Genetic Polymorphisms of N-Acetyltransferases 1 and 2 and Gene-Gene Interaction in the Susceptibility to Childhood Acute Lymphoblastic Leukemia¹

ALL³is the most frequent form of cancer affecting children. As a sporadic cancer, it can be considered as a complex disease in which the effect of a series of low penetrance genes, modulated by external factors,modify the individual’s risk of cancer (1). Despite many efforts, little is known about leukemogenesis, particularly with respect to genetic susceptibility and environmental factors (2, 3, 4, 5). Individuals having a modified ability to metabolize carcinogens seem to be at increased risk of cancer (see review by Perera; Ref. 6). Furthermore, infants and children may be at greater risk than adults from a variety of environmental toxicants because of differential exposure and/or physiological immaturity (7, 8, 9). Therefore, functional polymorphisms in genes encoding carcinogen-metabolizing enzymes may have relevance in determining susceptibility to pediatric cancer. Recently, we showed that the GSTM1 null and CYP1A1*2A genotypes were both significant predictors of ALL risk in children (10), which suggests that polymorphisms in genes that encode carcinogen-metabolizing enzymes may indeed play a role in leukemogenesis.

N-acetyltransferases 1 (NAT1) and 2 (NAT2) are conjugating enzymes involved in the metabolism of aryl- and heterocyclic amines (11, 12). Genetic polymorphisms that have been described in both NAT1 and NAT2 genes correlate with biochemical phenotypes ranging from slow to fast acetylators (11, 12, 13, 14, 15, 16). Associations have been reported between these DNA variants and the risk of a number of cancers including head and neck, lung, breast, laryngeal and bladder cancers (14, 17, 18, 19, 20, 21) as well as colorectal carcinomas (16, 22). Interestingly, both fast and slow NAT2 acetylators were shown to represent susceptibility factors in different carcinomas (13, 14, 19, 21, 22, 23, 24, 25). Such a dual effect is presumably attributable to differences in biochemical pathways of carcinogen activation in the liver and in other organs that become eventually affected (14, 21, 22). The prevalence of acetylator variants varies remarkably among different populations (11, 12). Therefore, special care has to be taken to avoid interpretation errors attributable to population heterogeneity. We recently proposed that French-Canadians who represent a population founded in the 17th century by immigrants from France (26, 27) constitute an appropriate genetic model for such genetic epidemiology studies (10).

Here we report a case-control study on the relationship between DNA variants in NAT1 and NAT2 genes and the susceptibility to childhood ALL in the French-Canadian population from the Province of Quebec, Canada. We also examined the effect of the combined multilocus genotypes on childhood ALL susceptibility.

Childhood ALL patients (n = 176) were diagnosed in the Division of Hematology-Oncology of Ste-Justine Hospital, Montreal,Canada, between August 1988 and September 1997. They comprised 108 males and 68 females with a median age of 6.0 years. The distribution of ALL subtypes was as follows: 142 pre-B and 21 T-cell ALLs and 13 with undetermined lineage. A control group (n = 306)was randomly selected from a large institutional DNA bank. All of the participants were of French origin and resided in the Province of Quebec, Canada. The inclusion criteria for patients and controls have been described previously (10). The Institutional Review Board approved the research protocol, and informed consent was obtained from all of the participating individuals and/or their parents.

DNA was isolated from buccal epithelial cells, peripheral blood, or bone marrow in remission as described in Baccichet et al.(28). All of the selected polymorphisms in NAT1(C559T, G560A, T640G, T1088A, and C1095A) and NAT2 (T341C,C481T, G590A, A803G, and G857A) were revealed by allele-specific oligonucleotide (ASO) hybridization assays (29). In some cases the NAT1 and NAT2 genotypes were confirmed by PCR-RFLP as described in Deitz et al. (30)and Cascorbi et al. (14), respectively. The alleles defined by these polymorphisms are given in Fig. 1,2.

The test for case-control differences in the distribution of the genotypes was based on χ² statistics. The level of significance was calculated by Fisher’s exact test. ORs were used to measure the strength of association between the tested genotypes and ALL risk. Crude ORs are given with 95% CIs. Unconditional logistic analysis was used to compute age and gender as covariables as well as the effect of combined genotypes at risk of ALL susceptibility. All of the statistical tests were based on two-tailed probability and were performed using SPSS version 7.5. Linkage disequilibrium between NAT1 and NAT2 alleles was tested with the software package GenePop (version 3.1).

The genotypes at 10 polymorphic sites in NAT1 and NAT2 genes (Fig. 1,2) were obtained for 176 children with ALL and 306 healthy controls, both groups consisted of French-Canadians residing in the Province of Quebec, Canada. The selected polymorphisms define the most common allelic variants found in populations of European descent (11, 31). Some individuals were not successfully genotyped for the whole set of tested polymorphisms, thus explaining a certain variation in the total number of samples listed in Tables. In this study, both pre-B and T-cell ALLs were considered together because no significant differences were observed in terms of the tested genotypes(data not shown).

The frequency of NAT2 alleles as well as the distribution of the genotypes in ALL patients and controls are given in Tables 1,2, respectively. The observed allelic frequencies were similar to those reported in other populations of European descent (11, 29). Of note, we did not observe linkage disequilibrium between NAT1 and NAT2 loci as reported elsewhere (31). The allele frequency distribution among ALL patients differed from the control group: NAT2*5C and *7Bwere overrepresented in the patient group, whereas NAT2*4was underrepresented (Table 1). When genotypes were determined, we found that those predicting a slow NAT2 acetylator phenotype (so-called NAT2 slow-acetylation genotype), i.e., two copies of any “slow” allele (*5A, *5B, *5C, *6A, or *7B), was present in 64.8% of the patients as compared with 54.6% of controls (Table 2). These results suggest that slow-acetylation activity of NAT2 is associated with an increased risk of ALL (OR, 1.5; 95%CI, 1.0–2.2; P, 0.03). At the same time, a significant decrease in NAT2 rapid-acetylation genotypes (presence of at least one copy of allele *4 or *12A) among cases pointed to a protective role of NAT2 activity (OR, 0.7; 95% CI, 0.4–1.0; P, 0.03; Table 2), particularly in individuals homozygous for the NAT2*4 allele (Table 2). Multivariate analysis that included age and gender as covariables did not change the interpretation of these results.

The NAT1 genotyping data for the patient and the control group are reported in Tables 3,4. A modest increase of NAT1*4 allele is observed among ALL patients (OR, 1.4; 95% CI, 1.0–1.9; P, 0.06) compared with the controls (Table 3), which could explain a higher, although not significant, prevalence (OR, 1.4; 95% CI, 0.9–2.1; P, 0.1)of patients homozygous for this allele (Table 4). The patients and controls were also arbitrarily grouped based on the presence or absence of the putative rapid allele *10 (16). The frequency of NAT1*10 carriers and noncarriers did not differ between cases and controls (Table 4), which suggests that the NAT1 acetylation genotype is not independently related to overall risk of childhood ALL. Involvement of NAT1 and NAT2 enzymes in similar metabolic pathways prompted us to examine the effect of the combined genotypes for these loci (Table 5). An increase in the risk of ALL (OR, 1.9; 95%CI, 1.1–3.4; P, 0.03) for individuals with NAT2 slow acetylators that are homozygous for the NAT1*4 allele compared with the risk for individuals without these genotypes was observed.

From a previous study of the same case-control group, we reported that children carrying the GSTM1 null genotype or at least one CYP1A1*2A allele had an increased risk of ALL (10). Here, we investigated whether this risk was increased further by additionally considering NAT2genotypes. The reference group (OR, 1.0) was defined as children having the following “low-risk” genotypes: NAT2 rapid acetylation, presence of GSTM1, and absence of CYP1A1*2A allele. The estimated risk of ALL for all of the combinations of genotypes at risk is given in Table 6. The presence of only one risk-elevating genotype did not influence the risk. However, the overall risk of ALL increased with the number of risk-increasing genotypes (P for trend, 0.0001), with ORs of 2.7 (95%CI, 1.4–4.9; P, 0.002) and 3.1 (95%CI, 1.1–8.4; P, 0.03) for two and three genotypes, respectively (Table 6).

ALL in children offers a unique opportunity to examine the effect of carcinogen-metabolism genes in the risk of pediatric cancers (10). The young ages of patients and, thus, a short latency period between the appearance of the initiating mutation and the detection of tumor cells should facilitate the identification of risk factors, as compared with adult cancer patients in whom many factors come into play because of long latency periods. Here, we determined the frequencies of NAT1 and NAT2allelic variants in French-Canadians from Quebec, Canada. This population was founded by a few thousand immigrants from France in the 17th century (26, 27). After the British conquest in the middle of 18th century, the flow of French immigrants practically ceased. However, because of a large demographic growth, this population grew naturally and expanded to about 10 million in Quebec and elsewhere in North America. Despite the founder effect (26), the overall frequencies of the genotypes tested in controls agree with those reported for other populations of European descent (11, 12, 13, 14, 16, 29, 32), which suggests that the implications of our results could be extended to these groups as well.

We found that children carrying NAT2 slow-acetylation genotypes were at increased risk of developing ALL (OR, 1.5; 95% CI,1.0–2.2), mainly because of the overrepresentation of alleles NAT2*5C and *7B. We observed a lower prevalence of the NAT2*4 allele in the patients’ group, which may suggest a protective role for this allele. The effect was more obvious in homozygous individuals as it was also reported by others (14). To our knowledge, this is the first study documenting an association between NAT2 variants and the risk of hematological malignancies, particularly in children. However,these data are in agreement with the studies of solid neoplasias in adults that report that NAT2 slow-acetylator individuals are at greater risk of head and neck, breast, and laryngeal cancers (17, 18, 19) as well as bladder cancer (20, 21). When NAT1*10 was considered alone or combined with NAT2 slow-acetylation genotype, our data provided no evidence of involvement of this allele in ALL susceptibility. In other studies, NAT1*10 has been associated with an increase of N-acetyltransferase activity (16), higher levels of DNA adducts (33), and an elevated risk of colon, bladder, or breast carcinomas (16, 23, 32). The risk of cancer was accentuated when NAT1*10 genotypes were combined with NAT2 at-risk genotypes (16, 32). However, in light of conflicting data in the literature, the role of NAT1*10 in tumorigenesis still remains unclear (34, 35, 36). On the other hand, we noted a slight increase in the frequency of NAT1*4 among cases, which, when combined with NAT2 slow-acetylator genotypes, further increases the risk of ALL (OR, 1.9; 95% CI, 1.1–3.4). Taken together,the results for NAT1 should be interpreted with caution because the functional significance of NAT1 variants remains to be clarified.

It is difficult to predict how the genotypes at risk will modify the infant’s (or fetus’) response to different exposures. Epidemiological studies have led to the suggestion that in utero and postnatal exposures to various biological, physical and chemical factors may be important determinants of childhood ALL (7, 37, 38). Children may be at greater risk than adults from toxic substances because of differential exposure or physiological immaturity (7, 9, 19). We have recently shown that the risk of ALL among children carrying certain CYP1A1 variants was modified by maternal exposure to pesticides during the pregnancy (39). Both NAT1 and NAT2 enzyme activities are detectable in human placentas (40, 41), although it seems that placental N-acetylation activity is predominantly attributable to NAT1 (11, 40, 41, 42). Thus, it is possible that genetically determined variability in N-acetylation of aromatic amines present in different environmental and occupational pollutants like dyes, tobacco smoke, antioxidants, medications, or pesticides may confer cancer susceptibility on children (13, 14, 19, 21, 22, 23, 24, 25). Predominance of the competing N-oxidation in subjects with NAT2 slow acetylation would lead to the formation of aryl hydroxylamines. Because NAT1 is widely expressed in different tissue (11, 43), it can act locally, performing O-acetylation of hydroxilated arylamines, which will result in the formation of arylnitrenium ions and DNA adducts (44).

Recently, we reported that the GSTM1 null genotype and the presence of at least one CYP1A1*2A allele were significant predictors of ALL risk (10). When these two genotypes were combined with genotypes predicting NAT2 slow acetylators, we found that the presence of only one risk-elevating genotype had little effect on OR estimates. However, the combination of risk-elevating genotypes seemed to confer an increased risk of ALL among the carriers compared with those with the more beneficial genotypes (OR, 2.7; 95% CI,1.4–4.9; and OR, 3.1, 95% CI, 1.1–8.4, respectively). These results are similar to those obtained in nonhematological malignancies for the combined effect of NAT2 and GSTM1 (19, 25, 45). However, the basis of synergy between CYP1A1 and NAT2 is less obvious because they are involved in different metabolic pathways. This effect could be mediated through CYP1A2 that is involved in the first step of the activation of the aromatic amines (13). It has been recently shown that CYP1A2 activity was higher in individuals with GSTM1 null genotype or heterozygous for CYP1A1 variants compared with individuals with GSTM1 or no CYP1A1 polymorphisms (46). Consequently, slow acetylators may not be able to compete with high activity of CYP1A2 for aromatic amines.

In this study, the simultaneous analysis of several loci suggests that the combination of risk-elevating genotypes is more predictive of ALL risk than when they are taken independently. In other words, the etiology of childhood ALL cannot be explained by allelic variability at a single locus probably, because of the complexity of xenobiotics metabolism and the diversity of chemicals to which individuals are exposed.

We thank all of the patients and control subjects who participated in this study as well as the physicians and the clinical staff for their collaboration and Ulrike Brockstedt for critical reading of the manuscript. D. S. is a scholar of the Fonds de la Recherche en Santé du Québec.