* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project
Download PersPectIves
Compounding wikipedia , lookup
Polysubstance dependence wikipedia , lookup
Neuropharmacology wikipedia , lookup
Drug design wikipedia , lookup
Neuropsychopharmacology wikipedia , lookup
Prescription drug prices in the United States wikipedia , lookup
Pharmaceutical industry wikipedia , lookup
Pharmacognosy wikipedia , lookup
Prescription costs wikipedia , lookup
Drug interaction wikipedia , lookup
Drug discovery wikipedia , lookup
Theralizumab wikipedia , lookup
Perspectives opinion Defining drug disposition determinants: a pharmacogenetic– pharmacokinetic strategy David A. Katz, Bernard Murray, Anahita Bhathena and Leonardo Sahelijo Abstract | In preclinical and early clinical drug development, information about the factors influencing drug disposition is used to predict drug interaction potential, estimate and understand population pharmacokinetic variability, and select doses for clinical trials. However, both in vitro drug metabolism studies and pharmacogenetic association studies on human pharmacokinetic parameters have focused on a limited subset of the proteins involved in drug disposition. Furthermore, there has been a one-way information flow, solely using results of in vitro studies to select candidate genes for pharmacogenetic studies. Here, we propose a twoway pharmacogenetic–pharmacokinetic strategy that exploits the dramatic recent expansion in knowledge of functional genetic variation in proteins that influence drug disposition, and discuss how it could improve drug development. Drug disposition is influenced by drug metabolizing enzymes (DMEs), drug transport proteins (DTPs), serum binding proteins and transcription factors that regulate DME and DTP expression (BOX 1). Foreknowledge of the specific proteins that influence the disposition of a new chemical entity (NCE) is an important goal of preclinical and early clinical drug develop‑ ment. Early availability of this information enables mathematical modelling of the drug interaction potential of an NCE using quantitative kinetic parameters of specific DMEs1. This can therefore lead to better projection of doses for subsequent studies. Although tools exist to assess the role of most of these proteins for the disposi‑ tion of an NCE, drug developers typically learn only about a limited number of them (that is, several cytochrome P450s (CYPs), a few additional DMEs and the DTP P‑glycoprotein), and generally do not know the relative clinical importance of most of the more than 170 drug disposition pathways (TABLES 1,2; BOX 2; Supplementary information S1 (table), S2 (table), S3 (table), S4 (table)). Much of the knowledge about drug disposition determinants comes from academic laboratories after a drug is marketed. Such knowledge has been highly beneficial for patients; for example, when the prevalence of severe adverse events in thiopurine S-methyl transferase (TPMT) poor metabolizers dosed with thiopurines was identified2,3. However, for many drugs, the pathways that determine pharmacokinetic variation remain unknown even years after regulatory approval. This limits the ability of drug developers to identify, manage and understand the consequences of pharmacokinetic variability for efficacy, safety and drug–drug interactions. For example, statins were used by many millions of people before it was discovered that their pharmaco kinetic variability, drug interactions, efficacy and safety might be dependent on the DTP solute carrier organic anion transport protein 1B1 (OATP1B1)4–6. Association between a gene variant and drug pharmacokinetics implies a mechanistic role of the gene product in drug disposition, and the potential outcome of making such associations is to learn the nature reviews | drug discovery extent to which any of a wide range of factors influences the disposition of a drug. Increased knowledge and more accessible technology should now make it easier for drug developers to study which pathways are responsible for the disposition of a drug. In this article, after a brief overview of the preclinical characterization of drug disposition, we summarize current know ledge on pharmacogenetics and drug dispo sition. We then propose a new approach in which pharmacogenetic results derived from early clinical studies can both feed back to additional targeted in vitro studies and feed forward to optimize later-stage, larger clinical trials for NCEs, contribute to more informative drug labels, and thereby potentially enable better drug use. Preclinical drug disposition assessment Preclinical characterization of drug dis position generally involves the assessment of individual proteins for their role in the disposition of the NCE. A summary of currently available tools for such studies, including purified or recombinant proteins, selective substrates and inhibitors is provided in Supplementary information S5 (table). Although many purified or recombinant DMEs — particularly members of the human CYP, flavin monooxygenase (FMO), monoamine oxidase (MAO), UDP glucu ronosyltransferase (UGT), sulphotransferase (SULT), N-acetyltransferase (NAT) and glutathione S-transferase (GST) families — are commercially available, there are substantial gaps in the availability of purified or recombinant DMEs from other families. Several DTPs have been functionally expressed in recombinant cell-based assays, but these are generally not commercially available. When purified or recombinant protein is not available, selective inhibitors (if known and commercially available) can be useful to deconstruct the disposition process in perfused organs, tissue slices, isolated cells or subcellular fractions. However, sufficiently selective inhibitors that distinguish between closely related proteins have only been established for a few DME or DTP families. Even among the CYPs, it is still not possible to fully distinguish between volume 7 | april 2008 | 293 © 2008 Nature Publishing Group Perspectives Box 1 | Primer on drug disposition determinants Four main types of drug disposition determinants are known, and these are outlined below. Drug metabolizing enzymes Drug metabolizing enzymes (DMEs) are proteins that catalyse the chemical transformation of drugs and other small molecules. DMEs can be thought of as the body’s chemical defence system (although there are certainly examples of DMEs converting a drug to a more toxic metabolite), but also function in the synthesis and degradation of endogenous chemicals (for example, neurotransmitters, nucleotides). Drugs are mainly metabolized in the liver and intestines, but also in other organs. Broadly, DMEs can be divided into two categories: functionalizing enzymes and conjugating enzymes. Functionalizing enzymes introduce or reveal functional groups in a substrate through oxidation (for example, cytochrome P450s), reduction (for example, aldo-keto reductases) or hydrolysis (for example, epoxide hydrolases). Conjugating enzymes transfer moieties from a cofactor to a substrate (for example, N‑acetyltransferases). Previously, functionalizing and conjugating enzymes were referred to as catalysing ‘phase I’ and ‘phase II’ reactions. This terminology is losing popularity, in part because it leads to the erroneous expectation that the processes are ordered and sequential. Drug transport proteins Drug transport proteins (DTPs) facilitate the movement of drugs and other small molecules across cellular membranes. The intestines, liver, kidney and brain are the organs where facilitated drug transport most affects pharmacology, but DTPs are present in all tissues. There are two main families of DTPs. Members of the ATP-binding cassette (ABC) family utilize the energy of ATP hydrolysis for the purpose of carrying small molecules across biological membranes, either against or with concentration gradients. This process is known as active transport. Examples include P‑glycoprotein and the multidrug resistance proteins. Members of the solute-linked carrier (SLC) family enable the passage of small molecules across biological membranes along concentration gradients. This process is known as facilitated diffusion. Examples include organic ion transporters, organic anion transport proteins and nucleoside transporters. Serum binding proteins Serum binding proteins often bind a large fraction of a drug that is in central circulation — sometimes more than 99%. Protein binding affects pharmacology as only unbound (free) drug can enter tissues. The main serum binding proteins are albumin and α‑1 acid glycoprotein. Transcription factors Transcription factors are proteins that regulate gene expression. When bound to a range of exogenous chemicals, certain transcription factors induce the expression of DMEs and DTPs. Pregnane X receptor and aryl hydrocarbon receptor are among the best-characterized transcription factors that are relevant for drug disposition. the four CYP3A enzymes, and selective inhibitors for some CYPs (for example, CYP2C18, CYP2G, CYP2R1, CYP2S1) have not been identified. Indeed, little is known at all about these and a number of other DMEs. When a selective inhibitor is not available, it may be possible to take advantage of tissue-selective expression of particular isoforms within a protein family to learn the extent to which each can influence NCE disposition (for example, the roles of FMO1, FMO2 and FMO3 can be assessed separately in kidney, lung and liver using the same inhibitor). When neither recombinant protein nor selective inhibitor for a DME or DTP is available, observing that an NCE is a competitive inhibitor (in perfused organs, tissue slices, isolated cells or subcellular fractions) towards a known selective sub‑ strate (again, if known and commercially available) might indicate that a particular DME or DTP is important for the NCE’s disposition. If a drug induces DME or DTP expression by a transcription factor binding in a heterologous transcription activation assay, it might autoinduce its own disposition as well as that of other drugs. To provide a comprehensive in vitro survey of drug disposition determinants, a laboratory needs the capability to per‑ form the diverse assay types mentioned above, many of which must be developed in-house. Because of the current lack of a comprehensive toolset and the resources required — and also because there has not been a regulatory imperative for additional investigation — a typical pre‑ clinical in vitro survey has covered only several DMEs (mainly CYPs) and the DTP P‑glycoprotein. Knowing whether these factors can influence the disposition of an NCE provides valuable but far from 294 | april 2008 | volume 7 comprehensive information to predict drug interaction potential, estimate and understand population pharmacokinetic variability, and select doses for subsequent clinical trials. Elucidation of the specific set of disposition pathways that are important for a particular NCE’s disposition has not been achieved, mainly because the neces‑ sary resources have not been available. There is a need for an improved toolset to identify the most important proteins that influence the disposition of an NCE. This toolset should be more comprehensive in coverage of DMEs, DTPs and other factors; less diverse in assay types; feasible during preclinical or early clinical development; and affordable. We propose that a strategy based on the growing knowledge of the influence of pharmacogenetic factors on drug disposi‑ tion (summarized in the following section) can help provide that toolset. Pharmacogenetics and drug disposition A genetic component of pharmacokinetic variability was postulated more than 100 years ago by Archibald Garrod in studies of patients with alkaptonuria7. Half a century later, several drugs were shown to have indistinguishable disposition in mono zygotic twins, but often distinct disposition in dizygotic twins (for example, phenyl butazone8). These results established drug disposition as a heritable trait. Deficiencies of the DMEs NAT and butyrylcholin esterase (BCHE) were later identified as risk factors for adverse effects of isoniazid9 and succinylcholine10, respectively, and the genetic basis for these11–15 and other DME poor metabolizer phenotypes (for example, CYP2D616–20, TPMT21–23) were discovered around 1990. By the 1990s, there was an understand‑ able reticence in the pharmaceutical industry to develop drugs that were substrates of these few polymorphic DMEs because of the likelihood of variable pharmacokinetic and drug–drug interactions. However, there were exceptions: atomoxetine, a sensitive CYP2D6 substrate, was approved for use in 2002 (REF. 24). Also in the 1990s, drug developers began to utilize pharmacogenetic studies to learn the magnitude of pharmaco kinetic variability of NCEs that could be attributed to genetic variation. In general, these studies focused on the few DMEs in which there were known polymorphisms, and were undertaken only when they were considered necessary. That is, when in vitro results indicated that the NCE was a substrate of the polymorphic DME (hypothesis-based experiments). www.nature.com/reviews/drugdisc © 2008 Nature Publishing Group Perspectives Table 1 | Consistently replicated associations between genotype and clinical pharmacokinetics Protein name (gene name) Representative affected drugs References‡ Alcohol dehydrogenase 2 (ADH2) Ethanol OMIM entry for ADH2 and references therein Alcohol dehydrogenase 4 (ADH4) Ethanol 64 Aldehyde dehydrogenase 2 (ALDH2) Ethanol OMIM entry for ALDH2 and references therein Cytochrome P450 2A6 (CYP2A6) Nicotine 65–70 Cytochrome P450 2B6 (CYP2B6) Efavirenz, nelfinavir 71–73 Cytochrome P450 2C19 (CYP2C19) Omeprazole, lansoprazole, pantoprazole, rabeprazole, trimipramine, amitryptyline, imipramine, clomipramine 44,74,75 and references therein Cytochrome P450 2C9 (CYP2C9) Phenytoin, warfarin, tolbutamide, glipizide, celecoxib, fluvastatin 76 and references therein Cytochrome P450 2D6 (CYP2D6) Imipramine, perphenazine, trimipramine, desipramine, nortryptyline, clomipramine, tamoxifen 44 and references therein Cytochrome P450 3A5 (CYP3A5) Tacrolimus, saquinavir 77–84 Dihydropyrimidine dehydrogenase (DPYD) 5-Fluorouracil OMIM entry for DPYD and references therein Flavin monooxygenase 3 (FMO3) Sulindac 85,86 N-acetyltransferase 2 (NAT2) Isoniazid, hydralazine, procainamide, dapsone, sulphamethazine OMIM entry for NAT2 and references therein Organic anion transport protein 1B1* (SLCO1B1) Pravastatin, rosuvastatin 30–33 Pseudocholinesterase (BCHE) Suxamethonium OMIM entry for BCHE and references therein Thiopurine S-methyltransferase (TPMT) 6-Mercaptopurine, 6-thioguanine, azathioprine 2 and references therein Uridine glucuronosyltransferase 1A1 (UGT1A1) Irinotecan 87 and references therein *Also known as OATP-C or OATP2. ‡Online Mendelian Inheritance in Man (OMIM) database web site: http://www.ncbi.nlm.nih.gov/sites/entrez?db=omim. Until recently, limited knowledge about functional genetic variation in DMEs (and none about polymorphism in other drug disposition factors) substantially limited the scope of pharmacogenetics–pharmaco kinetics research. However, in the past few years, substantial knowledge in this field has been accumulated, and a review of the literature reveals that there are now over 170 gene products known or expected to have a role in drug disposition (BOX 3). These include not only numerous DMEs and DTPs, but also abundant serum binding proteins and regulatory (transcription) factors that control the expression of DMEs and DTPs. More than half of the corresponding genes are known to be poly‑ morphic (TABLES 1,2; BOX 2; Supplementary information S1 (table)); most that are not known to contain common functional poly‑ morphisms (Supplementary information S2 (table), S3 (table), S4 (table)) have not been adequately studied to state with certainty that they do not. The 16 proteins involved in drug disposition for which consistently replicated associations between variants in the corresponding genes and the human pharmacokinetics of at least one drug have been published are shown in TABLE 1. With the exception of FMO3, for which only two reports showing relationship to sulindac pharmacokinetics were found, there were three or more consistent reports for at least one drug for each gene. FMO3 was included in this group because the established associ‑ ation of variants in this gene with fish-odour syndrome25 is additional evidence supporting its relevance for xenobiotic disposition. Nearly all the genes listed in TABLE 1 encode DMEs, although there is also a gene that encodes a DTP (OATP1B1). There is robust scientific evidence showing that each gene contains common variants (combined minor allele frequency of variants sharing a phenotype ≥5%), which have substantial effects on human pharmacokinetics of one or more drugs (see table 1 for references). However, this does not mean that the rele vance of these variants for clinical practice has been established; in fact, dose adjust‑ ments or contraindications based on only CYP2D6, CYP2C9, TPMT and UGT1A1 are currently included in US drug labels26. TABLE 2 shows the 18 genes for which common variants are likely to have a role in nature reviews | drug discovery drug disposition, having been shown to be associated with the human pharmacokinetics of one or more drugs, albeit in single studies. Of these, 15 encode DMEs, 2 encode DTPs and 1 encodes a serum binding protein (α‑1 acid glycoprotein, gene = ORM1). The preponderance of DMEs in TABLE 1 and TABLE 2 reflects that DMEs were the sole focus of pharmacogenetics research related to drug disposition until recently. Many of the studies cited in TABLE 2 are relatively recent, and might therefore be replicated in the near future. As this occurs, the range of genes established as polymorphic deter‑ minants of human pharmacokinetics will expand. In addition to pharmacokinetics, several of these genes have also been associated with drug efficacy or safety. Common variants of at least 55 genes encoding drug disposition factors have func‑ tional effects on protein activity or expression (Supplementary information S1 (table); BOX 2), but have not yet been associated with human pharmacokinetics for any drug. It is currently unknown how many of these genes will be found to have meaningful influence on human pharmacokinetics. Several of them encode DMEs such as CYP3A4 and the volume 7 | april 2008 | 295 © 2008 Nature Publishing Group Perspectives Table 2 | Associations between genotype and clinical pharmacokinetics* Protein name (gene name) α1-Acid glycoprotein (ORM1) Alcohol dehydrogenase 3 (ADH3) Breast cancer resistance protein (ABCG2) Catechol-O-methyltransferase (COMT) Cytidine deaminase (CDA) Cytochrome P450 1A2 (CYP1A2) Cytochrome P450 2C8 (CYP2C8) Deoxycytidine kinase (DCK) Epoxide hydrolase 1 (EPHX1) Glutathione-S-transferase µ (GSTM1) Glutathione-S-transferase π (GSTP1) Glutathione-S-transferase θ (GSTT1) Inosine triphosphatase (ITPA) Multidrug resistance-associated protein 2 (ABCC2) Nicotinamide-Nmethyltransferase (NNMT) Uridine glucuronosyltransferase 1A6 (UGT1A6) Uridine glucuronosyltransferase 1A9 (UGT1A9) Uridine glucuronosyltransferase 2B15 (UGT2B15) Notes • Variants observed to be associated with the unbound serum fraction of quinidine • Associations related to ethanol are controversial • ADH3 is adjacent to ADH2 (Table 1) in the genome, so positive results for ADH3 may reflect effects of ADH2 • ADH2 protein has a much lower Km for ethanol and has a greater role in its metabolism • Variant observed to be associated with the pharmacokinetics of diflomotecan • In one study, low activity variant was associated with low levodopa dose in patients with Parkinson’s disease • The same variant was not associated with levodopa pharmacokinetic parameters in a similar study • Earlier work associated levodopa to erythrocyte COMT enzyme activity but not to genotype • Progression rate and survival of patients with cancer treated with gemcitabine have been associated with CDA activity • A CDA variant has been shown to reduce CDA activity in vitro • One promoter variant has been associated with increased enzyme inducibility and another with reduced theophylline clearance • However, a CYP1A2 that can unequivocally be used to predict the metabolic phenotype in any individual patient variant has yet to be identified • Reduced function variants have been associated in single studies with reduced clearance of r-ibuprofen and repaglinide • The most common reduced-function variant is in strong linkage disequilibrium with reduced-function variants of CYP2C9; this complicates interpretations of CYP2C8 pharmacogenetic results • DCK deficiency may be important in AraC resistance; two promoter variants have been associated with increased DCK expression and favourable response to therapy • EPHX1 genotype has been associated with carbamazepine metabolic ratio • Low epoxide hydrolase activity appears to be a risk factor for congenital malformations in infants of mothers who take phenytoin; however, association of a specific genetic variant has not been established • Dual GST M1–T1 null (gene deletion) genotype was associated with increased risk for tacrine hepatotoxicity • Gene duplication (probably uncommon) affecting activity has been reported in a Saudi population • A variant has been shown to alter catalytic activity • Associations with progression-free surivival in multiple myeloma and colorectal cancer, susceptibility to therapy-related acute myelogenous leukaemia, and CNS relapse-free survival in acute lymphocytic leukaemia, have each been reported in single studies • Dual GST M1–T1 null (gene deletion) genotype was associated with increased risk for tacrine hepatotoxicity • Two research groups have observed an association of the 94A allele with adverse events of azathioprine in patients with inflammatory bowel disease, but this finding was not replicated by three other groups • An ABCC2 genotype was associated with higher areas under the curve for irinotecan and its metabolites • An intronic variant was linked to plasma homocysteine levels • Promoter and coding region variants are associated with reduced hepatic expression levels and decreased enzymatic activity for most substrates • An interaction of UGT1A6 variant genotype and use of aspirin/NSAID to lower risk for colorectal adenoma has been observed in two studies • In another study, the same UGT1A6 genotype was associated with increased colorectal cancer risk; so, it is possible that the interaction is due either to genetic modulation of drug effect, or drug modulation of a genetic role in disease • An amino-acid change has reduced in vitro activity • Several promoter polymorphisms have been identified; one that was associated with higher transcription in a heterologous expression assay was not found to alter protein level or activity in human livers, but was associated with efficacy and safety of combination irinotecan/capecitabine • Protein level and activity in human livers was associated with two other promoter variants (one higher, one lower) • A variant has been associated with decreased metabolism of lorazepam and estradiol, and with increased progression-free survival in patients with breast cancer treated with tamoxifen; this last observation might be explained by effects on the drug or on oestrogen levels. • A different non-synonymous SNP is common in Japanese individuals References 88 89 90 91–93 94,95 96–100 101,102 103 104; OMIM‡ entry for EPHX1 and refs therein 105–107 108 and refs therein; 109–114 107,115 116–120 121 122,123 124–131 40,132–136 137–141 *Consistent replication is so far lacking. ‡Online Mendelian Inheritance in Man (OMIM) database web site: http://www.ncbi.nlm.nih.gov/sites/entrez?db=omim. NSAID, non-steroidal anti-inflammatory drug; SNP, single nucleotide polymorphism. 296 | april 2008 | volume 7 www.nature.com/reviews/drugdisc © 2008 Nature Publishing Group Perspectives UGTs, which are known to metabolize a wide range of drugs27,28, and these may be among those most likely to be found to be relevant for human pharmacokinetic variability. Finally, DMEs, DTPs and regulatory factors for which the relevance of genetic variation to drug disposition has not been established are shown in Supplementary information S2 (table), S3 (table) and S4 (table). Proposal for a new PG–PK strategy The increase in knowledge about phar‑ macogenetics and drug disposition sum‑ marized above — coupled with technology advancements that have made genotyping more affordable (with costs of less than US$1 per genotype for multiple available technologies) in the past decade — have made broad application of pharmacogenet‑ ics in most drug development programmes more feasible. Several platforms have been developed to screen the known functional variants that might influence drug disposi‑ tion in the context of early clinical trials from which high-quality pharmacokinetic data are also available. The approach we propose includes a broad evaluation of genotype–pharmaco kinetic relationships during early drug development, together with in vitro studies, to first generate and then confirm hypoth‑ eses about the pathways that are major disposition determinants of an NCE. Our core strategy for incorporation in NCE development comprises five steps (FIG. 1). The first step is to conduct in vitro experiments before clinical trials to assess whether, and to what extent, selected DMEs and DTPs might influence disposition of the NCE. This activity provides valuable information about the potential role of proteins that are known to influence the disposition of many drugs and to mediate a number of clinically important drug inter‑ actions, such as CYP3A and P‑glycoprotein. These experiments have demonstrated utility in drug development planning and decision-making. Additional in vitro assays may become standard on the basis of new information about drug disposition and interaction factors. For example, recent results from our group29 and others30–33 have demonstrated that the hepatic uptake DTP OATP1B1 (SLCO1B1) is a meaningful determinant of drug disposition for statins and other drugs, and that OATP1B1 inhibi‑ tors may have drug interaction potential. Recombinant OATP1B1 can be expressed in cell culture, and selective substrates and inhibitors are available (Supplementary Box 2 | Genes with common functional variants not yet associated with pharmacokinetics Drug metabolizing enzymes • Cytochrome P450 (CYP) family. CYP1A1, CYP1B1, CYP2A13, CYP2E1, CYP2G1, CYP2G2, CYP2J2, CYP3A4, CYP3A7, CYP3A43, CYP4B1, CYP4F12. • Other oxidoreductases. CBR3 (carbonyl reductase 3), FMO1 (flavin monooxygenase 1), FMO2, MAOA (monoamine oxidase A), MAOB, PON1 (paraoxonase 1), NQO1 (quinone NADPH dehydrogenase 1), NMOR2 (quinone NADPH dehydrogenase 2), ACADS (short chain acyl-coA dehydrogenase). • Glutathione‑S-transferase (GST) family. GSTA1, GSTA2, GSTM3, GSTO1, GSTO2, GSTZ1. • Sulphotransferase (SULT) family. SULT1A1, SULT1A2, SULT2A1. • Uridine glucuronosyltransferase (UGT) family. UGT1A3, UGT1A4, UGT1A7, UGT1A10, UGT2B7, UGT2B17. • Other transferases. HNMT (histamine‑N-methyltransferase); NAT1 (N‑acetyltransferase 1); PNMT (phenylethanolamine‑N-methyltransferase). • Hydrolases. GUSB (β‑glucuronidase), CES2 (carboxylesterase 2). Drug transport proteins • ATP binding cassette (ABC) proteins. ABCC3 (multidrug resistance-associated protein 3); ABCC11 (multidrug resistance-associated protein 8). • Solute-linked carrier (SLC) proteins.SLC28A1 (concentrative nucleoside transporter 1); SLC29A1 (equilibrative nucleoside transporter 1); SLC15A2 (oligopeptide transporter 2; also known as PEPT2); SLCO1A2 (organic anion transport protein 1A2; also known as OATP-A); SLCO2B1 (organic anion transport protein 2B1; also known as OATP-B); SLC22A12 (organic anion transporter 4‑like; also known as URAT1); SLC22A1 (organic cation transporter 1); SLC22A4 (organic cation transporter N1); SLC22A5 (organic cation transporter N2); SLC10A1 (sodium-taurocholate cotransporting polypeptide). • Nuclear receptors. AHR (aryl hydrocarbon receptor); PXR (pregnane X receptor; also known as NR1I2). Additional information and literature references for each protein are provided in Supplementary information S1 (table). information S5 (table)). Learning whether an NCE is an OATP1B1 substrate or inhibi‑ tor could become an ordinary preclinical activity. The second step is to conduct a broad search for associations of pharmacokinetics with genotypes during the first-in-human study (BOX 4). The scope of this search could include all relevant genes for which there is reasonable expectation that a positive result can be obtained (based on study power) and readily interpreted based on current information about functional variants. This includes genes for which there are well-established (TABLE 1) or observed (TABLE 2) associations between genetic vari‑ ants and human pharmacokinetics, as well as genes for which in vitro evidence indicates that common variants alter the activity or expression of the gene product (BOX 2; Supplementary information S1 (table)). Defining the search scope in this way will include over half of known or suspected drug disposition determinants. For genes in which common variants are known but their function is not (Supplementary information S2 (table)), associations between genotype nature reviews | drug discovery and human pharmacokinetics might not be readily interpreted without subsequent experimentation to define the phenotype of the associated variant(s). These genes might be screened if there is preclinical evidence that the NCE is a substrate or inhibitor, and perhaps not otherwise. Because firstin-human studies are generally not large, the likelihood of finding an association between an uncommon functional variant (Supplementary information S3 (table)) and human pharmacokinetics is low. If there is preclinical evidence that the NCE is a substrate or inhibitor, a special study to determine the clinical relevance of genotype therein may be a better approach to assess the effect of uncommon functional variants than the general strategy described here. Genes for which there is no published information regarding common genetic variation (Supplementary information S3 (table)) might be screened using single nucleotide polymorphisms (SNPs) found in online databases. Including these SNPs would enable some assessment of essentially every known or expected drug disposition determinant. volume 7 | april 2008 | 297 © 2008 Nature Publishing Group Perspectives Box 3 | Survey of genetic variants in drug disposition factors A search of the scientific literature and online resources was conducted to provide an overview of the variation in genes that encode drug metabolizing enzymes (DMEs), the drug transport proteins (DTPs), abundant plasma binding proteins, and factors that regulate DME and DTP expression. For each gene, at least the following sources were searched: •Online Mendelian Inheritance in Man (OMIM) (http://www.ncbi.nlm.nih.gov/entrez/query. fcgi?db=OMIM) •Medline (accessed through Dialog DataStar) was searched using the OMIM gene symbol. If this simple search yielded >100 articles, it was limited by applying the condition AND (pharmacogenetics OR polymorphism-genetic). •The Pharmacogenetics and Pharmacogenomics Knowledge Base (http://www.pharmgkb.org) Genes were categorized as follows: •Consistently replicated association of variants with human pharmacokinetics of at least one drug (TABLE 1). •Association of variants with human pharmacokinetics of one or more drugs, but without consistent replication for any drug (TABLE 2). •Functionality of common variants (≥5% combined frequency of variants of similar phenotype) demonstrated by in vitro methods, but no published association with human pharmacokinetics (BOX 2; see Supplementary information S1 (table)). •Common variants have been published, but functionality or association with human pharmacokinetics have not (see Supplementary information S2 (table)). •Functionality of one or more rare (or unreported frequency) variants demonstrated (see Supplementary information S3 (table)). •No information on variant functionality or association related to human pharmacokinetics; rare mutations in some of these genes have been linked to Mendelian metabolic disorders (see Supplementary information S4 (table)). The aim of this overview is to provide a sense of the diversity of pharmacogenetics– pharmacokinetics knowledge. Although extensive, the tables are not necessarily comprehensive. We generally did not include reports of association unrelated to the pharmacokinetics of a specific drug, even when a phenotype (such as cancer susceptibility) might be related to xenobiotic disposition. Literature citations are either to our choice of review articles or to primary literature supporting what we considered to be the most important sort of information available. For example, if variants in a gene were associated with human pharmacokinetics in a clinical study we do not also cite work showing the molecular phenotype of those variants; if the phenotype of common variants in a gene is understood we do not also cite work concerning the phenotype of rare variants. We apologize in advance to any scientists whose relevant publications are not cited. The third step is to replicate any observed associations during the multiple rising-dose study (BOX 4). As large numbers of genes may be screened, replication of any observed association from a first-in-human study is essential to minimize the risk of making a decision based on false-positive results. The fourth step is to confirm the gene product’s role using an in vitro assay (if the role was not already known from preclinical work) before or concurrently with Phase II clinical studies (BOX 4). When functional genetic variants affect protein activity (rather than expression), the activities of those variants towards the NCE should also be measured in the in vitro assay, as some variants have substrate-dependent effects. For example, the *17 allele of CYP2D6 has normal catalytic activity towards codeine but reduced activity towards dextromethorphan and debrisoquin34. If different variants in the same gene were used jointly to classify individuals (for example, into poor metabo‑ lizer or non-poor metabolizer) for clinical pharmacogenetic–pharmacokinetic associa‑ tions, the classification should be confirmed by showing that the different variants share a similar phenotype toward the compound (for example, SLCO1B1 and atrasentan29). The fifth step is to estimate the magni tude of genotype effect in a population pharmacokinetics model during Phase II clinical studies. At this point, genotype is used as a covariate in the population phar‑ macokinetics model, just as sex or weight might be used. It is often not feasible to use genotype as a population pharmacokinetics model covariate before Phase II, because the number of subjects in Phase I clinical studies (BOX 4) is generally not large enough to ensure an accurate estimate of the magni‑ tude of genetic effect in a diverse population. 298 | april 2008 | volume 7 Sufficient numbers of individuals having rare genotypes might not be dosed with an NCE until Phase III clinical studies (BOX 4), or it might be necessary to conduct an enriched clinical study to appropriately inform the model. Knowledge about genetic variability rele vant to drug disposition can be applied in several ways, and these are discussed in the following section. Applications of PG–PK knowledge Potent inhibitors or inducers of a polymor‑ phic DME or DTP are likely to have effects that are proportional to the magnitude of effect of genotype (for example, CYP2C19 and ticlopidine35). Thus, associations between human pharmacokinetics of an NCE and genetic variation identifies a potential pathway for drug–drug inter actions, and hence the need for specific and additional drug–drug interaction studies. By contrast, little or no effect of a wellestablished genetic variant might show that certain drug–drug interaction studies are not necessary. For example, a CYP2D6 inhib‑ itor is unlikely to have a meaningful effect on a CYP2D6 substrate if that substrate’s pharmacokinetics are not meaningfully influenced by CYP2D6 genotype. Prioritizing drug–drug interaction studies on the basis of clinical pharmacogenetic effects will improve on the current approach of doing so on the basis of in vitro experiments alone1. CYP2D6 can be used to exemplify another application of clinical pharmaco genetics: to increase confidence that pharmacokinetic outliers are not likely to be an issue in later development. Lack of a significant effect of any genotype suggests that multiple distribution pathways are contributing to drug disposition equally, and that genetic pharmacokinetic outliers are at most very rare (as they would have to carry multiple rare genotypes). Pharmacogenetic– pharmacokinetic relationships can help to distinguish between ordinary population variability and true outliers in a limited dataset — a relevant factor in decisions of how (or whether) to move a programme forward. For example, in one study we iden‑ tified that unusual pharmacokinetics of a CYP2D6 substrate was observed in a person whose CYP2D6 genotype had a population frequency of <0.4% and was thus reasonably considered an outlier36. In other situations, learning that one or more apparent outliers share genetic constitution with other subjects suggests a greater level of inter-individual variation rather than the existence of two separate populations. www.nature.com/reviews/drugdisc © 2008 Nature Publishing Group Perspectives Dose verification and proof of concept Assess safety and tolerability in healthy volunteers Preclinical development Study metabolism of NCE by selected DMEs and DTPs Phase I First in human Long-term safety and efficacy Phase II Phase III Multiple rising dose Broad search for associations of PK with genotype (include all PK-relevant genes) Replication of observed associations Confirm the role of gene product in metabolism of NCE; measure activity of variants Build population PK model to estimate magnitude of genotype effect Use information for follow-up compounds Figure 1 | Flow chart of the proposed pharmacogenetic–pharmaco kinetic strategy. In vitro experiments are conducted before clinical trials to assess whether, and to what extent, selected drug metabolizing enzymes (DMEs) and drug transport proteins (DTPs) might influence the disposition of a new chemical entity (NCE). Then, during the first-in-human study, a broad search for associations of pharmacokinetic (PK) properties The use of genotype–pharmacokinetic associations can also enhance the design of special population or regional bridging studies (BOX 4). If a drug’s pharmacokinetic properties are sensitive to a polymorphism in a gene, success in special population studies may depend on including that gene in the design. How this is most efficiently done may vary between situations, depend‑ ing in part on the strength of the effect and frequency of the genotype. For some studies, retrospective determination of whether an unbalanced representation of genotypes contributed to group differences may be sufficient. Sometimes, recruitment of geno‑ type-balanced cohorts or separate matched cohorts of different genotypes, or excluding individuals of a certain genotype, may be desirable to address the key clinical phar‑ macology question. These approaches can be particularly relevant in regional bridging studies, as genetic variant frequencies are known to differ substantially between populations of different geographic origins (for example, CYP2C9 (REF. 37), CYP2C19 (REF. 38), CYP2A6 (REF. 39), UGT1A1 and UGT1A9 (REF. 40), NAT2 (REF. 41), OATP1B1 (REF. 42)). For example, suppose that pharmaco genetics–pharmacokinetics research in both first-in-human and multiple rising-dose studies showed that, on average, individu‑ als heterozygous for a low activity allele of CYP2A6 (intermediate metabolizers) had higher levels of an NCE than those homozygous for the wild-type allele of the gene (extensive metabolizers). Furthermore, with genotypes is conducted. Any observed associations are replicated during the multiple rising-dose study. Before or concurrently with Phase II Nature Reviews | Drug Discovery clinical studies, the gene product’s role is confirmed using an in vitro assay (if the role was not already known from preclinical work). The magnitude of genotype effect in a population pharmacokinetics model is then estimated during Phase II clinical studies. in vitro experiments conducted subsequently showed that the NCE is a CYP2A6 substrate and population pharmacokinetic analysis of Phase II clinical trial results confirm the influence of the CYP2A6 genotype on the pharmacokinetic properties of the NCE. To facilitate global development of the NCE, a pharmacokinetic bridging study between Japanese and Caucasians is a next step2. Successful conduct of this study could elimi‑ nate the need for a separate full development programme in Japan43. Yet, because of highly different variant frequencies, individuals homozygous for low activity alleles (poor metabolizers) are common among Japanese but not in Caucasians. Hence, random recruitment of Japanese and Caucasians (the standard practice for regional bridging studies) is almost certain to fail to show equivalent pharmacokinetics between the two groups. There are several possible regional bridging trial designs that might provide the evidence to avert a requirement to conduct similar full development programmes in both groups, including recruitment of genotype-matched cohorts of Japanese and Caucasians. However, there is currently a lack of public experience, and hence uncer‑ tainty, as to whether this or other designs will be acceptable to regulatory agencies. Another way to use this information is in dose selection for pivotal studies (BOX 4). Understanding that the dose–exposure relationship differs between identifiable groups may lead to a decision to move forward with a dose or doses that differ from what might have been selected considering nature reviews | drug discovery a homogeneous population. For example, it may be desirable to increase the pivotal study dose to enhance efficacy among individuals who can be expected to have lower exposures (FIG. 2). The drug levels of some leading antidepressants (for example, paroxetine, venlafaxine) or antipsychotics (for example, olanzapine, aripiprazole) are moderately influenced by CYP2D6 genotype44. These drugs are generally safe at a range of doses45–48; however, there have been reports of poor efficacy related to low drug levels in CYP2D6 ultrarapid metabolizers49–51. FIGURE 2a illustrates how pharmacogenetics might have been applied during the development of these drugs. Here, the ‘default’ dose represents the lowest dose that showed efficacy in pivotal studies and is generally the recommended starting dose in the drug’s labels. Suppose that the association between CYP2D6 genotype and pharmacokinetics had been established during Phase I and II clinical studies. Then, the developers of these drugs could have predicted that using a somewhat higher dose (the pharmacogenetics-based dose) in pivotal studies would improve the efficacy profile in ultrarapid metabolizers and not meaningfully diminish the safety profile in other groups (including poor metabolizers). Perhaps they would have selected the phar‑ macogenetics-based doses for their pivotal studies. Because the drug label indicates doses studied in controlled efficacy trials, such a decision would have been reflected in the label and perhaps improved clinical practice using these drugs. volume 7 | april 2008 | 299 © 2008 Nature Publishing Group Perspectives Box 4 | Phases of clinical development and clinical study types The clinical trials that lead to approval of a new chemical entity (NCE) as a drug are typically divided into three phases, each of which commonly comprises several individual clinical trials. While the distinctions are not absolute, the goals of each phase and some of the types of clinical trials typically considered as part of each phase are described below. Phase I. Phase I drug development includes clinical studies in which an NCE is dosed in a small group of (healthy) volunteers (usually <150). It includes first-in-human and multiple rising-dose (MRD) studies. The main goal of Phase I studies is the assessment of the compound’s safety, tolerability (dose-limiting toxicity, maximum tolerated dose) and pharmacokinetics. Other goals such as early assessment of efficacy (generally using biomarkers) or clinical pharmacology issues (such as food effect) may also be included. First in human. In a first-in-human study, healthy volunteers (patients in some therapeutic areas) are given single doses of an NCE. Generally, a low dose is administered to the first group of subjects and the dose is progressively increased until a maximum tolerated dose is reached, or the highest intended dose is given without adverse effects. The purpose of this study is to begin learning the pharmacokinetics, safety and tolerability of an NCE. Multiple rising dose. A multiple rising-dose study is generally conducted in the same manner as a first-in-human study, except that multiple doses of an NCE (or placebo) are given to each subject. The purposes of multiple rising-dose studies are to further characterize pharmacokinetics, safety and tolerability; obtain dosage and administration information for Phase II studies; and frequently to investigate concentration– or dose–response relationships using biomarkers. Phase II. Phase II drug development includes clinical studies aimed towards dose verification and optimization in a defined population of patients (usually <500). Studies to identify unique or major metabolites, understand drug interaction liabilities and observe the NCE’s safety in special populations may also be considered part of this phase. Proof of concept. A proof-of-concept study is typically the first in which patients, rather than healthy volunteers, are enrolled. This study provides the first evaluation of clinical efficacy for an NCE. If the pharmacology is unprecedented, this study also provides the first evidence for clinical efficacy based on the pharmacological target. Dose-ranging. In a dose-ranging study, the efficacy and safety of several doses or regimens of an NCE are compared in patients. Results from a dose-ranging study are used to choose the dose(s) for pivotal studies. Phase III. Phase III drug development is the most costly part. This is when hundreds or thousands of patients are studied in controlled, multicentre and often multinational trials to assess longterm safety and efficacy of the drug (benefit–risk ratio). The usual comparator is the established standard of care, and study end points must be of clinical relevance. Pivotal study. A pivotal study is an essential part of the regulatory submission package that is meant to demonstrate the safety and efficacy of the NCE in a broad patient population. The results of pivotal studies typically determine the indications for which a new drug is labelled. Special population study. A special population study assesses an NCE’s pharmacokinetics, efficacy or safety in a group that may be excluded from pivotal studies (for example, hepatically impaired, paediatric subjects). Bridging study. A bridging study is performed to allow extrapolation of clinical data from one geographical region (for example, North America/Europe) to the population in a different geographical region (for example, East Asia). The goal of a bridging study is to demonstrate that an NCE has similar characteristics, often pharmacokinetics, in groups of subjects drawn from each of the regions. In another situation, if a drug’s efficacy continues to increase as the dose approaches the amount that is the maximally tolerated dose, it may be desirable to sacrifice some efficacy across the population by lowering the pivotal study dose in order to be able to safely dose the entire patient population, including those who can be expected to have higher exposures based on their genetic constitutions (FIG. 2b). Additionally, pharmacogenetics– pharmacokinetics research during Phase I and IIa drug development might reveal dose–exposure relationships that are so different between genotype groups that there is no single dose of an NCE that can be predicted to be generally safe and effective for the entire patient population (FIG. 2c). If clinicians would not be likely to accept required pre-prescription testing to select 300 | april 2008 | volume 7 a dose for an individual, this information might be used to cease development of an NCE before larger investment in Phase IIb and III studies. Moreover, identifying a specific gene product that is related to an unacceptable pharmacokinetic profile provides the opportunity to enhance the probability of future success by screening against this gene product before advanc‑ ing chemically or mechanistically similar follow-up compounds into clinical trials. However, if unmet medical need allows for pairing of a drug and a genetic test, the pharmacogenetic results might provide a path forward in which pivotal studies would be conducted using different doses for each genotype group. Thiopurine drugs (for example, azathioprine) are effective for treatment of certain leukaemias, yet can lead to life-threatening adverse events52. Thiopurine drug levels are strongly influ‑ enced by TPMT genotype53,54. Suppose that this association had been established during Phase I and II clinical studies of azathio‑ prine. Then, different doses could have been tested for extensive, intermediate and poor metabolizer groups. Instead of the current situation, in which extensive metabolizers are probably under-dosed and physicians are left to their best judgment to select doses for individual patients, azathioprine might have been developed in a way that its label could recommend specific doses (or dose ranges) according to TPMT genotype. Early clinical pharmacogenetics data might also identify rare genotypes that should be studied during later development. Because of the limited number of subjects in Phase I/ IIa trials, less frequent genotypes might not be represented. Even when there is a numeri‑ cal but not clinically meaningful difference between wild-type and heterozygous geno‑ type groups, a special study that enrols only healthy volunteers or patients with the variant genotype, or in which the study cohort is deliberately enriched with individuals with the variant genotype, may be justified. This might apply when a drug is being tested in global clinical trials that are likely to include variant genotypes that were uncommon in the Phase I population but are more common among other geographical ancestries. As described above, in the context of NCE development only incremental resource commitment is necessary to reap the potential benefits of pharmacogenetics– pharmacokinetics research. The interest of regulatory agencies in this research26 provides additional incentive for the pharma‑ ceutical industry to make such investments. The approach described herein could also be www.nature.com/reviews/drugdisc © 2008 Nature Publishing Group Perspectives a Minimum for efficacy b c Maximum for safety Population Dose Unselected Genotype 1 Genotype 2 Default Unselected Genotype 1 Genotype 2 PG-based Unselected Genotype 1 Genotype 2 Default Unselected Genotype 1 Genotype 2 PG-based Genotype 1 Genotype 2 Figure 2 | Scenarios in which pharmacogenetic information related to pharmacokinetics in early drug development can support decision-making for later development. Each horizontal Nature Reviews | Drug Discovery line represents a predicted range of exposures for a given dose, with red lines representing doses that extend outside the clinical window for safety and efficacy. In (a) and (b), the predicted range of concentrations for the total patient population is misestimated if patients with genotype 2 (probably a relatively small percentage) are not modelled separately. In both of these scenarios, pharmaco genetic (PG) information enables dose adjustment to enhance the probability of success of subsequent studies. In (c), separate doses for genotypes 1 and 2 are necessary for the successful development of a drug. applied to further characterize the disposi‑ tion pathways of currently available drugs. It is not clear how this research would be funded, particularly for generic and overthe-counter drugs. The large investments to conduct well-controlled clinical studies are difficult to justify solely for pharmaco genetics purposes. The US National Institutes of Health and the Food and Drug Administration are funding clinical trials for some drugs (for example, irinotecan, tamoxifen, warfarin), yet additional funding from government and other sources are needed to fully recognize the potential of pharmacogenetics–pharmacokinetics research for already marketed drugs. Limitations of the strategy The exploratory pharmacogenetic– pharmacokinetic strategy we have outlined can readily be applied today for many but not all genes that are relevant for drug disposition. For some genes, the functional variants will typically be observed only in heterozygous individuals and so only particularly strong pharmacokinetic pheno‑ types are likely to be detected during early clinical development. For example, if nine individuals (15%) of a Phase I cohort of 60 individuals are heterozygous for a certain genotype, there is 80% power (assuming a variance of one in each genotype group) to detect a twofold difference between the average clearance of the wild-type and heterozygous groups. These numbers are adequate, as the goal is to make decisions based on large effects of genotype on pharmacokinetics. The strategy is most readily applied for genes in TABLES 1,2; BOX 2 and in Supplementary information S1 (table), as functional variants are sufficiently understood to enable interpretation of either positive or negative results of clinical pharmacogenetic–pharmacokinetic associations. For many potentially phar‑ macokinetic-relevant genes, the molecular or cellular phenotypes of variants have not been reported (Supplementary information S2 (table), S4 (table)). Without phenotype information, interpretation of clinical phar‑ macogenetic–pharmacokinetic associations is tenuous. Many drug developers might not wish to perform exploratory clinical phar‑ macogenetics research for these genes until variant phenotype information is publicly available. Research to establish the pheno‑ types of common variants in these genes will enable a more comprehensive applica‑ tion of pharmacogenetics–pharmacokinetics nature reviews | drug discovery research in drug development. For other pharmacokinetic-relevant genes, the only known functional variants are uncommon (Supplementary information S3 (table)). Exploratory pharmacogenetic–pharmaco kinetic association analysis during Phase I clinical studies usually will not have adequate power to assess the role of these in an NCE’s disposition. A more effective approach may be to conduct an additional pharmacokinetics study recruiting subjects having the uncommon genotype(s), when in vitro experiments indicate that the NCE is a substrate. There will be some genes for which exploratory pharmaco genetic–pharmacokinetic association studies will probably never be useful. This might already be the situation for the gene encoding P‑glycoprotein (ABCB1); associations of common ABCB1 variants to any drug disposition phenotype have been inconsistently replicated or are even contradictory26,55, suggesting that the known common genetic variants may be nonfunctional. The role of this important DTP in the disposition of new drugs may need to continue to be established by in vitro and drug–drug interaction studies. Some may have a perception that substantial changes to early development processes, trial design or resource invest‑ ment would be necessary to implement this strategy. This is not the case. Indeed, one impetus for a drug developer to adopt it is as an alternative to a larger resource invest‑ ment to build a wider range of in vitro tech‑ nologies for preclinical evaluation of diverse drug disposition determinants. Our strategy has as its first step the same preclinical analy sis of an NCE that is ordinary in today’s pharmaceutical industry drug metabolism laboratory. To help confirm observed phar‑ macogenetic–pharmacokinetic associations, it may sometimes be necessary to develop in-house capability to perform specific assays related to the pathway that appears meaningful for an NCE’s disposition. Alternatively, a drug developer may choose to collaborate with an academic laboratory that has the appropriate assays, as we did in our work related to the role of OATP1B1 in the disposition of atrasentan29. The same genotyping technologies can be applied to any gene, so investment in a single platform is sufficient to enable the genetic experi‑ ments. Several vendors provide high-quality reagents and equipment to generate all of the relevant genetic data. In addition, several service providers are available for those drug developers who do not have or wish to build in-house genotyping capability. volume 7 | april 2008 | 301 © 2008 Nature Publishing Group Perspectives Glossary Alternating group design An alternating group design is similar to an escalating-dose design, except that each group of subjects participates in several study periods. The same subjects in each group receive placebo in every period, but the dose of study drug escalates at each subsequent period. Crossover design In a crossover design, each group of subjects is randomized to receive all treatments (including placebo), usually separated by a washout period to eliminate carry-over effects between treatments. A crossover design is generally considered to provide increased sensitivity to treatment effects as it virtually eliminates inter-individual variability (each study subject functions as their own control), and hence requires fewer participants than a parallel arm design. However, the informativeness of pharmacogenetic analysis is minimized in this study design. drug transport protein. This classification applies to a specific enzyme or protein; a single individual can be an extensive metabolizer for one enzyme and not for another. Fish-odour syndrome This is an inborn error of metabolism accompanied by fish-like body odour. The offensive odour is due to the build-up of amino-trimethylamine (TMA) derived from foodstuffs such as egg yolk, liver, kidney, legumes, soy beans, peas and saltwater fishes. TMA is normally oxidized by the drug metabolizing enzyme flavin monooxygenase 3 (FMO3); genetic deficiency in this enzyme is the syndrome’s cause. Functional genetic variants New chemical entity (NCE). This is a common term in the pharmaceutical industry for a chemical that is being tested to learn whether it is useful as a drug. Generally (and in this paper) NCE refers only to small molecules and not to large molecules such as peptides, proteins and nucleic acids that also may be developed as drugs. Pharmacogenetics (PG). This is the study of how genetic variation influences drug response. In this paper, we focus on the subset of PG that relates genetic and pharmacokinetic variation. Pharmacokinetics These are variant gene sequences that alter the expression, activity or substrate specificity of the corresponding protein. (PK). Is the study of how the body affects drugs. In this paper, we frequently talk about PK variation (or variability), which refers to differences between individuals of the concentration–time profile of a drug. Drug disposition Heterologous transcription activation assays Polymorphic Drug disposition is what happens to a chemical after it enters the body. After oral dosing, drug disposition includes absorption from the gastrointestinal tract into central circulation; distribution to and between various tissues; metabolism to different chemicals; and excretion (usually in urine or faeces). These are experimental tools used to measure the interaction of a compound with nuclear receptors. These can either be cell-based or cell-free systems engineered so that a readily detectable product (RNA or protein) is made in larger amounts when a ligand for the nuclear receptor is present. In the context of pharmacogenetics, polymorphic refers to when a DNA sequence exists in two or more forms in human populations. Polymorphism refers to the DNA sequence that is polymorphic. Escalating-dose design Intermediate metabolizers/intermediate transporters In an escalating-dose design, a first group of subjects is randomized to either the lowest dose of study drug or matching placebo, and subsequent groups (composed of different subjects) are each randomized to the next higher dose of study drug or matching placebo. Poor metabolizers/poor transporters These are individuals who have very low or no activity of a drug metabolizing enzyme or drug transport protein, respectively. Poor-metabolizer alleles include those that lead to premature termination of the protein, and gene deletions. These are individuals who are homozygous for a wild-type (normal activity) form of a drug metabolizing enzyme or These are individuals who have reduced activity of a drug metabolizing enzyme or drug transport protein. Reduced activity may result from a structural change in the enzyme or a lower level of protein expression. An intermediate phenotype can result either from heterozygosity (for example, one extensive metabolizer allele + one poor metabolizer allele) or homozygosity (for example, for a reduced expression allele). These are individuals who have unusually high activity of a drug metabolizing enzyme. Most commonly this is a result of gene duplication. We are not aware of an ultrarapid transporter phenotype. We expect that discovery of pharmaco‑ genetic–pharmacokinetic associations will generally be done as piggy-back activities to Phase I and II clinical studies that are conducted as they are today. Collecting DNA samples with appropriate informed consent from every subject in Phase I clinical studies is already an ordinary activity at many large pharmaceutical companies, and most at least incorporate optional consent for pharmaco genetics–pharmacokinetics research into Phase II clinical studies (William, J. A. et al., personal communication). Both DNA sample collection and genotyping are inexpensive in the context of clinical trial costs (William, J. A. et al., personal communication). Because sample sizes might be increased in combined analyses using samples and data from first in human or multiple rising dose and other Phase I studies, drug developers should ensure that inclusion and exclusion criteria are consistent across all Phase I studies of an NCE. This is also already normal practice at major pharmaceutical companies. So, there are few operational hurdles to implementing our proposed pharmaco genetic strategy in drug development. However, an ongoing trend in the phar‑ maceutical industry that may limit the future application of our pharmacogenetic strategy is a move away from the traditional firstin-human and multiple rising-dose Phase I clinical studies with escalating dose design. Alternatives, including crossover designs and alternating group designs, are growing in popularity. The attractions of these designs are reduced cost, exposing fewer healthy volunteers (who can gain no health benefit from the study) to potential health risk, and reduced inter-individual variation (which further reduces the necessary number of subjects in the study). While crossover and alternating group designs are more efficient for assessing certain safety parameters (for example, QT prolongation, chronotropic effects) of an NCE than the escalating dose design, the trade-off is that they provide less information about population variability in pharmacokinetics and adverse event profile of an NCE. The power of a pharmacogenetic study rests on the degree of pharmacokinetic variability in the study population. Hence, the growing popularity of smaller first-inhuman and multiple rising-dose studies in the pharmaceutical industry potentially makes our strategy more difficult to implement during the development of an NCE. If sufficient subjects are not dosed during Phase I trials, it may be necessary to use Phase II trials to both replicate genetic associations as well as quantitate the mag‑ nitude of genotype effect in a population pharmacokinetics model. One of the potential applications of our proposed strategy is prediction of an NCE’s drug interaction potential through any of a wide range of pathways (see below). However, clinical drug-interaction models are not available for most disposition path‑ ways. This is a key technology development need. There are few widely accepted, specific probe drugs for use in drug interaction clini‑ cal studies56. Although selective substrates and inhibitors of many pharmacokinetic relevant factors are established for use in vitro (Supplementary information S1 (table)), many of these compounds are not (or have not been shown to be) safe for human use. Furthermore, probe drugs currently con‑ sidered as gold standards for certain types of drug interactions might not have the degree Extensive metabolizers/extensive transporters 302 | april 2008 | volume 7 Ultrarapid metabolizers www.nature.com/reviews/drugdisc © 2008 Nature Publishing Group Perspectives of in vivo specificity as commonly believed. For example, rifampin (also known as rifampicin) is considered a standard probe to study the effect of inducing several DMEs and DTPs on the pharmacokinetics of an NCE56. When an NCE is given after several days of rifampin dosing, there have been two expected outcomes. If rifampin induces a DME that is important for the NCE’s metabolism, or a DTP that is important for limiting its absorption or increasing its excretion, NCE levels will be decreased (relative to levels before rifampin dosing). Otherwise, the levels should be similar before and after rifampin dosing. Yet, we recently observed that rifampin had the unexpected effect of increasing atrasentan concentrations57. Pharmacogenetics– pharmacokinetics research, following the strategy proposed herein, established that this occurred via inhibition of the hepatic uptake DTP OATP1B1 (REF. 29). Several other drug–drug interactions have also recently been recognized as mediated by OATP1B1 (REFS 58–63). As the range of DMEs, DTPs and other factors known to mediate drug interactions grows beyond the few that are known today, we expect that the dual effect of rifampin will not be the only surprise from widely accepted clinical ‘victims’ or ‘perpetrators’ of drug–drug interactions. Our work showing that the effect of rifampin on atrasentan levels are mediated via OATP1B1 exemplifies that using clinical pharmacogenetics in concert with in vitro methods can be more effective than using in vitro methods alone for learning which of the many disposition pathways known are relevant for a particular drug. opportunities if drug developers enter later phases of clinical development with a better understanding of the pathways that influence exposure variability between individuals or populations, as well as any potential drug–drug interactions that might occur. The approach we have proposed is well suited to exploit the wealth of informa‑ tion that has been generated along with completion of the human genome project. David A. Katz and Anahita Bhathena are at Abbott Global Pharmaceutical Research & Development, 100 Abbott Park Road, Abbott Park, Illinois 60064-3500, USA. Bernard Murray is at Gilead Sciences, 333 Lakeside Drive, Foster City, California 94404, USA. Leonardo Sahelijo is at Takeda Global Research & Development, One Takeda Parkway, Deerfield, Illinois 60015, USA. Correspondence to D.A.K. e-mail: [email protected] doi:10.1038/nrd2486 1. 2. 3. 4. 5. 6. 7. Conclusions Genetic variation is only one of several factors contributing to variability in drug disposi‑ tion. Other factors, such as dose and disease state, may turn out to be more important determinants of drug disposition for many drugs. However, that fact does not lessen the impetus to discover the situations for which genetic variation is a major determinant for the disposition of an NCE, or limit the potential utility of such information when it is learned. Practical limitations to implementation of our proposed strategy are few, although changes to study designs in the pharma ceutical industry could limit its value in Phase I drug development. For some applications, the value may also be limited because we do not have the ability to act on the knowledge gained. Nevertheless, we believe that there are many useful 8. 9. 10. 11. 12. 13. 14. 15. Obach, R. S., Walsky, R. L., Venkatakrishnan, K., Houston, J. B. & Tremaine, L. M. In vitro cytochrome P450 inhibition data and the prediction of drug–drug interactions: qualitative relationships, quantitative predictions, and the rank-order approach. Clin. Pharmacol. Ther. 78, 582–592 (2005). Weinshilboum, R. Thiopurine pharmacogenetics: clinical and molecular studies of thiopurine methyltransferase. Drug Metab. Dispos. 29, 601–605 (2001). Evans, W. E. Pharmacogenetics of thiopurine S‑methyltransferase and thiopurine therapy. Ther. Drug Monit. 26, 186–191 (2004). Maeda, K. et al. Effects of organic anion transporting polypeptide 1B1 haplotype on pharmacokinetics of pravastatin, valsartan, and temocapril. Clin. Pharmacol. Ther. 79, 427–439 (2006). Hirano, M., Maeda, K., Shitara, Y. & Sugiyama, Y. Drug–drug interaction between pitavastatin and various drugs via OATP1B1. Drug Metab. Dispos. 34, 1229–1236 (2006). Chung, J. Y. et al. Effect of OATP1B1 (SLCO1B1) variant alleles on the pharmacokinetics of pitavastatin in healthy volunteers. Clin. Pharmacol. Ther. 78, 342–350 (2005). Garrod, A. The incidence of alkaptonuria: a study in chemical individuality. Lancet ii, 1616–1620 (1902). Vesell, E. S. & Page, J. G. Genetic control of drug levels in man: phenylbutazone. Science 159, 1479–1480 (1968). Price Evans, D., Manley, K. & McKusick, V. Genetic control of isoniazid metabolism in man. BMJ 485–491 (1960). Evans, F. T., Gray, P. W., Lehmann, H. & Silk, E. Sensitivity to succinylcholine in relation to serumcholinesterase. Lancet 1, 1229–1230 (1952). Blum, M., Grant, D., McBride, W., Heim, M. & Meyer, U. Human arylamine N‑acetyltransferase genes: isolation, chromosomal localization, and functional expression. DNA Cell Biol. 9, 193–203 (1990). Deguchi, T., Mashimo, M. & Suzuki, T. Correlation between acetylator phenotypes and genotypes of polymorphic arylamine N‑acetyltransferase in human liver. J. Biol. Chem. 265, 12757–12760 (1990). McGuire, M. et al. Identification of the structural mutation responsible for the dibucaine-resistant (atypical) variant form of human serum cholinesterase. Proc. Natl Acad. Sci. USA 86, 953–957 (1989). McTiernan, C. et al. Brain cDNA clone for human cholinesterase. Proc. Natl Acad. Sci. USA 84, 6682–6686 (1987). Prody, C., Zevin-Sonkin, D., Gnatt, A., Goldberg, O. & Soreq, H. Isolation and characterization of full-length cDNA clones coding for cholinesterase from fetal human tissues. Proc. Natl Acad. Sci. USA 84, 3555–3559 (1987). nature reviews | drug discovery 16. Bertilson, L. et al. Molecular basis for rational megaprescribing in ultrarapid hydroxylators of debrisoquine. Lancet 341, 63 (1993). 17. Gonzalez, F. J. et al. Characterization of the common genetic defect in humans deficient in debrisoquine metabolism. Nature 331, 442–446 (1988). 18. Gonzalez, F. J. et al. Human debrisoquine 4‑hydroxylase (P450IID1): cDNA and deduced amino acid sequence and assignment of the CYP2D locus to chromosome 22. Genomics 2, 174–179 (1988). 19. Gough, A. C. et al. Identification of the primary gene defect at the cytochrome P450 CYP2D locus. Nature 347, 773–776 (1990). 20. Johansson, I., Lundqvist, E., Dahl, M. L. & Ingelman-Sundberg, M. PCR-based genotyping for duplicated and deleted CYP2D6 genes. Pharmacogenetics 6, 351–355 (1996). 21. Honchel, R. et al. Human thiopurine methyltransferase: molecular cloning and expression of T84 colon carcinoma cell cDNA. Mol. Pharmacol. 43, 878–887 (1993). 22. Krynetski, E. Y. et al. A single point mutation leading to loss of catalytic activity in human thiopurine S‑methyltransferase. Proc. Natl Acad. Sci. USA 92, 949–953 (1995). 23. Szumlanski, C. et al. Thiopurine methyltransferase pharmacogenetics: human gene cloning and characterization of a common polymorphism. DNA Cell Biol. 15, 17–30 (1996). 24. FDA. Package Insert for Strattera (atomoxetine HCl). FDA web site [online], <http://www.fda.gov/cder/foi/ label/2002/21411_strattera_lbl.pdf> (2002). 25. Dolphin, C., Janmohamed, A., Smith, R., Shephard, E. & Phillips, I. Missense mutation in flavin-containing mono-oxygenase 3 gene, FMO3, underlies fish-odour syndrome. Nature Genet. 17, 491–494 (1997). 26. Andersson, T. et al. Drug-metabolizing enzymes: evidence for clinical utility of pharmacogenomic tests. Clin. Pharmacol. Ther. 78, 559–581 (2005). 27. Rendic, S. & DiCarlo, F. Human cytochrome P450 enzymes: a status report summarizing their reactions, substrates, inducers and inhibitors. Drug Metab. Rev. 29, 413–580 (1997). 28. Tukey, R. & Strassbug, C. Human UDPglucuronosyltransferases: metabolism, expression, and disease. Annu. Rev. Pharmacol. Toxicol. 40, 581–616 (2000). 29. Katz, D. A. et al. Organic anion transporting polypeptide 1B1 activity classified by SLCO1B1 genotype influences atrasentan pharmacokinetics. Clin. Pharmacol. Ther. 79, 186–196 (2006). 30. Lee, E. et al. Rosuvastatin pharmacokinetics and pharmacogenetics in white and Asian subjects residing in the same environment. Clin. Pharmacol. Ther. 78, 330–341 (2005). 31. Mwinyi, J., Johne, A., Bauer, S., Roots, I. & Gerloff, T. Evidence for inverse effects of OATP‑C (SLC21A6) *5 and *1b haplotypes on pravastatin kinetics. Clin. Pharmacol. Ther. 75, 415–421 (2004). 32. Niemi, M. et al. High plasma pravastatin concentrations are associated with single nucleotide polymorphisms and haplotypes of organic anion transporting polypeptide‑C (OATP‑C, SLCO1B1). Pharmacogenetics 14, 429–440 (2004). 33. Nishizato, Y. et al. Polymorphisms of OATP‑C (SLC21A6) and OAT3 (SLC22A8) genes: consequences for pravastatin pharmacokinetics. Clin. Pharmacol. Ther. 73, 554–565 (2003). 34. Wennerholm, A. et al. The African-specific CYP2D6*17 allele encodes an enzyme with changed substrate specificity. Clin. Pharmacol. Ther. 71, 77–88 (2002). 35. Ieiri, I. et al. Interaction magnitude, pharmacokinetics and pharmacodynamics of ticlopidine in relation to CYP2C19 genotypic status. Pharmacogenet. Genomics 15, 851–859 (2005). 36. Furman, K. et al. Impact of CYP2D6 intermediate metabolizer alleles on single-dose desipramine pharmacokinetics. Pharmacogenetics 14, 279–284 (2004). 37. Takahashi, H. et al. Different contributions of polymorphisms in VKORC1 and CYP2C9 to intra- and inter-population differences in maintenance dose of warfarin in Japanese, Caucasians and African– Americans. Pharmacogenet. Genomics 16, 101–110 (2006). 38. Blaisdell, J. et al. Identification and functional characterization of new potentially defective alleles of human CYP2C19. Pharmacogenetics 12, 703–711 (2002). volume 7 | april 2008 | 303 © 2008 Nature Publishing Group Perspectives 39. Xu, C., Goodz, S., Sellers, E. & Tyndale, R. CYP2A6 genetic variation and potential consequences. Adv. Drug Deliv. Rev. 54, 1245–1256 (2002). 40. Innocenti, F. et al. Haplotypes of variants in the UDP-glucuronosyltransferase1A9 and 1A1 genes. Pharmacogenet. Genomics 15, 295–301 (2005). 41. Spielberg, S. N‑Acetyltransferases: pharmacogenetics and clinical consequences of polymorphic drug metabolism. J. Pharmacokinet. Biopharm. 24, 509–519 (1996). 42. Tirona, R., Leake, B., Merino, G. & Kim, R. Polymorphisms in OATP‑C. J. Biol. Chem. 276, 35669–35675 (2001). 43. International Conference on Harmonization (ICH). ICH Harmonized Guidance E5(R1): Ethnic Factors in the Acceptability of Foreign Clinical Data. ICH web site [online], <http://www.ich.org/LOB/media/MEDIA481. pdf> (1998). 44. Kirchheiner J. et al. Pharmacogenetics of antidepressants and antipsychotics: the contribution of allelic variations to the phenotype of drug response. Mol. Psychiatry 9, 442–473 (2004). 45. FDA. Package Insert for Paxil (paroxetine hydrochloride). FDA web site [online], <http://www.fda. gov/cder/foi/label/2001/20031s23lbl.pdf> (1999). 46. FDA. Package Insert for Effexor (venlafaxine hydrochloride). FDA web site [online], <http://www.fda.gov/cder/foi/label/2004/20151slr 028,030,032,20699slr041,048,052_effexor_lbl. pdf> (2004). 47. FDA. Package Insert for Zyprexa/Zydis (olanzapine). FDA web site [online], <http://www.fda.gov/cder/foi/ label/2000/21086lbl.pdf> (2000). 48. FDA. Package Insert for Abilify (aripiprazole). FDA web site [online], <http://www.fda.gov/cder/foi/label/2005/ 021713s004,021436s007lbl.pdf> (2005). 49. Baumann, P., Broly, F., Kosel, M. & Eap, C. Ultrarapid metabolism of clomipramine in a therapy-resistant depressive patient, as confirmed by CYP2D6 genotyping. Pharmacopsychiatry 31, 72 (1997). 50. Kawanishi, C., Lundgren, S., Agren, H. & Bertilson, L. Increased incidence of CYP2D6 gene duplication in patients with persistent mood disorders: ultrarapid metabolism of antidepressants as a cause of nonresponse. A pilot study. Eur. J. Pharmacol. 59, 803–807 (2004). 51. Rau, T. et al. CYP2D6 genotype: impact on adverse effects and nonresponse during treatment with antidepressants — a pilot study. Clin. Pharmacol. Ther. 75, 386–393 (2004). 52. FDA Package Insert for Imuran (azathioprine). FDA web site [online], <http://www.fda.gov/cder/foi/label/ 2005/016324s030,017391s013lbl.pdf> (2005). 53. Otterness, D. et al. Human thiopurine methyltransferase pharmacogenetics: gene sequence polymorphisms. Clin. Pharmacol. Ther. 62, 60–73 (1997). 54. Yates, C. et al. Molecular diagnosis of thiopurine S‑methyltransferase deficiency: genetic basis for azathioprine and mercaptopurine intolerance. Ann. Intern. Med. 126, 608–614 (1997). 55. Chowbay, B., Li, H., David, M., Cheung, Y. B. & Lee, E. J. Meta-analysis of the influence of MDR1 C3435T polymorphism on digoxin pharmacokinetics and MDR1 gene expression. Br. J. Clin. Pharmacol. 60, 159–171 (2005). 56. FDA/CDER/CBER. In Vivo Drug Metabolism/Drug Interaction Studies — Study Design, Data Analysis, and Recommendations for Dosing and Labeling. FDA web site [online], <http://www.fda.gov/cder/ guidance/2635fnl.pdf> (1999). 57. Xiong, H. et al. Effect of rifampin (RIF) on the pharmacokinetics (PK) of atrasentan (ABT‑627, ATN). J. Clin. Oncol. 22 (Suppl. 14), 4728 (2004). 58. Kyrklund, C., Backman, J., Neuvonen, M. & Neuvonen, P. Genfibrozil increases plasma pravastatin concentrations and reduces pravastatin renal clearance. Clin. Pharmacol. Ther. 73, 538–544 (2003). 59. Schneck, D. et al. The effect of gemfibrozil on the pharmacokinetics of rosuvastatin. Clin. Pharmacol. Ther. 75, 455–463 (2004). 60. Shitara, Y., Itoh, T., Sato, H., Li, A. P. & Sugiyama, Y. Inhibition of transporter-mediated hepatic uptake as a mechanism for drug–drug interaction between cerivastatin and cyclosporin A. J. Pharmacol. Exp. Ther. 304, 610–616 (2003). 61. Simonson, S. et al. Rosuvastatin pharmacokinetics in heart transplant recipients administered an antirejection regimen including cyclosporine. Clin. Pharmacol. Ther. 76, 167–177 (2004). 62. Tannergren, C. et al. Multiple transport mechanisms involved in the intestinal absorption and first-pass extraction of fexofenadine. Clin. Pharmacol. Ther. 74, 423–436 (2003). 63. Treiber, A., Schneiter, R., Delahaye, S. & Clozel, M. Inhibition of organic anion transporting polypeptidemediated hepatic uptake is the major determinant in the pharmacokinetic interaction between bosentan and cyclosporin A in the rat. J. Pharmacol. Exp. Ther. 308, 1121–1129 (2004). 64. Luo, X. et al. ADH4 gene variation is associated with alcohol and drug dependence: results from family controlled and population-structured association studies. Pharmacogenet. Genomics 15, 755–768 (2005). 65. Fukami, T. et al. A novel polymorphism of human CYP2A6 gene CYP2A6*17 has an amino acid substitution (V365M) that decreases enzymatic activity in vitro and in vivo. Clin. Pharmacol. Ther. 76, 519–527 (2004). 66. Huang, S. et al. CYP2A6, MAOA, DBH, DRD4, and 5HT2A genotypes, smoking behaviour and cotinine levels in 1518 UK adolescents. Pharmacogenet. Genomics 15, 839–850 (2005). 67. Pianezza, M. L., Sellers, E. M. & Tyndale, R. F. Nicotine metabolism defect reduces smoking. Nature 393, 750 (1998). 68. Schoedel, K. A., Hoffmann, E. B., Rao, Y., Sellers, E. M. & Tyndale, R. F. Ethnic variation in CYP2A6 and association of genetically slow nicotine metabolism and smoking in adult Caucasians. Pharmacogenetics 14, 615–626 (2004). 69. Swan, G. E. et al. Nicotine metabolism: the impact of CYP2A6 on estimates of additive genetic influence. Pharmacogenet. Genomics 15, 115–125 (2005). 70. Yoshida, R. et al. Effects of polymorphism in promoter region of human CYP2A6 gene (CYP2A6*9) on expression level of messenger ribonucleic acid and enzymatic activity in vivo and in vitro. Clin. Pharmacol. Ther. 74, 69–76 (2003). 71. Haas, D. W. et al. Pharmacogenetics of long-term responses to antiretroviral regimens containing efavirenz and/or nelfinavir: an Adult AIDS Clinical Trials Group study. J. Infect. Dis. 192, 1931–1942 (2005). 72. Haas, D. W. et al. Pharmacogenetics of efavirenz and central nervous system side effects: an Adult AIDS Clinical Trials Group study. AIDS 18, 2391–2400 (2004). 73. Rotger, M. et al. Influence of CYP2B6 polymorphism on plasma and intracellular concentrations and toxicity of efavirenz and nevirapine in HIV-infected patients. Pharmacogenet. Genomics 15, 1–5 (2005). 74. Desta, Z., Zhao, X., Shin, J.‑G. & Flockhart, D. A. Clinical significance of the cytochrome P450 2C19 genetic polymorphism. Clin. Pharmacokinet. 41, 913–958 (2002). 75. Sim, S. C. et al. A common novel CYP2C19 gene variant causes ultrarapid drug metabolism relevant for the drug response to proton pump inhibitors and antidepressants. Clin. Pharmacol. Ther. 79, 103–113 (2006). 76. Kirchheiner, J. & Brockmoller, J. Clinical consequences of cytochrome P450 2C9 polymorphisms. Clin. Pharmacol. Ther. 77, 1–16 (2005). 77. Frohlich, M. et al. Association of the CYP3A5 A6986G (CYP3A5*3) polymorphism with saquinavir pharmacokinetics. Br. J. Clin. Pharmacol. 58, 443–444 (2004). 78. Goto, M. et al. CYP3A5*1-carrying graft liver reduces the concentration/oral dose ratio of tacrolimus in recipients of living-donor liver transplantation. Pharmacogenetics 14, 471–478 (2004). 79. Haufroid, V. et al. The effect of CYP3A5 and MDR1 (ABCB1) polymorphisms on cyclosporine and tacrolimus dose requirements and trough blood levels in stable renal transplant patients. Pharmacogenetics 14, 147–154 (2004). 80. Macphee, I. A. M. et al. Tacrolimus pharmacogenetics: polymorphisms associated with expression of cytochrome P4503A5 and p-glycoprotein correlate with dose requirement. Transplantation 74, 1486–1489 (2002). 81. Mouly, S. J. et al. Variation in oral clearance of saquinavir is predicted by CYP3A5*1 genotype but not by enterocyte content of cytochrome P450 3A5. Clin. Pharmacol. Ther. 78, 605–618 (2005). 82. Tsuchiya, N. et al. Influence of CYP3A5 and MDR1 (ABCB1) polymorphisms on the pharmacokinetics of tacrolimus in renal transplant recipients. Transplantation 78, 1182–1187 (2004). 304 | april 2008 | volume 7 83. Zheng, H. et al. Tacrolimus dosing in pediatric heart transplant patients is related to CYP3A5 and MDR1 gene polymorphisms. Am. J. Transplant. 3, 477–483 (2003). 84. Zheng, H. et al. Tacrolimus dosing in adult lung transplant patients is related to cytochrome P4503A5 gene polymorphism. J. Clin. Pharmacol. 44, 135–140 (2004). 85. Hisamuddin, I. M. et al. Genetic polymorphisms of human flavin monooxygenase 3 in sulindacmediated primary chemoprevention of familial adenomatous polyposis. Clin. Cancer Res. 10, 8357–8362 (2004). 86. Hisamuddin, I. M. et al. Genetic polymorphisms of flavin monooxygenase 3 in sulindac-induced regression of colorectal adenomas in familial adenomatous polyposis. Cancer Epidemiol. Biomarkers Prev. 14, 2366–2369 (2005). 87. Innocenti, F. & Ratain, M. J. “Irinogenetics” and UGT1A: from genotypes to haplotypes. Clin. Pharmacol. Ther. 75, 495–500 (2004). 88. Li, J. H. et al. Influence of the ORM1 phenotypes on serum unbound concentration and protein binding of quinidine. Clin. Chim. Acta 317, 85–92 (2002). 89. Quertemont, E. Genetic polymorphism in ethanol metabolism: acetaldehyde contribution to alcohol abuse and alcoholism. Mol. Psychiatry 9, 570–581 (2004). 90. Sparreboom, A. et al. Diflomotecan pharmacokinetics in relation to ABCG2 421C>A genotype. Clin. Pharmacol. Ther. 76, 38–44 (2004). 91. Bialecka, M. et al. The effect of monoamine oxidase B (MAOB) and catechol‑O‑ methyltransferase (COMT) polymorphisms on levodopa therapy in patients with sporadic Parkinson’s disease. Acta Neurol. Scand. 110, 260–266 (2004). 92. Contin, M. et al. Genetic polymorphism of catechol‑O‑methyltransferase and levodopa pharmacokinetic–pharmacodynamic pattern in patients with Parkinson’s disease. Mov. Disord. 20, 734–739 (2005). 93. Weinshilboum, R. M., Otterness, D. M. & Szumlanski, C. L. Methylation pharmacogenetics: catechol O-methyltransferase, thiopurine methyl transferase, and histamine N-methyltransferase. Ann. Rev. Pharmacol. Toxicol. 39, 19–52 (1999). 94. Bengala, C. et al. Prolonged fixed dose rate infusion of gemcitabine with autologous haemopoietic support in advanced pancreatic adenocarcinoma. Br. J. Cancer 93, 35–40 (2005). 95. Yue, L. et al. A functional single-nucleotide polymorphism in the human cytidine deaminase gene contributing to ara-C sensitivity. Pharmacogenetics 13, 29–38 (2003). 96. Aklillu, E. et al. 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 (2003). 97. Chida, M. et al. Detection of three genetic polymorphisms in the 5′-flanking region and intron 1 of human CYP1A2 in the Japanese population. Jpn. J. Cancer Res. 90, 899–902 (1999). 98. Jiang, Z. et al. Search for an association between the human CYP1A2 genotype and CYP1A2 metabolic phenotype. Pharmacogenet. Genomics 16, 359–367 (2006). 99. Obase, Y. et al. Polymorphisms in the CYP1A2 gene and theophylline metabolism in patients with asthma. Clin. Pharmacol. Ther. 73, 468–474 (2003). 100.Sachse, C., Brockmoller, J., Bauer, S. & Roots, I. Functional significance of a C→A polymorphism in intron 1 of the cytochrome P450 CYP1A2 gene tested with caffeine. Br. J. Clin. Pharmacol. 47, 445–449 (1999). 101. Garcia-Martin, E., Martinez, C., Tabares, B., FrIas, J. & Agundez, J. A. Interindividual variability in ibuprofen pharmacokinetics is related to interaction of cytochrome P450 2C8 and 2C9 amino acid polymorphisms. Clin. Pharmacol. Ther. 76, 119–127 (2004). 102. Niemi, M. et al. Polymorphism in CYP2C8 is associated with reduced plasma concentrations of repaglinide. Clin. Pharmacol. Ther. 74, 380–387 (2003). 103.Shi, J. Y. et al. Association between single nucleotide polymorphisms in deoxycytidine kinase and treatment response among acute myeloid leukaemia patients. Pharmacogenetics 14, 759–768 (2004). 104.Nakajima, Y. et al. Haplotype structures of EPHX1 and their effects on the metabolism of carbamazepine10,11-epoxide in Japanese epileptic patients. Eur. J. Clin. Pharmacol. 61, 25–34 (2005). www.nature.com/reviews/drugdisc © 2008 Nature Publishing Group Perspectives 105.McLellan, R. A. et al. Characterization of a human glutathione S-transferase µ cluster containing a duplicated GSTM1 gene that causes ultrarapid enzyme activity. Mol. Pharmacol. 52, 958–965 (1997). 106. Seidegard, J., Vorachek, W. R., Pero, R. W. & Pearson, W. R. Hereditary differences in the expression of the human glutathione transferase active on transstilbene oxide are due to a gene deletion. Proc. Natl Acad. Sci. USA 85, 7293–7297 (1988). 107. Simon, T. et al. Combined glutathione-S-transferase M1 and T1 genetic polymorphism and tacrine hepatotoxicity. Clin. Pharmacol. Ther. 67, 432–437 (2000). 108.Hayes, J. D., Flanagan, J. U. & Jowsey, I. R. Glutathione transferases. Annu. Rev. Pharmacol. Toxicol. 45, 51–88 (2005). 109.Allan, J. M. et al. Polymorphism in glutathione S-transferase P1 is associated with susceptibility to chemotherapy-induced leukemia. Proc. Natl Acad. Sci. USA 98, 11592–11597 (2001). 110. Anderer, G. et al. Polymorphisms within glutathione S-transferase genes and initial response to glucocorticoids in childhood acute lymphoblastic leukaemia. Pharmacogenetics 10, 715–726 (2000). 111. Dasgupta, R. K. et al. Polymorphic variation in GSTP1 modulates outcome following therapy for multiple myeloma. Blood 102, 2345–2350 (2003). 112. Harries, L. W., Stubbins, M. J., Forman, D., Howard, G. C. & Wolf, C. R. Identification of genetic polymorphisms at the glutathione S-transferase Pi locus and association with susceptibility to bladder, testicular and prostate cancer. Carcinogenesis 18, 641–644 (1997). 113. Stoehlmacher, J. et al. A multivariate analysis of genomic polymorphisms: prediction of clinical outcome to 5-FU/oxaliplatin combination chemotherapy in refractory colorectal cancer. Br. J. Cancer 91, 344–354 (2004). 114. Zimniak, P. et al. Naturally occurring human glutathione S-transferase GSTP1-1 isoforms with isoleucine and valine in position 104 differ in enzymic properties. Eur. J. Biochem. 224, 893–899 (1994). 115.Sprenger, R. et al. Characterization of the glutathione S-transferase GSTT1 deletion: discrimination of all genotypes by polymerase chain reaction indicates a trimodular genotype–phenotype correlation. Pharmacogenetics 10, 557–565 (2000). 116. Allorge, D., Hamdan, R., Broly, F., Libersa, C. & Colombel, J. F. ITPA genotyping test does not improve detection of Crohn’s disease patients at risk of azathioprine/6-mercaptopurine induced myelosuppression. Gut 54, 565 (2005). 117. Gearry, R. B., Roberts, R. L., Barclay, M. L. & Kennedy, M. A. Lack of association between the ITPA 94C>A polymorphism and adverse effects from azathioprine. Pharmacogenetics 14, 779–781 (2004). 118. Marinaki, A. M. et al. Adverse drug reactions to azathioprine therapy are associated with polymorphism in the gene encoding inosine triphosphate pyrophosphatase (ITPase). Pharmacogenetics 14, 181–187 (2004). 119. van Dieren, J. M. et al. ITPA genotyping is not predictive for the development of side effects in AZA treated inflammatory bowel disease patients. Gut 54, 1664 (2005). 120.von Ahsen, N. et al. Association of inosine triphosphatase 94C>A and thiopurine S-methyltransferase deficiency with adverse events and study drop-outs under azathioprine therapy in a prospective Crohn disease study. Clin. Chem. 51, 2282–2288 (2005). 121.Innocenti, F. et al. Pharmacogenetic analysis of interindividual irinotecan (CPT‑11) pharmacokinetic (PK) variability: evidence for a functional variant of ABCC2. J. Clin. Oncol. 22 (Suppl. 14), 2010 (2004). 122.Souto, J. C. et al. A genomewide exploration suggests a new candidate gene at chromosome 11q23 as the major determinant of plasma homocysteine levels: results from the GAIT project. Am. J. Hum. Genet. 76, 925–933 (2005). 123.Yan, L., Otterness, D. M. & Weinshilboum, R. M. Human nicotinamide N-methyltransferase pharmacogenetics: gene sequence analysis and promoter characterization. Pharmacogenetics 9, 307–316 (1999). 124.Bigler, J. et al. CYP2C9 and UGT1A6 genotypes modulate the protective effect of aspirin on colon adenoma risk. Cancer Res. 61, 3566–3569 (2001). 125.Chan, A. T., Tranah, G. J., Giovannucci, E. L., Hunter, D. J. & Fuchs, C. S. Genetic variants in the UGT1A6 enzyme, aspirin use, and the risk of colorectal adenoma. J. Natl Cancer Inst. 97, 457–460 (2005). 126.Ciotti, M., Marrone, A., Potter, C. & Owens, I. S. Genetic polymorphism in the human UGT1A6 (planar phenol) UDP-glucuronosyltransferase: pharmacological implications. Pharmacogenetics 7, 485–495 (1997). 127.Krishnaswamy, S. et al. UDP glucuronosyltransferase (UGT) 1A6 pharmacogenetics: I. identification of polymorphisms in the 5′-regulatory and exon 1 regions, and association with human liver UGT1A6 gene expression and glucuronidation. J. Pharmacol. Exp. Ther. 313, 1331–1339 (2005). 128.Krishnaswamy, S. et al. UDP Glucuronosyltransferase (UGT) 1A6 pharmacogenetics: II. functional impact of the three most common nonsynonymous ugt1a6 polymorphisms (S7A, T181A, and R184S). J. Pharmacol. Exp. Ther. 313, 1340–1346 (2005). 129.Lampe, J. W., Bigler, J., Horner, N. K. & Potter, J. D. UDP-glucuronosyltransferase (UGT1A1*28 and UGT1A6*2) polymorphisms in Caucasians and Asians: relationships to serum bilirubin concentrations. Pharmacogenetics 9, 341–349 (1999). 130.Nagar, S., Zalatoris, J. J. & Blanchard, R. L. Human UGT1A6 pharmacogenetics: identification of a novel SNP, characterization of allele frequencies and functional analysis of recombinant allozymes in human liver tissue and in cultured cells. Pharmacogenetics 14, 487–499 (2004). 131.van der Logt, E. M. J. et al. Genetic polymorphisms in UDP-glucuronosyltransferases and glutathione S‑transferases and colorectal cancer risk. Carcinogenesis 25, 2407–2015 (2004). 132.Carlini, L. E. et al. UGT1A7 and UGT1A9 poly morphisms predict response and toxicity in colorectal cancer patients treated with capecitabine/irinotecan. Clin. Cancer Res. 11, 1226–1236 (2005). 133.Girard, H. et al. Identification of common polymorphisms in the promoter of the UGT1A9 gene: evidence that UGT1A9 protein and activity levels are strongly genetically controlled in the liver. Pharmacogenetics 14, 501–515 (2004). nature reviews | drug discovery 134.Thibaudeau, J. et al. Characterization of common UGT1A8, UGT1A9, and UGT2B7 variants with different capacities to inactivate mutagenic 4‑hydroxylated metabolites of estradiol and estrone. Cancer Res. 66, 125–133 (2006). 135.Villeneuve, L., Girard, H., Fortier, L.‑C., Gagne, J.‑F. & Guillemette, C. Novel functional polymorphisms in the ugt1a7 and ugt1a9 glucuronidating enzymes in Caucasian and African–American subjects and their impact on the metabolism of 7‑ethyl‑10hydroxycamptothecin and flavopiridol anticancer drugs. J. Pharmacol. Exp. Ther. 307, 117–128 (2003). 136.Yamanaka, H. et al. A novel polymorphism in the promoter region of human UGT1A9 gene (UGT1A9*22) and its effects on the transcriptional activity. Pharmacogenetics 14, 329–332 (2004). 137.Chung, J.‑Y. et al. Effect of the UGT2B15 genotype on the pharmacokinetics, pharmacodynamics, and drug interactions of intravenous lorazepam in healthy volunteers. Clin. Pharmacol. Ther. 77, 486–494 (2005). 138.Levesque, E. et al. Isolation and characterization of UGT2B15(Y85): a UDP- glucuronosyltransferase encoded by a polymorphic gene. Pharmacogenetics 7, 317–325 (1997). 139.Nowell, S. A. et al. Association of genetic variation in tamoxifen-metabolizing enzymes with overall survival and recurrence of disease in breast cancer patients. Breast Cancer Res. Treat. 91, 249–258 (2005). 140.Sparks, R. et al. UDP-glucuronosyltransferase and sulfotransferase polymorphisms, sex hormone concentrations, and tumor receptor status in breast cancer patients. Breast Cancer Res. 6, R488–R498 (2004). 141.Toide, K. et al. A major genotype in UDPglucuronosyltransferase 2B15. Drug Metab. Pharmacokinet. 17, 164–166 (2002). Acknowledgements The authors would like thank B. Spear for his leadership and mentoring, as well as C. Locke and numerous colleagues in the Department of Clinical Pharmacokinetics at Abbott for helpful discussions and collaboration. Competing interests statement The authors declare competing financial interests: see web version for details. DATABASES Entrez Gene: http://www.ncbi.nlm.nih.gov/entrez/query. fcgi?db=gene ABCB1 | CYP2A6 | CYP2C19 | CYP2C9 | NAT2 | ORM1 | SLCO1B1 | UGT1A1 | UGT1A9 UniProtKB: http://ca.expasy.org/sprot BCHE | CYP2C18 | CY2D6 | CYP2R1 | CYP2S1 | CYP3A4 | FMO1 | FMO2 | FMO3 | OATP1B1 | TPMT FURTHER INFORMATION OMIM: http://www.ncbi.nlm.nih.gov/entrez/query. fcgi?db=OMIM The Pharmacogenetics and Pharmacogenomics Knowledge Base: http://www.pharmgkb.org/ SUPPLEMENTARY INFORMATION See online article: S1 (table) | S2 (table) | S3 (table) | S4 (table) | S5 (table) All links are active in the online pdf volume 7 | april 2008 | 305 © 2008 Nature Publishing Group