Age-associated Risk of Cancer among Individuals with N-Acetyltransferase 2 (NAT2) Mutations and Mutations in DNA Mismatch Repair Genes¹

NAT2³is a polymorphic isozyme of N-acetyltransferase and is found in a variety of tissues including the colorectal mucosa (1, 2). It catalyzes the metabolism of xenobiotics and carcinogens by transferring an acetyl group to these agents. Phenotypic variation in the rate at which the acetylation of these agents occurs is attributable in part to the polymorphic nature of the NAT2gene (3).

We assessed the role of NAT2 as a modifier gene in HNPCC, an autosomal dominant disorder accounting for 3–14% of all of the cases of CRC (4, 5, 6). HNPCC has been associated with germ-line mutations in DNA MMR genes, specifically hMSH2, hMLH1, hMSH6,hPMS1, and hPMS2; hMLH1 and hMSH2 are the most commonly mutated (7, 8).

Testing for mutations in the NAT2 gene by PCR assays has made it possible to predict the rate at which drugs and chemicals containing primary aromatic amine or hydrazine groups will be metabolized (9). Humans can be classified as either slow or rapid acetylators, based on a DNA amplification assay developed by Bell et al. (10) that was 100%concordant with rapid acetylator phenotype (as measured by caffeine metabolite excretion) and 90% concordant overall.

Results from previous studies of the association between NAT2 genotype and CRC have been mixed; some groups(11, 12) have reported a positive association between fast acetylator phenotype and risk of CRC, whereas another report(13) did not. The aim of our study was to determine the extent to which polymorphisms in the NAT2 gene affect risk for CRC and other cancers among individuals with DNA MMR mutations. We hypothesized that some of the observed variation in time to onset of CRC among carriers of mutations in DNA mismatch repair genes may be because of genetic variation in the NAT2 gene at these polymorphic loci. Using Kaplan-Meier product limit estimates and Cox proportional hazards modeling, we compared time to onset of CRC and HNPCC-related cancers between carriers and noncarriers of mutations in NAT2, all of whom were predisposed to colon cancer as a result of inheriting mutations in the DNA MMR genes hMLH1, hMSH2, or hPMS1.

The patients selected for DNA MMR mutation testing came from either the M. D. Anderson Hereditary Colorectal Cancer Registry or a consecutive series of CRC patients evaluated at the M. D. Anderson Cancer Center. The patients in the registry came from HNPCC or HNPCC-like families or were very young (<45 years of age) at CRC diagnosis. The consecutive series of CRC patients included all of the newly registered patients with adenocarcinoma of the colon or rectum evaluated at the M. D. Anderson Cancer Center during an 11-month period beginning in September 1994. From these two sources, we identified 86 individuals with germ-line mutations in hMLH1or hMSH2. Of these, 43 were index cases, 37 were first-degree relatives of those cases, and the other 6 were more distant relatives. Of the 86 carriers, 49 were affected with CRC. There also was one case each of breast, cervical, uterine, and oral cancer,one unspecified malignant neoplasm of the female genitalia, and one malignant neoplasm of unspecified origin.

Blood was drawn from each study subject in Vacutainer tubes containing EDTA (Becton Dickinson Vacutainer System; Becton Dickinson, Rutherford,NJ). DNA was isolated from the blood with a 341 Nucleic Acid Purification System (Applied Biosystems, Foster City, CA) according to the manufacturer’s instructions.

The DNA was subjected to PCR with the primers for each exon of the hMLH1 and hMSH2 genes used by Wijnen et al. (14, 15), except that the GC clamp and M13 sequences were not included. PCR was performed on 500 ng of DNA in a 20-μl reaction mixture of 50 mm KCl; 10 mm Tris-HCl (pH 8.3); 1.5 mm MgCl₂; 0.2 mm dATP, dGTP, and dTTP; 0.1 mm dCTP; 20 pmol of each primer; 1 μCi of[³²P]dCTP (3000 Ci/mmol); and 1.0 unit of Taq polymerase (Perkin-Perkin-Elmer Corp., Norwalk, CT). Each PCR mixture was incubated at 94°C for 3 min and then subjected to 25 cycles of 94°C for 30 s, 58°C for 30 s, and 72° for 1 min. A 3-min final extension was performed at 72°C. For heteroduplex analysis, 4 μl of each PCR product was mixed with equal volumes of loading buffer containing 95% formamide, 20 mmEDTA, 0.05% xylene cyanol, and 0.05% bromphenol blue. This was heated at 94° for 10 min, slowly cooled to room temperature overnight, and subjected to electrophoresis on Mutation Detection Enhancement gels (J. T. Baker, Phillipsburg, NJ) as described previously by Jeon et al. (16). For SSCP analysis, the same procedure was followed except that after heating at 94°C, the DNA was promptly chilled on ice and then loaded onto the gel. The gels were then vacuum dried and subjected to autoradiography overnight at −80°C. Nucleotide sequence analysis was performed using one primer that was used to generate the PCR product.

Once an altered banding pattern was detected, exon-specific nucleotide sequence analysis was performed to determine the sequence. To accomplish this, the exon was subjected to PCR as described above except that the [³²P]dCTP was omitted. Before nucleotide sequencing, the PCR products were cleaned up by mixing them with 20 units of exonuclease 1 and four units of shrimp alkaline phosphatase followed by incubation at 37°C for 15 min and then at 85°C for 15 min to remove the unused primers and residual deoxynucleotide triphosphates. The products were then subjected to electrophoresis in 1.5% agarose gels in 1 × Tris-borate EDTA. Ethidium bromide was used to visualize the PCR products, verify the fragment length, and determine the concentration. The DNA sequences of PCR products were then determined using an Applied Biosystems model 377 sequencer.

The subjects were genotyped by PCR followed by SSCP analysis. Briefly,the PCR reaction conditions are the same as described above with the exception of the primers. The PCR primers used were: 2590F(5′-GGACCAAATCAGGAGAGAGCAG 3′) and 2590R (5′-GTTGGAGACGTCTGCAGGTATG-3′)for NAT2*5; 2857F (5′-GAAGAGGTTGAAGAAGTG-CTG-3′) and 2857R(5′-GTTGGGTGATACATACACAAGG-3′) for NAT2*6; and 2481F(5′-AAGGATCAGCCTCAGGTGCCTT-3′) and 2481R(5′-CTGCTCTCTCCTGATTTGGTCC-3′) for NAT2*7. The PCR was performed for 10 min at 94°C followed by 28 cycles of 94°C for 30 s, 65°C for 30 s, and 72°C for 10 min. SSCP analysis was performed as described above. The identities of the NAT2 alleles were confirmed by automated sequencing of several samples with an Applied Biosystems Model 377 sequencer as described above.

To assess differences in time to onset of cancer for individuals of different NAT2 genotypes, we used survival-analysis procedures (17). We calculated the median time to onset for cancer from the Product-Limit Kaplan-Meier estimator. To test for differences in the time to onset in individuals of different NAT2 genotypes, we used Cox proportional hazard regression analysis (18). The estimates from Cox modeling were exponentiated to give an estimate of the increased risk conferred by having mutant genotypes at each of the three NAT2 loci. Alleles NAT2*5, NAT2*6, and NAT2*7were assigned as indicator variables, and the wild-type genotype was used as reference. The time to onset to cancer for individuals in the same family may be correlated because of unmeasured covariables such as shared household environment and shared dietary exposures or because they share the same mutation. To allow for this possible correlation,we used the cluster function of S+ in the Cox proportional hazard model that we fitted. This action corrected the SEs of the robust estimates for familial correlation using a sandwich estimator of the variance. We analyzed the fast and slow acetylators of the NAT2 phenotypes by the same approach (19).

Twenty-eight of the subjects had mutations in hMLH1, 57 had mutations in hMSH2, and 1 patient had a mutation in hPMS1. These 86 individuals were analyzed by NAT2genotype and age of onset for CRC and all of the cancers (Table 1 and Table 2). No subjects were homozygous for the NAT2*7 mutant allele. In multivariate analysis of NAT2*5, NAT2*6, and NAT2*7, we found significantly increased risks for CRC among NAT2*5(P = 0.0034) and NAT2*7(P = 0.012) heterozygotes with HRs of 2.38 and 2.96, respectively. When the outcome was all of the cancers, only NAT2*7 heterozygotes had a significantly elevated HR of 3.37(P = 0.00004). The median times to onset for NAT2*7 mutants for both CRC and all of the cancers were decreased compared with wild-type NAT2, but consistent effects were not seen for NAT2*5 and NAT2*6. Subjects heterozygous for the NAT2*7 mutant allele had increased HRs for both CRC (HR,2.1; P = 0.053) and for all of the cancers(HR, 2.8; P = 0.00001) compared with wild-type homozygotes (Table 3).

Of the 86 individuals tested, 38 (47%) had two of any of the three mutant NAT2 alleles and, therefore, were classified as slow acetylators. The remaining 44 (53%) had only one or no mutant alleles and were classified as rapid acetylators. We used Cox proportional hazard regression of risk of CRC and cancer stratified by acetylator phenotype. When the rapid acetylator genotype was used as the reference genotype, the risk for CRC in slow acetylators versus rapid acetylators was not significantly different (HR, 1.53; 95% CI,0.087–2.69; P = 0.14). Similarly, the risk for all of the cancers in slow acetylators versus fast acetylators was not significantly different either (HR, 1.14; CI,0.69–1.91; P = 0.61).

Some studies have indicated that slow acetylators have a lower risk of CRC (11, 12). In contrast, a recent study by Heinimann et al. (20) suggested an increased risk of cancer among slow acetylators with HNPCC, who had CRC. The NAT2 genotype was determined for 26 unaffected hMLH1/hMSH2 mutation carriers and 52 with cancer in 21 Swiss HNPCC families. Slow acetylators were significantly more prevalent among affected MMR mutation carriers than among unaffected mutation carriers. In our study on HNPCC, we did not observe an increase in prevalence of slow acetylators or fast acetylators among the mutation carriers with cancer. However, we did find that the NAT2*7mutant allele of NAT2, which confers the slow acetylator phenotype, increased the risk for CRC and all of the cancers in HNPCC carriers. We also found that the NAT2*5 mutant allele of NAT2 increased the risk for CRC in heterozygotes, although the level of significance was not so great as for NAT2*7.

The different conclusions regarding the association of slow acetylation phenotypes in different populations might be because of different genetic and environmental influences. Patients with HNPCC may be more sensitive to certain environmental influences than subjects without MMR gene defects. The detrimental effects of such environmental factors might be enhanced by the NAT2*7 mutant allele in HNPCC while having little or no effect in other subjects.

Our findings also suggest that there may be differences in the slow acetylation phenotypes produced by different NAT2 genotypes,perhaps because of variation in the enzyme activity depending on the substrate. A recent review by Hein et al. (21)discusses the complexities of assessing phenotype with the different NAT2 genotypes. They point out that multiple mechanisms for reduction in N-acetyltransferase activity are associated with various nucleotide substitutions present on NAT2alleles and that the ability to distinguish acetylator phenotypes is complex and is a function of sensitivity and specificity of the phenotyping method.

The different results could also be because of different NAT2*5, NAT2*6, and NAT2*7 mutant allele frequencies in different populations. In the group of subjects studied by Heinimann et al. (20), the frequency of the NAT2*6 mutant allele is much lower than in our population. Our study suggested that although the NAT2*5, NAT2*6, and NAT2*7 mutant alleles may all confer the slow acetylation phenotype, only NAT2*5 and NAT2*7 were associated with increased risk for cancer. Therefore, the higher frequency of the NAT2*6 mutant alleles(which were not associated with increased risk) in our study may have reduced the actual effects of slow acetylation on cancer risk.

How the NAT2 mutant alleles modify MMR defects to increase age-associated risk for cancer is not clear. A more complete understanding is needed of the tissue-specific action of NAT2 and the drugs and compounds it metabolizes in the colon. Additional studies that include environmental exposures in NAT2 mutation analysis might increase our knowledge of how variation in the NAT2 gene affects time to onset and risk of CRC in carriers of mutations in DNA MMR.

We thank Dr. Maureen Goode for her editorial comments.