Genetic Polymorphisms of CYP2E1 and Risk of Colorectal Adenomas in the Self Defense Forces Health Study

Cytochrome P450 2E1 (CYP2E1) is a key enzyme that is involved in metabolic activation of a variety of carcinogenic and toxic compounds including N-nitrosamines (1, 2). It is well documented that CYP2E1 is induced through high alcohol consumption and obesity (3-8). Of many known genetic polymorphisms of CYP2E1, two functional polymorphisms, the RsaI polymorphism with C-1054T substitution (rs2031920) and the 96-bp insertion in the 5′-flanking region, have drawn much interest (9-13). The CYP2E1 activity measured by chlorzoxazone clearance was progressively lower with the increasing number of the variant c2 allele of RsaI polymorphism in Japanese in Hawaii (9), although such a relationship was not replicated in another human study of Japanese and Caucasians (12). A study of alcoholics reported that CYP2E1 activity was lower in those with the c2 allele, suggesting a lower inducibility of the enzyme by alcohol in individuals homozygous and heterozygous for the variant allele (10). The variant allele of the 96-bp insertion polymorphism was shown to express greater transcriptional activity (13). The 96-bp insertion polymorphism was also found to be associated with increased induction of the enzyme by obesity and alcohol consumption (11).

Few studies have addressed the role of the RsaI polymorphism in colorectal carcinogenesis, producing inconsistent findings. Individuals with RsaI c2 allele tended to have decreased risks of colon and rectal cancers in Hawaii (14) but a statistically significant increase in the risk of colorectal cancer in Hungary (15). No measurable association was found between RsaI polymorphism and the risk of colorectal cancer in Australia (16) and The Netherlands (17). The RsaI c2 allele is rather rare in Caucasians, and results from studies of Caucasians may have been subject to chance. The case-control study in Hawaii also examined the relation of 96-bp insertion polymorphism to colorectal cancer and showed an increased risk of colorectal cancer, especially of rectal cancer in those having at least one 96-bp insertion (14). We investigated the relation of these genetic polymorphisms to colorectal adenomas, a well-established precursor lesion of colorectal cancer (18, 19), in the Self Defense Forces Health Study, with particular attention to location and size of adenoma. We also examined effect modification of alcohol use and body mass index on the association between the CYP2E1 polymorphisms and colorectal adenomas, because induction of CYP2E1 by alcohol consumption and obesity may differ by these polymorphisms (10, 11).

Study subjects were male officials in the Self Defense Forces who received a preretirement health examination at the Self Defense Forces Fukuoka or Kumamoto Hospital from January 1997 to March 2001. The preretirement health examination is a nationwide program offering a comprehensive medical examination for those retiring from the Self Defense Forces. Details of the preretirement health examination have been described elsewhere (20, 21). Routine medical examinations included colonoscopy and blood biochemical measurements during a 5-day admission. In addition to blood samples for routine use in the health examination, a sample of 7 mL fasting venous blood was obtained for the purpose of medical research in general, not specifically for genetic study, with written informed consent. The study was approved by the Ethical Committee of Kyushu University, and genotyping was carried out after individuals were made completely anonymous.

The present study included 461 cases of histologically confirmed colorectal adenomas and 1,067 controls with no polyp who underwent total colonoscopy. In a consecutive series of 2,459 men, 5 men refused to participate in the survey and 77 men did not receive colonoscopy. Furthermore, 242 men were excluded for having a prior history of colectomy (n = 17), colorectal polypectomy (n = 212), malignant neoplasms (n = 27), or inflammatory bowel disease (n = 1); some of the men had two or more reasons for exclusion. In the remaining 2,135 men, colonoscopic findings were classified as colorectal cancer (n = 1), polyp (n = 938), nonpolyp benign lesion such as diverticula (n = 123), and normal (n = 1073). The cases comprised 461 men of 938 with colorectal polyps who were found to have adenomas without in situ or invasive carcinoma, and controls were 1,067 men of the 1,196 with normal or nonpolyp benign lesions who underwent total colonoscopy. DNA sample was not available with 21 men (6 cases and 15 controls). The number of remaining cases was 455, and controls numbered 1,052.

A self-administered questionnaire was used to ascertain smoking habits, alcohol use, leisure-time physical activity, and other lifestyle factors before colonoscopy. Details have been described previously (20, 21). In brief, smokers were defined as those who had ever smoked cigarettes daily for at least 1 year. Lifetime exposure to cigarette smoking was expressed as cigarette-years, the average number of cigarettes smoked per day multiplied by the total years of smoking, and classified into four levels of 0, 1 to 399, 400 to 799, and ≥800 cigarette-years. Alcohol drinkers were defined as those having drunk alcoholic beverages at least once a week for ≥1 year, and former alcohol use was separated from lifetime nonuse of alcohol. Daily ethanol intake was estimated for current drinkers based on consumption frequencies and amounts of five types of alcoholic beverages (sake, shochu, beer, whisky/brandy, and wine) on average in the past year. Alcohol use was categorized into never, former, and current use with a consumption of <30, 30 to 59, or ≥60 mL ethanol/d. Body mass index (kg/m²) was calculated and categorized into four levels of <22.5, 22.5 to 24.9, 25.0 to 27.4, and ≥ 27.5 kg/m². Questions on leisure-time physical activity were slightly changed in April 1999. In the earlier version, subjects were first asked about the frequency of regular participation in exercise and sport during leisure time on average in the past year using a closed-ended question (none, 1-2, 3-4, 5-6 times per week and daily). If the subjects participated in recreational physical activity at least once a week, they reported type of activity and time spent per occasion regarding at most three types of regular exercise. In the revised questionnaire, the subjects were first asked whether they participated in recreational activity regularly (one or more times a week) in the past year. Those with a regular participation reported at most three types of physical activities together with frequency per week and time spent per occasion for each activity. Type of physical activity was classified into light, moderate, heavy, or very heavy activity in terms of metabolic equivalent. The time spent in recreational exercise was multiplied by the corresponding metabolic equivalent value (light 2, moderate 4, heavy 6, and very heavy 8) to yield a metabolic equivalent-hour score a week. Individuals were classified into four groups with the quartiles in the control group as cutoff points. Parental history of colorectal cancer was also elicited.

DNA was extracted from buffy coat stored at -80°C by using a commercial kit (Qiagen). The following procedures of genotyping used 0.5 μL template DNA with a concentration of ∼50 to 150 ng/μL. Genotyping for the CYP2E1 RsaI polymorphism was determined by the PCR-RFLP method described by Le Marchand et al. (14) using primers 5′-CCAGTCGAGTCTACATTGTCA-3′ (forward) and 5′-TTCATTCTGTCTTCTAACTGG-3′ (reverse). The PCR was done in a reaction mixture of 10 μL containing 0.5 units Ex Taq (Takara Bio). After the initial denaturation at 94°C for 4 min, 30 cycles of PCR were carried out for 1 min at 94°C, for 1 min at 55°C, and for 1 min at 72°C and final annealing and extension at 55°C for 1 min and 72°C for 7 min. The PCR product of 413 bp was digested with 5 units RsaI in a reaction mixture of 20 μL for 3 h at 37°C. The digestion resulted in fragments of 352 and 61 bp for the c1 allele. The digested fragments were electrophoresed on a 3% MetaPhor agarose (Cambrex Bio Science) gel and visualized using ethidium bromide.

The 96-bp insertion genotype was determined by the PCR method, as described previously by Fritsche et al. (22), using primers 5′-GTGATGGAAGCCTGAAGAACA-3′ (forward) and 5′-CTTTGGTGGGGTGAGAACAG-3′ (reverse). The PCR was done in a reaction mixture of 20 μL containing 0.5 units Ex Taq (Takara Bio). After the initial denaturation at 94°C for 5 min, 25 cycles of PCR were carried out for 30 s at 94°C, for 30 s at 66°C, and for 1 min at 72°C. The fragments were electrophoresed on a 2% Ultra pure agarose (Invitrogen) gel and visualized using ethidium bromide or SYBR Gold stain (Molecular Probes). The PCR product of the 96-bp insertion allele is 729 bp in length and the noninsertion allele is 633 bp in length.

Genotyping of the two polymorphisms was done by one of the authors (M.M.). Photographs of the gel electrophoresis were kept and also inspected by another author (G.Y.). Genotyping was routinely repeated when there was any uncertainty in determining a genotype. Regarding RsaI polymorphism, genotyping was repeated when the band for an undigested fragment was less thick than the band for a digested fragment in the case of possible heterozygotes. Each run of experiments always included the c1/c1 genotype with both alleles digested by the restriction enzyme. After the completion of genotyping in the whole study subjects, a repeat sample of 90 subjects were randomly selected for each polymorphism with 30 subjects for each genotype. Genotypes in the repeat sample were in a complete agreement with those in the main study regarding the two polymorphisms.

The association of CYP2E1 polymorphisms with colorectal adenomas was examined in terms of odds ratio (OR) and 95% confidence interval (95% CI), which were obtained from logistic regression or polytomous logistic regression. The 95% CI was derived from the SE for the logistic regression coefficient. Statistical adjustment was made for age (years), hospital, Self Defense Forces rank (three classes), body mass index, cigarette smoking, alcohol consumption, parental colorectal cancer, and physical activity. Polytomous logistic regression was applied to the analysis by location or size of adenoma, and the Wald statistics was used to test for the difference (heterogeneity) in estimated regression coefficients between subtypes of adenomas. Interactions of the CYP2E1 polymorphisms with alcohol consumption and body mass index were evaluated by the likelihood ratio test.

Statistical significance was declared if two-sided P < 0.05 or if the 95% CI did not include unity. The SAS/Statistics version 8.2 was used (SAS Institute). Polytomous logistic regression analysis was done using Stata Statistical Software release 8.0 (Stata).

The age ranges were 50 to 57 years in the cases and 47 to 59 years in the controls, but the mean ages (52 years for both cases and controls) were identical in the two groups. Regarding the 96-bp insertion polymorphism, genotypes were undetermined for three controls. Whereas RsaI c1/c2 and c2/c2 genotypes were slightly less frequent in adenoma cases than in controls, genotype frequencies of 96-bp insertion polymorphism did not differ between cases and controls (Table 1). Frequencies of the variant allele of RsaI c2 were 0.201 in cases and 0.225 in controls, and those of the 96-bp insertion allele in cases and controls were 0.246 and 0.233, respectively. The distributions of the CYP2E1 RsaI and 96-bp insertion genotypes were each in agreement with the Hardy-Weinberg equilibrium in both cases and controls (RsaI: P = 0.91 for cases and 0.58 for controls; 96-bp insertion: P = 0.69 for cases and 0.49 for controls).

OR for the c1/c2 and c2/c2 genotypes of RsaI polymorphism, compared with the c1/c1 genotype, were lower than unity, and the decrease in the adjusted OR for c1/c2 and c2/c2 combined was nearly significant (P = 0.07). No measurable association was observed for the CYP2E1 96-bp insertion polymorphism and colorectal adenomas (Table 1).

Cases of adenomas with sizes of <5 and ≥5 mm (the largest size for multiple adenomas) numbered 265 and 190, respectively. Numbers of cases having adenomas at the proximal colon alone and the distal colon and/or rectum alone were 149 and 239, respectively; rectal adenomas and distal colon adenomas were combined because cases with rectal adenomas alone were few (n = 42). When the association with the two polymorphisms was examined for adenomas of the proximal colon and of the distal segment by polytomous logistic regression (Table 2), a statistically significant decrease in the adjusted OR of proximal colon adenomas, but not of distal adenomas, was observed among individuals with the RsaI c2 allele. The heterogeneity in the OR for c1/c2 and c2/c2 genotypes combined between the two types of adenomas was also statistically significant. The 96-bp insertion polymorphism was not associated with either proximal or distal adenomas. The OR of large adenomas, not of small adenomas, showed a statistically significant increase among individuals with at least one 96-bp insertion (Table 3). Again, the difference in the OR between large and small adenomas was statistically significant.

We further explored whether a decreased OR of proximal adenomas associated with RsaI c2 allele differed by size of adenoma and whether an increased OR of large adenomas in individuals with 96-bp insertion allele differed by location of adenoma. No appreciable heterogeneity was noted in either analysis. Adjusted OR (95% CI) associated with RsaI c2 allele for proximal large and small adenomas were 0.55 (0.31-0.99) and 0.65 (0.40-1.05), respectively (P_{heterogeneity} = 0.68). Adjusted OR (95% CI) associated with 96-bp insertion allele were 1.25 (0.73-2.13) for large proximal adenomas and 1.67 (1.05-2.65) for large distal adenomas (P_{heterogeneity} = 0.40).

Alcohol use and overweight were each related to colorectal adenomas. In the multivariate models without variables for genotypes, OR (95% CI) for nonalcohol drinkers, former drinkers, and drinkers with a consumption of <30, 30 to 59, or >60 mL alcohol/d were 1.0 (reference), 1.1 (0.5-2.2), 0.9 (0.6-1.3), 1.5 (1.0-2.2), and 1.6 (1.1-2.3), respectively. OR (95% CI) for body mass index of <22.5, 22.5 to 24.9, 25.0 to 27.4, and ≥ 27.5 kg/m² were 1.0 (reference), 1.2 (0.9-1.6), 1.5 (1.1-2.1), and 1.9 (1.3-3.0), respectively. In the analysis of effect modification of these factors, dichotomous variables were defined for high alcohol use (≥30 mL ethanol/d) and overweight (body mass index ≥ 25.0 kg/m²). Individuals homozygous for the variant allele of each polymorphism were always combined with those heterozygous for the variant allele.

The interaction between RsaI genotype and alcohol intake was not statistically significant regarding either all adenomas combined or subtypes of adenomas with respect to site and size (Table 4). However, OR associated with RsaI c2 allele were generally lower than unity in those with no or light consumption of alcohol (<30 mL/d), with statistically significant decreases for proximal adenomas and large adenomas as well as for all adenomas. The overall heterogeneity with respect to the OR for the combination of RsaI polymorphism and alcohol intake was nearly statistically significant between proximal and distal adenomas (P_{heterogeneity} = 0.053), and the OR of proximal and distal adenomas in the low category of alcohol intake also differed with the statistical significance (P = 0.03). No such heterogeneity was noted for large and small adenomas. There was an almost statistically significant effect modification of alcohol consumption on the association between 96-bp polymorphism and large adenomas (P_interaction = 0.054). Compared with individuals with null/light alcohol consumption and null insertion allele, those with 96-bp insertion showed a 2-fold increase in the OR of large adenomas in the absence of high alcohol consumption (≥30 mL/d), whereas similar increases in the OR were noted for those with high alcohol consumption regardless of the genotypes (Table 5). The overall heterogeneity was statistically significant between large and small adenomas, and the difference in the OR between the two in the low category of alcohol consumption was statistically highly significant (P = 0.002). Overweight did not show any measurable effect modification on the association with either RsaI or 96-bp insertion genotype (data not shown), except for large adenomas and the insertion polymorphism. A statistically significant increase in the OR of large adenomas associated with the 96-bp insertion allele was seen only in individuals with overweight; adjusted OR (95% CI) was 2.03 (1.29-3.21) compared with the combination of nonoverweight and the noninsertion genotype (P_interaction = 0.13).

The present study was the first that examined the relation between CYP2E1 RsaI and 96-bp insertion polymorphisms and colorectal adenomas and showed a statistically significant decrease in the risk of proximal colon adenomas associated with the RsaI c2 allele and also an evident increase in the risk of large adenomas in individuals harboring the 96-bp insertion allele. The latter finding is consistent with the recent observation in Hawaii that individuals with the 96-bp insertion allele was associated with 1.6-fold increased risk of rectal cancer (14). In that study, OR (95% CI) of colon and rectal cancer for individuals with the RsaI c2 allele versus those without were 0.8 (0.6-1.1) and 0.8 (0.6-1.3), respectively (14). A few other studies have also examined the relation between RsaI polymorphism and colorectal cancer (15-17), but none of the previous studies have addressed the relation between RsaI polymorphism and proximal colon cancer specifically.

The analysis by location of adenoma is of interest because different molecular alterations have been implicated in carcinogenesis of the proximal and distal sites of the colorectum. Genetic alterations such as K-ras and p53 mutations are more frequent in the distal site, whereas microsatellite instability was almost exclusively associated with the proximal colon cancer (23). Of particular interest with respect to the role of CYP2E1 is the observation that levels of the promutagenic lesion O⁶-methyldeoxyguanosine, a marker of exposure to N-nitroso compounds, were higher in normal mucosa of the distal colon than of the proximal colon (24). Large adenomas are at higher risk of developing carcinoma and are at advanced stage in the adenoma-carcinoma sequence (18, 19). In this regard, a differential increase in the risk of large adenomas associated with the 96-bp insertion polymorphism is a notable finding. On the other hand, a decreased risk of proximal adenomas associated with RsaI c2 allele is rather difficult to interpret. Site-specific associations observed differently in relation to the two polymorphisms may have been ascribed to chance because the number of cases was much smaller in the analysis by subsite.

A decreased risk of colorectal adenomas associated with the RsaI c2 allele was observed when alcohol consumption was null or low (<30 mL/d). The 96-bp insertion polymorphism was related to a statistically significant increase in the risk of large adenomas, again, in the absence of high alcohol consumption. Increased risks of large adenomas in individuals with high alcohol consumption did not differ by the 96-bp insertion polymorphism. Genetic effect on CYP2E1 inducibility may be exerted more evidently in individuals with low alcohol consumption. High alcohol consumption may also enhance the growth of colorectal adenomas via mechanisms other than the activation of carcinogens by CYP2E1. Bacterial conversion of ethanol to acetaldehyde in the colonic lumen and alcohol-related depletion of folate are alternative mechanisms by which alcohol may enhance colorectal carcinogenesis (25).

The present findings on the 96-bp insertion polymorphism are compatible with in vitro and in vivo experimental findings with respect to transcriptional activity and inducibility (11, 13). Evidence regarding the functional role of RsaI polymorphism is inconsistent, however. The RsaI variant c2 allele was shown to be associated with lowered activity of the enzyme in one human study (9) but not in other human studies (10, 12). One of the latter studies, however, suggested lower alcohol-related induction of the enzyme in individuals carrying the variant c2 allele (10). There are also studies showing no clear difference in the enzyme activity of human liver tissue according to RsaI genotypes (26, 27). On the other hand, in vitro studies have suggested an enhanced transcription of the CYP2E1 gene associated with the variant c2 allele (28, 29). CYP2E1 expression itself is regulated at multiple levels of transcription, mRNA stabilization, translation, and protein stabilization (1, 3). Interestingly, CYP2E1 mRNA expression was not correlated with protein expression or functional activity in human liver tissue (26). Although CYP2E1 is predominantly expressed in the liver especially after alcohol consumption (1, 30), the enzyme is also detected in colon mucosa (31) and induced in the colon of rodents administered with alcohol (30). It would be mechanistically important whether CYP2E1 is expressed sufficiently in colorectal mucosa. Because diazonium ion derived from N-nitrosamines is highly reactive for DNA alkylation locally, it is unlikely that the reactive compound produced in the liver is transported to other organs in significant amounts (32).

The role of CYP2E1 RsaI polymorphism has been investigated for various cancers other than colorectal cancer. Several case-control studies found a decreased risk of lung cancer associated with RsaI c2 allele in a multiethnic population in Hawaii (33), Swedish (34), and Chinese (35), but others found no association in Caucasians and African Americans (36) and in Japanese (37). Likewise, results from 13 studies of stomach cancer were found to be quite variable in a meta-analysis (38), although the investigators reported an increased risk in individuals with c2 allele based on three studies of high quality in Asian populations (38). As for esophageal cancer, some studies in China reported an inverse association with c2 allele, whereas Japanese studies found no association (39). Inconsistency is also noted for the relation between RsaI polymorphism and hepatocellular carcinoma (40, 41). Few studies have examined the relation between CYP2E1 96-bp insertion polymorphism and cancer risk. An increased risk of esophageal cancer, but not of lung cancer, was found in individuals who had two variant 96-bp insertion alleles in a small Japanese study (42).

Both RsaI c2 allele and 96-bp insertion allele are fairly common in Asian populations compared with Caucasians. In control subjects of the present study, RsaI c2 and 96-bp insertion alleles were present at frequencies of 22% and 23%, respectively. These frequencies are almost the same as reported for Japanese elsewhere. Frequencies of the RsaI c2 allele were 23% in Japanese, 4% in Caucasians, and 15% in Hawaiians in the United States (14). The 96-bp insertion allele accounted for 15% in Taiwanese, 10% in African Americans, and 2% in Caucasians (22). The frequency of the 96-bp insertion allele was reported to be 23% in Japanese in Hawaii (14).

The present study had methodologic advantages in that colonoscopy was done almost nonselectively in a defined population and that the absence of polyp lesions was confirmed in the control subjects by total colonoscopy. There were also several limitations. We excluded a fairly large number of men with a prior history of colorectal polypectomy. A total number of such men was 218 including those who did not receive colonoscopy. If they had been more susceptible to colorectal adenomas with respect to the CYP2E1 polymorphisms, the present results may have been biased. Parental colorectal cancer was slightly more frequent in the adenoma cases (4.6%) than in the controls (3.6%), with an adjusted OR (95% CI) of 1.35 (0.77-2.37). Of men with a history of colorectal polypectomy, 6.4% had a history of parental colorectal cancer; this proportion did not much differ from that of adenomas cases (P = 0.32). Genotypes of the CYP2E1 polymorphisms were also determined for men with a history of polypectomy (RsaI, n = 214; 96-bp insertion, n = 213). RsaI c1/c1, c1/c2, and c2/c2 genotypes numbered 127 (59.3%), 76 (35.5%), and 11 (5.1%), respectively. Numbers of men with null, one, and two alleles of 96-bp insertion were 129 (60.6%), 75 (35.2%), and 9 (4.2%), respectively. These distributions were similar to those observed for adenoma cases (P = 0.50 for RsaI and 0.52 for 96-bp insertion). We did not take into account dietary factors because validated data were not available for specific foods and nutrients. The interaction with red meat would be of particular interest. High intake of red meat was shown to increase endogenous production of N-nitroso compounds in the intestine (43), and increased risk of rectal cancer associated with the 96-bp insertion allele was more evident in those with high intakes of red med and processed meat in Hawaii (14). The study subjects were not representative of Japanese men in the general population, but selection was unlikely to exist with regard to the genetic polymorphisms under study. As noted above, frequencies of both RsaI c2 allele and 96-bp insertion allele were similar to those reported previously for Japanese (14).

In summary, a case-control study of Japanese men showed an evident increase in the risk of large adenomas of the colorectum in individuals having the CYP2E1 96-bp insertion allele and a moderately decreased risk of proximal colon adenomas associated with the CYP2E1 RsaI c2 allele. A suggestive effect modification of alcohol was noted for the association with these polymorphisms. These findings indicate that variation in activity and inducibility of CYP2E1 contribute to the development of colorectal carcinogenesis and add to evidence that substances activated via CYP2E1 are involved in colorectal carcinogenesis.

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.

We thank the ward nurses at the Self Defense Forces Fukuoka and Kumamoto Hospitals for supportive work.