Meat Intake, Heterocyclic Amine Exposure, and Metabolizing Enzyme Polymorphisms in Relation to Colorectal Polyp Risk

Meat consumption, particularly red and processed meat consumption, has been linked to the increased risk of colorectal cancer in many previous epidemiologic studies (1). Mutagens, such as heterocyclic amines (HCA) and polycyclic aromatic hydrocarbons (PAH), are formed during high-temperature cooking of meats (2). These compounds are mutagenic in Ames/Salmonella assays and cause colon tumors in laboratory animals (3). To exert their mutagenic action, HCAs require enzyme-catalyzed activation consisting of N-oxidation by hepatic cytochrome P450 1A2 (CYP1A2) and other extrahepatic P450 isozymes followed by O-acetylation by N-acetyltransferase 1 (NAT1) and N-acetyltransferase 2 (NAT2; ref. 4). The aryl hydrocarbon receptor (AhR or dioxin receptor) is an important mediator for xenobiotic signaling to enhance the expression of phase I and II enzymes (5), which affects HCA metabolism. HCAs bind strongly to DNA after activation, and detectable levels of 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP)-DNA adducts in the colon mucosa of rats and humans have been found (3, 6). Rapid NAT1 or NAT2 acetylators, in conjunction with a preference for well-done meat, have been shown to be associated with colorectal cancer risk in some but not in other studies (7, 8).

Colorectal adenomatous polyps are well-established precursors of colorectal cancer (9). The association of red meat intake with the risk (10-12) and recurrence (13, 14) of colorectal adenomas has been evaluated in several previous studies with inconsistent results. Studies on the effects of genetic polymorphisms in the CYP1A2, NAT1, and NAT2 genes and their combined effects on colorectal adenomas with well-done meat or HCA intake and these genetic polymorphisms in these genes have also been inconsistent (15-18). Hyperplastic polyps, formerly considered innocuous polyps, may have malignant potential through pathways other than the adenoma-carcinoma sequence and may be a precursor for microsatellite-instable (MSI) colorectal cancers (19). To our knowledge, only one study has evaluated the modifying effect of genes on the environmental factors in risk of hyperplastic polyps (20). In this study, we evaluated genetic polymorphisms of genes involved in HCA metabolism (CYP1A2, NAT1, NAT2, and AhR) and possible interactive effects of these genes with dietary intake of meat and HCAs in the risk of colorectal adenomatous and hyperplastic polyps in a large-scale colonoscopy-based study.

The Tennessee Colorectal Polyp Study (TCPS) is an ongoing colonoscopy-based study being conducted in Nashville, TN. Eligible participants, ages between 40 and 75 years, were identified from patients scheduled for colonoscopy at the Vanderbilt Gastroenterology Clinic and the Veteran's Affairs Tennessee Valley Health System Nashville campus between February 1, 2003 and December 31, 2005. Excluded from the study were participants who had genetic colorectal cancer syndromes (e.g., hereditary nonpolyposis colorectal cancer or familial adenomatous polyposis), prior history of inflammatory bowel disease, adenomatous polyps, or any cancer other than nonmelanoma skin cancer. The study was approved by the relevant committees for the use of human participants in research. Of these 4,617 eligible participants, 3,083 provided written informed consent (67%), and the majority of them (n = 2,752; 89.3%) were recruited either before or on the day of scheduled colonoscopy. A subset of participants (n = 331; 10.7%) were recruited after the colonoscopy largely because they were not timely identified to allow time for recruitment before the colonoscopy. Among the 3,083 participants, 2,678 (87%) completed a telephone interview and 2,355 (76%) completed a self-administered food frequency questionnaire (FFQ) developed for a similar population in the southern United States. Details for the FFQ used in this study have been described elsewhere (21). The median time between the colonoscopy and the completion of the survey was 13 days, and interviewers were blinded to the results of the colonoscopy. Of those who completed the telephone interviews, 98% of cases and 98% of controls donated a blood or buccal cell sample. For all participants recruited at the time of colonoscopy, blood or buccal cells were collected before the start of the colonoscopy. For participants recruited after colonoscopy, buccal cells were collected by the participant at home using a mouthwash kit and sent through the mail to the study laboratory for processing. These samples were processed on the same day, typically within 4 h of sample collection, and were stored at -80°C until relevant bioassays could be conducted.

Colonoscopic procedures were done and reported using standard clinical protocols by the patient's gastroenterologist. Likewise, all pathology diagnoses were determined by hospital pathologists and reported as part of routine care. Data were abstracted from these colonoscopy and pathology reports to classify study participants into one of the following study groups: adenomatous polyps only, hyperplastic polyps only, presence of both adenomatous and hyperplastic polyps, and polyp-free patients. Polyp-free patients are used as the controls for this study. To be diagnosed as a polyp-free control, the participant had to have a complete colonoscopy reaching the cecum and be polyp-free at colonoscopy.

Details for the meat consumption questionnaire and meat mutagen calculation methods have been described previously (22). In brief, a telephone interview was conducted to obtain information on medication use, demographics, medical history, and lifestyle, including questions on usual intake frequency, portion size, and preparation of 11 meats (hamburgers or cheeseburgers from fast food, hamburgers or cheeseburgers not from fast food, beef steaks, pork chops or ham steaks, bacon, sausage, hotdogs or franks, chicken, fish, meat gravies made with drippings, and short ribs or spareribs). For each meat item, we also asked how often the participant ate the meat according to different cooking methods; that is, oven broiled or oven baked, grilled or barbecued, pan fried, deep fried (for chicken and fish), and all other ways, for example, boiled, steamed, or microwaved.

Participants reported their usual preference level of meat doneness by using a series of three-color photographs representing increasing levels of doneness for each of the seven meat items evaluated in the study (hamburger patties, beef steaks, pork chops, bacon, grilled chicken, pan fried chicken, and pan fried or grilled fish). Each photograph was labeled with a number and represented a range of doneness from medium, well done, to very well done (hamburger patties and beef steaks) or from rare/just done, well done, to very well done (all other items). Participants were asked when they usually ate each food item, whether the item usually looked less done than photograph one, about the same as one, two, or three, or more than three. Participants were asked to have the food picture booklet, which was typically given to the participant in-person on the day of colonoscopy or in front of them during the telephone interview. At the beginning of the telephone interview, the trained interviewers verified that the participant had the food picture booklet in hand. In case that a participant had misplaced the booklet, that portion of the interview would be rescheduled for another time when a replacement booklet was received, generally within a few days.

Using the software application known as the Computerized Heterocyclic Amines Resource for Research in Epidemiology of Disease (2), we estimated the exposure level of 2-amino-3,8-dimethylimidazo[4,5-f]quinoxaline (MeIQx), 2-amino-3,4,8-trimethylimidazo[4,5-f]quinoxaline (DiMeIQx), PhIP, benzo[a]pyrene (BaP), and mutagenic activity for each study participant.

Genotyping assays were processed using genomic DNA extracted from blood or buccal cells. The allelic discrimination of the NAT2, CYP1A2, and AhR gene polymorphisms were assessed with the ABI PRISM 7900 Sequence Detection Systems (Applied Biosystems) using TaqMan assay. The TaqMan Assay-on-Demand reagents are available for NAT2 G590A (rs1799930), G857A (rs1799931), T341C (rs1801280), CYP1A2 A-164C (rs762551), and AhR Arg⁵⁵⁴Lys (rs2066853) from ABI. The reagents for NAT2 G191A (rs1801279) polymorphisms were obtained from ABI's Assay-by-Design service (primers were GGAGTTGGGCTTAGAGGCTATTTT and CAGAAGTTGATTGACCTGGAGACA; probes were VIC-CCACCCCGGTTTC and FAM-CCCACCCTGGTTTC). The final volume for each reaction was 5 μL, consisting of 2.5 μL TaqMan Universal PCR Master Mix, 0.25 μL primers/TaqMan probes mix, and 5 ng genomic DNA. The PCR profile consisted of an initial denaturation step at 95°C for 10 min and 40 cycles of 92°C for 15 s and 60°C for 1 min. The fluorescence level was measured with the ABI PRISM 7900HT sequence detector. Allele frequencies were determined by ABI SDS software. The laboratory staff was blinded to the identity of the participants. Quality-control samples were included in the genotyping assays. Each 384-well plate contained 4 water blanks, 8 CEPH 1347-02 DNA, and 16 blinded quality-control samples. The blinded quality-control samples were taken from the second tube of study samples included in the study. Quality-control samples were distributed across separate 384-well plates. The concordance rate for the blinded quality-control samples was 100% for all of these single nucleotide polymorphisms (SNP). In addition, DNA of 45 Caucasian samples that were used in the HapMap and/or Perlegen projects was purchased from Coriell Cell Repositories (http://locus.umdnj.edu/ccr/) and genotyped for all five SNPs. The average consistency rate of the five SNPs was 99.3% compared with data from HapMap (http://www.hapmap.org) and Perlegen (http://genome.perlegen.com).

Allelic discrimination assays for the NAT1 gene also used TaqMan primers and fluorogenic probes as described previously (23). The assay identifies a nucleotide at eight polymorphisms in the NAT1 coding region or 3′ untranslated region: C97T (R33Stop), C190T (R64W), G445A (V149I), C559T (R187Stop), G560A (R187Q), A752T (D251V), T1088A (3′ untranslated region), and C1095A (3′ untranslated region).

NAT2 phenotypes were imputed by standard methods based on the functional activities of each allele (24). In addition to the SNPs examined in this study, there are other NAT2 SNPs that were not checked. This has the potential to lead to genotype misclassification. However, the G590A, G857A, T341C, or G191A SNPs are signature SNPs of haplotype clusters associated with the slow acetylator phenotype (24). Thus, the following phenotype imputations were made based on codominant expression of rapid and slow acetylator NAT2 alleles: slow acetylators were either homozygous for any of the four SNPs or heterozygous for two or more of the four SNPs, intermediate acetylators were heterozygous for one of the four SNPs, and rapid acetylators were homozygous for the common nucleotide at each of the four SNPs.

Unlike NAT2, there is no clear consensus for imputing NAT1 phenotypes. However, in comparison with NAT1 *3 and *4, the NAT1 *14, *15, *17, *19, and *22 alleles have been associated with slow acetylation (25). We additionally hypothesized that the *10 and *11 alleles are rapid acetylation alleles based on previous literature (25, 26). However, these alleles are probably only slightly rapid, do not compensate for the slow alleles, and may only be rapid if in combination with or heterozygous with a NAT1 *3 and *4 allele. Therefore, we considered any participants possessing slow alleles (NAT1 *14, *15, *17, *19, and *22) to be a slow acetylator, any participants homozygous for rapid alleles (NAT1 *10 and *11) or in combination with a NAT1 *3 and *4 allele to be a rapid acetylator, and all others as intermediate acetylators.

Further details for classification of NAT1 and NAT2 genotypes and imputed phenotypes are provided in Appendices 1 and 2.

Generalized linear models and Mantel-Haenszel χ² tests were used to compare the distribution of demographic characteristics and risk factors for colorectal polyps among case and control groups with adjustment for age and sex differences. Polytomous logistic regression models were used to estimate odds ratios (OR) and 95% confidence intervals (95% CI) for the association between exposures and types of polyp by comparing multiple case groups with one common control group. ORs were adjusted for known risk factors for colorectal polyps and variables that showed different distributions among comparison groups [age, sex, study site, education, indication for colonoscopy, smoking, alcohol consumption, regularly exercised in the past 10 years, regular use of nonsteroidal anti-inflammatory drugs (NSAID), and total energy intake]. Total energy intake was derived using data from the FFQ. For those who did not provide FFQ information, energy intake level was imputed with age-specific (40-49, 50-59, 60-64, and ≥65 years) and sex-specific mean values. Stratified analyses for meat and HCA intake by genotype were done. ORs were estimated based on per decile increment of meat, HCA and BaP intake, and mutagenic activity stratified by genotype. The interactions between dietary variables (in decile increment) and genotypes were evaluated using the log-likelihood ratio test by comparing the model with and without the interaction terms. All analyses were carried out using SAS software (version 9.1).

For the current analysis, participants were limited to those with a confirmed diagnosis of adenomatous or hyperplastic polyp or clean colon. Excluded from the analysis were 7 participants diagnosed with colorectal cancer and 40 participants with a diagnosis other than polyps, an ambiguous colonoscopic procedure, or an unclear pathology evaluation. Also excluded were 91 participants with missing data in any of the meat consumption and doneness questions and one with missing genotyping information in any of the four genes. These exclusions were not mutually exclusive, and some participants met two or more the exclusion criteria. The final number of participants included in the analysis was 557 cases with adenomatous polyps only, 250 cases with hyperplastic polyps only, 195 cases with both adenomatous and hyperplastic polyps, and 1,493 polyp-free controls.

The distribution of demographic and other characteristics for the three case groups and polyp-free controls are presented in Table 1. Compared with controls, patients in any of the case groups were more likely to be male, to be a smoker, to have lower educational attainment, and to have a lower household income. Additionally, cases with adenomatous polyps were older than controls, whereas cases with hyperplastic polyps only were similar in age to controls.

Table 2 provides median and interquartile ranges for intake levels of total meat, red meat, HCAs, BaP, mutagenic activity, and total energy intake. In general, cases, particularly cases with hyperplastic polyps and cases with concurrent adenomatous and hyperplastic polyps, had a higher intake of meat and meat carcinogens, such as PhIP, MeIQx, and DiMeIQx, than polyp-free controls.

Shown in Table 3 are the distributions of AhR (rs2066853), CYP1A2 (rs762551), NAT2, and NAT1 genetic polymorphisms among colorectal polyp cases and polyp-free controls and their associations with polyp risk. With the exception of the NAT1 genotype, no other genetic polymorphisms showed a statistically significant association with colorectal polyps. Rapid acetylator NAT1 genotypes were associated with an elevated risk of adenomas (OR, 1.8; 95% CI, 1.0-3.2) compared with participants with slow acetylator genotypes. Participants with NAT1 intermediate or rapid acetylator genotypes also showed a nonsignificantly increased risk for concurrent adenomatous and hyperplastic polyps in comparison with NAT1 slow acetylators.

Table 4 presents associations of meat and meat mutagen intake with polyp risk stratified by genotypes. Due to a small number of participants with a NAT2 rapid acetylating status, this group of participants was combined with the NAT2 intermediate acetylator group. Similarly, NAT1 slow acetylators were combined with NAT1 intermediate acetylators to enhance statistical power in interaction analyses. Positive associations with intake of meat and meat mutagens were observed primarily for the case group with concurrent adenomatous and hyperplastic polyps and predominantly among participants with the GA/AA genotypes in AhR (rs2066853) or NAT1 rapid acetylator status. With the exception of the BaP, all other six exposure variables showed a significant dose-response relationship with the risk of concurrent adenomatous and hyperplastic polyps in participants with an A allele in AhR (rs2066853) or who were NAT1 rapid acetylators. On the other hand, none of the seven exposure variables evaluated in the study showed any significant dose-response relationship with concurrent adenomatous and hyperplastic polyps among participants with AhR GG genotype or who were NAT1 slow or intermediate acetylators. Statistical tests were significant for several gene-diet interactions evaluated in this study. For example, the OR (95% CI) for a decile increment in total meat intake was 1.32 (1.14-1.52) for participants carrying the A allele of the AhR gene and only 0.98 (0.91-1.05) for participants homozygous for the G allele (P for interaction = 0.002). Several positive interactions were also noted for NAT2 genotypes. However, the pattern of the association was less consistent than that observed for AhR and NAT1 genotypes.

Because a significant modifying effect was mainly observed for the AhR and NAT1 genes, we combined the risk genotypes for these two genes (Table 5). Participants who had risk genotypes for both genes (AhR GA/AA and NAT1 rapid acetylator) had the highest risk for concurrent adenomatous and hyperplastic polyps if they also consumed a high amount of total meat (OR, 1.57; 95% CI, 1.23-2.01; P for interaction = 0.019) or red meat (OR, 1.60; 95% CI, 1.23-2.09; P for interaction = 0.023). They were also at an increased risk if they had high intake of MeIQx, PhIP, or DiMeIQx (P for interaction = 0.048, 0.055, and 0.115, respectively).

Our study evaluated not only colorectal adenomas but also hyperplastic polyps in relation to well-done meat intake and genetic polymorphisms in carcinogen-metabolizing enzymes. We found that a high intake of meat and HCAs was associated with an increased risk of polyps, primarily for the status of concurrent adenomatous and hyperplastic polyps among individuals who carried high-activity genotypes for the NAT1 or NAT2 genes or the AhR A allele (rs2066853). However, we did not observe any consistent modifying effects of these genotypes on the association of meat and HCA exposure with the risk of polyps among patients who had only adenomas.

The NAT2 or NAT1 genotypes were not found to be directly associated with colorectal cancer risk in previous studies (7). However, high red meat or MeIQx intake in combination with rapid acetylator genotypes was related to increased risk for colorectal cancer in several studies (8, 17, 19, 27-29). The functional significance for specific NAT2 polymorphisms has been well established through measurement of catalytic activity for the N-acetylation of a sulfonamide drug and an aromatic amine carcinogen (30). NAT1 appears to be expressed in many extrahepatic tissues, including the colon (31). HCAs, which are produced during high-temperature cooking of meats, are activated by O-acetylation (32). The role of N-acetyltransferase as a determinant of colorectal cancer risk has been supported by animal studies. Purewal et al. reported a 2-fold higher number of aberrant crypt foci and higher colon PhIP-DNA adduct levels in rapid acetylator rats compared with slow acetylator rats when they were fed 0.04% PhIP (33). In addition, Nerurkar et al. also showed that rapid acetylator mice had 3-fold more DNA adducts formed by the food carcinogen 2-amino-3-methyl-imidazo[4,5-f]quinoline in the colon mucosa than slow acetylator mice (34). O-acetyltransferase activity has been observed in colon tissues, and colon tissues obtained from cancer patients have been shown to possess high levels of O-acetyltransferase, suggesting an etiologic association between acetylation phenotype and colorectal carcinoma (6). Our results, which suggest that a high intake of meat and HCAs increase the risk for polyps among participants with high activity NAT1 or NAT2 genotypes, are consistent with a hypothesis based on the metabolic pathway of HCAs.

CYP1A2 catalyzes the initial activation of HCA to produce N-hydroxy-HCA for O-acetylation by NAT1 and NAT2 (4). The high activity CYP1A2 phenotype has shown a nonsignificant increased risk for colorectal adenoma (15) or cancer (8, 35). Although CYP1A2 is highly inducible by environmental and dietary factors, such as smoking and HCAs, CYP1A2 activity is under strong genetic control (36). Correlations between CYP1A2 genotypes and phenotypes, which are often assessed using the caffeine urinary metabolite ratio, have been inconsistent. The CYP1A2 *1F (rs762551) polymorphism, a C > A substitution in intron 1 at position 734, has been associated with CYP1A2 activity in several studies (37, 38). However, it remains unclear which of the A and C alleles are associated with a high activity of this enzyme (37-39), and one study reported no difference in enzyme activity by genotypes (40). The CYP1A2 *1F polymorphism was not related to colorectal cancer risk in one study (35), but another study suggested an increased risk for colorectal cancer among those with the C allele (41). Elevated risk for colorectal cancer was observed among participants who had high intake of well-done meat, who smoked cigarettes, or who carried the NAT2 genotype or the rapid CYP1A2 phenotype (8). Yet, no interactive effect of MeIQx and the CYP1A2 phenotype was observed for colorectal adenoma risk (15). We did not find any significant modifying effect of the CYP1A2 (rs762551) polymorphism on association between meat, HCA intake, or colorectal polyp risk despite the suggested functional significance of this polymorphism on HCA metabolism.

AhR is a key regulator of transcriptional expression for the CYP1 family of genes (5). The Arg⁵⁵⁴Lys polymorphism (rs2066853) is located in exon 10 of this gene, a region associated with transactivity of other genes (42). No study has evaluated the AhR gene polymorphism for association with colorectal cancer risk. A recent report, which determined the level of DNA damage in peripheral blood lymphocytes using the alkaline comet assay, observed a higher level of DNA damage in participants with a variant rs2066853 genotype than those with the GG genotype among coke-oven workers (43). This result is consistent with our finding that participants with an A allele were more vulnerable to environmental carcinogen exposure.

Few studies have compared the risk factors associated with adenomatous and hyperplastic polyps (44, 45). Only one study has investigated the association of metabolizing gene polymorphisms and their interactions with meat intake in the risk adenomatous and hyperplastic polyps and found no evidence for an interaction (20). Hyperplastic polyps are commonly detected in the large intestine and have generally not been regarded as having malignant potential (19). However, recent studies have suggested that some hyperplastic polyps may develop into cancer via pathways other than the adenoma-carcinoma sequence (19). Ten percent to 15% of colorectal cancers show MSI and do not share common molecular pathways with the adenoma-carcinoma sequence (19). This group of colorectal cancers is characterized by defective nucleotide mismatch repair and aberrant DNA methylation. It has been reported that individuals with colorectal cancers with MSI are more likely to harbor serrated or hyperplastic polyps than individuals with cancers without MSI (46). Epidemiologic studies have reported that the risk factor profiles for MSI tumors, and tumor without MSI are different (47-50). Patients with MSI colon cancers had significantly higher dietary exposure to HCAs than patients with microsatellite-stable colon cancers, and the risk of MSI colon cancer was increased 3-fold among patients who preferred to eat very well done red meat (50). In contrast, Diergaarde et al. found that red meat consumption was higher among patients with a microsatellite stable tumor than patients with a MSI tumor (47). Slattery et al. did not find any association between red meat intake and MSI status of colon cancer (49), whereas Lüchtenborg et al. found mixed results dependent on the type of meat (48). The reasons for the inconsistent finding are unclear, and additional research is needed.

In our study, most of the statistically significant interactions between carcinogen-metabolizing genotypes and meat carcinogen exposures were observed for the status of concurrent hyperplastic and adenomatous polyps. Our results are consistent with a previous study of Goode et al. (20), in which the increased risk associated with meat intake was most prominent among patients with both types of polyps than patients with only one type of polyps. The reasons for these findings are not clear. It is possible that HCA exposures may induce multiple premalignant loci in the colon, which is characterized by the presence of both adenomatous and hyperplastic polyps. It is also possible that individuals with certain characteristics may be more likely to develop both adenomatous and hyperplastic polyps in the colon after HCA exposure than participants without these characteristics. Sessile serrated adenomas have attracted considerable attention recently, given their potential to develop invasive MSI colorectal cancer. These adenomas show a mixed hyperplastic and adenomatous morphology (51). It is possible that HCA exposure may be more closely related to the risk of sessile serrated adenomas than other polyps. Additional research is needed in the future to reveal potential mechanisms for the findings observed in our study.

Strengths of our study include a large sample size, a detailed exposure assessment, and the use of a colonoscopy-based approach to classify study groups. Dietary surveys, however, were conducted after the colonoscopy; thus, similar to other studies, recall bias could be a concern. However, because the diagnosis of polyps cannot influence the genotype of study participants and the genotype is unlikely to be related to meat intake and meat carcinogen exposure, out study resembles closely the case scenario of “Mendelian randomization,” a concept that has attracted considerable attention in studying gene-environment interaction (52). It is believed that observational studies that meet Mendelian randomization criteria are akin to randomized control trials, in which potentials for recall bias and confounding are minimized. In other words, our results regarding gene-environment interactions are unlikely to be explained by information bias and confounding effects (51). Multiple comparisons involving the evaluation of several meat and meat mutagen intakes could increase type I errors. However, the patterns of the association between different meats, HCA items, and colorectal polyp risks were consistent across genotype strata; therefore, the effect of chance related to multiple comparisons can not entirely explain our findings. Because only one polymorphism for the AhR and CYP1A2 genes was evaluated, a more comprehensive evaluation of the genetic information could provide further information on the modifying effect of these genes on meat and HCA intake. There was a suggestive main gene effect of NAT1 gene on adenomatous polyp risk; however, the number of participants with NAT1 slow acetylating status was small, affecting the stability of risk estimate. Because participants from the Veteran's Affairs clinic are mainly veterans, the difference in sex distribution is inevitable. However, we adjusted the analyses by study site as well as sex. Relatively low response rates for the FFQ could introduce a selection bias; however, the results excluding participants without FFQ showed very similar associations with the results from all participants who responded to the meat questions.

In summary, our results suggest that genetic polymorphisms involved in metabolic activation of HCAs may modify the risk of colorectal polyps. Our study provides strong evidence for a role of gene-diet interaction in the etiology of colorectal tumors.

We thank Ming Li, Huiyun Wu, and Heidi Chen for contributions to data management, Hongmei Wu and Regina Courtney for help with the genotyping assay, Brandy Sue Venuti for technical assistance in the preparation of this article, the research staff and study participants for contribution to the study, and the patient advocates for support.