Association of genetic polymorphisms in CYP2E1, MPO, NQO1, GSTM1, and GSTT1 genes with benzene poisoning.

Metabolic enzymes involved in benzene activation or detoxification, including NAD(P)H, quinone oxidoreductase 1 (NQO1), cytochrome P450 2E1 (CYP2E1), myeloperoxidase (MPO), glutathione-S-transferase mu-1 (GSTM1), and glutathione-S-transferase theta-1 (GSTT1), were studied for their roles in human susceptibility to benzene poisoning. The potential interactions of these metabolic enzymes with lifestyle factors such as cigarette smoking and alcohol consumption were also explored. We studied 156 benzene-poisoning patients and 152 workers occupationally exposed to benzene in South China. Sequencing, denaturing HPLC, restriction fragment-length polymorphism, and polymerase chain reaction were used to detect polymorphisms on the promoters and complete coding regions of NQO1, CYP2E1, MPO, and the null genotypes of GSTM1 and GSTT1. Seventeen single nucleotide polymorphisms (SNPs) were identified in NQO1, CYP2E1, and MPO genes, including 6 novel SNPs in CYP2E1 and MPO. Of the subjects who smoked and drank alcohol, an 8.15-fold [95% confidence interval (CI), 1.43-46.50] and a 21.50-fold (95% CI, 2.79-165.79) increased risk of benzene poisoning, respectively, were observed among the subjects with two copies of NQO1 with a C-to-T substitution in cDNA at nucleotide 609 (c.609 C>T variation; i.e., NQO1 c.609 T/T) compared to those with the heterozygous or wild (NQO1 c.609 C/T and c.609 C/C) genotypes. Our data also indicated that individuals with CYP2E1 c.-1293 C/C and c.-1293 G/C, and NQO1 c.609 T/T, and GSTT1 null genotypes tended to be more susceptible to benzene toxicity. Our results suggest that the combined effect of polymorphisms in NQO1, CYP2E1, and GSTT1 genes and lifestyle factors might contribute to benzene poisoning.

Benzene is commonly used to synthesize organic chemicals and is an important component of many organic solvents. Workers exposed to benzene may potentially suffer chronic benzene poisoning (BP). Clinical reports have shown that exposure to benzene can result in a variety of blood and bone marrow disorders, including leukopenia, anemia, myelodysplastic syndrome, aplastic anemia, acute myeloid leukemia, and acute lymphocytic leukemia (Aksoy et al. 1972;Linet et al. 1996;Yin et al. 1987).
Previous studies have indicated that benzene toxicity mainly results from its intermediate reactive metabolites (Irons and Stillman 1996;Kolachana et al. 1993). Benzene is initially oxidized to benzene oxide by hepatic CYP2E1 in the liver (Koop et al. 1989;Valentine et al. 1996). Benzene oxide forms phenol spontaneously or conjugates with glutathione to form less toxic or nontoxic derivates via glutathione-S-transferases (GSTs). Phenol is catalyzed by CYP2E1 to potentially toxic di-or trihydroxybenzenes such as hydroquinone, catechol, and 1,2,4-benzentriol (Eastmond et al. 1987;Smith et al. 1989;Subrahmanyam et al. 1991). The di-or trihydroxy metabolites are further oxidized in the bone marrow by myeloperoxidase (MPO) to benzoquinones (Schattenberg et al. 1994), a potent hematotoxic and genotic agent, which can be detoxified by NAD(P)H:quinone oxidoreductase 1 (NQO1) to less harmful hydroxybenzenes (Joseph et al. 2000;Ross et al. 1996). Thus we hypothesized that the deficient or altered activity of enzymes involved in benzene metabolism such as CYP2E1, MPO, NQO1, and GSTs would significantly affect susceptibility to benzene toxicity.
Genetic polymorphisms in genes encoding CYP2E1, MPO, NQO1, and GSTs (Hirvonen et al. 1993;Piedrafita et al. 1996;Puga et al. 1997;Traver et al. 1997) might be responsible for human susceptibility to BP because they might have an effect on enzyme activity. NQO1 with a C-to-T substitution in cDNA at nucleotide 609 (NQO1 c.609C>T variation) causes reduced or lost enzyme activity (Traver et al. 1992). A case-control study revealed that workers exposed to benzene with both the higher CYP2E1 enzyme activity and the NQO1 c.609 T/T genotype had a 7.6-fold increased risk of BP (Rothman et al. 1997). A single nucleotide polymorphism (SNP) in an Alu repeat in the MPO gene promoter, c.-463G>A, could decrease the expression of MPO by destroying an SP1 transcriptional factor binding site (Piedrafita et al. 1996). It has been reported that the GSTT1 null genotype increased susceptibility to myelodysplastic syndrome [odds ratio (OR) = 4.3] (Chen et al. 1996), although confirmation of this finding is needed.
Most chronic and complex diseases are caused by interactions among environment, genes, and lifestyle (Mucci et al. 2001). We conducted a case-control study to explore the effects of polymorphisms in genes involved in benzene metabolism on human susceptibility to BP and to explore the potential effect of lifestyle modification on BP. Our results indicated that NQO1 c.609C>T variation and lifestyle contributed to the risks of BP, and the combined effect of NQO1 c.609 T/T, CYP2E1 c.-1293 C/C and c.-1293 G/C, and the GSTT1 null genotype significantly increased the risk of BP.

Materials and Methods
Subjects. The workers with BP who enrolled in this study came from Shanghai, Hangzhou, Maanshan, and Guangzhou, China, where clusters of cases were reported. Benzene poisoning was diagnosed from 1980 to 1998 by the local authorized Occupational Disease Diagnostic Team, and patients were registered in the hospitals of prevention and treatment for occupational diseases, which cooperated with us. The diagnostic criteria for occupational BP, according to the Ministry of Health, China, include a) total white blood cell count < 4,000/µL or white blood cell count between 4,000 and 4,500/µL and platelet count < 80,000/µL, with repeated confirmation of this count in a few months in a peripheral blood examination; b) the individual with documented benzene exposure has been employed for at least 6 months in the factory; and c) exclusion of other causes of abnormal blood counts such as chloromycetin use and ionizing radiation. The medical records of patients were independently reviewed, especially those with white blood cell counts > 3,500 to confirm the BP diagnosis. Of the 171 eligible patients, 156 (91%) agreed to participate in this study. About 72% of BP patients (112 patients) were from Shanghai Second Fabric Machine Factory (19 patients), Hangzhou Tool Machine Factory (13), Maanshan Iron & Steel Group (68), and Guangzhou Piano-Making Factory (12). Another 44 patients, who returned to the hospital periodically for health examinations, were from 12 other factories that had been closed down. BP cases from the factories registered in the hospitals were also clustered (more than five reports).
We chose 152 workers in the four major factories, who had been occupationally exposed to benzene, as controls. Control subjects were frequency-matched to cases by age within 5 years, exposure duration within 3 years, exposure level, and sex. All the eligible controls agreed to participate in this study.
The subjects were interviewed by trained personnel, and a questionnaire was used to obtain general information including ethnic background, nutrition, cigarette smoking, alcohol consumption, protective measures, self-reported symptoms, medical history, and occupational history such as work unit (department), type of work, and exposure duration. Exposure estimation was based on monitoring data or industrial hygienists and long-term employees' evaluation considering historical changes (Dosemeci et al. 1996). The intensity of benzene exposure (milligrams per cubic meter) for the patients was taken as the benzene level of workplaces while diagnoses were made; the intensity of benzene exposure in controls was taken as the current level monitored by organic vapor passive dosimetry badges during collection of the blood samples from controls. Those who smoked at least one cigarette per day for more than 1 year were considered regular smokers. Alcohol consumption was defined as drinking at least 7 standard units of alcohol on average per week [1 standard unit = 10 g of alcohol equivalent; e.g., a glass/can/bottle (330 mL) of regular beer (5%), a measure (40 mL) of liquor, a glass (120 mL) of wine] for more than 6 months. The subjects were administrated a rigorous physical examination at a local occupational disease hospital. Alanine aminotransferase level in serum was examined to indicate liver function.
Collection of blood samples. Blood samples of the subjects enrolled in this study were collected only after informed consent was obtained. Blood was immediately frozen at -80°after collection and was sent to the laboratory later in dry ice.
Amplification of DNA samples. We extracted genomic DNA from blood samples by a routine phenol-chloroform method. We performed polymerase chain reaction (PCR) using 50 ng of genomic DNA, 0.2 µM of each primer, 100 µM dNTPs, 20 mM Tris-Cl (pH 8.8), 10 mM KCl, 1.0-1.5 mM MgCl 2 , and 2.5 U Pfu polymerase (Stratagene, La Jolla, CA, USA) in a 25-µL reaction volume. DNA samples were amplified for 35 cycles at 94°C for 45 sec, 63°C for 1 min, and 72°C for 2 min. The primer sequences are listed in Table 1.
Detection of genetic polymorphisms. DNA sequencing was performed on an Automated DNA Sequencer ABI 377 (PE Applied Biosystems, Foster City, CA, USA) according to the manufacturer's instructions. We used the PolyPhred computer program (PolyPhred, available online) to indicate possible SNP loci. Verification of each candidate SNP was carried out by visual inspection.
The null genotype of GSTM1 and GSTT1 (Arand et al. 1996) and 96-bp insertion in the promoter of CYP2E1 (Fritsche et al. 2000) were detected by amplifying the target DNA regions and electrophoresis visualization on agarose gel. We used albumin as the internal control.
Statistical analyses. Our analysis was designed to examine the relationships of genotypes of NQO1, CYP2E1, MPO, GSTM1, and GSTT1 with the risks of BP controlling potential confounding factors and evaluate whether lifestyle factors such as cigarette smoking and alcohol consumption modified the relationships of genotypes with risks of BP. SPSS 8.0 software (SPSS Inc., Chicago, IL, USA) was used to set up a database and analyze data. We used chi-square tests to examine the association between genetic polymorphisms and individual susceptibility to benzene hematotoxicity. If the value for a cell was < 5 in the chi-square test, we applied Fisher's exact test. To evaluate whether the lifestyle modified the relationship between genetic polymorphisms and susceptibility to BP, we examined the associations by applying chi-square tests after stratification according to cigarette smoking or alcohol consumption. The test for homogeneity of ORs was examined by the Breslow-Day method. The heterogeneity of ORs indicated there could be interaction (p < 0.05). To control potential confounding factors such as sex, intensity of benzene exposure, and exposure duration, we applied unconditional logistic regression to examine the relations of genetic polymorphisms with BP. Two-tailed p-values < 0.05 were considered statistically significant. ORs and 95% confidence intervals (95% CIs) were calculated to estimate the individual risk of BP. ORs adjusted for the potential confounding factors are also reported. We also used multiple-variables unconditional logistic regression analysis to analyze the data. We applied stepwise forward logistic regression selection to screen the covariates including sex, exposure duration, genetic polymorphisms, and interactions among them (criterion for acceptance: p ≤ 0.05; criterion for removal: p ≥ 0.10). The screened covariates and intensity of benzene exposure were used to set up a saturated model.

Demographics of cases and controls.
The distribution of age, sex, exposure duration, intensity of benzene exposure, type of work, self-reported symptoms, cigarette smoking, and alcohol consumption in the cases and controls is shown in in BP cases and 62.50% vs. 37.50% in controls). The higher female ratio may also explain the relatively low frequency of cigarette smoking (17.67%) and alcohol consumption (11.59%) in the subjects. Exposure duration was highly correlated with age (Spearman rank correlation, p < 0.05). Genetic polymorphisms of NQO1, CYP2E1, MPO, GSTM1, and GSTT1. We initially screened 24 cases and 24 controls randomly for possible genetic variations in NQO1, CYP2E1, and MPO genes by direct sequencing. More than 99% of SNPs with frequencies ≥ 5% will be observed among the normal population [1 -(1 -5%) 48 individuals × 2 chromosomes = 99.27%]. We identified 17 SNPs by sequencing in the screened regions of NQO1, CYP2E1, and MPO and 6 SNPs were genotyped by DHPLC in all subjects. Two SNPs in MPO promoter, MPO c.-581T>C and MPO c.-463G>A were completely linked; thus only the MPO c.-463G>A was analyzed. The SNPs on the CYP2E1 promoter region were too complicated (10 SNPs in 1,000 bp) to be detected by DHPLC (Table 3). We only determined 1 of 10 SNPs, CYP2E1 c.-1293G>C, because the G-to-C transition forms a PstI site. Table 3 shows the allele and genotype frequencies of polymorphisms on NQO1, CYP2E1, MPO, GSTM1, and GSTT1 genes. Genotype frequencies of these genetic polymorphisms calculated from the control group were in Hardy-Weinberg equilibrium, making selection bias less likely (chi-square test, p > 0.05).

Effect of genetic polymorphisms of NQO1, MPO, CYP2E1, GSTM1, and GSTT1 on the risks of BP.
The distribution of eight independent polymorphisms of the studied genes was compared in cases and controls (Table 4). Due to their small number, some genotypes were grouped with other genotypes according to previous reported research on the function on these genes. No association of genetic polymorphisms and susceptibility to risks of BP was found between BP cases and benzene-exposed workers (p > 0.05). Although the frequency of the cases with two copies of NQO1 c.609C>T variation (NQO1 c.609 T/T genotype) was slightly higher than that of controls (25.71% vs 19.58%), there was no statistical difference between them. There was little variation in OR values when adjusted for sex, exposure duration, and intensity of benzene exposure. Among the alcohol drinkers, the frequency of BP cases with NQO1 c.609 T/T was 61.11%, which was about five times as much as that of controls (Table 6). Compared with those of NQO1 c.609 C/T and C/C genotypes, the subjects with NQO1 c.609 T/T had an 11.00-fold increase for BP (OR = 11.00; 95% CI, 1.89-63.86; Fisher's exact test, p = 0.005), and this risk increased even higher, to 21.50fold, after adjustment for sex, exposure duration, intensity of benzene exposure, and cigarette smoking (OR = 21.50;95% CI,). The frequency of the BP cases with GSTM1 null genotype among the alcohol drinkers was higher than that of the controls (66.67% vs. 33.33%), but no significant difference between them was observed in this study. There was, however, a 4.21-fold increased risk of BP for the alcohol drinkers with GSTM1 null genotype compared with those with GSTM1 non-null genotype (Table 6).

Relations of genetic polymorphisms of NQO1, MPO, CYP2E1, GSTM1, and GSTT1 with the risks of BP modified by lifestyle. The test for homogeneity (H) of ORs indicated a possible interaction between
Multiple-variables unconditional logistic regression analysis. The covariates and crossproduct terms examined in the logistic regression model included intensity of exposure, alcohol, NQO1 c.609C>T, CYP2E1 c.-1293G>C, and GSTT1 genotypes ( Table 7). The model suggested there was a joint action between alcohol consumption and NQO1 c.609 C>T variation (p = 0.007) and among NQO1 c.609 C>T, CYP2E1 c.-1293G>C, and GSTT1 null genotypes (p = 0.019). We examined the combined effects by stratification according to NQO1, CYP2E1, and GSTT1 genotypes. The result showed individuals with NQO1 c.609 T/T, CYP2E1 c.-1293 C/C or C/G, and GSTT1 null genotypes were more susceptible to BP with a 5.64fold increased risk compared with individuals carrying NQO1 c.609 C/T or C/C, CYP2E1 c.-1293 G/G, and GSTT1 non-null genotypes.

Discussion
By examining the polymorphisms of the promoter and coding regions of NQO1, CYP2E1, and MPO and the null genotype of GSTM1 and GSTT1 genes, we studied the relationship between genetic polymorphism and the human susceptibility to risks of BP. The interaction of genetic diversities of these genes with lifestyle on BP was also explored.
Though no association was suggested between genetic polymorphisms of these genes and risks of BP from our study, Rothman et al. (1997) reported a 2.6-fold increased risk of BP in the workers with NQO1 c.609 T/T genotype. There was, however, an 8.15-fold or 21.50-fold increased risk of BP in the individuals with NQO1 c.609 T/T genotype compared with NQO1 C/C or C/T genotypes after stratification by cigarette smoking and alcohol consumption, respectively (Tables 5 and 6). Cigarette smoking and alcohol consumption were considered risk factors contributing to many diseases such as lung cancer and bladder cancer. Moreover, because benzene is a component of cigarette smoke, the cumulative exposures of regular smokers should be higher than those of nonsmokers who never smoked while exposed to benzene. Our results suggested that there might be an association between NQO1 c.609 T/T genotype and the risks of BP with modification by cigarette smoking and alcohol consumption. The C-to-T point variation of this SNP, which causes a proline  Table 4. Effect of NQO1, CYP2E1, MPO, GSTT1, and GSTM1 genotypes on the risks of BP in benzeneexposed workers. to serine change, is associated with a loss activity of NQO1 (Traver et al. 1992). Moran et al. (1999) demonstrated that benzene metabolite hydroquinone induced high levels of NQO1 activity in bone marrow CD34 + cells with the wild genotype (NQO1 c.609 C/C), and noncytotoxic doses of hydroquinone induced intermediate levels of NQO1 activity in heterozygous (NQO1 c.609 C/T) cells but exerted no effect on cells with the NQO1 c.609 T/T genotype. It is possible that failure to induce functional NQO1 enzyme activity in NQO1 c.609 T/T individuals might be responsible for BP. In addition to the interaction between genetic polymorphisms and lifestyle, different genetic polymorphisms could contribute to the risk of BP. We speculated that individuals with Articles • Association of genetic polymorphisms with benzene poisoning Environmental Health Perspectives • VOLUME 110 | NUMBER 12 | December 2002 1217 Table 5. Effects of genotypes of NQO1, CYP2E1, MPO, GSTT1, and GSTM1 modified by smoking on the risks of BP in benzene-exposed workers.

Smoking
Nonsmoking  high-level activity of metabolic enzymes involved in oxidizing benzene to more toxic metabolites such as CYP2E1 and low-level activity of metabolic enzymes participated in detoxification pathway as GSTT1 and GSTM1 were less resistant to benzene toxicity. There was no detectable enzyme activity in the individuals with GSTT1 null genotypes. Using an in vitro transfection system with CAT as a reporter gene, it has been observed that the RasI polymorphism, which was completely linked with PstI in the 5´-flanking sequence of the CYP2E1 gene, caused a 10-fold increase in transcription activity compared with the wild type sequence (Hayashi et al. 1991). However, another study showed the decreased oral clearance of chlorzoxazone with RasI homozygous variants in Japanese residents of Hawaii (Marchand et al. 1999). The higher inducible enzyme activity was also observed in individuals with a 96 bp insertion (Ins 96 +/-and Ins 96 +/+) on the CYP2E1 promoter region (McCarver et al. 1998). Analysis of the combined effects of various genotype on risks of BP showed that the individuals with the NQO1 c.609 T/T, CYP2E1 c.-1293 G/C and C/C, and GSTT1 null genotype had an increased risk of BP with exposure to benzene (Table 8). The results suggest that individuals with lower NQO1 and GSTT1 activity and higher CYP2E1 activity tend to be more susceptible to benzene toxicity. Joint action between genetic polymorphisms and environment on specific diseases is complicated. A more comprehensive, larger scale study should be warranted to confirm the gene-environment interaction on susceptibility to BP. VOLUME 110 | NUMBER 12 | December 2002 • Environmental Health Perspectives  Non-null C/T and C/C 30 29 1.00 1.00 a Data missing due to inability to amplify DNA. b ORs were adjusted (adj) for potential confounding variables including sex, exposure duration, and intensity of benzene exposure. *p < 0.05.