No Association between Cytochrome P450 and Glutathione S-Transferase Gene Polymorphisms and Risk of Colorectal Adenoma: Results from the UK Flexible Sigmoidoscopy Screening Trial

Genetic variation in carcinogen metabolizing enzymes has been proposed as a susceptibility marker for colorectal neoplasia. The cytochrome P450 (CYP) and glutathione S-transferase (GST) enzymes metabolize several classes of carcinogen in the human diet and tobacco smoke. Epidemiologic studies that have sought to investigate the relation between variants in CYP and GST genes and colorectal neoplasia have thus far yielded conflicting results and a consensus regarding their etiologic importance has yet to be reached (1). In light of this, we conducted a case-control study, nested within a large randomized controlled trial, to determine whether functionally characterized variants of the CYP1A1, CYP2E1, GSTM1, GSTT1, and GSTM3 genes are associated with risk of colorectal adenoma. In addition, we investigated the interaction between specific dietary components, smoking, genotype, and colorectal adenoma risk.

The study included 1,899 Caucasian individuals (1,267 males and 632 females), ages 55 to 64 years, who had undergone screening for polyps in the distal colorectum as part of the UK Flexible Sigmoidoscopy Screening Trial (2). The UK Flexible Sigmoidoscopy Screening Trial is a randomized controlled trial of 368,583 participants from 14 geographic regions, designed to test the efficacy of a once-only flexible sigmoidoscopy in the prevention of colorectal cancer. Individuals were invited to participate via their general practitioner. Of the 40,674 screened, 131 colorectal cancers were detected and excluded from the analysis. Other exclusion criteria included a history of colorectal cancer, adenoma, or inflammatory bowel disease; a severe or terminal disease with life expectancy of <5 years; and a sigmoidoscopy or colonoscopy within the past 2 years or incapability of providing informed consent. In the study presented here, cases were individuals with histologically confirmed adenoma of the distal bowel from three of the study centers (Leeds, Norwich, and Portsmouth). All adenoma cases were asked to provide a blood specimen, of which 94% agreed. Controls were age- and sex-matched individuals with a negative flexible sigmoidoscopy result (no adenomatous or hyperplastic polyps). Overall, blood samples were available for 918 cases and 981 controls.

Before screening, participants completed a questionnaire regarding how often 26 selected food items were eaten. The dietary items analyzed in this study comprise components known to interact with the cytochrome P450 and GST enzyme systems and included subtypes of cruciferous vegetables, different types of red and processed meat, and usual meat cooking methods. Participants were assigned as never smokers, former smokers, and current smokers.

Polymorphisms were selected on the basis of whether they lead to functional changes in the translated protein, their prevalence in Caucasian populations, and any previous association with colorectal neoplasia (Table 1). The methods used to discriminate the GSTM1, GSTT1, GSTM3, and CYP1A1 alleles have been described elsewhere (3, 4). The CYP2E1*5B allele was distinguished by amplification of a 480-bp fragment that carries two linked variants: a C(−1091)T transition recognized by PstI and a G(−1259)C transversion recognized by RsaI. The CYP2E1*5B allele is determined by a PstI cut, RsaI noncut. Primer sequences were 5′-ACTGGAAAGGAAAGAGAGGAG-3′ (sense) and 5′-CATTCTGTCTTCTAACTGGCA-3′ (antisense).

Differences in genotype distributions between cases and controls were ascertained by the χ² statistic. Risks were calculated as odds ratios with 95% confidence intervals by unconditional logistic regression and were adjusted for age, sex, and sigmoidoscopy center. To test for modification of the association between the dietary variables, smoking, and adenoma risk by genotype, a stratified analysis was conducted by genotype. Potential two-way interactions between genotype and the dietary variables were assessed by comparing models with and without the interaction term using the likelihood ratio test. Interactions were considered to be statistically significant at the 1% level.

None of the genotype distributions in the controls differed significantly from those expected under Hardy-Weinberg equilibrium and all were in the range reported previously for Caucasians (ref. 5; Table 2). Carriage of the CYP1A1*2C allele was inversely associated with adenoma risk (odds ratio, 0.7; 95% confidence interval, 0.5-0.9). We did not detect any association with alleles of the other genes and colorectal adenoma risk. Subgroup analyses revealed no effect of gender on genotypic risks. Risk estimates did not differ according to genotype and we found no evidence for multiplicative interaction between diet and smoking, genotype, and adenoma (data not shown). Overall results for diet and smoking from the UK Flexible Sigmoidoscopy Screening Trial study will be published elsewhere.

This is the largest study, to date, to examine the interaction of diet, smoking, and metabolic gene polymorphisms and colorectal adenoma risk. We detected an inverse association between carriage of the CYP1A1*2C allele and risk of colorectal adenoma but no association with the other genotypes. We found no evidence for an interaction between diet, smoking, and genotype in adenoma risk.

The finding that the CYP1A1*2C allele was inversely related to colorectal adenoma was unexpected and conflicts with previous studies where a positive association or no association was identified (6-12). Given the large number of genotypes examined in this study, this finding may have arisen by chance.

One of the strengths of this study is a control group known to be free of distal lesions. However, some degree of misclassification may have occurred as adenomas in the proximal colon cannot be detected during sigmoidoscopy.

Although this is the largest study of colorectal adenoma and CYP/GST polymorphisms to date, statistical power was compromised in several analyses; we had ∼50% power to detect the 30% reduction in risk associated with the CYP1A1*2C-carrier genotype. Larger studies are necessary to confirm the modest reduction in colorectal adenoma risk associated with this allele. However, the low prevalence of the CYP1A1*2C allele raises questions as to the overall impact this variant may have on colorectal neoplasia risk in the Caucasian population.

In conclusion, we report an overall lack of association between common variants of these xenobiotic metabolism genes and colorectal adenoma risk. Given the bipartite mechanism of carcinogen metabolism, studies should be done with adequate statistical power to assess such variants in a combinatorial manner.

We thank Pauline Rogers at Cancer Research UK for assistance with data management and for statistical advice.