Pharmacogenomics, Ancestry and Clinical Decision Making for Global Populations

ORIGINAL ARTICLE Pharmacogenomics, ancestry and clinical decision making for global populations

E Ramos1, A Doumatey1, AG Elkahloun2, D Shriner1, H Huang1, G Chen1, J Zhou1, H McLeod3, A Adeyemo1 and CN Rotimi1

Pharmacogenomically relevant markers of drug response and adverse drug reactions are known to vary in frequency across populations. We examined minor allele frequencies (MAFs), genetic diversity (FST) and population structure of 1156 genetic variants (including 42 clinically actionable variants) in 212 genes involved in drug absorption, distribution, metabolism and excretion (ADME) in 19 populations (n ¼ 1478). There was wide population differentiation in these ADME variants, reflected in the range of mean MAF (DMAF) and FST. The largest mean DMAF was observed in African ancestry populations (0.10) and the smallest mean DMAF in East Asian ancestry populations (0.04). MAFs ranged widely, for example, from 0.93 for single-nucleotide polymorphism (SNP) rs9923231, which influences warfarin dosing to 0.01 for SNP rs3918290 associated with capecitabine metabolism. ADME genetic variants show marked variation between and within continental groupings of populations. Enlarging the scope of pharmacogenomics research to include multiple global populations can improve the evidence base for clinical translation to benefit all peoples.

The Pharmacogenomics Journal (2014) 14, 217–222; doi:10.1038/tpj.2013.24; published online 9 July 2013 Keywords: ADME; clinical decision making; global populations; pharmacogenomics

INTRODUCTION study, we provide a comprehensive analysis of human genetic Rapid advances in genomic science have led to identiﬁcation by variation on B200 ADME genes in 19 global populations, regulatory agencies of a growing list of clinically important including the largest set of African ancestry populations studied biomarkers for drug response and toxicity. For example, the US for pharmacogenomics. We describe the range of variation Food and Drug Administration (FDA) now maintains a table of observed at multiple layers spanning from continental groups to pharmacogenomic biomarkers in drug labels,1 while the UK the individual. We discuss the impact of the observed variation on Medicines and HealthCare Products Agency and the European clinical decision making, as well as the utility of such data for Medicines Agency both have mechanisms for considering such regulatory purposes, including testing recommendations. biomarkers for targeted therapy and drug safety warnings. Clinically validated pharmacogenomic biomarkers can help physicians optimize drug selection, dose and treatment duration MATERIALS AND METHODS while averting adverse drug reactions.2 However, the drive to Study populations position pharmacogenomics as a core element in personalized A total of 1478 individuals from 19 populations with ancestry from medicine still suffers from limited data. For example, it is different parts of the world were included in this study (Supplementary estimated that over 90% of drugs currently used in clinical Table 1). Fifteen of these populations were from the 1000 Genomes Project practice lack valid and predictive biomarkers for therapeutic (http://www.1000genomes.org/) sample collection. The populations (and 3 their designated labels) were: Yoruba in Ibadan, Nigeria (YRI); Luhya in effects and/or avoiding severe side effects. Another limitation Webuye, Kenya (LWK); Maasai in Kinyawa, Kenya (MKK); African ancestry in is that our understanding of the distribution of human Southwest USA (ASW); Utah residents with Northern and Western pharmacogenomic variation remains limited due in part to the European ancestry from the Centre d’Etude du Polymorphisme Humain poor representation of ethnically diverse samples from various (CEPH) collection (CEU); Toscans in Italy (TSI); British from England and parts of the world in such studies. A more comprehensive Scotland (GBR); Finnish from Finland (FIN); Iberian populations in Spain understanding of the genetic landscape of the absorption, (IBS); Han Chinese in Beijing, China (CHB); Han Chinese South (CHS); distribution, metabolism and excretion (ADME) genes across Japanese in Tokyo, Japan (JPT); Mexican ancestry in Los Angeles, California global populations with different ancestral backgrounds can (MXL); Puerto Rican in Puerto Rico (PUR); and Columbians in Medellin (CLM; facilitate the translation of pharmacogenomics data to clinical Figure 1). The remaining four populations were obtained from ongoing studies in West Africa and the United States as follows: three groups—Igbo practice and public health policy. Achieving this task is now more from Nigeria (IGBO); Akan from Ghana (AKAN) and Gaa-Adangbe from feasible with increasing access to genotyping and sequencing Ghana (GAA)—were obtained from participants in the Africa America technologies, as well as the availability of gene chips speciﬁcally Diabetes Mellitus study6 and the fourth group comprised African designed to assess polymorphic alleles of drug metabolizing Americans from the metropolitan Washington, DC area that participated enzymes and other genes involved in the ADME of drugs.4,5 In this in the Howard University Family Study (HUFS).7

1Center for Research on Genomics and Global Health, National Human Genome Research Institute, National Institutes of Health, Bethesda, MD, USA; 2Cancer Genetics Branch, National Human Genome Research Institute, National Institutes of Health, Bethesda, MD, USA and 3Institute for Pharmacogenomics and Individualized Therapy, University of North Carolina, Chapel Hill, NC, USA. Correspondence: Dr E Ramos, Center for Research on Genomics and Global Health, National Human Genome Research Institute, National Institutes of Health, Bethesda, MD 20892, USA. E-mail: [email protected] Received 5 February 2013; revised 1 May 2013; accepted 13 May 2013; published online 9 July 2013 Pharmacogenomics and global populations E Ramos et al 218

Figure 1. Map of populations analyzed in this study. Positions of populations marked on the map indicate location of sample collection. African ancestry populations are indicated by blue markers A–H (YRI, IGBO, GAA, AKAN, LWK, MKK, ASW and HUFS, respectively), European ancestry populations by green markers I–M (CEU, TSI, GBR, FIN and IBS, respectively), East Asian ancestry populations by purple markers N–P (CHB, CHS and JPT, respectively) and Latin American populations by pink markers Q–S (MXL, PUR and CLM, respectively). AKAN, Akan from Ghana; ASW, African ancestry in Southwest USA; CEU, Centre d’Etude du Polymorphisme Humain collection; CHS, Han Chinese South; CLM, Columbians in Medellin; GAA, Gaa-Adangbe from Ghana; HUFS, Howard University Family Study; IGBO, Igbo from Nigeria; JPT, Japanese in Tokyo, Japan; LWK, Luhya in Webuye, Kenya; MKK, Maasai in Kinyawa, Kenya; MXL, Mexican ancestry in Los Angeles, California; PUR, Puerto Rican in Puerto Rico; YRI, Yoruba in Ibadan, Nigeria.

Samples from five population groups (IGBO, AKAN, GAA, MKK and HUFS) the minimum MAF for a given allele within a specified group of were directly genotyped using the Affymetrix DMET Plus platform (Santa populations was calculated to define the range of difference (DMAF). Clara, CA, USA) at the National Human Genome Research Institute Microarray Density plots for comparing DMAF were drawn using the R software Core laboratory at the National Institutes of Health (Bethesda, MD, USA) as package (www.r-project.org). Pairwise FST was used as a measure of described in the Supplementary Methods. For the other 14 groups, DMET population differentiation for a given marker between populations. FST Plus markers were extracted from the 1000 Genomes as described (Supple- values range from zero to one with one meaning that the two populations mentary Methods). To facilitate continental level and other comparisons, being compared are completely separated and zero means no divergence populations were grouped as follows: continental African samples (YRI, IGBO, (that is, the populations are freely sharing genetic materials through GAA, AKAN, LWK and MKK) were designated as AFR; continental African interbreeding). More information on FST estimates is described samples plus the African-American populations (ASW and HUFS) were (Supplementary Methods). designated AFR þ AA; continental European and Centre d’Etude du Polymor- Principal components of ancestry were computed by decomposing the phisme Humain samples (CEU, TSI, GBR, FIN and IBS) were designated as EUR; centered genotype matrix of the entire data set (1478 individuals and 1156 continental East Asian samples (CHB, CHS and JPT) were designated as EAS; markers). The number of significant principal components was estimated and Latin American samples (MXL, PUR and CLM) were designated as AMR. using the minimum average partial test.8 ADME variants are a subset of Data management is further described (Supplementary Methods). overall human genetic variation and there is abundant evidence that they are under selection.9 Therefore, we evaluated the similarity between the distribution of the studied variants and a random set of non-ADME markers Data analysis across the genome. Procrustes analysis,10 a method for comparing spatial The intersection of the DMET assay markers with the 1000 Genomes phase maps of human population genetic variation, was used to conduct a 1 data set yielded a final analytic data set comprising 1156 markers from statistical comparison of the shape of the distribution of the ADME markers 212 ADME-related genes shared across all 19 populations. In addition, we against that of B13 000 randomly sampled single-nucleotide analyzed an ‘actionable’ subset of 42 variants identified by the FDA or the polymorphisms (SNPs; Supplementary Methods). Pharmacogenetics for Every Nation Initiative (www.pgeni.org) as important pharmacogenomic biomarkers. Markers in this subset are listed in various drug labels1 and have supporting evidence of clinical utility in the RESULTS pharmacogenomics database: PharmGKB (www.pharmgkb.org). Additional Global variation of ADME SNPs information describing the actionable pharmacogenomics markers included in this study is available in the Supplementary Methods. The mean DMAF for all 1156 SNPs tested was 0.25 across all Minor allele frequencies (MAFs) were computed using the variant allele populations; similarly, the mean DMAF was 0.29 for the selected of pharmacogenomic effect as the reference. The maximum MAF minus 42 ‘actionable’ variants across all samples (Table 1). About 80% of

The Pharmacogenomics Journal (2014), 217 – 222 & 2014 Macmillan Publishers Limited Pharmacogenomics and global populations E Ramos et al 219 the 1156 SNPs studied showed a DMAF of 0.05 across all population groups regardless of geography, ancestry or ethnicity; populations, illustrating the diversity in allele frequencies across SNP rs3918290 is located in the dihydropyrimidine dehydrogen- these pharmacogenomically relevant variants. Notably, frequen- ase gene (DPYD) and is associated with adverse effects from the cies for the reference alleles (that is, MAFs) of the selected 42 chemotherapeutic agent capecitabine.12 Capecitabine is clinically ‘actionable’ markers varied widely (Figure 2 and converted to 5-fluorouracil, which inhibits DNA synthesis in the Supplementary Figure 1). Some markers covered nearly the entire targeted tumor. frequency range. A good example of such a marker is SNP rs9923231 with a global MAF that ranged from 0.02 to 0.95; this SNP has been shown to influence warfarin dosing11—an Inter- and intra-continental variation anticoagulant prescribed to prevent blood clots. In contrast, Comparison of all variants or just the clinically ‘actionable’ variants some variants (for example, SNP rs3918290 with global MAF range revealed that the AFR þ AA populations displayed the greatest of 0–0.01), displayed similar allele frequencies across all difference in DMAF as expected from the well-documented rich genetic diversity of African ancestry populations. Although X90% of all ADME markers had similar frequency differences (less than a Table 1. Comparison of DMAF of ADME markers across global DMAF 0.2) within continental ancestral groups (Supplementary populations Table 2 and Supplementary Figure 2), the distribution of MAF differed when comparing the number of markers for a given All variants Actionable variants DMAF interval between continental ancestral groups. For example, 361 (31%) variants had less than a 0.05 difference in frequency Mean DMAF Mean DMAF across AFR þ AA populations compared with 553 (48%) and 799 Group (min, max) P-valuea (min, max) P-valuea (69%) for EUR or EAS populations, respectively (Supplementary Table 2). These differences in proportions were statistically ALL 0.25 (0, 1.0) — 0.29 (0, 0.93) — significant (Po1 10 4 for all pairwise comparisons: AFR þ AA AFR þ AA 0.10 (0, 0.64) — 0.11 (0, 0.34) — versus EUR, AFR þ AA versus EAS, EUR versus EAS). For variants EUR 0.07 (0, 0.37) o0.0001 0.08 (0, 0.29) 0.0855 with a DMAFX0.20, the majority were observed in AFR þ AA EAS 0.04 (0, 0.26) o0.0001 0.04 (0, 0.17) o0.0001 populations (129 SNPs, 11%) compared with EUR, EAS or AMR AMR 0.06 (0, 0.30) o0.0001 0.06 (0, 0.22) 0.01 populations, which had only 63 (5%), 7 (o1%) or 43 (4%) SNPs Abbreviations: ADME, absorption, distribution, metabolism and excretion; within that range, respectively. Moreover, for a given DMAF AFR þ AA, continental African samples plus the African American popula- interval, markers were not necessarily shared across ancestral tions (ASW and HUFS); AMR, Latin American samples (MXL, PUR and CLM); groups. Of the 63 SNPs in EUR populations with a DMAFX0.20, EUR, continental European and Centre d’Etude du Polymorphisme Humain only 19 SNPs were also observed with the same frequency range samples (CEU, TSI, GBR, FIN and IBS); EAS, continental East Asian samples in AFR þ AA samples, 5 SNPs in AMR samples and none in EAS (CHB, CHS and JPT); MAF, minor allele frequency. a samples. This implies that for a given continental ancestry Pairwise comparison of each continental group’s mean DMAF to AFR þ AA by non-parametric Mann–Whitney U-test. grouping, different sets of ADME markers were responsible for a given range of allele frequency differences between ethnic

Figure 2. Reference allele frequencies of actionable absorption, distribution, metabolism and excretion (ADME) single-nucleotide polymorphisms (SNPs). The subset of variants listed is considered clinically useful in the context of identifying variable drug response. Nineteen global populations are represented.

& 2014 Macmillan Publishers Limited The Pharmacogenomics Journal (2014), 217 – 222 Pharmacogenomics and global populations E Ramos et al 220 groups. Furthermore, there was no correlation between inter- detected from 277 individuals (an average of 2.2 homozygote continental allele frequency distribution and intra-continental genotypes per individual) indicating some individuals were allele frequency distribution among the ADME markers studied. homozygous for more than one risk allele. Ten individuals were homozygous for at least five actionable risk alleles and over 200 Population structure and pharmacogenomic variants individuals were homozygous for at least two risk variants demonstrating the limitation of extrapolating from population to Global populations are known to show genetic population the individual level. structure.13 We investigated the hypothesis that pharmacogenomic variants recapitulate this population structure. We computed principal components of the genotypes for the set of DISCUSSION ADME markers across all population samples (Supplementary Knowledge of drug target genes and genes involved in drug Figure 3) and compared this with PCs computed for a random set ADME remains critical in predicting therapeutic effect and/or of markers of equivalent size for the same number of chromo- 14,15 adverse drug response. Here, we present data on one of the somes from each population. As expected, all AFR populations largest sets of pharmacogenomic variants so far studied on 19 cluster tightly together with the notable exception of the Masai population groups from around the globe. Previous studies have ethnic group—MKK. The African-American samples (HUFS and either focused on single genes or a handful of genes and/or ASW) were anchored by the AFR populations as well as EUR utilized samples with small numbers from each population group populations consistent with the history of African and European (for example, the Human Genome Diversity Project panel in which admixture in African Americans. The AMR populations showed some populations have fewer than 30 individuals). Notably, we some separation from the EUR samples in the direction of the conducted de novo genotyping to increase the representation of AFR samples and the EAS populations constituted another major Sub-Saharan Africa populations and African Americans given the cluster. 17 10 paucity of comparative genetic variation data currently available Procrustes analysis verified concordance between population from these groups in public databases despite the fact that they and genetic data from ADME markers and nearly 13 000 randomly display the highest degree of genetic diversity compared with sampled genotypes for each population. The principal other human population groups. We have focused on analyzing components analysis plot illustrated separation between the spectrum of diversity across individuals, ethnic groups and ancestral groups but also indicated genetic diversity within a continental ancestry to provide new insights into some of the given ancestry. Highlighting this point were the differences in potential challenges that lie in the ongoing global effort to move MAF observed within closely related groups (Supplementary from group labels such as ancestry, ‘race’ and ethnicity to drug D Figure 1). For example, the MAF for rs3211371 was 0.20 for prescription tailored to an individual’s genetic background AFR þ AA populations; however, much of this frequency range (‘personalized medicine’). Overall, our data demonstrate that was attributable to MAF differences within a given country as ADME genetic variants show considerable differences in allele opposed to population samples between countries. This was frequency among global populations in general (Figure 2) as well evident within the two Nigerian samples; YRI had a MAF as among populations that are often grouped together by approaching monomorphic compared with IGBO, which had a continental origin, ancestry or ‘race’ (Figure 3). MAF of 0.16 (FST ¼ 0.07). The two Ghanian samples, GAA and At one level, our findings illustrate the utility of population data D AKAN, also showed a large MAF for that allele (MAF of 0.21 and for guiding clinical decision making in the absence of individual- 0.07, respectively) with a slightly smaller FST of 0.04. The Kenyan level genetic data. Actionable variants that are monomorphic (that samples, MKK and LWK, did not differ at that position (both were is, no variation) across all samples are a good example. Three of monomorphic). Interestingly, the African-American samples, ASW the 42 actionable SNPs we investigated were either monomorphic and HUFS also showed similar frequency differences (that is, large or showed a MAF ofo0.01 for all global populations tested. These D 18 MAF) with MAF of 0.02 and 0.15, respectively, (FST ¼ 0.05). Given SNPs have implications on toxicity of thiopurines (rs1800462), the variability observed in the AFR samples, which represent a toxicity of capecitabine and other cancer drugs (rs3918290),19 and large component of parental ancestry for these admixed groups, 20 clopidogrel responsiveness (rs28399504). We also saw examples differences between two African-American groups sampled from at the continental level such as AFR populations, which showed different parts of the United States may not be surprising although 16 seven SNPs that were monomorphic (African Americans (ASW and there is also the potential for assay artifact. Population HUFS) had MAF 0.02) each having direct clinical actionability. differentiation was also observed within the other continental o Similarly, 11 actionable SNPs were monomorphic across EUR samples. Over 90 SNPs had allele frequencies that generated FST populations with the exceptions of GBR and FIN for just one SNP X values 0.05 for at least one EUR sample pair and 49 SNPs for each and 17 SNPs in EAS populations, again with nominal AMR sample pairs. In summary, the principal components analysis exceptions (that is, MAFo0.01) for a handful of SNPs. Pending the showed the clustering of individuals relative to continental routine use of individualized genetic testing at point-of-care, ancestry (geography) and ethnic grouping. group data remains very useful in a number of ways. Regulatory bodies can use such data (in combination with clinical and Individual-level variation in burden of pharmacogenomic risk functional data) to formulate guidelines for genetic testing variants of pharmacogenomic variants (see the FDA’s Table of Group-level data are useful for understanding population Pharmacogenomic Biomarkers in Drug Labels21). In addition, frequencies. However, the individual is the subject at the clinical guidelines and drug labels can be fine grained using such group level. To illustrate the spectrum of variation of individual burden of data, for example: testing certain polymorphisms only in people of risk for ADME variants, we examine the set of clinically ‘actionable’ specific ancestry (for example, dermatologic reactions from SNPs in EAS populations (the ancestral group with the smallest carbamazepine for individuals carrying the HLA-B*1502 allele).21 DMAF for all actionable SNPs). Among EAS populations (n ¼ 286), Second, such data can be used to guide specific the average MAF across all 42 actionable SNPs for these three pharmacogenetic-based dosing guidelines, for example: warfarin populations was just 0.13. We focused on individuals homozygous dosing guidelines based on VKORC1 rs99923231, CYP2C9*2 and for a risk allele given that these persons are likely to experience CYP2C9*3 alleles do not explain much of the variation of the more severe phenotype than heterozygotes. We identified at dose in African ancestry populations because these variants least one individual homozygous for the risk allele in 20 of the are monomorphic or near-monomorphic.22,23 Third, national 42 actionable SNPs. In total, 632 homozygous genotypes were regulatory bodies can use such data to guide their policies

The Pharmacogenomics Journal (2014), 217 – 222 & 2014 Macmillan Publishers Limited Pharmacogenomics and global populations E Ramos et al 221

Figure 3. Continental-level differences of absorption, distribution, metabolism and excretion (ADME) single-nucleotide polymorphisms (SNPs). The markers are sorted by increasing DMAF for the full ADME data set (a) and the actionable SNP subset (b). MAF, minor allele frequency.

Figure 4. Snapshot of actionable single-nucleotide polymorphisms (SNPs) minor allele frequencies (MAFs) and associated clinical implications. Examples of clinical importance are dosage (D), toxicity (T), efficacy (E) and drug clearance (C). The vertical bars indicate prevalence of MAF for each SNP listed. All bars represent real MAFs values. tailored to their specific populations. A good example of this is EUR, EAS and AMR populations. Many of the extreme examples of how the Singapore Health Services Authority used group genetic population differentiation observed have been shown to be data on the country’s main ethnic groups to request revision of driven by recent positive selection in ADME genes.9 the package insert for irinotecan (to include the pharmacogenetic Finally, findings from this study illustrate the potential pitfalls in association with severe neutropenia) and to publicize the the use of demographic labels such as ‘black’ or ‘white’ in the association and availability of a genotyping test.24 Thus, these practice of medicine as we have alluded to in previous reports.27 kinds of data are of high public health and clinical relevance, For example, CYP2D6 enzymatic activity has been linked to serving as part of the necessary evidence base for translation of tamoxifen efficacy in breast cancer patients. The CYP2D6*2A pharmacogenomics findings into routine clinical care (Figure 4). haplotype contains the SNP rs16947,19,28 which has the SNP has a The findings of this study also serve as a powerful reminder of MAF of 0.43 in GAA to 0.76 in MKK samples (range ¼ 33%; the stark differences in allele frequencies among population FST ¼ 0.11), two populations generally referred to as ‘black’. In groups and the direct clinical relevance of those differences. For addition, acetylator phenotypes of NAT2 are predicted in part by example, three core variants (rs179983, rs1057910 and rs9923231) rs1801280;29 this SNP has a MAF of 0.71 in IBS samples but are typically used for estimating warfarin sensitivity. In the case of observed to be as low as 0.42 in CEU (FST ¼ 0.09). Interestingly, the rs9923231, the mean MAF for EAS populations is 0.92 compared AMR samples all fell below a MAF of 0.40 for this SNP making with 0.06 for AFR populations indicating strong population ‘white’ or even ‘Latino’ or ‘Hispanic’ poor proxies for peoples of 11,25 differentiation (FST 0.51–0.87). Using current recommended Spanish Iberia for an allele that is found in less than half of the dosing algorithms, the dosing range for AFR þ AA populations will population of some European and Latin American populations but be 5.0–7.0 mg of warfarin per day with only 3% of individuals from nearly three-quarters of the IBS population. Traditional labels of these groups deviating from the recommended range.26 In sharp race and ethnicity are often used in research studies, clinical contrast, nearly 25% of the EAS individuals sampled in this study medicine and public health as proxies (albeit imperfect ones) for differ from the majority recommended dosing of 3.0–4.0 mg of unmeasured environmental and social covariates. When used to warfarin per day based on their genotype data with some guide drug choice and dosage, they are also imperfect proxies for individuals expected to respond better to lower doses (0.5–2.0 mg the unmeasured genotype. Until genotyping for pharmaco- per day) or higher doses (5.0–7.0 mg per day). In addition, two genomic biomarkers becomes universal and incorporated into variants (rs2256871 and rs28371685 in CYP2C9*9 and CYP2C9*11, routine clinical practice, traditional classifications of race/ethnicity respectively) with enzymatic activities observed in AFR þ AA will continue to be used for categorization of individuals. We population at a high frequency of 0.15 are monomorphic in the anticipate that improvements in technology, falling costs and

& 2014 Macmillan Publishers Limited The Pharmacogenomics Journal (2014), 217 – 222 Pharmacogenomics and global populations E Ramos et al 222 better guidelines for use of pharmacogenomic biomarkers in 4 Deeken J. The Affymetrix DMET platform and pharmacogenetics in drug clinical decision-making will gradually lead to replacement of the development. Curr Opin Mol Therapeut 2009; 11: 260–268. race/ethnicity label (a blunt tool) with the more precise genotype. 5 VeraCode ADME core panel. http://res.illumina.com/documents/products/ As we take steps toward the integration of genomic medicine datasheets/datasheet_veracode_adme_core_panel.pdf. into day-to-day clinical care, the practice of medicine will benefit 6 Rotimi CN, Dunston GM, Berg K, Akinsete O, Amoah A, Owusu S et al. In search of from studies that incorporate pharmacogenomics data from susceptibility genes for type 2 diabetes in West Africa: the design and results of individuals sampled from multiple ancestral backgrounds across the first phase of the AADM study. Ann Epidemiol 2001; 11: 51–58. the world.23,30 Expanding the evidence base to include 7 Adeyemo A, Gerry N, Chen G, Herbert A, Doumatey A, Huang H et al. A genome- wide association study of hypertension and blood pressure in African Americans. multiple global populations will facilitate clinical decision PLoS Genet 2009; 5: e1000564. making and provide useful data for regulatory bodies to utilize 8 Shriner D. Investigating population stratification and admixture using eigenana- in policy recommendations about drug labels and genetic testing lysis of dense genotypes. Heredity (Edinb) 2011; 107: 413–420. recommendations. 9 Li J, Zhang L, Zhou H, Stoneking M, Tang K. Global patterns of genetic diversity and signals of natural selection for human ADME genes. Hum Mol Genet 2011; 20: 528–540. PANEL: RESEARCH IN CONTEXT 10 Wang C, Szpiech ZA, Degnan JH, Jakobsson M, Pemberton TJ, Hardy JA et al. Systematic review Comparing spatial maps of human population-genetic variation using Procrustes analysis. Stat Appl Genet Mol Biol 2010; 9: Article 13. We did a PubMed search for ‘pharmacogenomics’ or ‘pharmaco- 11 Limdi NA, Veenstra DL. Warfarin pharmacogenetics. Pharmacotherapy 2008; 28: genetics’ and ‘global’ and ‘populations’, which yielded 49 citations. 1084–1097. However, most were either of a single gene, genetic variants for a 12 Schwab M, Zanger UM, Marx C, Schaeffeler E, Klein K, Dippon J et al. Role of single drug or a limited number of populations. One study did genetic and nongenetic factors for fluorouracil treatment-related severe toxicity: a include a large number of pharmacogenes as well as a wide range prospective clinical trial by the German 5-FU Toxicity Study Group. J Clin Oncol of global populations.9 However, many of the included populations 2008; 26: 2131–2138. in that study had few individuals and the study focused primarily 13 1000 Genomes Project Consortium et al. A map of human genome variation from on population genetic parameters and signals of natural selection. population-scale sequencing. Nature 2010; 467: 1061–1073. 14 Tishkoff SA, Reed FA, Friedlaender FR, Ehret C, Ranciaro A, Froment A et al. The genetic structure and history of Africans and African Americans. Science (New Interpretation York, NY) 2009; 324: 1035–1044. We examined 1156 genetic variants in 212 genes involved in drug 15 Shriner D. Improved eigenanalysis of discrete subpopulations and admixture ADME of drugs in 19 populations (n ¼ 1478). These pharmacoge- using the minimum average partial test. Human Heredity 2012; 73:73–83. 16 Clayton DG, Walker NM, Smyth DJ, Pask R, Cooper JD, Maier LM et al. Population nomic variants showed marked variation between and within structure, differential bias and genomic control in a large-scale, case-control continental groupings of the populations studied. Group labels association study. Nat Genet 2005; 37: 1243–1246. often used in clinical settings (that is, race labels) did not 17 Rosenberg NA, Huang L, Jewett EM, Szpiech ZA, Jankovic I, Boehnke M. accurately portray the underlining genetics of an individual. This Genome-wide association studies in diverse populations. Nat Rev Genet 2010; 11: implies that individual genotype data is the best way of evaluating 356–366. a patient’s pharmacogenetic risk profile. However, group data 18 Larussa T, Suraci E, Lentini M, Nazionale I, Gallo L, Abenavoli L et al. High remain essential for developing recommendations for genetic prevalence of polymorphism and low activity of thiopurine methyltransferase in testing, targeted therapeutics and drug labeling. Enlarging the patients with inflammatory bowel disease. Eur J Intern Med 2012; 23: 273–277. scope of pharmacogenomics research to include multiple global 19 Dai Z, Papp AC, Wang D, Hampel H, Sadee W. Genotyping panel for assessing response to cancer chemotherapy. BMC Med Genomics 2008; 1:24. populations can improve the evidence base for clinical translation 20 Santos PC, Soares RA, Santos DB, Nascimento RM, Coelho GL, Nicolau JC et al. and provide a starting point for studies that relate genotype to CYP2C19 and ABCB1 gene polymorphisms are differently distributed according to drug efficacy, toxicity and dosage guidelines. ethnicity in the Brazilian general population. BMC Med Genet 2011; 12:13. 21 Tegretol Drug Label http://www.accessdata.fda.gov/drugsatfda_docs/label/2012/ 016608s107,018281s055,018927s048,020234s040lbl.pdf. CONFLICT OF INTEREST 22 Langley MR, Booker JK, Evans JP, McLeod HL, Weck KE. Validation of clinical The authors declare no conflict of interest. testing for warfarin sensitivity: comparison of CYP2C9-VKORC1 genotyping assays and warfarin-dosing algorithms. J Mol Diagn 2009; 11: 216–225. 23 Ross KA, Bigham AW, Edwards M, Gozdzik A, Suarez-Kurtz G, Parra EJ. Worldwide ACKNOWLEDGEMENTS allele frequency distribution of four polymorphisms associated with warfarin dose requirements. J Hum Genet 2010; 55: 582–589. The informatics expertise of Kevin Long, University of North Carolina, Chapel Hill is 24 Sung C, Lee PL, Tan LL, Toh DS. Pharmacogenetic risk for adverse reactions to greatly appreciated. The study was supported by National Institutes of Health grants irinotecan in the major ethnic populations of Singapore: regulatory evaluation by S06GM008016-320107 to CNR and S06GM008016-380111 to AA. HUFS participants the health sciences authority. Drug Saf 2011; 34: 1167–1175. were enrolled at the Howard University General Clinical Research Center, which is 25 Bhatia G, Patterson N, Pasaniuc B, Zaitlen N, Genovese G, Pollack S et al. Genome- supported by grant 2M01RR010284 from the former National Center for Research wide comparison of African-ancestry populations from CARe and other cohorts Resources, National Institutes of Health. This research was supported in part by the reveals signals of natural selection. Am J Human Genet 2011; 89: 368–381. Intramural Research Program of the Center for Research on Genomics and Global 26 Johnson JA, Gong L, Whirl-Carrillo M, Gage BF, Scott SA, Stein CM et al. Clinical Health. The Center for Research on Genomics and Global Health is supported by the Pharmacogenetics Implementation Consortium Guidelines for CYP2C9 and National Human Genome Research Institute, the National Institute of Diabetes and VKORC1 genotypes and warfarin dosing. Clin Pharmacol Ther 2011; 90: 625–629. Digestive and Kidney Diseases, the Center for Information Technology and the Office 27 Rotimi CN, Jorde LB. Ancestry and disease in the age of genomic medicine. N Engl of the Director at the National Institutes of Health (Z01HG200362). JMed2010; 363: 1551–1558. 28 Serrano D, Lazzeroni M, Zambon CF, Macis D, Maisonneuve P, Johansson H et al. Efficacy of tamoxifen based on cytochrome P450 2D6, CYP2C19 and SULT1A1 REFERENCES genotype in the Italian Tamoxifen Prevention Trial. Pharmacogenomics J 2011; 11: 1 Table of pharmacogenomic biomarkers in drug labels http://www.fda.gov/drugs/ 100–107. scienceresearch/researchareas/pharmacogenetics/ucm083378.htm. 29 He YJ, Shapero MH, McLeod HL. Novel tagging SNP rs1495741 and 2-SNPs 2 Wang L, McLeod HL, Weinshilboum RM. Genomics and drug response. N Engl J (rs1041983 and rs1801280) yield a high prediction of the NAT2 genotype in Med 2011; 364: 1144–1153. HapMap samples. Pharmacogenet Genomics 2012; 22: 322–324. 3 Schwab M, Schaeffeler E. Pharmacogenomics: a key component of personalized 30 Ramos E, Callier SL, Rotimi CN. Why personalized medicine will fail if we stay the therapy. Genome Med 2012; 4: 93. course. Personalized Med 2012; 9: 839–847.

Supplementary Information accompanies the paper on the The Pharmacogenomics Journal website (http://www.nature.com/tpj)

The Pharmacogenomics Journal (2014), 217 – 222 & 2014 Macmillan Publishers Limited