Exploring the distribution of genetic markers of pharmacogenomics relevance in Brazilian and Mexican populations

Abstract

Studies of pharmacogenomics-related traits are increasingly being performed to identify loci that affect either drug response or susceptibility to adverse drug reactions. However, the effect of the polymorphisms can differ in magnitude or be absent depending on the population being assessed. We used the Affymetrix Drug Metabolizing Enzymes and Transporters (DMET) Plus array to characterize the distribution of polymorphisms of pharmacogenetics and pharmacogenomics (PGx) relevance in two samples from the most populous Latin American countries, Brazil and Mexico. The sample from Brazil included 268 individuals from the southeastern state of Rio de Janeiro, and was stratified into census categories. The sample from Mexico comprised 45 Native American Zapotecas and 224 self-identified Mestizo individuals from 5 states located in geographically distant regions in Mexico. We evaluated the admixture proportions in the Brazilian and Mexican samples using a panel of Ancestry Informative Markers extracted from the DMET array, which was validated with genome-wide data. A substantial variation in ancestral proportions across census categories in Brazil, and geographic regions in Mexico was identified. We evaluated the extent of genetic differentiation (measured as FST values) of the genetic markers of the DMET Plus array between the relevant parental populations. Although the average levels of genetic differentiation are low, there is a long tail of markers showing large frequency differences, including markers located in genes belonging to the Cytochrome P450, Solute Carrier (SLC) and UDP-glucuronyltransferase (UGT) families as well as other genes of PGx relevance such as ABCC8, ADH1A, CHST3, PON1, PPARD, PPARG, and VKORC1. We show how differences in admixture history may have an important impact in the distribution of allele and genotype frequencies at the population level.

Figure 1. Principal Component (PC) analysis.

The first two PCs were plotted. The plots are based on data for 1,647 SNPs available for two HapMap samples (EUR, AFR), Black, Brown, White Brazilians (BZ.BK, BZ.BN, BZ.WT) and Mexican Natives and Mestizos (NAT, MEX).

Figure 2. Surface plots describing the predicted relationship between the distribution of rs2470890 and parental ancestry, obtained by fitting piece-wise smooth logistic regression models to the 268 Brazilians and 224 Mexicans data.

The relative proportions of Native American and African ancestry are plotted in the x and y axes as appropriately labeled, whereas the relative proportion of European ancestry is inferred as the remaining proportion: 1- (af+nat). The plotted surface corresponds to the natural logarithm of the odds of having the variant rs2470890 T allele, depending on the relative admixture proportions of the parental populations. The odds refer to the ratio of the variant:wild-type alleles. For example the odds of having a variant allele frequency of 0.1 (wild-type frequency = 0.9) is 01/09 or 0.111. The corresponding Ln odds is −2.197. The circles correspond to the average ancestral proportions for Black Brazilians (black circle), Brown Braziilans (grey), White Brazilians (white), Mexicans (red), EUR (yellow), AFR (green) and NAT (blue).

Figure 3. Surface plots describing the predicted relationship between the frequency of rs1135840 and parental ancestry, obtained by fitting piece-wise smooth logistic regression models to the 268 Brazilians and 224 Mexicans data.

The relative proportions of Native American, African ancestry and European ancestry are presented as described in Figure 2. The plotted surfaces correspond to the natural logarithm of the odds of having the variant rs1135840 G allele, depending on the relative admixture proportions of the parental populations. The circles correspond to the average ancestral proportions for Black Brazilians (black circle), Brown Braziilans (grey), White Brazilians (white), Mexicans (red), EUR (yellow), AFR (green) and NAT (blue).

Figure 4. Surface plots describing the predicted relationship between the frequency of rs9923231 and parental ancestry, obtained by fitting piece-wise smooth logistic regression models to the 268 Brazilians and 224 Mexicans data.

The relative proportions of Native American, African and European ancestry are presented as described in Figure 2. The plotted surfaces correspond to the natural logarithm of the odds of having the variant rs9923231 G allele, depending on the relative admixture proportions of the parental populations. The circles correspond to the average ancestral proportions for Black Brazilians (black circle), Brown Brazilians (grey), White Brazilians (white), Mexicans (red), EUR (yellow), AFR (green) and NAT (blue).

Figure 5. Frequency distributions of DMET Plus markers by region within Mexico.

We analyzed the distribution of the DMET Plus markers by region. CPIC marker rs12248560 shows much lower frequency for those states with stronger Native American component (GUE, VER) almost 4-fold when compared to that of SON. Whilst, rs1135840 shows that frequency for the state of SON is almost 2-fold when compared to that of GUE, a state with a high Native American component. This also correspond to what is observed at the ancestral population levels, over 4-fold difference between European and Native American.