Ethnicity is an important variable that influences an individual's health in several ways, in particular increasing risks for the development of chronic diseases and unresponsiveness or adverse reactions to drug treatment . The influence of the ethnic component in the distribution of NAT2 genetic polymorphism is well established. An example is the ethnic-specific 191A allele, mainly identified in Africans (7-20%) and with lower frequencies in Euro-Caucasian groups (less than 2%) [28–31]. Another example is the 857A allele, mainly identified in eastern Asians .
Following our initial purpose of investigating the frequency of NAT2 SNPs in a Brazilian admixed population, the five most common NAT2 SNPs known from published research were selected. Except for C481T SNP, which does not alter the Nat2 enzymatic function, the other four SNPs are associated with slow acetylator status and had high frequencies in our whole sample as well as among ethnic groups (Table 1 and 2). Although the 191A allele has been described as relatively common in African populations but not in Caucasians, no significant difference was observed between both groups in this study (Table 2). Similar results were obtained for G590A. The ethnic similarity in the distribution of NAT2 SNPs observed in this study could be due to the high degree of admixture between Afro-descendents and Euro-Caucasian groups (mainly Portuguese settlers) that has occurred in Brazil over the centuries since colonization . However, the bias from the self-reported ancestry classification method can not be totally excluded.
Interestingly, meaningful results from the distribution of NAT2 alleles in Amerindian descendents were observed. The 590A allele was significantly more frequent in Amerindians (42.5%) than in White or Afro-Brazilian descendents (19.4% and 20.0%, respectively), even after permutation tests to decrease the risk of a type I error (Table 2). Multiple logistic regression analysis confirmed that Amerindians have the highest frequency of the 590A allele with OR (odds ratio) of 3.714 (95% confidence interval - CI = 0.284-8.545; p = 0.040) (Table 3). Significant difference in the distribution of the 191A allele between Amerindians and other groups was revealed by multiple logistic regression analysis, but not in UNPHASED association. The difference in results may be attributed to the small sample size of 191A carriers among Amerindians. Moreover, the frequency of the 590A allele in Amerindians is higher than what has been reported in studies with Amerindians from Panama (0% and 3.7% in Ngawbe and Embera Amerindians, respectively) [17, 34]. Such unexpected frequency may have originated by phylogeographical differences among Amerindians populations in South-America, miscegenation, and genetic drift.
Despite Fuselli et al. (2007) having found that the NAT2 variants are homogeneously distributed across native populations, the Amerindian sample studied here showed a lower frequency of the 857A allele (10.3%; OR = 0.391; CI= 0.053-2.898; p= 0.003) than those observed in two other Amerindians groups (23.3% and 22.8%)  (Table 3). To date, the frequency of the 857A allele observed in this study is similar to Asian and Central America Amerindian populations , which corroborates the hypothesis that native Americans descend from people who migrated from Siberia thousands of years ago and therefore share their genetic background [36, 37].
To elicit further information about the relationship of SNPs, a haplotype analysis was performed. Although previous studies have shown the efficiency of the PHASE method, we relied on the work of Sabbagh and Darlu (2005), which shows the effectiveness of the EM method for NAT2 haplotype reconstruction and suggests that there is no impact on phenotype prediction compared to results given by PHASE analysis . We observed significant differences in the haplotype structure and frequency among the descendents of the three ethnic groups (Figure 1 and Table 4). Using haplotype analysis based on LD data, a haplotype block between G590A-A803G (Block 1; Figure 1C) was detected in Amerindians but was not found in the other two ethnic groups. This result may help to explain the highest frequency of slow acetylation haplotypes in Amerindians (Table 4). Consistent with the hypothesis that Amerindians may not be under a high selective pressure for fast metabolism, we have previously reported different distribution patterns of GSTP1 low activity polymorphism in this same Amerindian population . Different distributions found in Amerindians, when compared with other groups, may be attributed to their low degree of admixture despite the high degree of miscegenation in the whole population. This occurs for historical reasons related to the particular way Brazil was colonized. In this way, the Amerindian group still maintains its socio-economic distinction that contributes to low degrees of admixture.
Due to our limited sample size, we suggest a careful matching of ethnicity for future larger genetic investigations. Except among the Amerindian descendents, our results suggest that self reported ethnicity might not have significant effects on the distribution of these NAT2 genetic variants studied in the Brazilian population. This data is relevant due to the classic role of Nat2 on isoniazid metabolism in tuberculosis treatment, which still remains an important problem of public health. In fact, several reports indicate that the acetylator status is associated with drug-induced hepatitis and Mycobacterium-resistance [40, 41]. Furthermore, as observed in other phase II metabolizing enzyme polymorphisms, NAT2 genetic variants have been used as a genetic marker in different diseases like bladder and colon-rectal cancers (fast acetylator and slow acetylator, respectively) [42, 43].