Effect of NAT1 and NAT2 Genetic Polymorphisms on Colorectal Cancer Risk Associated with Exposure to Tobacco Smoke and Meat Consumption

Only a small percentage of colorectal cancers are attributable to inherited genetic syndromes whereas the majority of cases are thought to be caused by multiple factors including environmental exposures and mild genetic predisposition. Both cigarette smoking and consumption of red meat have been associated with an increased risk of colorectal cancer in several previous studies (reviewed in refs. 1, 2). Cigarette smoke and red meat cooked by high-temperature cooking techniques contain a plethora of potential carcinogens such as heterocyclic aromatic amines, aromatic amines, and polycyclic aromatic hydrocarbons (3, 4). These procarcinogens can be activated in vivo to reactive metabolites that may readily bind to DNA and thus induce DNA damage, which may subsequently lead to tumor initiation.

The phase II metabolizing enzymes N-acetyltransferases 1 and 2 (NAT1 and NAT2), both being expressed in colorectal tissue (5, 6), catalyze N- and O-acetylation of several xenobiotics, such as heterocyclic aromatic amines and aromatic amines, which may entail detoxification or activation of the substrate (7). Both genes, NAT1 and NAT2, exhibit several genetic polymorphisms with >25 different alleles known to date (http://www.louisville.edu/medschool/pharmacology). The NAT1*10 allele is the most common variant allele among Caucasians and, although not consistently, has been associated with increased enzyme activity in vitro and in vivo (8, 9). In European populations, ∼60% of individuals have genotypes associated with NAT2 slow acetylation (10). Hence, genetically determined differences in acetylation capacity may alter exposure to carcinogens and thus influence individual susceptibility to environmental exposures.

Overall, the epidemiologic evidence provides little support for a main effect of NAT1 or NAT2 genotype on colorectal cancer risk (10, 11). Previous studies that investigated a modification of risk of colorectal neoplasms associated with cigarette smoking by NAT2 acetylator status yielded inconsistent results (12-22). However, some of the previous studies were based on a small number of cases (12, 13, 15, 20, 22) or used only qualitative measures for smoking (15, 16, 20-22). Although aromatic amines and heterocyclic aromatic amines are known to be substrates for both NAT1 and NAT2 (7, 23, 24), the joint effect of NAT1 and NAT2 genotypes on colorectal cancer risk was considered only in one small study (22). A possible modifying effect of NAT2 acetylator status on colorectal cancer risk associated with exposure to environmental tobacco smoke (ETS), also containing potentially carcinogenic arylamines (4), was only investigated in one previous study on rectal cancer (17). Findings about colorectal cancer risk associated with red meat consumption point towards greater risk among NAT2 fast acetylators than among NAT2 slow acetylators (20, 25-27); however, not all studies support such an effect of NAT2 acetylator status (13, 28).

Therefore, we investigated the effect of NAT1 and NAT2 genotypes on colorectal cancer risk associated with exposure to tobacco smoke and meat consumption in a German case-control study.

A population-based case-control study on colorectal cancer was carried out in the Rhine-Neckar-Odenwald region in the southwest of Germany. Patients were eligible for the study if they had a first diagnosis of invasive colorectal cancer (ICD-10: C18-C20) between January 2003 and June 2004. All 22 hospitals treating this disease in the study region agreed to participate in the study.

Control individuals were selected randomly from lists of residents supplied by population registries and matched to cases according to sex, 5-year age groups, and county of residence. Cases as well as control persons were eligible if they were of ages ≥30 years, lived in the study region, spoke German, and were physically and mentally able to participate in a personal interview of ∼1 hour. Control individuals with a history of colorectal cancer were excluded from the study.

The study was approved by the ethics committees of the University of Heidelberg and the medical boards of Baden-Wuerttemberg and Rhineland-Palatinate. Written informed consent was obtained from all study participants.

Using a standardized questionnaire, information on demographic factors as well as on known and suspected risk factors for colorectal cancer was collected by trained interviewers. As part of the interview, participants were asked to provide a detailed lifetime history of active and passive smoke exposure as well as information on frequency of meat consumption, usual doneness, and mode of preparation in the 12 months preceding the interview (controls) or the onset of symptoms (cases). Smokers were asked to report details about their smoking history in phases to be able to take into account substantial changes of smoking behavior. Data on exposure to ETS in adulthood were also collected in phases. For cases, we additionally requested histology reports and discharge letters. Information on prior endoscopies of the large intestine was validated by contacting the physicians who carried out the examination. Blood or mouthwash samples were collected from both cases and controls.

Overall, 540 colorectal cancer patients agreed to participate in the study. Compared with the estimated number of colorectal cancer cases based on data from the population-based cancer registry of the nearby state of Saarland, the participating patients constitute ∼50% of the expected number of eligible cases in the study region. The relatively low participation rate among patients may be mainly attributable to incomplete recruitment by the attending physicians. Of all 1,391 eligible control persons, 614 (44%) participated in the interview and an additional 350 (25%) individuals completed a short questionnaire. Of the 540 cases and 614 controls who participated in the interview, 516 (96%) cases and 613 (>99%) controls provided either a blood or a mouthwash sample.

Genomic DNA was isolated from blood using the FlexiGene DNA Kit (Qiagen GmbH, Hilden, Germany) and for a subset of samples using the MagNA Pure LC instrument (Roche Diagnostics GmbH, Mannheim, Germany). For 3 cases and 31 control individuals (0.6% and 5.1%, respectively) who had provided only a mouthwash sample, the QIAmp DNA Mini Kit (Qiagen) was used for DNA isolation. Genotyping was carried out blinded to case-control status and a random selection of 10% of all samples was genotyped twice for quality control.

Determination of NAT2 acetylator status was based on genotyping of the single nucleotide polymorphisms at nucleotide positions 481, 590, 803, and 857 of the NAT2 gene using capillary-based real-time PCR followed by melting curve analysis (29). Primers and probes, as well as the conditions for PCR and melting curve analyses, have previously been described (29, 30). To reduce genotyping effort, the single nucleotide polymorphism A803G was only determined for samples in which only one or none of the other three rare alleles was present; i.e., only for those that would be classified as carriers of the NAT2 wild-type allele when considering solely single nucleotide polymorphisms at nucleotide positions 481, 590, and 857 (∼43% of samples). The NAT2*6A allele is distinguished by the single nucleotide polymorphism G590A, the NAT2*7B allele by G857A, and the NAT2*5 allele by C481T and/or A803G. Individuals carrying at least one NAT2*4 wild-type allele were classified as NAT2 fast acetylators whereas individuals possessing two NAT2 variant alleles were classified as NAT2 slow acetylators.

For the purpose of the present analysis, carriers of the NAT1*10 allele were compared with genotypes composed of any of the other NAT1 variant alleles or the NAT1*4 wild-type allele (http://www.louisville.edu/medschool/pharmacology). To identify the NAT1*10 allele, the single nucleotide polymorphisms T1088A and C1095A were investigated. Because these two single nucleotide polymorphisms are also present in the NAT1*14A allele, we additionally genotyped for G560A to distinguish the NAT1*14A allele from the NAT1*10 allele. Other alleles with the single nucleotide polymorphisms T1088A and C1095A [i.e., the NAT1*18A allele, which has been associated with high enzymatic activity (31), and the NAT1*29 allele with a reported allele frequency of 0.002 in a French population (32)] were not accounted for in the present study. The detection of the single nucleotide polymorphisms T1088A and C1095A using the LightCycler system (Roche Diagnostics) was carried out as described by Wikman et al. (29).

Primers and hybridization probes which were used for the genotyping are given in Table 1. PCR and melting curve analyses were done in 10-μL glass capillaries using 1× PCR buffer, 2.5 mmol/L MgCl₂, 200 μmol/L deoxynucleotide triphosphates each, 0.1% bovine serum albumin, 0.5 units of Taq polymerase, 0.15 μmol/L of each probe, 1 μmol/L of each primer, and ∼10 ng of genomic DNA as template. The cycling conditions were as follows: initial denaturation at 95°C for 2 minutes, followed by 45 cycles of denaturation at 95°C for 0 seconds, annealing at 54°C (T1088A and C1095A) or 53°C (G560A) for 5 seconds, and elongation at 72°C for 7 seconds (T1088A and C1095A) or 10s (G560A) with a ramping rate of 20°C/s. Melting curve analyses included initial denaturation at 95°C for 30 seconds, 30 seconds at 40°C followed by slow heating of the samples to 70°C (T1088A and C1095A) or 75°C (G560A) with a ramping rate of 0.1°C/s, and continuous fluorescence detection.

For a subset of samples wherein NAT1 genotype could not be determined without ambiguity (n = 30; 2.7%), we additionally tested for presence of the NAT1*11 allele, which exhibits a 9-bp deletion in the region between nucleotide positions 1,065 to 1,090, to ensure correct genotype assignment with respect to NAT1*10. For the detection of T640G, which is unique to the NAT1*11 allele, an asymmetrical PCR was carried out using 0.1 μmol/L of the forward primer NAT1*386F and 1 μmol/L of the reverse primer NAT1*22R. The cycling conditions were as follows: initial denaturation at 95°C for 2 minutes, followed by 45 cycles of denaturation at 95°C for 0 seconds, annealing at 55°C for 5 seconds, and elongation at 72°C for 19 seconds with a ramping rate of 20°C/s. Melting curve analysis included initial denaturation at 95°C for 30 seconds, 30 seconds at 40°C followed by slow heating of the samples to 75°C with a ramping rate of 0.1°C/s, and continuous fluorescence detection.

Results for the 10% of samples that were genotyped twice for quality control did not show any discrepancies.

Ever active smoking was defined as having smoked >100 cigarettes in lifetime. Subjects were termed current smokers if they smoked within the 2 years preceding the reference date, which was the date of hospital admission for surgery or neoadjuvant treatment of colorectal cancer for cases and the date of the interview for controls. Pack-years of smoking were calculated by dividing the average number of cigarettes smoked per day by 20 and multiplying the latter by the number of years of smoking for each phase and summing up the pack-years of the smoking phases to obtain the total number of pack-years.

Ever exposure to ETS was defined as having been exposed to ETS in childhood for at least 1 hour/wk for at least 1 year or as having been exposed in adult life for at least 1 hour/d for at least 1 year by the partner or at work. To quantify exposure to ETS in adulthood, duration and cumulative exposure in hours/day-years were calculated (33). For the latter, average hours per day of exposure to ETS from the partner were calculated separately for weekdays and weekend days, and then summed up and multiplied with the number of years of the respective phase. By summation of all phases of exposure, the total hours/day-years were obtained. The same procedure was used for exposure to ETS at workplace although the hours of exposure on working days were averaged over all 7 weekdays. Pack-years of exposure to ETS from the partner were calculated as described for active smoking. Missing data on duration and cumulative exposure to ETS (<5% of cases and controls) were replaced by the mean of nonsmoking controls for the respective variable.

Education level was defined according to recommendations of the German Association of Epidemiology by combining schooling and professional training (34). The continuous variable (levels 1-8 for low to high) was categorized into three groups for the descriptive analysis (levels 1-3, 4-5, and 6-8). Regular use of nonsteroidal anti-inflammatoiry drugs (NSAID) was defined as use of any drug containing NSAIDs at least once a month for at least 1 year regardless of indication or dosage. Previous endoscopic examinations included any rectoscopy, sigmoidoscopy, or colonoscopy that was not part of the diagnosis of colorectal cancer.

The present analysis was restricted to 1,109 European born participants (505 cases and 604 controls) who completed the interview and for whom DNA was available.

χ² tests were used to compare the distribution of potential risk factors between cases and controls. Multivariate conditional logistic regression analysis was used to estimate odds ratios (OR) and their 95% confidence intervals (95% CI) using the PHREG procedure of the statistical software package SAS 9.1 (SAS Institute, Inc., Cary, NC). The analyses were stratified for sex and age group and additionally adjusted for regular use of NSAIDs (yes/no), any previous endoscopic examinations of the large intestine (yes/no), first-degree family history of colorectal cancer (yes/no/unknown), average daily alcohol consumption (0, >0-7.3, 7.4-20.6, 20.7+ g/d; categorized according to tertiles of controls), frequency of red meat consumption (low/intermediate/high), and education level and body mass index (kg/m²) as continuous variables. Models to estimate risk associated with meat consumption were additionally adjusted for pack-years of smoking (0, 1-9, 10-19, 20-29, 30+). Statistical interaction on a multiplicative scale between genotypes and environmental exposures was evaluated with the likelihood ratio test. For case-only analyses, we used the LOGISTIC procedure (SAS 9.1), adjusting for the above-mentioned variables. Other factors, such as place of birth, county of residence, fruit and vegetable consumption, or physical activity, did not materially influence the estimates and were therefore not included in the multivariate models.

The distribution of sociodemographic factors and potential risk factors among the 505 colorectal cancer patients and 604 controls is given in Table 2. The mean age of the study participants was 67.9 (SD, 10.0) years among cases and 66.6 (SD, 10.2) years among controls. Controls were more likely than cases to have undergone an endoscopic examination of the large intestine, to use NSAIDs regularly, and to have a high level of education. The proportion of ever smokers did not differ significantly between cases (52.9%) and controls (52.2%). Among never active smokers, 30.7% of cases and 24.2% of controls were also not regularly exposed to ETS (P = 0.10). More cases (11.5%) than controls (8.3%) reported very frequent consumption of red meat (i.e., at least once per day). The distribution of NAT1 and NAT2 genotypes was not differential by case-control status.

The distribution of most covariates including sex, age, body mass index, regular use of NSAIDs, colorectal endoscopy, and smoking or alcohol consumption, as well as site, staging, and grading of the tumor, did not differ by NAT1 or NAT2 genotype. However, significantly more patients who were NAT2 fast acetylators reported to have a first-degree family history of colorectal cancer than NAT2 slow acetylators (18.0% versus 10.5%, P = 0.01), whereas among controls this difference was marginal (12.3% versus 10.7%, P = 0.80). Neither NAT2 acetylator status nor NAT1 genotype was an independent risk factor for colorectal cancer, the OR being 0.86 (95% CI, 0.66-1.11) for NAT2 slow versus fast acetylators and 0.84 (95% CI, 0.64-1.10) for carriers of the NAT1*10 allele versus noncarriers, respectively.

Table 3 shows the association between active smoking and colorectal cancer risk, both overall and stratified by NAT2 acetylator status. Overall, our data indicate a risk increase associated with smoking after accumulation of 30+ pack-years (OR, 1.38; 95% CI, 0.89-2.15; P_trend = 0.03; Table 3). An analysis stratified by sex showed that the association between smoking and colorectal cancer risk was pronounced among women but not among men [ORs for 30+ pack-years were 3.72 (95% CI, 1.36-10.15) and 1.02 (95% CI, 0.61-1.70), respectively]. In spite of the difference in risk estimates between males and females, there was no statistically significant heterogeneity by sex with respect to smoking-associated colorectal cancer risk (P = 0.16).

The analysis of risk associated with active smoking stratified by NAT2 acetylator status yielded slightly higher risk estimates among NAT2 slow acetylators than among NAT2 fast acetylators with a significant trend for pack-years among NAT2 slow acetylators (Table 3). With respect to NAT1 genotype, colorectal cancer risk associated with smoking was moderately elevated among noncarriers of the NAT1*10 allele but not among carriers of the NAT1*10 allele [ORs for 20+ pack-years of smoking were 1.37 (95% CI, 0.89-2.12) and 1.08 (95% CI, 0.56-2.06), respectively; data not shown]. When the effect of NAT1 and NAT2 genotypes as modifiers of smoking-associated colorectal cancer risk was analyzed jointly, NAT2 slow acetylators not carrying the NAT1*10 allele (i.e., those with presumably low acetylation capacity) seemed to be at higher risk than those with NAT1 and/or NAT2 fast genotype (carriers of the NAT1*10 allele and carriers of the NAT2*4 allele, respectively; Table 4). However, the effect modification by combined NAT1 and NAT2 genotypes was modest (P_interaction = 0.34).

Among the 527 nonsmoking participants (238 cases and 289 controls), exposure to ETS was not found to be associated with an increased risk of colorectal cancer (Table 5). For heavy exposure to ETS from the partner solely, we observed an indication for a risk increase [OR, 1.91 (95% CI, 0.74-4.95) for 24+ pack-years of exposure to ETS from the partner]. However, we observed heterogeneity by NAT2 acetylator status for the association between exposure to ETS and colorectal cancer risk (Table 5). Exposure to ETS in childhood and adulthood conferred an increase in risk among NAT2 fast acetylators whereas among NAT2 slow acetylators there seemed to be an inverse association. Likelihood ratio tests were supportive of a modification of risk associated with exposure to ETS by NAT2 acetylator status (P_interaction values were 0.06 for timing, 0.07 for duration, and 0.08 for cumulative exposure). After testing the independence between exposure to ETS and NAT2 acetylator status among controls (P > 0.3), we also carried out a case-only analysis of our data. There was a statistically significant interaction between NAT2 acetylator status and exposure to ETS in childhood and adulthood [OR (for interaction), 2.20; P = 0.03] and borderline significant interactions with duration and cumulative exposure to ETS (P = 0.11 and 0.09, respectively).

Consumption of red meat at least once per day was associated with a 74% increased risk of colorectal cancer (95% CI 0.96-3.16) when compared with meat consumption less than once per week. Risk associated with frequent red meat consumption tended to be higher for individuals preferring well-done meat than for those consuming mostly rare or medium-cooked meat [OR, 2.06 (95% CI, 0.88-4.81) and 1.46 (95% CI, 0.70-3.05), respectively] when compared with individuals consuming meat infrequently and preferably rare or medium cooked. Individuals being current smokers and consuming red meat very frequently were at a significantly 3-fold increased risk for colorectal cancer (OR, 3.11; 95% CI, 1.03-9.44) compared with nonsmokers reporting infrequent meat consumption. However, due to the small number of individuals in this high-risk group (13 cases, 8 controls), an investigation of gene-environment interactions was not possible.

NAT2 acetylator status alone did not markedly modify risk associated with frequent consumption of red meat [OR, 2.02 (95% CI, 0.71-5.73) for NAT2 fast acetylators and 1.77 (95% CI, 0.84-3.72) for NAT2 slow acetylators]. Risk estimates for frequent consumption of red meat were more pronounced among carriers of the NAT1*10 allele than among noncarriers [OR, 2.40 (95% CI, 0.78-7.35) and 1.67 (95% CI, 0.81-3.48), respectively]. However, there was no statistically significant interaction between NAT1 genotype and frequency of red meat consumption in colorectal cancer risk (P = 0.40). The joint analysis of NAT1 and NAT2 genotypes with respect to meat-associated colorectal cancer risk revealed a significantly increased risk associated with frequent consumption of red meat among individuals with NAT1 and/or NAT2 fast genotype but not among those with NAT1 and NAT2 slow genotypes (Table 4). There was a moderate effect modification by combined NAT1 and NAT2 genotypes for consumption of red meat (P_interaction = 0.14).

The findings of the present study support the hypothesis that both active smoking and frequent consumption of red meat are associated with increased risk of colorectal cancer. Although NAT1 and NAT2 genotypes were not independently associated with colorectal cancer risk, our findings suggest an effect modification of the association between exposure to xenobiotics from tobacco smoke or red meat consumption and colorectal cancer risk by NAT1 and NAT2 genotypes.

The magnitude of colorectal cancer risk associated with active smoking, as well as the fact that an increased risk was only apparent after smoking for a long duration and at a high intensity, is in accordance with the literature (4). Thus, our findings are consistent with the hypothesis of smoking acting at an early stage in carcinogenesis and many years being required for completion of the carcinogenic process (2). Results of the analysis stratified by NAT2 acetylators point towards a slightly stronger association among NAT2 slow acetylators, which is in line with several previous studies (14, 20, 35), two of which reported a significant gene-environment interaction for risk of colorectal adenomas (14, 35). However, there are also studies pointing towards a higher risk of colorectal neoplasms associated with smoking among NAT2 fast acetylators (15, 19) or not supporting an effect modification by NAT2 acetylator status (12, 16-18). In the present study, the tendency towards a stronger association with smoking among individuals with slow acetylation capacity was somewhat more pronounced when both NAT1 and NAT2 genotypes were considered although the gene-environment interaction was not statistically significant. An impaired detoxification of aromatic amines such as 4-aminobiphenyl, for which N-acetylation is an important detoxification pathway (7), may explain the findings of greater risk among NAT slow acetylators. This mechanism has also been suggested for smoking-associated bladder cancer (36-38) and is substantiated by the findings of higher 4-aminobiphenyl-hemoglobin adduct levels among NAT2 slow acetylators compared with NAT2 fast acetylators (38, 39).

Data on the association between ETS and colorectal cancer risk are scarce and inconsistent (17, 40-42). Overall, the present study does not support a strong association between exposure to ETS and colorectal cancer. However, because the effect of ETS was only investigated in a subgroup of the present study population, the power was limited to detect a weak or moderate increase in risk associated with exposure to ETS. The inverse association with exposure to ETS in childhood is likely to be a spurious finding, which may be due to difficulties in recalling childhood exposure in this relatively old study population.

However, in the subgroup of NAT2 fast acetylators, exposure to ETS in childhood and adulthood was associated with a significant 2.6-fold increase in risk but risk seemed to be decreased among NAT2 slow acetylators exposed to ETS. The latter seems to be driven by the composition of the reference group among NAT2 slow acetylators, which was composed of more cases than controls. This is contrary to the ratio of cases to controls in the overall study population and to subgroups of NAT2 fast and slow acetylators overall and among nonsmokers. We do not have a biologically plausible explanation for a protective effect of ETS with respect to colorectal cancer nor is there evidence for such an effect from previous studies on other cancer sites (4). Hence, we believe that the decreased risk among NAT2 slow acetylators is a chance finding. Nevertheless, the strong association between exposure to ETS and colorectal cancer risk among NAT2 fast acetylators is striking. To our knowledge, apart from a study on rectal cancer, which was, however, not restricted to nonactive smokers (17), this is the first study to investigate the effect of NAT2 genotype in this context. A case-only analysis of our data, which is not affected by any bias that may be introduced by the control group, such as differential recall of exposure, supports a statistical interaction between NAT2 genotype and exposure to ETS with respect to colorectal cancer risk. In addition, greater risk associated with ETS among NAT2 fast acetylators has also been reported for breast cancer (30, 43). However, our findings about a possible effect modification of the association with exposure to ETS by NAT2 acetylator status are based on a relatively small subgroup of our study population and should therefore be interpreted with caution.

An association between exposure to ETS and colorectal cancer risk is nevertheless biologically plausible. Although the large bowel is not in direct contact with tobacco smoke, carcinogens from active smoking as well as from ETS may reach the colonic mucosa by direct ingestion or after inhalation via the circulatory system (44). Exposure to carcinogens from ETS may be relatively low when compared with active smoking. Still, prolonged exposure to ETS from a smoking partner or at work may lead to considerable exposure to carcinogens. Several carcinogens are present at higher levels in sidestream than in mainstream smoke (45). In addition, there is accumulating evidence for a causal association between passive smoking and risk of breast cancer, a site that is also not directly exposed to ETS (46).

In line with previous reports and the hypothesis that heterocyclic aromatic amines are among the relevant carcinogens in red meat (1, 3), frequent consumption of red meat was associated with an increased risk of colorectal cancer in the present study, the association being more pronounced for consumption of well-done meat than for consumption of rare or medium-cooked meat. Whereas NAT2 acetylator status alone did not significantly modify this association, colorectal cancer risk tended to be higher among carriers of the NAT1*10 allele than among noncarriers. This observation corroborates the findings of previous studies and lends support to a fast acetylation capacity associated with the NAT1*10 allele that increases susceptibility to heterocyclic aromatic amine–induced carcinogenesis (27, 47).

The joint analysis of NAT1 and NAT2 genotypes strengthened the notion of increased susceptibility to meat-related carcinogens among individuals with presumably fast acetylation capacity. In accordance with previous studies (29, 48), we observed strong linkage disequilibrium between the NAT1*10 allele and the NAT2*4 allele (P < 0.0001). This linkage disequilibrium may partly explain the findings of the joint analysis of NAT1 and NAT2 genotypes. Differential effects by acetylation status may be diluted when only considering one gene at a time. A further differentiation of effects by all genotype combinations of NAT1 and NAT2 was not possible due to limitations in sample size. However, as mentioned above, the genotype-phenotype relationship for NAT1 is not entirely clear (9) and further investigations are needed to establish the functional properties of NAT1.

Taken together, the findings about exposure to ETS and meat consumption suggest that NAT enzymes play a greater role in bioactivation than in detoxification with respect to colorectal cancer. Both combustion of tobacco and cooking meat at high temperature generate heterocyclic aromatic amines (49). 2-Amino-α-carboline, which is present at high levels in sidestream smoke, is readily activated by both NAT1 and NAT2 after N-oxidation (24, 49). Thus, individuals with fast acetylation capacity may more efficiently activate heterocyclic aromatic amines to reactive metabolites, thereby increasing the likelihood of heterocyclic aromatic amine–induced DNA damage and, consequently, would be more susceptible to colorectal cancer associated with exposure to heterocyclic aromatic amines. Findings of significantly higher adduct levels and aberrant crypt foci in the colon of fast-acetylator rats than in slow-acetylator rats after administration of heterocyclic aromatic amines corroborate this hypothesis (50, 51).

The present study has several strengths: its population-based design, the extensive risk factor information which rendered possible the adjustment for potential confounder variables, the availability of biological samples for >95% of cases and controls, and the quality control procedures for genotyping. The observed proportion of NAT2 slow acetylators, 58% of cases and 62% of controls, as well as the frequencies of 0.16 and 0.17 for the NAT1*10 allele, are in accordance with previous studies in German populations (29, 30, 52).

The observation of a greater risk of colorectal cancer associated with active smoking among females than among males is unexpected. The majority of studies reported either no sex difference or a stronger association among males (4). Because there was no statistically significant heterogeneity between males and females in the present study and the findings were based on a relatively small number of smoking women, it cannot be ruled out that the observed difference between males and females is due to chance. Unfortunately, the power of the present study did not allow for a reliable investigation of gene-environment interactions in subgroups of the study population.

Recall bias on exposure to tobacco smoke is unlikely because tobacco smoke has not been publicized as a risk factor for colorectal cancer, and recall would have to be differential for active smoking and exposure to ETS to result in the observed effects with respect to NAT2 genotype. Meat as a potential risk factor for colorectal cancer drew public interest recently; however, the risk estimates of the present study are in line with those of several previous studies (1) and, hence, potential recall bias is likely to be marginal. Because individual dietary habits tend to be stable over time (53, 54), the fact that meat consumption was only assessed for the 12 months preceding the interview for controls or the onset of symptoms for cases is unlikely to have led to major misclassification. We cannot exclude that selection bias may have affected our results because the overall participation rate among controls was only 44%, which may partly be attributable to the inclusion of older subjects with a participation rate of only 37% for those of ages ≥70 years. However, a selection bias with respect to smoking is unlikely because the observed prevalences of smoking correspond with those from a retrospective birth cohort analysis of West Germany (55). The fact that only ∼50% of eligible cases participated in the study seems to be mainly due to incomplete recruitment in some hospitals because of work overload of the clinicians. A large variation in regional estimates of recruitment rates of cases supports this notion. Participating clinicians could not provide us with a detailed description about reasons for nonparticipation except that only a small proportion of patients who were informed about the study refused to participate. Unfortunately, we do not have access to characteristics of nonparticipants due to strict confidentiality rules that preclude the use of data from patients not participating in the study. However, because the participation of patients was largely dependent on the situation in the hospitals, we do not believe that this is an important source of bias. Although it is unlikely that population stratification is of major concern in our study population (56), we excluded individuals born outside of Europe from the analysis to enhance ethnic homogeneity. Because solely incident cases were recruited and the mean period between diagnosis and interview was <2 months, survival bias is negligible in the present study.

In conclusion, the present study strengthens prior evidence of an association between colorectal cancer and smoking as well as red meat consumption. Colorectal cancer risk associated with exposure to ETS may only be relevant among genetically susceptible individuals. Our findings indicate that NAT1 and NAT2 genotypes contribute jointly to individual susceptibility and suggest that bioactivation of heterocyclic aromatic amines by NAT enzymes plays an important role in colorectal carcinogenesis. Our hypothesis of a differential effect of acetylator status with respect to active smoking and exposure to ETS needs confirmation in larger studies.

We thank the many clinicians who endorsed the study; all patients and control individuals who participated in the study; and B. Collins, S. Todt, U. Handte-Daub, and B. Kaspereit for technical assistance.