Prospective study of N-acetyltransferase-2 genotypes, meat intake, smoking and risk of colorectal cancer

Authors

  • Andrew T. Chan,

    Corresponding author
    1. Gastrointestinal Unit, Massachusetts General Hospital and Harvard Medical School, Boston, MA, USA
    2. Channing Laboratory, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston, MA, USA
    • Gastrointestinal Unit, Massachusetts General Hospital GRJ-722, Boston, MA 02114
    Search for more papers by this author
    • Fax: 617 525 2008.

  • Gregory J. Tranah,

    1. Department of Epidemiology, Harvard School of Public Health, Boston, MA, USA
    Search for more papers by this author
  • Edward L. Giovannucci,

    1. Cancer Epidemiology Program, Dana-Farber/Harvard Cancer Center, Boston, MA, USA
    2. Department of Epidemiology, Harvard School of Public Health, Boston, MA, USA
    3. Department of Nutrition, Harvard School of Public Health, Boston, MA, USA
    Search for more papers by this author
  • Walter C. Willett,

    1. Channing Laboratory, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston, MA, USA
    2. Cancer Epidemiology Program, Dana-Farber/Harvard Cancer Center, Boston, MA, USA
    3. Department of Epidemiology, Harvard School of Public Health, Boston, MA, USA
    4. Department of Nutrition, Harvard School of Public Health, Boston, MA, USA
    Search for more papers by this author
  • David J. Hunter,

    1. Channing Laboratory, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston, MA, USA
    2. Cancer Epidemiology Program, Dana-Farber/Harvard Cancer Center, Boston, MA, USA
    3. Department of Epidemiology, Harvard School of Public Health, Boston, MA, USA
    4. Department of Nutrition, Harvard School of Public Health, Boston, MA, USA
    Search for more papers by this author
  • Charles S. Fuchs

    1. Channing Laboratory, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston, MA, USA
    2. Cancer Epidemiology Program, Dana-Farber/Harvard Cancer Center, Boston, MA, USA
    3. Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, MA, USA
    Search for more papers by this author

Abstract

Consumption of red meat has been associated with elevated risk of colorectal cancer; however, mechanisms underlying this relationship are not well established. N-acetyltransferase 2 (NAT2) appears to activate carcinogenic heterocyclic amines found in meat as well as cigarette smoke. Genetic variation in this enzyme, associated with rapid acetylation, may modulate the influence of meat intake on cancer risk. We examined the risk of incident colorectal cancer according to NAT2 genotypes, meat intake and smoking in a prospective, nested case-control study among 32,826 women enrolled in the Nurses' Health Study who provided prediagnostic blood specimens. We matched 183 women with colorectal cancer to 443 controls. Although acetylator genotype alone was not associated with the risk of colorectal cancer, women with rapid acetylator genotypes experienced a greater risk associated with intake of ≥0.5 serving of beef, pork or lamb as a main dish per day compared to intake of less meat (multivariate OR = 3.01; 95% CI = 1.10–8.18). In contrast, among slow acetylators, intake of beef, pork or lamb was not associated with risk of colorectal cancer (multivariate OR = 0.87; 95% CI = 0.35–2.17). The interaction between acetylator genotype and meat intake approached statistical significance (Pinteraction = 0.07). Moreover, compared to slow acetylators who smoked ≤35 pack-years and ate <0.5 serving/day of red meat, the OR for rapid acetylators who smoked >35 pack-years and ate ≥0.5 serving/day was 17.6 (95% CI 2.0–158.3). These prospective data suggest that red meat may increase the risk of colorectal cancer, particularly among genetically susceptible individuals. The influence of NAT2 genotype on this association supports a role for heterocyclic amines in mediating the effect of red meat on colorectal carcinogenesis. © 2005 Wiley-Liss, Inc.

Substantial epidemiological evidence suggests that intake of red meat is associated with an increased risk of colorectal cancer.1, 2, 3 However, underlying mechanisms are not well established. Experimental data have demonstrated that prolonged cooking of meat at high temperatures favors production of several potent carcinogens, including heterocyclic amines,4, 5 which are also found in cigarette smoke.6

N-acetyltransferase 2 (NAT2) is a critical enzyme in the activation of these carcinogens.7 Prior studies of inherited NAT2 acetylator phenotypes have suggested an association between increasing intake of meat and colorectal cancer in individuals with rapid but not slow acetylators of drug substrates.8, 9, 10 In predominantly Caucasian populations, 3 of the most common alleles have been associated with a slower acetylation phenotype compared to the rapid wild type.11 Although most studies have not demonstrated an overall association between rapid acetylation genotypes and colorectal cancer,12 such a relationship may only be evident in individuals with substantial exposure to heterocyclic amines in diet or cigarette smoke, which additionally produce carcinogenic polycyclic aromatic hydrocarbons.6

Present data on the interaction between diet, acetylator genotypes and risk of colorectal cancer are limited and conflicting.10, 13, 14, 15, 16, 17 Using prospectively collected blood specimens and validated exposure data,18, 19 we had the opportunity to examine these relationships in a nested case-control study of women participating in the Nurses' Health Study.

Material and methods

Study population

The Nurses' Health Study is a cohort of 121,701 female nurses, residing in 11 US states, who were 30 to 55 years of age when they returned a questionnaire in 1976.20 Using follow-up questionnaires mailed every 2 years, we have updated information and diagnoses of cancer that are subsequently confirmed through review of medical records or death reports. In 1989–1990, 32,826 participants provided a blood specimen; follow-up of this subcohort exceeds 96%. The methods used for follow-up, end-point ascertainment and blood collection has been previously detailed.21

Selection of colorectal cancer cases and controls

Eligible women for our study provided a blood specimen and were free from inflammatory bowel disease or cancer (except nonmelanoma skin). After blood collection through June 1, 2000, we confirmed 193 incident cases of colorectal cancer, whom we matched on year of birth and month/year of blood collection to 2–3 controls without colorectal cancer. Most cases (96%) were confirmed by medical record; cases for which medical records were unavailable were reconfirmed by the participant in writing or by telephone. The mean age at diagnosis was 65.5 years (range 46.0–77.8 years). Subsequently, we excluded 13 women without diet or smoking data and 44 women for whom there was insufficient DNA or genotyping could not be performed. Thus, we included 626 participants in the final analysis.

Genotyping

Using previously described methods,22 laboratory personnel blinded to case-control status extracted DNA from blood using QIAmp (Qiagen, Inc., Chatsworth, CA), performed genotyping of NAT2 I114T (rs1901280), R197Q (rs1799930) and G286E (rs1799931) polymorphisms using TaqMan (PE Applied Biosystems, Foster City, CA) analysis (primers/probes available upon request) and assigned genotypes using end-point fluorescence (Applied Biosystems 7900HT, SDS Software v1.7a). We inserted quality control samples to validate genotype identification procedures; concordance for blinded samples was 100%. Using the χ2 test, we verified that NAT2 genotypes in control subjects were in Hardy-Weinberg equilibrium. For the statistical analysis and consistent with prior studies, we classified individuals as rapid acetylators if they had no more than 1 NAT2 slow-acetylator allele.23

Assessment of meat intake and smoking status

We assessed diet using a validated semiquantitative food frequency questionnaire administered in 1980, 1984, 1986, 1990 and 1994.18, 19 For the present analysis, data on intake of meat, assessed as beef, pork or lamb as a main dish, was obtained prospectively from the questionnaire prior to diagnosis. We did not consistently assess cooking or preparation techniques; hence, we were unable to evaluate the influence of these factors on incident cancer risk. We determined smoking status at baseline in 1976 and updated this information on every subsequent biennial questionnaire.24 To calculate pack-years, smoking duration in years was multiplied by packs of cigarettes smoked per day (20 cigarettes per pack).

Statistical analysis

We estimated odds ratios (OR) and corresponding 95% confidence intervals (CI) for the associations between acetylator genotype and colorectal cancer using conditional logistic regression models. We obtained similar results using unconditional logistic regression models adjusting for matching factors. To increase statistical power in our analyses of smoking and meat intake stratified by genotype, we used unconditional logistic regression models adjusting for age, according to 5-year categories, and multiple known or suspected cancer risk factors simultaneously. For all covariates, we used updated data from the most recent questionnaire completed before diagnosis of cancer. We assessed interaction by creating indicator variables for the combination of acetylator genotype and meat or smoking variables, and slow acetylators with the lowest intake of meat or smoking were used as the reference categories. We calculated p values for the test for interaction by the log-likelihood test, comparing the model with these indicator variables to a model containing terms only for the main effect of genotype and dietary exposure.25 All P values are 2-sided.

Results

The mean age at blood draw was 60.6 (range 44.7–70.9) in the 183 cases of colorectal cancer and 443 controls. Over 98% of the population was Caucasian, and the genotype frequencies were similar to other predominantly Caucasian populations (Table I).14, 26 Baseline characteristics of this population have been previously described.21 Compared to slow acetylators, rapid acetylators did not have a significantly higher risk of colorectal cancer (Table II).

Table I. NAT2 Genotype Frequencies
 Cases N (%)1Controls N(%)Total N(%)
  • 1

    Percentages do not always sum to 100 due to rounding.

  • 2

    NAT*4 allele was defined as wild type, or the absence of amino acid changes isoleucine for threonine (I114T), arginine for glutamine (R197Q), or glycine for glutamic acid (G286E). NAT*5 allele was defined as the I114T amino acid change associated with a cytosine for thymine base pair substitution at position 341 (T341C). NAT*6 was defined as the R197Q amino acid change associated with an adenine for guanine base pair substitution at position 590 (G590A). NAT*7 allele was defined as the G286E amino acid change associated with an adenine for guanine base pair substitution at position 857 (G857A). Nomenclature based on www.louisville.edu/medschool/pharmacology/NAT.html.31

  • 3

    Acetylator genotypes based on previous classifications.23 Slow acetylators included participants with genotypes NAT2*5/NAT2*5, NAT2*5/NAT2*6, NAT2*5/NAT2*7, NAT2*6/NAT2*6, NAT2*6/NAT2*7. Rapid acetylators included participants with genotypes NAT2*4/NAT2*4, NAT2*4/NAT2*5, NAT2*4/NAT2*6, NAT2*4/NAT2*7.

NAT2 genotype2
 NAT2*4/NAT2*411 (6.0)24 (5.4)35 (5.6)
 NAT2*4/NAT2*542 (23.0)98 (22.1)140 (22.4)
 NAT2*4/NAT2*623 (12.6)48 (10.8)71 (11.3)
 NAT2*4/NAT2*70 (0.0)6 (1.4)6 (1.0)
 NAT2*5/NAT2*534 (18.6)94 (21.2)128 (20.4)
 NAT2*5/NAT2*647 (25.7)113 (25.5)160 (25.6)
 NAT2*5/NAT2*71 (0.6)15 (3.4)16 (2.6)
 NAT2*6/NAT2*622 (12.0)36 (8.1)58 (9.3)
 NAT2*6/NAT2*73 (1.6)9 (2.0)12 (1.9)
 Total183 (29.2)443 (70.8)626 (100)
Acetylator genotype3
 Slow acetylator107 (58.5)267 (60.3)374 (59.7)
 Rapid acetylator76 (41.5)176 (39.7)252 (40.3)
 Total183 (29.2)443 (70.8)626 (100)
Table II. Odds Ratio of Incident Colorectal Cancer According to NAT2 Acetylator Genotype, Meat Intake or Smoking History1
 Number of cases/Number of controlsAge-adjusted OR (95% CI)Multivariate2 OR (95% CI)
  • 1

    OR = odds ratio; CI = confidence interval.

  • 2

    Multivariate ORs are adjusted for age (5-year categories), body mass index (tertiles), colorectal cancer in a parent or sibling (yes or no), post-menopausal hormone use (premenopausal or never, past, current), previous endoscopy (yes or no), current multivitamin use (yes or no) and regular aspirin use (<2 standard tablets/week or ≥2 standard tablets/week). The body mass index is weight in kilograms divided by the square of the height in meters. For analyses of meat intake, multivariate ORs are additionally adjusted for smoking history (≤35 pack-years or >35 pack-years). Pack-years were calculated as smoking duration in years multiplied by packs of cigarettes smoked per day (20 cigarettes per pack). For analyses of smoking, multivariate ORs are additionally adjusted for beef, pork or lamb as a main dish (≤0.5 servings/week and >0.5 servings/week). All covariates were computed using the most recent questionnaire data prior to date of diagnosis for cases or the matched time interval for controls.

  • 3

    Acetylator genotypes are based on previous classifications.23 Slow acetylators included participants with genotypes NAT2*5/NAT2*5, NAT2*5/NAT2*6, NAT2*5/NAT2*7, NAT2*6/NAT2*6, NAT2*6/NAT2*7. Rapid acetylators included participants with genotypes NAT2*4/NAT2*4, NAT2*4/NAT2*5, NAT2*4/NAT2*6, NAT2*4/NAT2*7.

  • 4

    p for interaction between servings of beef, pork or lamb as a main dish and acetylator genotype was 0.07. Interaction was assessed by creating indicator variables for the combination of acetylator status and meat variables, and slow acetylators who consumed ≤0.5 servings/day comprised the reference category. p value for the interaction was calculated by the log-likelihood ratio test, comparing this model with a model containing indicator terms only for the main effect of genotype and meat intake.

  • 5

    p for interaction between smoking before age 30 and acetylator genotype was 0.39. Interaction was assessed by creating indicator variables for the combination of acetylator status and smoking before age 30, and slow acetylators who did not smoke before age 30 comprised the reference category. p value for the interaction was calculated by the log-likelihood ratio test, comparing this model with a model containing indicator terms only for the main effect of genotype and smoking before age 30.

  • 6

    p for interaction between pack-years of smoking and acetylator genotype was 0.87. Interaction was assessed by creating indicator variables for the combination of acetylator status and pack-years, and slow acetylators who smoked ≤35 pack-years comprised the reference category. p value for the interaction was calculated by the log-likelihood ratio test, comparing this model with a model containing indicator terms only for the main effect of genotype and pack-years of smoking.

Acetylator genotype3   
  Slow acetylator107/2671.00 (referent)1.00 (referent)
  Rapid acetylator76/1761.04 (0.73–1.49)1.06 (0.73–1.53)
Beef, pork or lamb as a main dish4   
 All women   
  ≤0.5 serving/day166/4131.00 (referent)1.00 (referent)
  >0.5 serving/day17/301.15 (0.82–1.63)1.21 (0.85–1.72)
 Slow acetylators   
  ≤0.5 serving/day100/2481.00 (referent)1.00 (referent)
  >0.5 serving/day7/190.91 (0.37–2.24)0.87 (0.35–2.17)
 Rapid acetylators   
  ≤0.5 serving/day66/1651.00 (referent)1.00 (referent)
  >0.5 serving/day10/112.40 (0.96–5.99)3.01 (1.10–8.18)
Smoking before age 305   
 All women   
  Nonsmoker before age 3084/2171.00 (referent)1.00 (referent)
  Any smoking before age 3099/2261.15 (0.82–1.63)1.21 (0.85–1.72)
 Slow acetylators   
  Nonsmoker before age 3047/1181.00 (referent)1.00 (referent)
  Any smoking before age 3060/1491.02 (0.65–1.61)1.06 (0.67–1.69)
 Rapid acetylators   
  Nonsmoker before age 3037/991.00 (referent)1.00 (referent)
  Any smoking before age 3039/771.40 (0.81–2.43)1.50 (0.85–2.65)
Pack-years of smoking6   
 All women   
  ≤35 pack-years143/3791.00 (referent)1.00 (referent)
  >35 pack-years40/641.56 (1.07–2.58)1.63 (1.04–2.55)
 Slow acetylators   
  ≤35 pack-years84/2281.00 (referent)1.00 (referent)
  >35 pack-years23/391.60 (0.90–2.85)1.64 (0.91–2.94)
 Rapid acetylators   
  ≤35 pack-years59/1511.00 (referent)1.00 (referent)
  >35 pack-years17/251.76 (0.88–3.51)1.67 (0.80–3.48)

In the overall cohort, among over 61 foods included in the baseline dietary questionnaire, the strongest risk of colon cancer was associated with servings of red meat (beef, pork or lamb) as a main dish.1 Thus, we examined this association within this nested cohort using previously described categories.14, 23 Compared to women who ate ≤1 serving of red meat as a main dish per week, the multivariate ORs of colorectal cancer were 0.96 (95% CI = 0.66–1.42) for women who ate >1 serving per week but <0.5 serving/day and 1.42 (95% CI = 0.74–2.74) for women who ate ≥0.5 serving/day. We therefore examined risk associated with eating ≥0.5 serving/day compared to eating <0.5 serving/day in analyses stratified by genotype. Risk was largely confined to rapid acetylators (multivariate OR = 3.01; 95% CI = 1.10–8.18) whereas slow acetylators experienced no significant increase in risk (multivariate OR = 0.87; 95% CI = 0.35–2.17). The interaction between acetylator genotype and meat intake approached statistical significance (Pinteraction = 0.07) (Table II).

Based on a prior analysis of the overall cohort,24 we also evaluated the risk of colorectal cancer according to both early and total lifetime exposure to smoking, another source of heterocyclic amines.6 Although there appeared to be a greater risk associated with smoking before age 30 among rapid acetylators, we did not observe a statistically significant interaction between genotype and smoking before age 30 (Pinteraction = 0.39) or genotype and pack-years (Pinteraction = 0.87) (Table II). However, compared to slow acetylators who did not smoke before age 30 and ate <0.5 serving/day of red meat, the OR for rapid acetylators who did smoke before age 30 and ate ≥0.5 serving/day was 5.85 (95% CI=1.54–22.2). Compared to slow acetylators who smoked ≤35 pack-years and ate ≤0.5 serving/day of red meat, the OR for rapid acetylators who smoked >35 pack-years and ate ≥0.5 serving/day was 17.6 (95% CI 2.0–158.3). Nonetheless, the interactions between genotype, meat intake and either early or total lifetime smoking failed to achieve statistical significance (Pinteraction = 0.16 for smoking before age 30; Pinteraction = 0.17 for pack-years).

Discussion

In our study, we found an association between high intake of red meat and incident colorectal cancer risk, particularly among women with rapid acetylator NAT2 genotypes. Although we did not observe a clear interaction between smoking and NAT2 genotype, it appeared that women with rapid acetylator genotypes who consumed the highest levels of red meat and had the greatest exposure to cigarette smoke had a particularly elevated risk of colorectal cancer. Controlling for other known or suspected risk factors for colorectal cancer did not alter these findings.

Although previous studies support our results,10, 13, 14 others have not found an association between NAT2 acetylator genotype, meat intake and risk of colorectal cancer.15, 17 These inconsistencies may be due to retrospective design, inaccurate recall and assessment of diet and selection of controls that are not representative of the population from which cases were derived. Moreover, analyses based on prevalent, rather than incident, cases may be biased if acetylator genotypes are differentially associated with survival.27 Our study minimized such potential biases through its prospective design, high follow-up rate, well-validated self-report,18, 19 repeated assessments of exposures, detailed data on potential confounders and analysis of incident cancers. A prospective study of similar size largely confirmed our results in US men.14 Smaller, prospective studies in Dutch populations also found the positive association of colorectal cancer with either red meat consumption or smoking was strongest among NAT2 fast acetylators, although the interactions were not significant16 or not reported.28

We acknowledge several limitations. First, we had limited power to examine interactions of genotype and exposure. Because the sample size required to detect an interaction is at least 4 times larger than that needed to detect the main effect of a single variable,29 studies maintaining a prospective design are inherently challenging. Indeed, we were restricted to those cases of cancer after blood collection and assessment of diet that could be reasonably expected over 10 years of follow-up. Second, during the study period we did not collect detailed data on cooking techniques to more directly estimate heterocyclic amine intake.30 Third, we did not characterize all of the numerous NAT2 variant alleles of varying functional significance,31 and we classified genotypes into 2 categories of rapid and slow acetylator status. Phenotypic studies have suggested further variation in acetylator status within individual genotype categories.7 However, within the confines of our sample size, we focused on the most common alleles in a Caucasian population that have been phenotypically validated32 and classified the most common genotypes according to the methodology of previous studies.23 Moreover, misclassification of genotype would have tended to bias the observed relationships toward unity. Fourth, it is possible that this nested cohort may not be entirely representative of the larger study population. In particular, there were a relatively small number of women in the highest category of meat intake and we did not observe a strong dose-relationship between meat intake and cancer risk as we did in our prior study of the larger cohort.1 However, the prevalence of other recognized colorectal cancer risk factors in our nested cohort is similar to that of the larger cohort,21 and the larger study was conducted at an earlier time period when meat consumption was higher among study participants (data not shown).

Finally, we did not consider other polymorphic enzymes other than NAT2 (e.g., NAT1, CYP1A2 and GSTM-1) hypothesized to influence the metabolism of xenobiotic exposures.7, 33 Although a few, but not all, studies have suggested a role for the combined effect of these enzymes,8, 10, 33, 34 we did not have sufficient power to examine such gene-gene interactions. Moreover, we focused on NAT2 because there are stronger data on genotype-phenotype correlations and animal models of acetylator status than are presently available for the other enzyme pathways.11

The growing potential to test a large number of associations between genetic variants and disease poses considerable challenges in the application and interpretation of standard statistical techniques. Many initial reports of significant genetic associations ultimately are further assessed to be false-positive findings.35 Although there is no widely accepted alternative statistical approach with which to evaluate genetic associations, particularly when assessing gene-environment interactions,36 we believe our results meet several important criterion: first, our results replicate, in an independent sample, the findings of previously well-designed, prospective studies; second, there is a strong biological a priori hypothesis supporting our results; and third, the genetic variants on which we focused have established physiological correlates.37 Although additional large, prospective studies of gene-environment interactions and colorectal cancer would be valuable, obtaining significantly larger sample sizes may not be feasible within a reasonable time frame. Alternatively, efforts at pooling data from several large cohorts is ongoing38, 39 and may prove particularly informative given the methodological limitations of existing meta-analyses.40

In summary, these prospective results support an association between high intake of red meat and colorectal cancer risk, particularly among genetically susceptible individuals. The influence of NAT2 genotype on this association supports the hypothesis that the effect of red meat on colorectal carcinogenesis may be mediated, in part, by the activation of carcinogens.

Acknowledgements

The authors acknowledge the continued dedication of the participants in the Nurses' Health Study. Supported by grants (HL 34594, CA 87969 and CA 55075) from the National Institutes of Health and funds from the National Colorectal Cancer Research Alliance. Dr. Chan is a recipient of the American Gastroenterological Association/Foundation for Digestive Health and Nutrition Research Scholar Award. Dr. Tranah receives support from CA 09001-27 from the National Institutes of Health.

Ancillary