Racial disparities for colon cancer cannot be explained solely by socioeconomic, behavior, environmental, lifestyle factors, or genetic factors, rather etiologic explanations for these disparities may include interactions between all of these factors [41
]. Using data from a population-based case-control study, we observed statistically significant interaction between meat-derived HCA intake and NAT1 genotype, regardless of race. In addition, our data suggest that NAT1 acetylation of HCAs differs by race, where PhIP increased risk of colon cancer among African Americans with “fast” acetylation, and MeIQx increased risk among whites with “slow” acetylation.
Although genetic variation is greater within race/ethnic groups than between, individual single nucleotide polymorphisms may occur uniquely within specific race/ethnic populations resulting in allele frequencies that differ between race/ethnic populations and could account for racial differences in disease susceptibility [42
]. To test this hypothesis we measured NAT1 and NAT2 polymorphisms in a population-based sample of African Americans and whites. We observed a greater frequency of the NAT1*10 and NAT2-rapid/intermediate genotypes among African-American controls, than among white controls. NAT genotype frequencies observed in our study population were similar to other case-control studies in the U.S. [43
]. Also similar to our findings, in a study of black South Africans, a greater frequency of NAT2 high activity alleles, compared to whites. Black South Africans have among the lowest incidence of colon cancer in the world [45
], suggesting that the magnitude of association between NAT2 rapid genotypes and colon cancer may be lower in populations of African versus European decent. In support of this hypothesis, we observed a weak positive association for colon cancer with the NAT2-rapid/intermediate genotype versus NAT2-slow, among whites but not African Americans. In addition no association was observed for NAT1*10, regardless of race. Our findings among whites were consistent with previous epidemiologic studies that reported null to weak positive associations for NAT1*10 versus NAT1-non*10, and NAT2-rapid/intermediate versus NAT2-slow and colon cancer [19
], with some exceptions from small studies [34
We observed that NAT1 genotype differentially modified the association between dietary sources of HCA exposure and colon cancer by race, where positive associations were present among African Americans with the NAT1*10 genotype, and among whites with the NAT1-non*10 genotype. Although NAT1 polymorphisms contribute to differential N
- and O
-acetylation of HCAs [15
] and NAT1 activity is greater than NAT2 in the colon [49
], few epidemiologic studies have evaluated NAT1 as a potential modifier of meat and colorectal cancer associations [50
]. From a case-control study in Germany, Lilla, et al.
reported stronger positive associations with frequency of red meat consumption among NAT1*10 than among NAT1-non*10, but interaction was not statistically significant [50
]. In a small (N=102 colorectal cases) case-control study, interaction was not observed between red meat and NAT1 genotype, but this may be due to inadequate power [51
There is epidemiologic evidence for differential effects of acetylation by race. For example, Probst, et al.
, reported an inverse association between NAT2 “rapid” genotypes and colorectal adenomas among African Americans, but an increased risk among whites [52
]. Mechanistic explanations for differences by race include the differential effect of individual NAT alleles on HCA acetylation [53
]. Our data support this explanation, since we observed differences in NAT allele frequencies by race. Another mechanistic explanation is that genetic variation by race exists in other genes that are relevant for HCA metabolism, such as GSTs [54
] and CYP1A2 [12
] that are upstream from NAT. Epidemiologic support for this explanation has been observed where increased risk with well-done red meat only among those with rapid CYP1A2 and NAT2 in a case-control study of colorectal cancer [55
Another explanation for our finding includes the effect of unmeasured heterogeneity in other xenobiotic metabolism genes and/or other environmental factors that may be related to differences in cancer risk between race/ethnic groups [56
]. For example, functional data suggest that there was differential mutagenic potential for PhIP depending on genetic variability [57
]. One likely suspect is differential glucuronidation of the PhIP intermediate, where greater activity makes the intermediate unavailable for further activation and eventually DNA damage [58
]. We have previously reported interaction between UDP-glucuronosyltransferase (UGT) 1A7 and DiMeIQx, although differences by race were not observed [32
Our findings of no modification by NAT2 genotype for HCA-colon cancer associations supports findings from a recent computational study in which observed differences in mutagenic activity between PhIP and MeIQx intermediates were not related to acetylation with NAT2 [59
]. There have been several epidemiologic studies that have investigated meat intake and modification by NAT2 genotype for colorectal cancer [44
]. However, few studies have incorporated information on meat cooking method and/or doneness level [21
]. Most relevant is that our findings support those of Le Marchand, et al.
, where no difference was reported for well-done red meat and colon cancer by NAT2 genotype, because this population-based, case-control study used similar methods of HCA estimation to our study [55
A strength of our study, and a possible reason for previous inconsistent findings of meat-NAT genotype interactions for colorectal cancer, is the use of rigorous methods to estimate dietary HCAs [63
]. High-quality exposure assessment is particularly important in estimating interaction with genetic polymorphisms, such as NAT1 and NAT2. By definition they have low-penetrance and are common in the population, so the association with disease is driven by the exposure and not by the presence of polymorphisms [64
]. However, poor exposure assessment does not automatically result in biased gene-environment interactions [65
Misclassification of gene effects may be a source of bias in these data, because genotype data was used to make assumptions about metabolic activity. Previously, epidemiologic studies have categorized NAT1 genotypes based on the presence of the *10 allele, because it contains a DNA sequence variant at the polyadenylation site which increases mRNA stability and possibly metabolic activity [36
]. However, this categorization for NAT1 may no longer be appropriate [67
]. One reason is that a novel NAT1 allele (NAT1*14A) was identified, and determined to be associated with a 15-20 fold reduction in affinity for p-aminosalicylic, a NAT1-selective substrate [68
], whereas NAT1*10 was not [69
]. The RFLP PCR-based genotyping methods used in this study, can not be used to distinguish between the NAT1*10 and NAT1*14A alleles. It has been estimated that using the RFLP genotyping methods misclassifies about 3% of the NAT1*14 alleles as NAT1*10 [71
], which would result in a bias of the observed odds ratio toward the null, given that NAT1*10 genotype is not associated with diet among controls in these data (data not shown).
Incomplete NAT2 genotype-phenotype correlation may be another source of bias in our study. We examined three NAT2 SNPs, and although strong evidence exists for their correlation with acetylation phenotype [72
], we have misclassified some rapid acetylators as “slow” and visa versa. An analysis of the potential bias introduced in measuring three versus eleven NAT2 SNPs concluded that minimal misclassification and only slight biases in interaction estimates would result [74
Our data indicate modification by NAT1 for HCA and colon cancer associations, regardless of race. Although the at-risk NAT1 genotype differs by race, the magnitude of the individual HCA-related associations in both race groups are similar. Therefore, our data do not support the hypothesis that NAT1 by HCA interactions contribute to differences in colorectal cancer incidence between African Americans and whites.
Reducing risk of colorectal cancer can be furthered with research on the genetic susceptibility of modifiable exposures, including dietary factors [75
]. For example, chemopreventive strategies are currently being investigated to reduce the effects of HCA exposure that activate detoxification pathways in general [76
], or specifically by activation CYP1A2 [77
] and GSTs [78
]. Reducing red meat intake, particularly pan fried meat, and increasing intake of foods known enhance detoxification enzymes remain to be important for reducing risk of colon cancer among both African Americans and whites.