0003) and squamous cell carcinoma (P = 0.0001) and not for small cell carcinoma (P = 0.40).

The GSTM1-null genotype significantly increased squamous cell lung cancer risk (OR, 1.36; 95% CI, 1.05-1.77), and the GSTT1-null genotype was moderately associated only with small cell lung cancer risk (OR, 1.36; 95% CI, 0.99-1.86; Table 4). Analysis of combined genotypes did not reveal associations beyond what was apparent in the single polymorphism analyses (data not shown).

When the interaction was evaluated with smoking, increasing trend of lung cancer risk as pack-year increased was much stronger among those with the CYP1A1 6235 TC/CC genotype compared with those with TT genotype (P_interaction = 0.001; Fig. 1). Although the association between smoking and lung cancer was stronger among those with the GSTM1-null genotype compared with the present type, it was only marginally significant with the assumption of no genotype effect in the absence of the smoking exposure (P_interaction = 0.08). Significant interactive effect with smoking has not been observed for GSTT1.

There was no evidence of significant heterogeneity among studies or of publication bias for all four polymorphisms investigated in our study; we found only moderate heterogeneity for the effect of CYP1A1 ⁴⁶²Val/Val compared with Ile/Ile (P = 0.08), and all Begg's and Egger's tests were not significant (P ≥ 0.2 and 0.3, respectively).

Our results suggest that the CYP1A1 polymorphisms (Ile⁴⁶²Val and T⁶²³⁵C) and the GSTM1-null genotype are associated with lung cancer risk, especially for squamous cell carcinoma, in Asian populations. In addition, the association of smoking with lung cancer was significantly modified by the CYP1A1 T⁶²³⁵C polymorphism in our study.

A significant interactive effect between the CYP1A1 ⁶²³⁵C allele and smoking is consistent with the results of previous pooled analysis that the stronger association between the ⁶²³⁵C allele and lung cancer was found among ever smokers (5). The previous pooled analysis for the GSTM1-null genotype conducted by Benhamou et al. (12) found a nonsignificant elevated lung cancer risk among Asians, especially among heavy smoker (>40 pack-years). Likewise, our extended analysis with additional Asian populations also observed a moderate elevation of overall lung cancer risk by the GSTM1 deletion and moderate interaction with smoking. On the other hand, stronger effect of CYP1A1 ⁴⁶²Val found in previous pooled analysis among nonsmokers (6) was not observed in Asian populations investigated in our study.

Le Marchand et al. (19) hypothesized that genetic susceptibility to polycyclic aromatic hydrocarbons (based on high-risk genotypes for CYP1A1 and GSTM1) predominantly causes squamous cell carcinoma. In the multiethnic study conducted by Le Marchand et al. (19), CYP1A1 ⁶²³⁵C allele was associated with a 3.1-fold risk of squamous cell carcinoma when combined with a GSTM1 deletion. Decreasing trend of squamous cell carcinoma, relative to the increase in adenocarcinoma, associated with filter-tipped cigarettes in developed country indirectly supports this hypothesis (33). The increased risk of squamous cell carcinoma in relation with the GSTM1-null genotype observed in our study is consistent with the results of previous studies, including those of a meta-analysis (10, 19, 34, 35). The effect of the CYP1A1 ⁶²³⁵C allele, especially when combined with a GSTM1-null genotype, also tended to be associated with a higher risk of squamous cell carcinoma among Asians (5); in our study, CYP1A1 TC or CC genotype was associated with significant elevation of squamous cell carcinoma risk compared with TT genotype (OR, 1.6) and Ile-C haplotype was significantly associated with squamous cell carcinoma risk (OR, 2.1).

BPDE is known to induce G:C to T:A transversion mutations in the hotspot codons of the p53 tumor suppressor gene (36), which is found more frequent in squamous cell carcinoma than in adenocarcinoma (37). Cigarette smoke is also known to be causally related to BPDE-DNA adducts (38, 39), and BPDE-DNA adduct level is elevated in the lung parenchyma of smokers with GSTM1-null genotype (40). Moreover, the combined genotypes of CYP1A1 ⁴⁶²Val and GSTM1 null have been associated with increased adduct level in lung tissues of squamous cell carcinoma patients (41). Thus, it is speculated that our finding of an association between GSTM1 and CYP1A1 polymorphisms with the risk of squamous cell carcinoma is related to polycyclic aromatic hydrocarbon exposure derived from smoking because polycyclic aromatic hydrocarbons are primarily metabolized by CYP1A1 and GSTM1. The greater effects observed among smokers also support this smoking-related etiology of squamous cell carcinoma in Asian population.

Our study is the largest pooled analysis conducted for Asian populations to evaluate the role of polymorphisms in carcinogen-metabolizing genes (i.e., CYP1A1, GSTM1, and GSTT1) in lung cancer development. We simultaneously evaluated the potential effect of four polymorphisms on lung cancer and the modification of those effects by smoking exposure reporting significant interaction between CYP1A1 ⁶²³⁵C allele and smoking. Subtype-specific results in Asian population are also noteworthy.

However, our study has several limitations to be considered. First, not all published Asian studies were included in this study. However, there was no evidence of significant publication bias for this pooled analysis. In terms of heterogeneity, only marginally significant heterogeneity was found for CYP1A1 ⁴⁶²Val/Val compared with Ile/Ile (P = 0.08). We note that when the adjusted values were considered, the heterogeneity did not remain. Other limitation of our study may be the relatively small sample size in subgroup analyses. We found that the GSTT1-null genotype was marginally associated only with small cell lung cancer risk, whereas no association with lung cancer was observed for either Asians or Caucasians in the previous pooled analysis for GSTT1-null genotype (13). Although relatively higher variant allele frequencies, compared with other ethnic groups (13, 32), may compensate for the relative small sample size in terms of statistical power, we cannot exclude chance for the explanation of the significant association between the GSTT1-null genotype and small cell lung cancer risk, considering that only 71 cases were available. Sizable exclusion of subjects for missing data on smoking and pathologic subtypes also limits the conclusion from our results for interactive effects between the polymorphisms and smoking, and subtype-specific analysis. Thus, our findings need to be replicated in a larger study. Future study should also include the measurement of dietary factors, such as isothiocyanates, which are involved in the detoxification of tobacco-related carcinogens (42) and may have protective effects on lung cancer especially among smokers or those with GST-null genotypes, as observed in a Chinese population (24, 43).

In summary, the results of our study suggest that genetic polymorphism in CYP1A1 and GSTM1 plays a role in lung cancer susceptibility in Asian populations and that the effects are strongest for squamous cell carcinoma. Although our results are generally consistent with previous studies and are supported by epidemiologic and experimental observations, additional large studies are needed to help to elucidate the role of genetic polymorphisms in xenobiotic-metabolizing genes in lung cancer development. The interaction between environmental exposure other than smoking (e.g., indoor coal combustion) and these polymorphisms still remains to be evaluated.

No potential conflicts of interest were disclosed.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.