To Genotype or Phenotype for Personalized Medicine? CYP450 Drug Metabolizing Enzyme Genotype–Phenotype Concordance and Discordance in the Ecuadorian Population

Abstract

Genetic variations within the cytochrome P450 (CYP450) superfamily of drug metabolizing enzymes confer substantial person-to-person and between-population differences in pharmacokinetics, and by extension, highly variable clinical effects of medicines. In this context, “personalized medicine,” “precision medicine,” and “stratified medicine” are related concepts attributed to what is essentially targeted therapeutics and companion diagnostics, aimed at improving safety and effectiveness of health interventions. We report here, to the best of our knowledge, the first comparative clinical pharmacogenomics study, in an Ecuadorian population sample, of five key CYP450s involved in drug metabolism: CYP1A2, CYP2C9, CYP2C19, CYP2D6, and CYP3A4. In 139 unrelated, medication-free, and healthy Ecuadorian subjects, we measured the phenotypic activity of these drug metabolism pathways using the CEIBA multiplexed phenotyping cocktail. The subjects were genotyped for each CYP450 enzyme gene as well. Notably, based on the CYP450 metabolic phenotypes estimated by the genotype data, 0.75% and 3.10% of the subjects were genotypic poor metabolizers (gPMs) for CYP2C19 and CYP2D6, respectively. Additionally, on the other extreme, genotype-estimated ultrarapid metabolizer (gUMs) phenotype was represented by 15.79% of CYP2C19, and 5.43% of CYP2D6. There was, however, considerable discordance between directly measured phenotypes (mPMs and mUMs) and the above genotype-estimated enzyme phenotypes. For example, among individuals genotypically carrying enhanced activity alleles (gUMs), many showed a lower actual drug metabolism capacity than expected by their genotypes, even lower than individuals with reduced or no activity alleles. In conclusion, for personalized medicine in the Ecuadorian population, we recommend CYP450 multiplexed phenotyping, or genotyping and phenotyping in tandem, rather than CYP450 genotypic tests alone. Additionally, we recommend, in consideration of equity, ethical, and inclusive representation in global science, further precision medicine research and funding in support of neglected or understudied populations worldwide.

Introduction

“Personalized medicine,” “precision medicine,” and “stratified medicine” are related concepts attributed to what is essentially targeted therapeutics and companion diagnostics, aimed at improving safety and effectiveness of health interventions, be they drugs, nutrition, vaccines, or others (Day et al., 2016; McCarthy, 2015). We shall hereafter use the term personalized medicine, which remains as one of the leading frontiers for omics technology and integrative biology applications in the new century.

Among the molecular targets examined with a view to personalized medicine, drug metabolizing enzymes within the cytochrome P450 (CYP450) superfamily are among the most notable, along with the existing and new drug targets (Nassan et al., 2016). Genetic variations within the CYP450 superfamily of drug metabolizing enzymes confer substantial person-to-person differences in pharmacokinetics, and by extension, highly variable clinical effects of medicines preventing personalized medicine. Because populations differ in their genetic background, there is a need for personalized medicine and rational therapeutics research on population genetic diversity of CYP450 enzymes. On the contrary, past research has tended to focus on developed country populations (Day et al., 2016), with relatively fewer works available in resource-constrained settings in the developing world (Singh et al., 2015; Sitole et al., 2014). Worrisomely, there are, therefore, neglected populations that are understudied or overlooked in this context.

The catalytic function of CYP450 enzymes is greatly influenced by person-to-person and between-population differences conferred, in part, by genetic polymorphisms within the genes encoding for these enzymes. Consequently, fully functional alleles, reduced function alleles, or nonfunctional alleles, convey a wide range of drug metabolism activity from ultrarapid to no metabolism (LLerena et al., 2014a; Meyer, 2004). However, their functional phenotype is the result of a complex summary of genetic factors and gene–environment interactions, with the influence of other genes, epigenomic regulation, drug–drug or drug–herb interactions, as well as physiological or environmental regulators (e.g., disease comorbidity states; De Andrés et al., 2013). The functional metabolic phenotypes are usually predicted from genotypes, according to the number of active copies of the gene analyzed (i.e., genotype-estimated drug metabolism phenotype or “the metabolic activity score”; Gaedigk et al., 2008). However, this estimated phenotype is most accurately inferred based on CYP450 genotyping in the case of individuals with null enzyme activity (LLerena et al., 2012, 2014a).

In a previous study by the Ibero-Latino-American Network of Pharmacogenetics and Pharmacogenomics (RIBEF), it was found a lack of correlation between the genotype of the main CYP enzymes and their actual drug metabolic capacity, as only 67% of poor metabolizers (PMs) carried zero active genes, whereas 23.5% of subjects with CYP2D6 gene multiplications were phenotypically ultrarapid metabolizers (UMs), and only 40% of individuals with UM phenotype showed duplications (LLerena et al., 2012). As a consequence, determining actual drug oxidation capacity and evaluating its correlation with genotype are essential to obtain accurate information on drug metabolism from each individual in a given moment. More indeed, these analyses can be carried out in patients under pharmacological treatment, as is the case of previous studies on drug metabolism capacity carried out in patients for whom their metabolic phenotype was determined with the prescribed treatment of thioridazine or risperidone (Berecz et al., 2002; Llerena et al., 2000).

In addition, ethnicity or population-to-population variability is a major factor responsible for highly variable drug metabolism capacity and pharmacokinetics. Although CYP450 genotypes and metabolic phenotypes have been widely studied worldwide (McGraw and Waller, 2012; Sistonen et al., 2009), this information is still scarce in certain populations such as in Ecuador (Dorado et al., 2012a, 2012b, 2014; Sinués et al., 2008; Soriano et al., 2011; Vicente et al., 2014), for whom only two previous studies have been carried out on actual drug metabolizing capacity for CYP2D6 and CYP2C9 (Dorado et al., 2012a, 2012b).

Consequently, the knowledge on the correlation between drug metabolizing capacity and the genetic polymorphism of drug-metabolizing enzymes (De Andrés et al., 2013) has to be considered to regulatory decision-making in certain populations, especially those for which pharmacogenetic data are lacking (De Andrés et al., 2013; Ramamoorthy et al., 2015). Although few studies in the Ecuadorian population have been conducted, they have focused on CYP450 genotypic variation (Dorado et al., 2012a, 2012b, 2014; Sinués et al., 2008; Soriano et al., 2011; Vicente et al., 2014) and only few included phenotypic analysis (Dorado et al., 2012a, 2012b, 2014). Moreover, a comparative head-to-head evaluation of the genotype–phenotype concordance and discordance of the key metabolic pathways for medicines is lacking in the Ecuadorian population.

Within the human CYP450 superfamily, CYP3A, CYP2D6, and CYP2C subfamilies catalyze phase I metabolism of more than 75% of the drugs currently prescribed (Wang and Chou, 2010; Zanger and Schwab, 2013). We report here, to the best of our knowledge, the first comparative clinical pharmacogenomics study, in an Ecuadorian population sample, of five key CYP450s (CYP1A2, CYP2C9, CYP2C19, CYP2D6, and CYP3A4) involved in drug metabolism, using the CEIBA multiplexed phenotyping cocktail.

Materials and Methods

Subjects and study protocol

Healthy unrelated and chronic medication-free Ecuadorian volunteers were included after routine clinical and physical examination was performed. Women who reported or suspected pregnancy were excluded. Volunteers with a history of adverse drug effects with any drug intake 2 weeks before the study was initiated were excluded. Information on alcohol intake, tobacco consumption, and concomitant medication was obtained by direct interview or from their medical record.

The study has been performed in accordance with the Declaration of Helsinki and all participants provided written and informed consent. This research was authorized by the Ethics and Research Committee of the Universidad San Francisco de Quito (Quito, Ecuador).

Genotyping procedure

The QIAmp^® DNA blood kit (Qiagen, Hilden, Germany) was utilized to extract DNA from 5 mL of peripheral blood drawn in a tube with EDTA for each participant. DNA integrity and concentration were evaluated through 1% agarose electrophoresis and spectrophotometry. Quantitative real-time polymerase chain reaction (PCR) using fluorescence-based TaqMan^® assays on a Fast 7300 Real-Time PCR System (Applied Biosystems) was additionally used to analyze the alleles for CYP2C9, CYP2C19, CYP2D6, and CYP3A4, by applying the PCR amplification conditions previously described (LLerena et al., 2014a; Table 1). To detect duplication or CYP2D6 gene, deletion (CYP2D6*5) or duplications were performed by long-range PCR, as indicated elsewhere (Dorado et al., 2005). If multiplications were detected, further analysis on gene copy number characterization was performed by RT-PCR (real-time polymerase chain reaction) (Villagra et al., 2011). Moreover, the CYP1A2*1F allele (−163C>A), was analyzed by PCR-RFLP (restriction fragment length polymorphism-PCR), as described elsewhere (De Andrés et al., 2016). Genotypes were determined by using the allelic discrimination software (Applied Biosystems).

Table 1.

Cytochrome P450 Allelic Variants (Single-Nucleotide Polymorphisms Analyzed) and Their Corresponding TaqMan Assays Utilized for Genotyping by RT-PCR (Real-Time PCR) in the Ecuadorian Population Analyzed

CYP450 gene	Allelic variant analyzed	SNP analyzed	Assay ID
CYP2C9	2C9^2*	430C>T	C_25625805_10
	2C9^3*	1075A>C	C_27104892_10
	2C9^6*	818delA	C_32287221_20
CYP2C19	2C19^2*	681G>A	C_25986767_70
	2C19^3*	636G>A	C_27861809_10
	2C19^4*	1A>G	C_30634136_10
	2C19^5*	1297C>T	C_27861810_10
	2C19^17*	-806C>T	C___469857_10
CYP2D6	2D6^2*	-1584C>G	C_32407252_30
	2D6^3*	2549delA	C_32407232_50
	2D6^4*	1864G>A	C_27102431_BO
	2D6^6*	1707T>del	C_32407243_20
	2D6^10*	100C>T	C_11484460_40
	2D6^17*	1023C>T	C__2222771_40
	2D6^29*	3183G>A	C_34816113_20
	2D6^35*	31G>A	C_27102444_80
	2D6^41*	2988G>A	C_34816116_20
CYP3A4	3A4^1B*	-392A>G	C___1837671_50

CYP450, cytochrome P450; SNP, single-nucleotide polymorphism.

Phenotyping procedure

Single oral doses of omeprazole (20 mg), losartan (25 mg), caffeine (100 mg), and dextromethorphan (30 mg) were administered to each participant after overnight fasts, and all of the subjects were asked to abstain from any intake of caffeine-containing products during the test, and at least 72 h before the probe drug administration. Then, blood samples were collected 4 h after administering the probe drugs and centrifuged for 10 min at 3500 rpm to obtain plasma aliquots that were stored at −20°C until LC-MS/MS (liquid chromatography-tandem mass spectrometry) analysis (De Andrés et al., 2013). The 4-h postdose time point metabolic ratios (MRs; De Andrés et al., 2016) were then calculated for CYP1A2 (caffeine/paraxanthine), CYP2C9 (losartan/losartan carboxylic acid), CYP2C19 (omeprazole/5-hydroxyomeprazole), CYP2D6 (dextromethorphan/dextrorphan), and CYP3A4 (dextromethorphan/3-methoxymorphinan) to determine the actual metabolic capacity at each individual.

Data analysis

The genotype frequencies obtained were compared with the expected values using a contingency table χ² statistic with Yates's correction to determine Hardy–Weinberg equilibrium. Moreover, the potential relationship between CYP450 genotypes and MRs (De Andrés et al., 2013; LLerena et al., 2012, 2014a) was further studied by assigning “activity score” values to the predicted enzymatic activities for each allelic variant (Thorn et al., 2012; Videau et al., 2010; Villagra et al., 2011).

A frequency distribution histogram of the number of subjects versus MR (in a log scale) and a probit plot were used to analyze the variations in the MR throughout the individuals studied. Together with the shape of the histogram, linear regression from the probit plot and the normal test variable (NTV) analysis plot was utilized to assess potential bimodality and to determine if an antimode was present. The normality of the MR distributions was tested using the Shapiro–Wilk test and multiple regression analyses, and carried out to evaluate the effects of genetic background (activity score), gender, and tobacco, alcohol, and caffeine consumption on the metabolism of the probe drugs. In addition, one-way analysis of variance followed by Bonferroni's post-test or the nonparametric Kruskal–Wallis test followed by Dunn's post-test for multiple comparisons was used to compare mean MRs for different genotypes and/or “activity score,” whereas t-test or Mann–Whitney test was used for the enzymes for which only two genotype groups could be compared.

Spearman's rank or Pearson's correlation coefficient was used to assess the potential correlation of the genetic polymorphism and the age on actual drug metabolism activity (MR_log). Fisher's exact test was used to compare frequencies of different metabolic groups among populations. Data are expressed as mean ± standard deviation where appropriate, and a p-value lower than 0.05 was considered as statistically significant. GraphPad Prism for Windows version 5.00 (GraphPad Software, Inc., San Diego, CA, USA) and SPSS, Inc., v15.0 (Chicago, IL, USA) were used for statistical analyses.

Results

A total of 139 individuals, 70 men and 69 women, with a mean age of 22.94 ± 2.08 years (range 18–32 years) completed the study, and none of them reported suffering from any adverse effect. The simultaneous information on genotype and drug metabolizing capacities of CYP1A2, CYP2C9, CYP2C19, CYP2D6, and CYP3A4 was obtained for 136, 132, 132, 128, and 76 subjects, respectively. 12.1% of the total subjects were occasional consumers of alcoholic beverages, whereas 8.59% smoked 1–5 cigarettes/day, and no subject smoked more than 10 cigarettes/day (heavy smokers).

CYP genetic polymorphisms

The frequencies of the genotypes analyzed in the studied populations corresponded with those predicted by the Hardy–Weinberg's law (Table 2). The frequency of CYP1A2*1F, a variant related to a higher CYP1A2 inducibility, was found to be 78.3%. Likewise, for CYP1A2 genotype, 33.1% were found to be heterozygous CYP1A2*1A/*1F and 61.8% homozygous CYP1A2*1F/*1F, whereas 5.2% of the individuals were CYP1A2*1A homozygous.

Table 2.

Frequencies of CYP1A2, CYP2C9, CYP2C19, CYP2D6, and CYP3A4 Genotypes Observed in the Ecuadorian Population Sample

CYP450 genotype	No. of active genes	Activity score for genotype-estimated phenotypic status	Sample size (N)	Genotype frequency	95% CIs for genotype frequency
CYP1A2
^1A/^1A	2	—	7	0.051	0.025–0.102
^1A/^1F	2	—	45	0.331	0.257–0.414
^1F/^1F	2	—	84	0.618	0.534–0.695
CYP2C9
^1/^3	1	1	8	0.061	0.031–0.115
^1/^2	2	1.5	11	0.083	0.047–0.143
^1/^1	2	2	113	0.856	0.786–0.906
CYP2C19
^2/^2	0	0	1	0.008	0.001–0.042
^1/^2	1	1	31	0.235	0.171–0.314
^2/^17	1	2	1	0.008	0.001–0.042
^1/^1	2	2	78	0.591	0.506–0.671
^1/^17	2	>2	18	0.136	0.088–0.205
^17/^17	2	>2	3	0.023	0.008–0.065
CYP2D6
^4/^4	0	0	2	0.016	0.004–0.055
^4/^5	0	0	1	0.008	0.001–0.043
^4/^6	0	0	1	0.008	0.001–0.043
^4/^17	1	0.5	1	0.008	0.001–0.043
^5/^41	1	0.5	1	0.008	0.001–0.043
wt/^4*	1	1	9	0.070	0.037–0.128
wt/^4xN*	1	1	1	0.008	0.001–0.043
wt/^5*	1	1	1	0.008	0.001–0.043
^2/^4	1	1	6	0.047	0.022–0.098
^2/^5	1	1	1	0.008	0.001–0.043
^4/^35	1	1	1	0.008	0.001–0.043
wt/^10*	2	1.5	2	0.016	0.004–0.055
wt/^17*	2	1.5	4	0.031	0.012–0.078
wt/^29*	2	1.5	1	0.008	0.001–0.043
wt/^41*	2	1.5	3	0.023	0.008–0.067
^2/^10	2	1.5	3	0.023	0.008–0.067
^2/^29	2	1.5	1	0.008	0.001–0.043
^2/^41	2	1.5	2	0.016	0.004–0.055
wt/wt	2	2	36	0.281	0.211–0.365
wt/^2*	2	2	30	0.234	0.169–0.315
wt/^35*	2	2	2	0.016	0.004–0.055
^2/^2	2	2	9	0.070	0.037–0.128
^2/^35	2	2	1	0.008	0.001–0.043
^2xN/^5	1	2	1	0.008	0.001–0.043
wtxN/^4*	2	2	1	0.008	0.001–0.043
wtxN/^17*	2	2.5	1	0.008	0.001–0.043
wtxN/^29*	2	2.5	1	0.008	0.001–0.043
wtxN/wt	2	≥3	2	0.016	0.004–0.055
^2xN/^2	2	≥3	1	0.008	0.001–0.043
wt/^2xN*	2	≥3	2	0.016	0.004–0.055
CYP3A4
wt/wt	2	—	55	0.724	0.614–0.812
wt/^1B*	2	—	20	0.263	0.177–0.372
^1B/^1B	2	—	1	0.013	0.002–0.071

CIs, confidence intervals; N, number of subjects.

Among the individuals genotyped for CYP2C9, no mutants homozygous were detected, but 19 individuals (13.7%) were heterozygous for genotypes associated with reduced activity (*1/*2 and *1/*3). Indeed, 11 subjects (7.9%) carried the CYP2C9*2 and 8 were heterozygous for the CYP2C9*3 allele. The null activity allele CYP2C9*6 was not found (Table 3).

Table 3.

Frequencies of CYP1A2, CYP2C9, CYP2C19, CYP2D6, and CYP3A4 Variant Alleles Within the Ecuadorian Population Analyzed

Allele	N	Frequency	95% CI
CYP1A2
^1A*	59	0.217	0.172–0.270
^1F*	213	0.783	0.730–0.828
CYP2C9
^1*	245	0.928	0.890–0.953
^2*	11	0.042	0.023–0.073
^3*	8	0.030	0.015–0.059
^6*	—	—	—
CYP2C19
^1*	205	0.777	0.722–0.823
^2*	34	0.129	0.094–0.175
^3*	—	—	—
^4*	—	—	—
^5*	—	—	—
^17*	25	0.095	0.065–0.136
CYP2D6
Wt	129	0.504	0.443–0.565
^2*	62	0.242	0.194–0.298
^4*	24	0.094	0.064–0.136
^5*	6	0.023	0.011–0.050
^6*	1	0.004	0.001–0.022
^10*	5	0.020	0.008–0.045
^17*	6	0.023	0.011–0.050
^29*	3	0.012	0.004–0.034
^35*	4	0.016	0.006–0.039
^41*	6	0.023	0.011–0.050
Multiplications
wtxN	5	0.020	0.008–0.045
^2xN*	4	0.016	0.006–0.039
^4xN*	1	0.004	0.001–0.022
CYP3A4
Wt	130	0.855	0.791–0.902
^1B*	22	0.145	0.098–0.209

With regard to the CYP2C19 genetic analysis, the CYP2C19*2/*2 genotype was only found in one individual (0.7%), although the CYP2C19*2 allele was additionally found in 31 (23.5%) heterozygous subjects. Another subject (0.7%) was heterozygous for two mutated alleles (CYP2C19*2/*17). Regarding the allele associated with CYP2C19 enhanced activity (*17), 18 individuals (13.6%) have been found to be *1/*17 heterozygous and, additionally, 3 subjects (2.3%) were homozygous for CYP2C19*17 mutated allele.

When the CYP2D6 genotype was analyzed, four (3.1%) individuals were found to carry a genotype associated with poor metabolism status (genotypic poor metabolizers [gPMs]), although these alleles are also present in several heterozygous individuals who are predicted to possess CYP2D6 activity to a certain extent (Table 2). Moreover, the alleles associated with reduced CYP2D6 activity (*10, *29, and *41) allele have been found in five (3.6%), three (2.2%), and six (4.3%) heterozygous individuals, respectively (Table 3). Likewise, CYP2D6*17 and CYP2D6*10, also related to reduced CYP2D6 activity, are present in six (4.3%) and five (3.6%) heterozygous individuals, respectively. Conversely, 10 individuals (7.2%) were found to carry the multiplication of CYP2D6 gene, and 1 (0.7%) of them carries multiplications of the CYP2D6*4 variant (Table 3). The remaining nine (6.5%) individuals carry multiplications of an active gene (wtxN or *2xN), but only seven subjects (5.5%) have genotypes that might be associated with ultrarapid metabolism (gUMs; Table 2). For CYP3A4, a frequency of 26.3% of the total individuals was heterozygous wt/*1B, while just an individual (0.7%) was *1B homozygous.

CYP hydroxylation phenotypes

A remarkable interindividual variability in the calculated MRs (probe drug/metabolite, MR) was found for the CYPs analyzed: 10.23-fold differences in the values of MRs were found for CYP1A2 phenotype, as well as the losartan/E-3174 MRs utilized for CYP2C9 phenotyping ranged from 0.014 to 2.833; the CYP2C19 MRs were found in the 0.139–26.56 range, and the CYP2D6 MRs ranged from 0.0001 to 6.878. Likewise, dextromethorphan/3-methoxymorphinan MRs ranged from 0.065 to 28.38. The values of MRs determined for CYP1A2 and CYP3A4 were normally distributed, contrarily to the distribution of MRs detected for the rest of the enzymes analyzed. Normal distribution was indicated in the overall population by the Shapiro–Wilk test for CYP1A2 (p = 0.229) and CYP3A4 (p = 0.058).

By the analysis of the MR (log₁₀) histograms, as well as for some deviations of the linearity of the probit plots observed for each CYP enzyme analyzed, the existence of bimodal distribution could be established, more particularly for CYP2C9, CYP2C19, and CYP2D6 (Fig. 1), although it could not be confirmed by the analysis of the NTV plots. Consequently, no exact antimode to discriminate PMs or UMs from the rest of the individuals was calculated for any of the enzymes analyzed.

FIG. 1.

Probit plots and frequency distributions of MRs (log₁₀) of the volunteers evaluated for CYP1A2 (n = 136), CYP2C9 (n = 132), CY2C19 (n = 132), CYP2D6 (n = 128), and CYP3A4 (n = 76). The continuous line represents the mean MR of the overall population. Red dots (blue for CYP1A2) indicate outliers from linearity of probit plots. MRs, metabolic ratios.

Nevertheless, several outliers with clearly enhanced or reduced actual metabolic capacity were found for CYP2C9, CYP2C19, and CYP2D6, which can be considered either as actual metabolic poor (mPMs) or ultrarapid metabolizers (mUMs). For CYP2C9, six individuals (4.5%) could be classified as mPMs as of their higher MRs (MR_log > 0), whereas four subjects (3.1%) were considered as mPMs for CYP2D6 (MR_log > 0; Fig. 1). Conversely, one individual (0.8%) was observed to show clearly enhanced CYP2D6 activity and therefore could be considered as mUM (Fig. 1), whereas no subject was identified as mUM for CYP2C19 within the population analyzed (Table 4).

Table 4.

Prevalence of Individuals Classified as Poor (mPM) or Ultrarapid Metabolizers (mUM) According to Their Directly Measured Metabolic Phenotype Within the Ecuadorian Population Sample Analyzed

CYP450 enzyme	mPMs (%)	mUMs (%)
CYP2C9	4.5	Not detected
CYP2C19	Not detected	Not detected
CYP2D6	3.1	0.8

mPM, directly (phenotyping test) measured poor metabolizers; mUM, directly (phenotyping test) measured ultrarapid metabolizers.

CYP genetic polymorphisms and actual hydroxylation phenotypes

The plasma MRs calculated for CYP2C9, CYP2C19, and CYP2D6 at each subject studied showed an inverse correlation with the number of active alleles (p < 0.01 in all cases; Fig. 2), although a remarkable overlap in the MRs was observed among the wild homozygous and heterozygous subjects for the mutated active allele groups. Specifically, a distinguishable difference was found for individuals homozygous for null activity alleles, especially for CYP2D6 (Fig. 3).

FIG. 2.

Relationship between activity score (phenotype predicted from genotype) and MRs (log₁₀) for each CYP450 enzyme evaluated in the present Ecuadorian population sample. For CYP1A2 and CYP3A4, genotype instead of the activity score was presented. CYP450, cytochrome P450.

FIG. 3.

CYP450 enzyme metabolic phenotype values in relation to number of subjects, stratified by activity scores (phenotype predicted from genotype) or genotype values in the present Ecuadorian population sample.

In addition, a statistically significant difference (p < 0.05) for CYP1A2 metabolic capacity was observed among the different genotypes analyzed: the *1F/*1F individuals showed a reduced catalytic activity when compared to the heterozygous or the CYP1A2*1A homozygous subjects (Fig. 2). However, no significant difference in CYP1A2 MRs among subjects carrying CYP1A2*1F allele (−163C>A) either in heterozygosis or homozygosis was observed (p < 0.05) when compared to the CYP1A2*1A homozygous individuals.

For CYP2C9 analysis, the group of individuals carrying two active alleles (activity score = 2) shows a lower mean MR_log than those carrying one null or highly reduced activity allele (−1.03 ± 0.40 versus −0.76 ± 0.50; p < 0.01). In addition, null allele homozygous individuals were absent in the population analyzed, while some individuals with two active alleles showed lower CYP2C9 enzymatic activity than subjects carrying one reduced activity allele (Fig. 3).

Regarding the results obtained for the CYP2C19 genotype and phenotype analysis, the MR_log of the only individual carrying two null enzymatic activity alleles was found to be higher (1.424) than those of the individuals with one (0.46 ± 0.35) or two active alleles (0.03 ± 0.35). The mean MR_log of the individuals carrying a *1/*2 genotype is significantly higher than the mean value calculated for the individuals with two normal (*1/*1) or enhanced activity alleles (*1/*17 or *17/*17) (p < 0.001). Nevertheless, 61.9% of the individuals with active CYP2C19 alleles associated with enhanced enzymatic activity (activity score >2) exhibit MRs in the range of the individuals with only one active gene (Fig. 3).

For CYP2D6, the individuals with two null activity alleles are considered as mPMs, as confirmed by the highest MRs calculated for all of them. However, among the individuals with higher MRs and, consequently, with a lower hydroxylation capacity, there were individuals with either one or two active CYP2D6 alleles. On the contrary, among those individuals who would be considered as faster metabolizers, only one individual carrying CYP2D6 active allele multiplications was also found to be phenotypically mUM. Indeed, the MRs calculated for three gUM individuals (2.3%) were within the range calculated for MRs at individuals with no multiplications of active alleles or even carriers of allelic variants associated with reduced CYP2D6 metabolic capacity (Fig. 3). Nevertheless, significant differences (p < 0.0001) in the CYP2D6 MRs were found among the subjects studied according to their genetic background (activity score) and, more specifically, significant differences were detected among the individuals either with one or no active allele (activity score ranging 0–1) and those subjects with two or more active genes (activity score ≥2; p < 0.01; Fig. 2). A significantly reduced metabolic capacity (p < 0.05) was found for those individuals carrying *4 and *41 allelic variants.

Likewise, the distribution of the calculated MRs for CYP3A4 was also overlapped between the wild homozygous, heterozygous, and subjects homozygous for the mutated allele groups (Fig. 3), and no significant difference was observed in the MRs calculated for subjects carrying CYP3A4*1B when compared to the rest of the participants (p = 0.618). However, a remarkable high MR_log value (1.06) was found for the only individual carrying *1B/*1B genotype when compared to the rest of the genotype groups, whereas three individuals with wt/wt or wt/*1B genotype still showed lower CYP3A4 activity.

No significant correlation was found with regard to age, alcohol intake, or tobacco consumption habits for any of the CYPs analyzed; although it is noteworthy the reduced rate of both alcohol consumption and tobacco consumption annotated for the participants. Nevertheless, a significant influence of caffeine consumption on CYP1A2 activity was observed as a reduced CYP1A2 activity was observed among those individuals not consuming caffeine when compared to those caffeine consumers (p = 0.004).

When MRs were compared between genders, no correlation was found for CYP2C19, CYP2D6, and CYP3A4, while a significant decreased CYP1A2 activity was found for female subjects (p = 0.004). On the contrary, slightly higher CYP2C9 MR_log values were calculated for male individuals (p = 0.011). However, when individuals were stratified by the number of active genes, differences in MRs between genders were not observed either for CYP1A2 or CYP2C9.

Discussion

This is the first report, to the best of our knowledge, performed in the Ecuadorian population for a comparative analysis of the five key drug metabolizing enzyme genotypes and phenotypes for CYP1A2, CYP2C9, CYP2C19, CYP2D6, and CYP3A4 simultaneously. We also assessed the concordance and discordance between the genotype-estimated and directly measured drug metabolism phenotypes for these crucial and clinically relevant drug metabolism pathways.

Interestingly, a wide overlap was found among the MRs calculated at each individual for each CYP enzyme analyzed, as previously observed in other studies analyzing the CYP450 genotype and phenotype (Dorado et al., 2012a, 2012b; Griese et al., 1998; Shah and Smith, 2015a).

The elevated frequency of CYP1A2*1F found in this study is in accordance with the previously detected in ethnically different populations (McGraw and Waller, 2012) and quite similar to a Chilean population (Roco et al., 2012). The CYP1A2*1F polymorphism has been associated with increased induction of expression by smoking or heavy coffee consumption (Djordjevic et al., 2010; Zanger and Schwab, 2013), as well as by pregnancy or the circadian rhythms (Rasmussen et al., 2002; Yu et al., 2016). In this study, in subjects carrying one CYP1A2*1F allele, the caffeine metabolizing activity is enhanced, as well as a remarkable reduction in the mean CYP1A2 MR_log observed among the CYP1A2*1F homozygous caffeine consumers (0.474 ± 0.027) and non-caffeine consumers (0.568 ± 0.040). Likewise, in the CYP1A2*1F homozygous individuals, a significantly higher CYP1A2 activity was observed for those smoker subjects, which was not observed in the *1A/*1F genotype group. Moreover, the enhanced CYP1A2 activity found in male subjects has also been found in previous studies in Chinese individuals (Chen et al., 2005).

Interestingly, the *1F/*1F individuals showed a reduced enzymatic activity (mean MR_log = 0.506 ± 0.210) when compared to *1A/*1F (0.435 ± 0.219) and *1A/*1A (0.331 ± 0.225) genotype groups, although the differences observed were not significant. In this sense, the rare variant CYP1A2*1K, which results in highly decreased CYP1A2 expression and caffeine metabolism, could explain interindividual differences among groups (Aklillu et al., 2003; Klein et al., 2010). No poor metabolizers were identified according to the metabolic phenotype, although 10-fold interindividual variability was found. This value is lower than in Swedish individuals, but higher than in Koreans (Chen et al., 2005; Ghotbi et al., 2007). A reduced caffeine metabolic capacity was found in the Ecuadorian population (mean MR_log = −0.475 ± 0.218) when compared with Chinese, Koreans, and Swedish individuals (Chen et al., 2005; Ghotbi et al., 2007). Dietary conditions could explain this difference in CYP1A2 activity as cruciferous vegetables and charcoal-broiled meat are known CYP1A2 inducers, while apiaceous vegetables exert a depressing effect on the enzyme activity of some CYP1A2 alleles (Peterson et al., 2009).

Genotypes related to CYP2C9 reduced activity (CYP2C9*2/*2, *3/*3, and *2/*3) were not present in the population here studied, which is in agreement with previous studies for these polymorphisms in Ecuadorian populations (Dorado et al., 2012a). Indeed, the prevalence of CYP2C9*2 (4.2%) and CYP2C9*3 (3.0%) is lower than the frequencies found in Caucasian, North African, and American admixed populations (Dorado et al., 2011), while the frequency of CYP2C9*2 was higher than in East Asians (0.6%), black Africans (0.46%), and Native Americans (1.25%; Dorado et al., 2011). Furthermore, a lower frequency was calculated for CYP2C9*3 when compared to South Asian populations (11.7%), but similar to the CYP2C9*3 prevalence found in East and Northern Asians (Céspedes-Garro, Fricke-Galindo, et al., 2015).

With regard to the CYP2C9 phenotype, no poor metabolizer phenotype (mPM) was present in this population. These results are additionally concordant with studies on CYP2C9 activity previously performed in Ecuadorians (Dorado et al., 2012a) and Spanish (Dorado et al., 2003), whereas mPMs were detected in other populations in which the frequency or CYP2C9*3 is much more elevated, as is the case of the 14.28% of mPMs described in an Indian population (Varshney et al., 2013). A significant difference was found among individuals with different number of active alleles (p < 0.01, Fig. 2). As previously reported in different populations (Babaoglu et al., 2004; Yasar et al., 2002), the MR_log in subjects with CYP2C9*3 was significantly higher (p < 0.001) than individuals not carrying this allele (Cabaleiro et al., 2013).

However, genotype–phenotype discordance was detected: 3.79% of the total individuals possessed an MR_log > 0, which is remarkably higher than the MR_log found in the rest of the population analyzed, and they could be considered as slower metabolizers, but these individuals were either genotyped as *1/*1 or *1/*3 (Fig. 3). No influence of smoking, alcohol consumption, caffeine intake, age, or gender on CYP2C9 activity was observed, which is not in accordance with a previous study on CYP2C9 drug metabolizing capacity also carried out in Ecuadorians (Dorado et al., 2012a), but these results are concordant with other previous studies (Cabaleiro et al., 2013; LLerena et al., 2014a).

In this study, the interethnic variability for CYP2C19 is confirmed as of the similar CYP2C19*2 and CYP2C19*3 frequencies detected in American admixed and Caucasian populations, but lower than in Asians and African Americans (Fricke-Galindo et al., 2016). Likewise, the low gPM frequency found in this population is similar to the frequency found in North Africans, but lower than the gPM frequencies found in the rest of the populations studied worldwide, especially in Asians (Fricke-Galindo et al., 2016).

On the contrary, the CYP2C19*17 allele has a frequency of 9.5%, which is only similar to the frequency previously found in Native Americans (10.95%), but lower than in Caucasians, black Africans, admixed, and South Asians, while it is quite more frequent than in East Asians (0.96%; Fricke-Galindo et al., 2016). Interestingly, 15.9% of the individuals were identified as gUMs (*1/*17 or *17/*17), which is lower than previously reported in Ecuadorians (Vicente et al., 2014), as well as in Caucasians, South Asians, and Middle Easterners. However, it is in accordance with studies performed in black Africans and African admixed populations, and much higher than in East Asians (Fricke-Galindo et al., 2016).

A relationship between the genotype determined and the MRs calculated for CYP2C19 is confirmed for this population (Fig. 2), although 13 individuals (9.85%) with enhanced activity-associated genotypes (activity score >2) showed similar CYP2C19 capacity (MRs) than individuals with just one active allele (Fig. 3). These differences could be related to unknown or genetic variants such as CYP2C19*4, *5, *6, *7, and *8 (Hirota et al., 2013), which have been reported to cause a loss of enzymatic activity although they have been scarcely analyzed and thereby not studied in this work (Shirasaka et al., 2016). Also, the assignment of CYP2C19*17 homozygotes as UMs is still controversial as the range of the omeprazole MR overlapped for all genotypes, including CYP2C19*1 and *17 detected in previous studies (Li-Wan-Po et al., 2010).

The frequency of CYP2D6 gPMs (3.12%) calculated for the present Ecuadorian population is not different from the gPM frequencies found in other populations, including Caucasians, Mediterranean South Europeans, American admixed, or Native Americans (LLerena et al., 2014b), but it is lower than in Scandinavians (8.69%, p < 0.05) and higher than in East Asians (0.26%, p < 0.01). The aforementioned interethnic variability for CYP2D6 is also confirmed through the prevalence of gUMs. In contrast, 5.47% of the individuals were considered as CYP2D6 gUMs, as not such a high frequency of active gene multiplications was found (3.91%). This value is similar to the values determined in Native Americans (7.63%), Caucasians from America (4.33%), in different African populations (4.04–6.74%), Middle Easterners (10.54%), and Mediterranean populations (5.82%). On the contrary, the prevalence of gUMs previously found in South Asians (0.30%), East Asians (2.20%), American admixed (3.34%), Central Europeans (2.12%), and Slavs (3.65%) is significantly lower than in the present study (p < 0.05 in all cases; LLerena et al., 2014b).

A correlation between activity score and MRs was detected as well as significant differences in MR_log values among subjects with different number of active alleles were found. Frequency of mPMs was 3.1%, determined as of the values of MR_log > 0 found in four individuals who additionally carried a PM genotype. This prevalence is higher than in East Asians (0.84%, p < 0.05), but lower than in Central Europeans (8.13%, p < 0.05) and in Slavs (8.32%, p < 0.05; LLerena et al., 2014b). Likewise, just an individual who carries four copies of active alleles (activity score >2) and with a significantly higher MR is considered as mUM. This frequency of mUMs (0.8%) is slightly lower than in other populations worldwide, including a previous study on CYP2D6 metabolic phenotype in Ecuador (Dorado et al., 2012b; LLerena et al., 2014b).

In this work, the Ecuadorian population showed a mean MR_log of −2.233 (95% confidence interval ranging from −2.358 to −2.109), which indicates higher CYP2D6 activity in the Ecuadorian population when compared to a study previously performed in Mexican Tepehuano Amerindians (Lares-Asseff et al., 2005), and remarkably lower than in Iranians (Afshar et al., 2005). On the contrary, there is a slight but not significant rightward shift when compared to the frequency distribution in previous works performed on Mexican American subjects (Casner, 2005; Mendoza et al., 2001).

Notwithstanding, within the group of individuals with a “normal” CYP2D6 drug metabolizing capacity, a great overlap is observed, as there are individuals not only carrying two copies of the active gene but also multiplications of active alleles and even individuals with reduced activity alleles (Fig. 3), which represents a discordance between the genotype and the actual metabolic phenotype obtained. These particular inaccuracies in the prediction of the actual metabolic phenotype from the genotype could be due to different factors, although no significant influence of smoking habits, alcohol consumption, or gender was found in this population.

Consequently, a more extensive research on the issues related to the genotype–phenotype correlation for CYP2D6 is required because of the importance of this enzyme not only at a clinical level but also for the analysis of behavior and brain function (Sim et al., 2013). Indeed, the CYP2D6 enzyme has been linked to a potential role in the vulnerability to schizophrenia (Llerena et al., 2007), to a delay on the clinical onset of Parkinson's disease in predisposed people (Benítez et al., 1990), and to the observed phenomenon of lower vulnerability to psychopathology and greater impulsivity of CYP2D6 PMs versus extensive metabolizers (EMs; Peñas-Lledó et al., 2009).

CYP3A4 is a key drug-metabolizing enzyme of ∼50% of all clinically prescribed drugs. The variation in the expression levels of CYP3A4 contributes up to 20% of the interindividual variability and, consequently, can be influenced by genetic polymorphisms, although loss-of-function variants have been reported as rare events. Among these polymorphisms, the CYP3A4*1B variant has been previously found to be responsible for a decrease in the CYP3A4 activity and, accordingly, a remarkably high MR was found for the only *1B/*1B individual present in this study (Fig. 2). However, no significant difference was detected among the *1A homozygous and the *1A/*1B subjects.

In accordance with the previously determined interethnic variability in the metabolic activity of CYP3A4 (McGraw and Waller, 2012), the*1B allele is less frequent (14.5%) in the Ecuadorian population than in Africans (55–69%) or African descents (37.9%) from Brazil (Kohlrausch et al., 2014), but similar to Hispanic Americans (9.3–12.5%) (Céspedes-Garro, Naranjo, et al., 2015). Mexican Mestizos (8.8%), or Mexican Amerindians (8%). Conversely, it is more frequent in Ecuadorians than in Caucasians (2–9%; Sosa-Macías and Llerena, 2013), European descents from Brazil (9.8%; Kohlrausch et al., 2014), or in Asians, for whom this allele was not found. Different external factors could potentially influence CYP3A4 drug metabolizing activity such as alcohol consumption, smoking habit, gender, or body mass index.

However, no effect of these factors was observed on the Ecuadorian population studied. Consequently, the overlap in CYP3A4 activity could be caused by recently found nonanalyzed polymorphisms (e.g., CYP3A4*8, *20, *22, *25, *26, and *27) identified as potential CYP3A4 decreasing activity alleles (Apellániz-Ruiz et al., 2015a, 2015b; Elens et al., 2013). However, in vivo evidences have not been obtained for all these variants, and further research on genotype–phenotype correlation for CYP3A4 is warranted.

This study was carried out in a modest number of individuals (N = 139). Despite this issue, most of the influential polymorphic variants present in this population and their frequencies were detected. Most of these frequencies are in accordance with most CYP450 allele frequencies found in other populations worldwide. Therefore, this study may be considered as suggestive for further studies in the Ecuadorian population.

The Ecuadorian population here analyzed showed no multimodal distribution for any of the CYP enzymes analyzed and reduced frequencies of gPMs or gUMs were detected. These results are concordant with the results obtained in other admixed Latin American populations, but some differences with respect to more distant populations such as Asians or Africans are detected. In addition, the correspondence between gPMs and the lowest actual hydroxylation capacities (highest MRs) reinforces the reliable prediction of the poor metabolizer status (mPMs) from genotype for CYP2C19 and CYP2D6. However, the genotypes analyzed in this study cannot accurately predict the precise metabolic capacity among those individuals who possess active genes and, considering the noninfluence of external factors such as smoking, alcohol consumption, or the age on the actual metabolic capacity for any of the enzymes analyzed, additional elements should be further studied to elucidate the causes of the genotype–phenotype mismatch observed.

This is the case of phenoconversion, which turns temporally genotypic extensive metabolizer (gEM) individuals into actually reduced drug metabolizing capacity subjects (Brøsen, 2015), due to a number of issues (i.e., epigenetics, dietary habits, the microbiome, not detected morbidities, and/or inflammatory states) that can somehow exert some effect on this variation on the actual enzymatic capacity on the subjects participating in these studies (Shah and Smith 2015a, 2015b; Shah et al., 2016). Further and extensive research on the multiple factors influencing the actual metabolic phenotype of the main CYP enzymes together with a complete genetic analysis (sequencing) of the polymorphisms present in these genes would enhance the knowledge for an accurate prediction on actual individuals’ drug response and, consequently, to improve clinical outcomes individualizing therapies and dosing for the population from Ecuador.

Conclusions and Expert Outlook

Drug dosage adjustment for the Ecuadorian population is desirable due to the existence of individuals with absent or enhanced drug metabolism function, especially for CYP2C19 and CYP2D6 metabolic pathways. However, the remarkable discordance between genotype-estimated phenotype and the actual measured drug metabolism phenotype for many individuals is noteworthy. Hence, for personalized medicine in the Ecuadorian population, we recommend CYP450 multiplexed phenotyping using a cocktail strategy, such as the CEIBA cocktail, or genotyping and phenotyping in tandem, rather than CYP450 genotypic tests alone. Also, we recommend further future research in support of equitable personalized medicine initiatives in neglected or understudied populations worldwide.

Footnotes

Acknowledgments

This study was supported by Junta de Extremadura, AEXCID 13IA001 (to SIFF) and coordinated by the network Red Iberoamericana de Farmacogenética y Farmacogenómica (). F.D.A. was supported by Instituto de Salud Carlos III—Sara Borrell program (CD13/00348). E.T. received a 2014 Collaboration Grant from Universidad San Francisco de Quito. The views expressed represent the personal opinions of the authors and do not necessarily reflect the positions of their affiliated institutions.

Author Disclosure Statement

The authors declare that no conflicting financial interests exist.

Abbreviations Used

References

Afshar

, Rouini

, and Ala

. (2005). Dextromethorphan metabolic phenotyping in an Iranian population. Eur J Clin Pharmacol, 60, 849–854.

Aklillu

, Carrillo

, Makonnen

, et al. (2003). Genetic polymorphism of CYP1A2 in Ethiopians affecting induction and expression: Characterization of novel haplotypes with single-nucleotide polymorphisms in intron 1. Mol Pharmacol, 64, 659–669.

Apellániz-Ruiz

, Inglada-Pérez

, Naranjo

MEG

, et al. (2015a). High frequency and founder effect of the CYP3A4*20 loss-of-function allele in the Spanish population classifies CYP3A4 as a polymorphic enzyme. Pharmacogenomics J, 15, 288–292.

Apellániz-Ruiz

, Lee

, Sánchez-Barroso

, et al. (2015b). Whole-exome sequencing reveals defective CYP3A4 variants predictive of paclitaxel dose-limiting neuropathy. Clin Cancer Res, 21, 322–328.

Babaoglu

, Yasar

, Sandberg

, et al. (2004). CYP2C9 genetic variants and losartan oxidation in a Turkish population. Eur J Clin Pharmacol, 60, 337–342.

Benítez

, Ladero

, Jimenez-Jimenez

, et al. (1990). Oxidative polymorphism of debrisoquine in Parkinson's disease. J Neurol Neurosurg Psychiatry, 53, 289–292.

Berecz

, LLerena

, de la Rubia

, et al. (2002). Relationship between risperidone and 9-hydroxy-risperidone plasma concentrations and CYP2D6 enzyme activity in psychiatric patients. Pharmacopsychiatry, 35, 231–234.

Brøsen

. (2015). Pharmacogenetics of drug oxidation via cytochrome P450 (CYP) in the populations of Denmark, Faroe Islands and Greenland. Drug Metab Pers Ther, 30, 147–163.

Cabaleiro

, Román

, Ochoa

, et al. (2013). Evaluation of the relationship between sex, polymorphisms in CYP2C8 and CYP2C9, and pharmacokinetics of angiotensin receptor blockers. Drug Metab Dispos, 41, 224–229.

10.

Casner

. (2005). The effect of CYP2D6 polymorphisms on dextromethorphan metabolism in Mexican Americans. J Clin Pharmacol, 45, 1230–1235.

11.

Céspedes-Garro

, Fricke-Galindo

, Naranjo

MEG

, et al. (2015). Worldwide interethnic variability and geographical distribution of CYP2C9 genotypes and phenotypes. Expert Opin Drug Metab Toxicol, 11, 1893–1905.

12.

Céspedes-Garro

, Naranjo

MEG

, Ramírez

, et al. (2015). Pharmacogenetics in Central American healthy volunteers: Interethnic variability. Drug Metab Pers Ther, 30, 19–31.

13.

Chen

, Wang

, Zhi

, et al. (2005). The G-113A polymorphism in CYP1A2 affects the caffeine metabolic ratio in a Chinese population. Clin Pharmacol Ther, 78, 249–259.

14.

Day

, Coombes

, McGrath-Lone

, Schoenborn

, and Ward

. (2016). Stratified, precision or personalised medicine? Cancer services in the ‘real world’ of a London hospital. Sociol Health Illn, 2016. [Epub ahead of print]; DOI: 10.1111/1467-9566.12457

15.

De Andrés

, Sosa-Macías

, Lazalde-Ramos

, et al. (2013). Evaluation of drug-metabolizing enzyme hydroxylation phenotypes in Hispanic populations: The CEIBA cocktail. Drug Metabol Drug Interact, 28, 135–146.

16.

De Andrés

, Terán

, Bovera

, Fariñas

, Terán

, and LLerena

. (2016). Multiplex phenotyping for systems medicine: A one-point optimized practical sampling strategy for simultaneous estimation of CYP1A2, CYP2C9, CYP2C19, and CYP2D6 activities using a cocktail approach. OMICS, 20, 88–96.

17.

Djordjevic

, Ghotbi

, Jankovic

, and Aklillu

. (2010). Induction of CYP1A2 by heavy coffee consumption is associated with the CYP1A2-163C>A polymorphism. Eur J Clin Pharmacol, 66, 697–703.

18.

Dorado

, Beltrán

, Machín

, Peñas-Lledó

, Terán

, and LLerena

. (2012a). Losartan hydroxylation phenotype in an Ecuadorian population: Influence of CYP2C9 genetic polymorphism, habits and gender. Pharmacogenomics, 13, 1711–1717.

19.

Dorado

, Berecz

, Norberto

, Yasar

, Dahl

, and LLerena

. (2003). CYP2C9 genotypes and diclofenac metabolism in Spanish healthy volunteers. Eur J Clin Pharmacol, 59, 221–225.

20.

Dorado

, Cáceres

, Pozo-Guisado

, Wong

, Licinio

, and LLerena

. (2005). Development of a PCR-based strategy for CYP2D6 genotyping including gene multiplication of worldwide potential use. Biotechniques, 39(10 Suppl), S571–S574.

21.

Dorado

, Gallego

, Peñas-Lledó

, Terán

, and LLerena

. (2014). Relationship between the CYP2C9 IVS8-109A>T polymorphism and high losartan hydroxylation in healthy Ecuadorian volunteers. Pharmacogenomics, 15, 1417–1421.

22.

Dorado

, Heras

, Machín

, Hernández

, Terán

, and LLerena

. (2012b). CYP2D6 genotype and dextromethorphan hydroxylation phenotype in an Ecuadorian population. Eur J Clin Pharmacol, 68, 637–644.

23.

Dorado

, Sosa-Macías

, Peñas-Lledó

, et al. (2011). CYP2C9 allele frequency differences between populations of Mexican-Mestizo, Mexican-Tepehuano, and Spaniards. Pharmacogenomics J, 11, 108–112.

24.

Elens

, van Gelder

, Hesselink

, Haufroid

, and van Schaik

RHN

. (2013). CYP3A4*22: Promising newly identified CYP3A4 variant allele for personalizing pharmacotherapy. Pharmacogenomics, 14, 47–62.

25.

Fricke-Galindo

, Céspedes-Garro

, Rodríguez-Soares

, et al. (2016). Interethnic variation of CYP2C19 alleles, “predicted” phenotypes and “measured” metabolic phenotypes across world populations. Pharmacogenomics J, 16, 113–123.

26.

Gaedigk

, Simon

, Pearce

, Bradford

, Kennedy

, and Leeder

. (2008). The CYP2D6 activity score: Translating genotype information into a qualitative measure of phenotype. Clin Pharmacol Ther, 83, 234–242.

27.

Ghotbi

, Christensen

, Roh

, Ingelman-Sundberg

, Aklillu

, and Bertilsson

. (2007). Comparisons of CYP1A2 genetic polymorphisms, enzyme activity and the genotype-phenotype relationship in Swedes and Koreans. Eur J Clin Pharmacol, 63, 537–546.

28.

Griese

, Zanger

, Brudermanns

, et al. (1998). Assessment of the predictive power of genotypes for the in-vivo catalytic function of CYP2D6 in a German population. Pharmacogenetics, 8, 15–26.

29.

Hirota

, Eguchi

, and Ieiri

. (2013). Impact of genetic polymorphisms in CYP2C9 and CYP2C19 on the pharmacokinetics of clinically used drugs. Drug Metab Pharmacokinet, 28, 28–37.

30.

Klein

, Winter

, Turpeinen

, Schwab

, and Zanger

. (2010). Pathway-targeted pharmacogenomics of CYP1A2 in human liver. Front Pharmacol, 1, 1–20.

31.

Kohlrausch

, Carracedo

, and Hutz

. (2014). Characterization of CYP1A2, CYP2C19, CYP3A4 and CYP3A5 polymorphisms in South Brazilians. Mol Biol Rep, 41, 1453–1460.

32.

Lares-Asseff

, Sosa-Macías

, Elizondo-Azuela

, Flores-Pérez

, and Bradley-Alvarez

. (2005). Phenotypical expression of CYP2D6 in Amerindians of Tepehuano origin from Durango, Mexico. Proc West Pharmacol Soc, 48, 102–107.

33.

Li-Wan-Po

, Girard

, Farndon

, Cooley

, and Lithgow

. (2010). Pharmacogenetics of CYP2C19: Functional and clinical implications of a new variant CYP2C19*17. Br J Clin Pharmacol, 69, 222–230.

34.

LLerena

, Álvarez

, Dorado

, et al. (2014a). Interethnic differences in the relevance of CYP2C9 genotype and environmental factors for diclofenac metabolism in Hispanics from Cuba and Spain. Pharmacogenomics J, 14, 229–234.

35.

LLerena

, Berecz

, de la Rubia

, Norberto

, and Benítez

. (2000). Use of the mesoridazine/thioridazine ratio as a marker for CYP2D6 enzyme activity. Ther Drug Monit, 22, 397–401.

36.

LLerena

, Dorado

, Peñas-Lledó

, Cáceres

, and De la Rubia

. (2007). Low frequency of CYP2D6 poor metabolizers among schizophrenia patients. Pharmacogenomics J, 7, 408–410.

37.

LLerena

, Dorado

, Ramírez

, et al. (2012). CYP2D6 genotype and debrisoquine hydroxylation phenotype in Cubans and Nicaraguans. Pharmacogenomics J, 12, 176–183.

38.

LLerena

, Naranjo

MEG

, Rodrigues-Soares

, Peñas-LLedó

, Fariñas

, and Tarazona-Santos

. (2014b). Interethnic variability of CYP2D6 alleles and of predicted and measured metabolic phenotypes across world populations. Expert Opin Drug Metab Toxicol, 10, 1569–1583.

39.

McCarthy

. (2015). Obama seeks $213m to fund “precision medicine.”. BMJ, 350, h587.

40.

McGraw

, and Waller

. (2012). Cytochrome P450 variations in different ethnic populations. Expert Opin Drug Metab Toxicol, 8, 371–382.

41.

Mendoza

, Wan

, Poland

, Smith

, and Lin

. (2001). CYP2D6 polymorphism in a Mexican American population. Clin Pharmacol Ther, 70, 552–560.

42.

Meyer

. (2004). Pharmacogenetics – Five decades of therapeutic lessons from genetic diversity. Nat Rev Genet, 5, 669–676.

43.

Nassan

, Nicholson

, Elliott

, Rohrer Vitek

, Black

, and Frye

. (2016). Pharmacokinetic pharmacogenetic prescribing guidelines for antidepressants: A template for psychiatric precision medicine. Mayo Clin Proc, 91, 897–907.

44.

Peñas-LLedó

, Dorado

, Pacheco

, González

, and LLerena

. (2009). Relation between CYP2D6 genotype, personality, neurocognition and overall psychopathology in healthy volunteers. Pharmacogenomics, 10, 1111–1120.

45.

Peterson

, Schwarz

, Li

, et al. (2009). CYP1A2, GSTM1, and GSTT1 polymorphisms and diet effects on CYP1A2 activity in a crossover feeding trial. Cancer Epidemiol Biomarkers Prev, 18, 3118–3125.

46.

Ramamoorthy

, Pacanowski

, Bull

, and Zhang

. (2015). Racial/ethnic differences in drug disposition and response: Review of recently approved drugs. Clin Pharmacol Ther, 97, 263–273.

47.

Rasmussen

, Brix

, Kyvik

, and Brøsen

. (2002). The interindividual differences in the 3-demthylation of caffeine alias CYP1A2 is determined by both genetic and environmental factors. Pharmacogenetics, 12, 473–478.

48.

Roco

, Quiñones

, Agúndez

JAG

, et al. (2012). Frequencies of 23 functionally significant variant alleles related with metabolism of antineoplastic drugs in the Chilean population: Comparison with Caucasian and Asian populations. Front Genet, 3, 1–9.

49.

Shah

, Gaedigk

, LLerena

, Eichelbaum

, and Smith

. (2016). CYP450 genotype and pharmacogenetic association studies: A critical appraisal. Pharmacogenomics, 17, 259–275.

50.

Shah

, and Smith

. (2015a). Addressing phenoconversion: The Achilles’ heel of personalized medicine. Br J Clin Pharmacol, 79, 222–240.

51.

Shah

, and Smith

. (2015b). Inflammation-induced phenoconversion of polymorphic drug metabolizing enzymes: Hypothesis with implications for personalized medicine. Drug Metab Dispos, 43, 400–410.

52.

Shirasaka

, Chaudhry

, McDonald

, et al. (2016). Interindividual variability of CYP2C19-catalyzed drug metabolism due to differences in gene diplotypes and cytochrome P450 oxidoreductase content. Pharmacogenomics J, 16, 375–387.

53.

Sim

, Kacevska

, and Ingelman-Sundberg

. (2013). Pharmacogenomics of drug-metabolizing enzymes: A recent update on clinical implications and endogenous effects. Pharmacogenomics J, 13, 1–11.

54.

Singh

, Singh

, Vasudevan

, et al. (2015). Esophageal cancer epigenomics and integrome analysis of genome-wide methylation and expression in high risk Northeast Indian population. OMICS, 19, 688–699.

55.

Sinués

, Vicente

, Fanlo

, et al. (2008). CYP3A5*3, CYP3A4*1B and MDR1 C3435T genotype distributions in Ecuadorians. Dis Markers, 24, 325–331.

56.

Sistonen

, Fuselli

, Palo

, Chauhan

, Padh

, and Sajantila

. (2009). Pharmacogenetic variation at CYP2C9, CYP2C19, and CYP2D6 at global and microgeographic scales. Pharmacogenet Genomics, 19, 170–179.

57.

Sitole

, Steffens

, Krüger

, and Meyer

. (2014). Mid-ATR-FTIR spectroscopic profiling of HIV/AIDS sera for novel systems diagnostics in global health. OMICS, 18, 513–523.

58.

Soriano

, Vicente

, Carcas

, et al. (2011). Differences between Spaniards and Ecuadorians in CYP2A6 allele frequencies: Comparison with other populations. Fundam Clin Pharmacol, 25, 627–632.

59.

Sosa-Macías

, and LLerena

. (2013). Cytochrome P450 genetic polymorphisms of Mexican indigenous populations. Drug Metabol Drug Interact, 28, 193–208.

60.

Thorn

, Aklillu

, Klein

, and Altman

. (2012). PharmGKB summary: Very important pharmacogene information for CYP1A2. Pharmacogenet Genomics, 22, 73–77.

61.

Varshney

, Saha

, Tandon

, Shrivastava

, and Ali

. (2013). Genotype-phenotype correlation of cytochrome P450 2C9 polymorphism in Indian National Capital Region. Eur J Drug Metab Pharmacokinet, 38, 275–282.

62.

Vicente

, González-Andrade

, Soriano

, Fanlo

, Martínez-Jarreta

, and Sinués

. (2014). Genetic polymorphisms of CYP2C8, CYP2C9 and CYP2C19 in Ecuadorian Mestizo and Spaniard populations: A comparative study. Mol Biol Rep, 41, 1267–1272.

63.

Videau

, Delaforge

, Levi

, et al. (2010). Biochemical and analytical development of the CIME cocktail for drug fate assessment in humans. Rapid Commun Mass Spectrom, 24, 2407–2419.

64.

Villagra

, Goethe

, Schwartz

, et al. (2011). Novel drug metabolism indices for pharmacogenetic functional status based on combinatory genotyping of CYP2C9, CYP2C19 and CYP2D6 genes. Biomark Med, 5, 427–438.

65.

Wang

, and Chou

. (2010). Molecular modeling of cytochrome P450 and drug metabolism. Curr Drug Metab, 11, 342–346.

66.

Yasar

, Forslund-Bergengren

, Tybring

, et al. (2002). Pharmacokinetics of losartan and its metabolite E-3174 in relation to the CYP2C9 genotype. Clin Pharmacol Ther, 71, 89–98.

67.

, Campbell

, Stockmann

, et al. (2016). Pregnancy-induced changes in the pharmacokinetics of caffeine and its metabolites. J Clin Pharmacol, 56, 590–596.

68.

Zanger

, and Schwab

. (2013). Cytochrome P450 enzymes in drug metabolism: Regulation of gene expression, enzyme activities, and impact of genetic variation. Pharmacol Ther, 138, 103–141.