The present study evaluates the worldwide frequency distribution of CYP2C19 alleles and CYP2C19 metabolic phenotypes ('predicted' from genotypes and 'measured' with a probe drug) among healthy volunteers from different ethnic groups and geographic regions, as well as the relationship between the 'predicted' and 'measured' CYP2C19 metabolic phenotypes. A total of 52 181 healthy volunteers were studied within 138 selected original research papers. CYP2C19*17 was 42- and 24-fold more frequent in Mediterranean-South Europeans and Middle Easterns than in East Asians (P<0.001, in both cases). Contrarily, CYP2C19*2 and CYP2C19*3 alleles were more frequent in East Asians (30.26% and 6.89%, respectively), and even a twofold higher frequency of these alleles was found in Native populations from Oceania (61.30% and 14.42%, respectively; P<0.001, in all cases), which may be a consequence of genetic drift process in the Pacific Islands. Regarding CYP2C19 metabolic phenotype, poor metabolizers (PMs) were more frequent among Asians than in Europeans, contrarily to the phenomenon reported for CYP2D6. A correlation has been found between the frequencies of CYP2C19 poor metabolism 'predicted' from CYP2C19 genotypes (gPMs) and the poor metabolic phenotype 'measured' with a probe drug (mPMs) when subjects are either classified by ethnicity (r=0.94, P<0.001) or geographic region (r=0.99, P=0.002). Nevertheless, further research is needed in African and Asian populations, which are under-represented, and additional CYP2C19 variants and the 'measured' phenotype should be studied.

As of today, Africa and Asia are under-represented in this area relative to the total number of their inhabitants, so that further studies in these regions are warranted. The CYP2D6*4 allele frequency was higher in Caucasians, CYP2D6*10 in East Asians, CYP2D6*41 and duplication/multiplication of active alleles in Middle Easterns, CYP2D6*17 in Black Africans and CYP2D6*29 in African Americans, than in other ethnic groups. Overall, gPMs and mPMs are more frequent among Caucasians, and gUMs among Middle Easterns and Ethiopians. However, mUMs could not be evaluated because only two studies were found presenting this information. Further studies including mUMs are thus warranted. There is a correspondence between gPMs and mPMs, but the few studies of mUMs meant that their relationship with gUMs could not be demonstrated. Finally, evolutionary aspects of the CYP2D6 allele distribution appear to support the Great Human Expansion model.

CYP2C9*2 allele is the most frequent in Caucasian populations (average 14%), with the lowest frequencies for Africans (0.46%), East Asians (0.56%) and Native Americans (1.25%), which is in agreement with the hypothesis about the low prevalence in Amerindians. CYP2C9*3 shows the highest frequency among South Asians (11.7%), while CYP2C9*5 (1.56%) and *8 (4.70%) in African Americans. The predicted poor metabolizers (gPMs) were found overall in a low frequency, with the highest frequency detected for South Asians, in accordance with the CYP2C9*3 frequency in these populations. This study shows the worldwide variability in the CYP2C9 allele frequencies across different ethnic and geographic groups. Data about CYP2C9 "measured" metabolic phenotypes is still limited.

Cytochrome P450 3A4 (CYP3A4) is a key drug-metabolizing enzyme. Loss-of-function variants have been reported as rare events, and the first demonstration of a CYP3A4 protein lacking functional activity is caused by CYP3A4*20 allele. Here we characterized the world distribution and origin of CYP3A4*20 mutation. CYP3A4*20 was determined in more than 4000 individuals representing different populations, and haplotype analysis was performed using CYP3A polymorphisms and microsatellite markers. CYP3A4*20 allele was present in 1.2% of the Spanish population (up to 3.8% in specific regions), and all CYP3A4*20 carriers had a common haplotype. This is compatible with a Spanish founder effect and classifies CYP3A4 as a polymorphic enzyme. This constitutes the first description of a CYP3A4 loss-of-function variant with high frequency in a population. CYP3A4*20 results together with the key role of CYP3A4 in drug metabolism support screening for rare CYP3A4 functional alleles among subjects with adverse drug events in certain populations.

Global precision medicine demands characterization of drug metabolism and phenotype variation in diverse populations, including the indigenous societies. A related question is the extent to which CYP450 drug metabolizing enzyme genotype and phenotype data are concordant and whether they can be used interchangeably. These issues are increasingly debated as precision medicine continues to expand as a popular research topic worldwide. We report here the first study in clinically relevant CYP450 drug metabolism phenotypes and genotypes in Mexican Amerindian indigenous subjects. In a large sample of 450 unrelated and medication free Mexican Amerindian indigenous healthy persons from four Mexican states (Chihuahua, Durango, Nayarit, and Sonora), we performed multiplexed phenotyping for the CYP1A2, CYP2C9, CYP2C19, CYP2D6, and CYP3A4 drug metabolizing enzymes using the CEIBA cocktail and genotyped the same pathways for functional polymorphic variation. Remarkable interindividual variability was found for the actual drug metabolizing capacity of all the enzymes analyzed, and, more specifically, the metabolic ratios calculated were significantly different across individuals with different number of active alleles for CYP2C9, CYP2C19, and CYP2D6. The drug metabolizing capacity "predicted" from the genotype determined was not in accordance with the actual capacity "measured" by phenotyping in several individuals for CYP2C9, CYP2C19, and CYP2D6. Consequently, a more extensive genotyping of the main CYP enzymes, including rare variants, together with the analysis of the actual drug metabolizing capacity using an appropriate phenotyping approach will add valuable information for accurate drug metabolism studies, especially useful in understudied populations such as Mexican Amerindians. In sum, this study demonstrates that current personalized medicine strategies based on "predicted" phenotype from genotyping of alleles with high frequency in European populations are not adequate for Mestizos and Native American populations.