Skip main navigation
×

Opportunities and Challenges in Cardiovascular Pharmacogenomics

From Discovery to Implementation
Originally publishedhttps://doi.org/10.1161/CIRCRESAHA.117.310965Circulation Research. 2018;122:1176–1190

    Abstract

    This review will provide an overview of the principles of pharmacogenomics from basic discovery to implementation, encompassing application of tools of contemporary genome science to the field (including areas of apparent divergence from disease-based genomics), a summary of lessons learned from the extensively studied drugs clopidogrel and warfarin, the current status of implementing pharmacogenetic testing in practice, the role of genomics and related tools in the drug development process, and a summary of future opportunities and challenges.

    The American Heart Association estimates that the annual cost of cardiovascular care in the United States is $316.6 billion, including $32.8 billion for prescription drugs.1 The extent to which any drug generates its desired pharmacological effects varies among patients (Figure 1), and indeed the spectrum of response to all drugs extends from benefit to lack of efficacy to side effects. Thus, any attempt to understand and minimize this variability should improve cardiovascular health. As we describe here, genetic factors contribute to this variability, and, thus, a long-term goal of the discipline of pharmacogenomics is to understand genetic mechanisms underlying this variability and to apply this information to optimize care. We use the term pharmacogenomics to convey the idea that discovery in the field uses contemporary tools of genome science, including DNA variation across large populations and regulation of gene expression, and pharmacogenetics to describe work confined to small numbers of variants (eg, with early implementation efforts).

    Figure 1.

    Figure 1. Variability in low-density lipoprotein (LDL) lowering in response to 40 mg of simvastatin daily for 6 wk in 1007 patients. Most patients achieve 20% to 60% LDL lowering; in a few, the effect is greater and in a few, there is little effect. The distribution does not show any clear subset, consistent with the idea that multiple genetic variants contribute36 (Figure 5, bottom). Adapted by permission from Simon et al, 2006.140

    Arguably, the most common side effect of a drug is failure of efficacy. Some side effects represent an extension of a drug’s known pharmacological actions, such as hypotension with antihypertensives or bleeding with anticoagulants. In other cases, side effects are off target and, thus, may be difficult to predict in a patient or a population during drug development or even after drug marketing.

    Serious adverse events (SAEs) represent a major problem in drug use. These are generally defined as threatening life, or causing death, hospitalization, prolongation of hospitalization, a congenital anomaly, or significant disability. Common causes of SAEs include dosing errors (by the prescriber, the pharmacy, or the consumer), variable compliance, drug interactions, and other environmental factors such as diet. One report suggests that genetic factors play a prominent role in 20% to 30% of cases.2 SAEs were recognized as a major public health problem in the late 1990s when it was estimated that they are the fourth to sixth leading cause of in-hospital mortality in the United States,3 accounting for >100 000 deaths annually. A reanalysis of these data >10 years later showed no major change in this trend.4 In 1 study in the United Kingdom, adverse drug events accounted for 6.5% of all admissions to 2 major medical centers, used 4% of available hospital bed capacity (with an overall fatality rate of 0.15%),5 and the majority was because of cardiovascular SAEs. The drugs most commonly implicated were diuretics, warfarin, and nonsteroidal anti-inflammatory drugs including aspirin; gastrointestinal bleeding was the most common SAE. Summaries of data showing the role of genetic variants across disease domains in cardiovascular medicine are available,6,7 so this review will focus on principles, lessons learned, and opportunities for future discovery and implementation (Figure 2).

    Figure 2.

    Figure 2. The spectrum of pharmacogenomic science, from discovery to implementation.

    Mechanisms Underlying Variable Drug Responses

    Drugs interact with specific receptors to effect changes in molecular, cellular, whole organ, and whole organism function. Drug response is determined by 2 conceptually distinct sources of variability (Figure 3). The first, pharmacokinetics, describes the relationship of drug concentration as a function of time in any body compartment including that containing the drug receptor. Variability in drug disposition, determined by the processes of absorption, distribution, metabolism, and excretion, underlies variable drug action attributable to pharmacokinetic factors. Variants that alter the function of drug-metabolizing enzymes (most commonly cytochrome P450 drug-oxidizing enzymes [CYPs] such as CYP2C9 for warfarin) or drug transporters (such as SLCO1B1 for simvastatin) lead to interindividual variability in drug response attributable to variable pharmacokinetics. The second source of variability is pharmacodynamic: this term describes variability in drug action in the face of constant drug concentration. Variants that alter function in drug target genes (such as VKORC1 for warfarin or the β1 adrenergic receptor gene ADRB1 for β-blockers) are one source of pharmacodynamic variability. Others include variability in downstream receptor signaling, in cellular or whole-organ effects because of factors such as changes in serum electrolytes or autonomic tone, or in the disease process being treated.

    Figure 3.

    Figure 3. Pharmacokinetic and pharmacodynamic sources of variability in drug action. Pharmacokinetic processes determine drug concentration at molecular targets that through multiple mechanisms broadly termed pharmacodynamics transduce beneficial and undesirable drug effects.

    For each marketed and new drug, specific genes encoding the drug-metabolizing enzymes, drug transporters, and other proteins that mediate the absorption, distribution, metabolism, and excretion processes are increasingly well defined. Similarly, defining the genes encoding drug receptors and their downstream effector mechanisms provides a starting point for understanding pharmacodynamic mechanisms underlying variable drug responses; these gene sets can be much larger and more drug-specific than those mediating pharmacokinetics.

    Multiple types of variants affect gene function and drug response, including nonsynonymous single-nucleotide polymorphisms (SNPs), variants affecting splicing (Figure 4), promoter variants, small insertions or deletions, and copy number variants. In the case of drug-metabolizing enzymes, it is conventional to designate individuals with no recognized function-altering variants as extensive metabolizers; a more recent terminology8 suggests using normal metabolizers. Poor metabolizers are those with 2 loss-of-function alleles, intermediate metabolizers are those with one loss-of-function allele, and ultrarapid metabolizers (another proposed term is rapid metabolizers8) are those with alleles (SNPs or gene duplications) that result in enhanced enzymatic activity. As described below, the contribution of single common variants to variable drug actions can be so extensive that the distribution of drug effects is clearly multimodal; the resultant ability to identify subgroups with aberrant drug responses has provided much of the focus for contemporary pharmacogenetic science (Figure 5, top). It is important, however, to recognize that variation in multiple genes often underlies variable drug action, and in this case, multimodal distributions are usually not evident (Figure 5, bottom; also Figure 1); in this case, studying extremes of drug response or drug response across the whole population becomes a preferred approach.

    Figure 4.

    Figure 4. A common splice variant (arrow on the top) results in generation of an mRNA (bottom) that encodes a CYP3A5 with markedly reduced hepatic expression141and associated with reduced tacrolimus bioinactivation. Subjects of African origin, in whom this variant is uncommon, display lower tacrolimus concentrations,142 and this has been associated with an increased risk of rejection.

    Figure 5.

    Figure 5. Contrasting outcomes when variants in a single gene (top) or multiple genes (bottom) drive variability in drug response. Top, The distribution of drug responses is not normally distributed but shows distinct antimodes (heavy arrows) separating subjects with the poor metabolizer trait (because of 2 nonfunctional alleles) and those with the ultrarapid metabolizer trait (because of gene duplication or the presence of hyperfunctional alleles) from the majority (extensive and intermediate metabolizers) that are not readily distinguished. High-risk pharmacokinetic scenarios discussed in the text are examples. Bottom, A more usual distribution of drug responses without clear antimodes. Alleles associated with decreased drug responses are shown in darker colors, and the distribution reflects varying combinations of multiple alleles to overall drug response.

    One feature unique to pharmacogenomics is the use of a star allele system to convey information for each variant allele. For well-established genes affecting drug response (pharmacogenes), notably those encoding drug-metabolizing enzymes and drug transporters, the fully functional reference allele is generally referred to as *1 (Figure 4; Table 1), and each variant or combination of variants subsequently discovered designated by a new star allele number. Star alleles are numbered consecutively in order of discovery irrespective of impact on function; thus, the *2 allele for any gene may represent 1 (or sometimes >1) variant compared with the reference and may or may not confer altered functional status of the gene product. It is likely that as sequencing discovers large numbers of variants across pharmacogenes (Table 2), the star numbering system must be revised to a more standard nomenclature.9

    Table 1. Allele Frequencies for Common Pharmacogene Variants

    Table 1.

    Table 2. Numbers of Variants in Selected Pharmacogenes*

    GeneMissense or loss of function variantspresent in all ancestries at any frequencyNumber of variants with allele frequency > 1%
    AfricanAshkenaziEast AsianFinnishOther EuropeanLatinoOtherSouth AsianAll 8 ancestries
    CYP2D648710121051412121073
    CYP2C94362721222231
    CYP2C194421522322221
    VKORC12311230112220
    ADRB11652522222222
    SLCO1B151051042444462
    HLA-B2871049896979097961009073

    Other indicates ancestry was not assigned by the contributing site or by genomic analysis.

    *From the Genome Aggregation Database at http://gnomad.broadinstitute.org/.

    Application of the Tools of Genome Science to Pharmacogenomics: Apparent Exceptions to Some General Rules

    Since the completion of the human genome project in 2003, the cost of large-scale genotyping and sequencing has plummeted, analysis pipelines for quality control and variant identification are becoming standardized, and a fuller understanding of the complexity of the genetic underpinnings of human traits has emerged. Several themes from disease-focused genomic studies also apply in pharmacogenomics. For example, sequencing consistently identifies large numbers of predicted deleterious variants in each sequenced genome, even from healthy individuals.10,11 Similarly, in pharmacogenes where (as described below) the emphasis to date has been on common variants, rare variants are frequent (Table 2), and genetic variation is often ancestry specific (Tables 1 and 2). However, there are several areas where pharmacogenomic investigations have defined themes that diverge from those of disease-focused genomics.

    Family Studies

    From early linkage studies to modern whole-exome sequencing, parent–child trios and large pedigrees have provided sentinel discoveries in disease-focused genomics. In pharmacogenomics, family studies have been used to define patterns of inheritance and point to candidate genes for unusual drug responses; early studies of malignant hyperthermia12 and of prolonged paralysis following succinylcholine because of pseudocholinesterase deficiency13 are examples. Studies of drug disposition have also compared monozygotic versus dizygotic twins to demonstrate a familial component to variable drug actions.1417 However, kindreds with well-defined variable drug responses are not frequently collected because it is unusual to have multiple members of a family exposed to the same drug. Thus, candidate gene and unbiased methods such as genome-wide association studies (GWAS) have more commonly been used to define genes contributing to variable drug actions.

    Common Alleles Can Be Associated With Large Effects

    It is virtually axiomatic in genome science that highly prevalent genetic variants do not generate large effect sizes for human traits with adverse health consequences because evolutionary pressures would rapidly eliminate variants that unfavorably impact survival. Pharmacogenomics provides one exception to this generalization, because common alleles with large effect sizes in the presence of a drug are well established as described below for clopidogrel, warfarin, simvastatin, and tacrolimus (Figure 4), for example. The likely explanation is that there is no evolutionary pressure on variants in these genes because until recently drug exposures were not part of the human condition. It is this set of high-frequency, large effect size alleles that has generated the tantalizing idea of widespread use of pharmacogenetic variant information to tailor the choice of drugs or drug dose.

    Candidate Gene Studies Can Replicate

    Another way in which pharmacogenomic variants break with conventional wisdom in genome science is in the success of candidate gene approaches. The early days of genome science saw many studies that reported apparent effects on human traits of common variants in genes chosen because they make physiological sense; a myriad of studies describing the relationship between the common insertion–deletion polymorphism in ACE and a range of vascular effects are excellent examples. However, the vast majority of positive results from such candidate gene studies did not ultimately replicate.18,19 The most likely cause is that while variability in many common human traits has a prominent genetic component, that component is itself highly polygenic, so the expectation that variants in any single gene would have a major contribution to a phenotype is (at least in retrospect) naive. In the case of drug disposition, there are now examples (described in the following section) in which loss- or gain-of-function of a single drug elimination pathway can result in dramatic (orders of magnitude) and readily replicated changes in drug concentration and thus effect. These functional alterations can occur because of variation in genes encoding drug-metabolizing or drug transport molecules, drug interactions that affect function of these molecules, or excretory organ dysfunction.

    High-Risk Pharmacokinetics

    There are 2 scenarios under which single genetic variants can exert large effects on drug concentration. In the first, administration of a prodrug requires bioactivation by a single pathway to achieve therapeutic effects, so genetic variants that reduce function of the bioactivating enzyme system can produce drastically reduced or absent drug action. Clopidogrel, bioactivated by CYP2C19 as discussed further below,20 is an example. Occasionally, genetic variants can enhance metabolism, and in this case, exaggerated drug effects may arise as a result of unusually high active drug metabolite concentrations. CYP2D6 ultrarapid metabolizer individuals (because of gene duplications) who receive the prodrug codeine may generate high concentrations of the active metabolite morphine and respiratory depression and fatalities have been reported.21 Similarly, individuals with the CYP2C19*17 variant display enhanced enzymatic activity and may be at risk for increased bleeding complications with clopidogrel.22

    In the second scenario, an active drug is eliminated to inactive metabolites by a single pathway at usual doses. If the activity of that drug-metabolizing pathway is reduced (because of genetic variation or drug interactions), parent drug concentrations can rise and produce excess drug action and toxicity. S-warfarin, metabolized by CYP2C9 and also discussed further below, is an example. These scenarios have been termed high-risk pharmacokinetics23 and represent an unusual example in which variation in a common candidate genetic polymorphism reproducibly affects the trait of variable drug response. The antiarrhythmic flecainide represents an extension of this concept. The drug is excreted unchanged by the kidneys and metabolized by CYP2D6. Thus, in patients with 2 hits (renal dysfunction and decreased CYP2D6 function caused by the poor metabolizer trait or by interacting drugs), serious toxicity attributable to flecainide accumulation can result.24

    Pharmacogenetics Traits Have Large Effect Sizes, Thus Successful GWAS With Small Sample Sizes Are Possible

    The last decade has seen the development of technologies and analytic methodologies to interrogate the whole genome and identify loci, or specific variants, contributing to the complex genetic architecture of human traits. When this approach, notably GWAS, has been applied to common traits such as coronary artery disease or atrial fibrillation, or to physiological markers such as serum lipids or electrocardiographic intervals, dozens of associated loci have been identified. Studies involving hundreds of thousands of cases and controls have identified new pathways and mechanisms modulating disease susceptibility or trait variability, but individual variants identified by GWAS generally have small effect sizes. GWAS for some drug response traits represent exceptions to these general rules in that large signals have sometimes been detected in relatively small numbers of cases.25 One explanation is large contributions by single common gene variants to specific drug response phenotypes. Further, a recent systematic comparison confirmed larger effect sizes for pharmacogenetic variants detected by GWAS compared with effect sizes for other traits26 and suggested that the mechanism was that although many complex traits represent interactions between genetic variation and the environment, pharmacogenetic traits were unusual in that the environmental component (drug exposure) was well measured. For example, case–control GWAS of immunologically mediated SAEs, such as hepatotoxicity or the Stevens–Johnson syndrome/toxic epidermal necrolysis (SJS/TEN), have used small case numbers (several dozen or less) to generate strong, genome-wide signals often at the human leukocyte antigen (HLA) locus, and specific HLA alleles required to generate these SAEs have been described.27,28 Early studies of simvastatin pharmacokinetics demonstrated that the organic anion transporter protein molecule OAT1B1 regulates liver uptake of the drug, and variants in SLCO1B1, which encodes the transporter, were associated with higher simvastatin concentrations.29 A GWAS of 85 biochemically confirmed cases of simvastatin-induced myopathy occurring with high doses of the drug (80 mg/d) identified a single SNP in SLCO1B1 that was associated with a 16.9-fold increase in odds ratio for myopathy in homozygote carriers versus homozygote controls; the odds ratio for heterozygote was 4.5, and >65% of myopathy cases could be attributed to the variant.30 Although myopathy is a feature of therapy with all statins, the effect of the SLCO1B1 variant is greater with high-dose simvastatin than other agents of the class,31,32 so other genes are likely involved in the genesis of this SAE.

    Although GWAS has successfully validated candidate genes for traits with high effect size single variants, including immunologically mediated adverse drug reactions and some high-risk pharmacokinetics drugs such as clopidogrel or warfarin discussed further below,3335 other GWASs have studied larger numbers of subjects and have generated more modest signals. For example, a meta-analysis of GWAS of low-density lipoprotein response to statin therapy in >40 000 subjects studied replicated previous signals at APOE and LPA and found new signals near SORT1 and SLCO1B1,36 implicated in variable simvastatin disposition and myotoxicity. More recently, a preliminary report of a GWAS in 10 780 statin-exposed subjects identified variants at LPA as risk alleles for cardiovascular events during statin therapy.37

    GWAS results have provided the starting point for further studies of drug mechanisms in disease. Thus, for example, variants at 4q25 have been implicated in atrial fibrillation susceptibility38 and have also been used as candidate variants to probe variability in response to drug and other therapies for atrial fibrillation.3941 As discussed further below, GWAS results have also provided a starting point for the development of genetic risk scores (GRS) that may be useful in assessing some drug responses.

    Applications: Warfarin and Clopidogrel—What Have We Learned?

    Exploration of the genetic basis of variability in response to these 2 drugs has provided new knowledge and at the same time has highlighted controversies in how such genetic information can be used clinically. Genetic variants have been repetitively shown to influence outcomes with these drugs. Outcomes have included intermediate phenotypes, such as drug dose, and harder end points such as major adverse cardiovascular events.

    Initial Studies Showing That Genetic Variation Affects Response to These Drugs

    The initial identification of common variants42 in CYP2C9, the gene responsible for the bioinactivation of the active S-enantiomer of warfarin, led to studies demonstrating that European ancestry individuals with common loss-of-function variants had lower dose requirements43 and were nevertheless at higher risk for bleeding events, an observation that has been replicated but has not been fully explained. One possibility is that loss-of-function variation leads to more intraindividual variability in warfarin concentration and thus more susceptibility to swings in anticoagulation efficacy, although this has not been definitively proven.

    Rare families demonstrate the phenomenon of warfarin resistance requiring high dosages to achieve therapeutic anticoagulation. The genetic basis of this syndrome is loss-of-function variation in VKORC1, encoding the component of the vitamin K complex with which warfarin interacts (the warfarin target).44 Common variants in VKORC1 promoter have been associated with variability in hepatic VKORC1 transcripts and with warfarin dose requirements.45 Nonsynonymous coding region variants in the gene have also been described in specific populations (eg, Ashkenazi Jewish populations46) and have been associated with relatively high warfarin dose requirements. Studies in registries or in electronic health records (EHRs) have associated the CYP2C9*3 loss-of-function variant or a variant in CYP4F2 (responsible for vitamin K metabolism) as risk factors for warfarin-associated bleeding.47,48 These studies have involved hundreds of subjects with bleeding events; notably, large randomized clinical trials (RCTs) assessing the utility of using warfarin pharmacogenetics discussed further below included few bleeding events.

    Although clopidogrel was approved for marketing in 1998 and was known to be a prodrug, the bioactivation pathway was only identified in 2006 in studies demonstrating that individuals with a single copy of the *2 loss-of-function variant in CYP2C19 demonstrated less antiplatelet activity during standard-dose clopidogrel than did subjects lacking the variant.20 This finding was followed by 3 large retrospective analyses that demonstrated increased major adverse cardiovascular event and similar events in subjects with the CYP2C19*2 variant compared with controls, and, in some studies, the effect was larger in homozygotes than in heterozygotes.4951

    Influences of Genetics on Drug Response Can Be Indication Specific

    One large meta-analysis of the role of genetic variation in clopidogrel response suggested no effect,52 but that analysis included a large number of subjects in trials of clopidogrel in atrial fibrillation, a setting in which the drug exerts minimal, if any, therapeutic effect. Thus, any effect of genetics would be expected to be minuscule in such a setting. It is in the acute coronary syndrome setting that the data to date have been strongest for an allele-specific effect of clopidogrel.53 Information on the influence of CYP2C19 variants on clopidogrel efficacy in cerebrovascular disease has been hampered by small numbers of patients. However, one recent study in 2933 Chinese patients with minor stroke or transient ischemic attack identified a strong effect of CYP2C19 loss-of-function variants (including not only *2 but also a variant common in Asian populations, *3): only subjects lacking these alleles (41.6% of this study population) showed a protective effect of clopidogrel on recurrent events.54 Factors that were important for this result included large numbers, a population enriched in loss-of-function variants (the frequency in European ancestry subjects is ≈25%), and a high event rate. These data show that demonstrating the influence of pharmacogenetic variation will depend on the efficacy of the drug for the target indication, the size of the population studied, and the effect size and population frequency of the variant(s) studied.

    Variants in Single Genes May Be Insufficient

    In the case of warfarin, the major effects of variants in CYP2C9 and in VKORC1 illustrate the way in which pharmacokinetic and pharmacodynamic genes can together have major influences on drug outcomes. Multiple other genes in the clopidogrel pathway have been implicated in variability in clopidogrel action55 but have not been replicated in large studies.56 A small GWAS33 (n=429) of clopidogrel’s effect to inhibit ADP-induced platelet aggregation identified a signal at genome-wide significance in the CYP2C19 locus, and subsequent analyses demonstrated that CYP2C19*2 accounted for the entire signal. However, this allele accounted for only 12% of the variability in the trait of clopidogrel inhibition of ADP-induced platelet aggregation. Although this is a large effect for a single variant, it also illustrates the fact that much of the variability in the trait continues to be unexplained. Similarly, a GWAS of steady-state warfarin dose requirements in 181 European ancestry patients (with replication in 374) identified variants at CYP2C9 and VKORC1 at genome-wide significance as mediators of warfarin dose requirement.34 A larger GWAS on the same trait in 1053 Swedish subjects35 also implicated a variant in CYP4F2. These common variants account for ≈50% of the variability in warfarin dose requirement, again a large proportion, but the remainder is unexplained. A GWAS examining warfarin dose requirements in African ancestry subjects identified an association, at genome-wide significance, for a SNP in the CYP2C9/19 gene cluster, but the actual functional SNP has yet to be identified.57

    Ancestry Matters

    Allele frequencies for a large number of functional polymorphisms in absorption, distribution, metabolism, and excretion genes vary strikingly by ancestry (Table 1).58 The allele frequency of common loss-of-function variants in CYP2C19 varies from 15% (European ancestry) to >50% (Asians) in the stroke study above. Similarly, the population frequencies of variants in CYP2C9 that lead to decreased warfarin requirements because of decreased S-warfarin bioinactivation are different in African subjects compared with European ancestry subjects. Therefore, studies that do not genotype African alleles will necessarily fail to predict dose requirement and outcome in subjects of African origin.59,60 This was highlighted in the COAG trial (Clarification of Optimal Anticoagulation Through Genetics) discussed further61 below in which the genetic algorithms used to predict dose actually performed worse than an empirical clinical algorithm in African ancestry subjects.

    Randomized Clinical Trials

    In 2013, 3 large RCTs, 1 from the United State and 2 from Europe, reported the results of randomizing subjects to genetically guided or clinically guided warfarin dose selection.6163 The trials were not powered to examine hard outcomes such as bleeding or recurrent thrombosis but rather focused on time in the therapeutic range in first 30 to 90 days of therapy. The US trial (COAG) used an algorithm that accounted for multiple features (age, sex, interacting drugs, etc) in the control arm and incorporated variants known to be important for warfarin therapy (in European ancestry) in the genetically guided arm. There was no difference in time in therapeutic range and, as discussed above, time in therapeutic range was actually decreased in black subjects randomized to the genetic arm. The European trials used fixed loading doses compared with a genetically guided arm and showed statistically significant improvement in time in therapeutic range. Most recently, the GIFT (Genetic Informatics Trial) of warfarin therapy to prevent deep vein thrombosis64 randomized 1650 subjects after hip or knee surgery to pharmacogenetically guided or conventionally guided warfarin therapy. Unlike the three 2013 trials, the composite primary end point focused on adverse events: major bleeding at 30 days, an international normalized ratio >4 at 30 days, death within 30 days, or a venous thromboembolic event within 60 days of surgery. The use of genetically guided warfarin dose selection resulted in a 10.8% primary outcome rate versus 14.7% in the conventional arm (relative risk, 0.73; 95% confidence interval, 0.56–0.95).

    Organizing an RCT for clopidogrel genetics faces the problem of equipoise. Available data show that even quadrupling the usual clopidogrel dose in individuals homozygous for loss-of-function variants will not alter platelet reactivity.65 This makes some prescribers reluctant to include such individuals in RCTs where they might be randomized to clopidogrel. There is currently a 5200-patient RCT (TAILOR-PCI [Tailored Antiplatelet Therapy Following PCI]) underway comparing clopidogrel in individuals lacking variant alleles to ticagrelor in individuals with variant alleles.66

    Thus, trials in this area have been difficult because of issues around sample size (especially with respect to rarer SAEs), equipoise, expected effect sizes, difficulty in defining and accruing subjects with clinically meaningful end points, and the fact that genetic testing is unlikely to produce a meaningful dose adjustment or clinical effect in most patients. A nice demonstration of these issues was highlighted in a study examining the added value of using thiopurine methyltransferase pharmacogenetics in a trial of azathioprine in inflammatory bowel disease.67 Azathioprine is a thiopurine methyltransferase substrate, and loss-of-function variants in thiopurine methyltransferase have been associated with striking elevations of toxic metabolites of the drug. This trial randomized 378 subjects to conventional therapy and 405 subjects to a pharmacogenetically based intervention, downward dose adjustment in heterozygote and homozygote carriers. After dose adjustment, there was no difference in disease activity across groups, and there was no difference in the percentage of subjects with immunologic or adverse drug reactions (30/405 versus 30/378) by intention-to-treat analysis. However, in a planned secondary analysis focusing only on thiopurine methyltransferase variant carriers, the hematologic SAE rate was 1/39 in the intervention group versus 8/35 in the conventional group (relative risk, 0.11; 95% confidence interval, 0.01–0.85). This result demonstrates the fact that pharmacogenetic variants will only affect outcome in a small number of subjects, and, therefore, large RCTs that randomize an entire ungenotyped population are likely to produce negative results. One trial underway in Europe is examining many drug–gene pairs and includes a primary end point that focuses on individuals known to have genetic variants.68

    Implementing Pharmacogenomics in Practice

    One stance (endorsed in the case of clopidogrel by the American Heart Association69) has been that the evidence base to implement genetic testing for all subjects is not sufficiently robust when drugs such as warfarin and clopidogrel are started. The converse attitude is that the biology is strong and that where genetic data widely and readily available, it would be useful to incorporate into clinical decision making. It is important to acknowledge that because well-studied common polymorphisms leave considerable variability in drug effects unexplained, the goal of implementation is to improve the likelihood of a favorable outcome, and not to guarantee it.

    Two approaches to such adoption have been described.70,71 One uses rapid point-of-care genotyping to deliver results within an hour or 2. These results (eg, CYP2C19 genotype for clopidogrel prescribing) can then be used to tailor drug selection or drug dose at the point of care.72 The second is a preemptive approach that embeds pharmacogenetic variant information in EHRs with clinical decision support to be used when a drug with established genetically based variability is prescribed to a patient with a variant in the relevant genetic pathway.73,74 An advantage of the preemptive approach is that multiple genetic variants for many drugs can be tested simultaneously, but a disadvantage is that the time lag is generally longer than the point of care approach and clinical decision support needs to be implemented (and in some cases developed) for each drug–gene pair tested. In some settings (eg, patients scheduled for cardiac catheterization and patients scheduled for hip or knee surgery), subjects at increased risk for receiving drugs with known drug–gene interactions can be identified and preemptively screened. Both approaches have been implemented at early adopter institutions (Figure 6), and initial results are being reported. A consortium of centers in the Pharmacogenetics Working Group of the NIH’s IGNITE (Implementing Genomics in Practice) network75 examined major adverse cardiovascular event during antiplatelet therapy after percutaneous coronary intervention in 1815 patients (572 of whom had a CYP2C19 loss-of-function variant) across 7 medical centers.76 They found that events were significantly more common in those with a loss-of-function allele given clopidogrel versus those prescribed alternate therapy, whereas event rates were similar in those with loss-of-function variants prescribed alternate therapy and those with no loss-of-function variants prescribed clopidogrel. In another study, implementation of a preemptive pharmacogenetic program was associated with continuation of ordinary-dose clopidogrel in 96% of subjects without genetic variants, compared with 43% of homozygotes and 65% of heterozygotes.77 Reasons physicians gave for not switching individuals with genetic variants were delay in delivery of genetic information and lack of suitable alternatives (eg, in the elderly or those with renal dysfunction). The argument is also made that such settings can be used to generate data on real-world effects because they include subjects, such as the elderly, children, those with multiple comorbidities, and those of non-European ancestries, who are often not well studied in drug development and clinical trials.

    Figure 6.

    Figure 6. Point-of-care clinical decision support. This pop-up appears in the Vanderbilt electronic health record when clopidogrel is prescribed to a patient who has had pharmacogenetic data deposited in their record and is known to carry CYP2C19 loss-of-function allele(s).

    Clinical Pharmacogenetics Implementation Consortium

    The development of multigene panels and requirements for clinical decision support for each drug–gene pair has been facilitated by the Clinical Pharmacogenetics Implementation Consortium, a group of pharmacogenetic and content experts that has developed standardized templates to provide literature reviews and responses to the question “If the prescriber knows this variant is present in a patient receiving this drug, what action should be taken?” The Clinical Pharmacogenetics Implementation Consortium does not address the issue of who should have genetic testing, but rather if the results of genetic testing are available (eg, from individuals who have undergone commercial testing for other reasons, or from individuals participating in point of care of preemptive pharmacogenetic programs), what response could be envisioned.78 To date, Clinical Pharmacogenetics Implementation Consortium has produced guidelines for 27 drug–gene pairs. The guidelines include data on drug metabolizer phenotypes (eg, extensive metabolizer [or normal metabolizer], poor metabolizer, ultrarapid metabolizer [or rapid metabolizer]) expected with specific diplotypes, anticipated effect size, the availability of alternate therapies, and the clinical consequences of drug inefficacy or toxicity. A Dutch consortium is producing similar guidelines.79

    One area where pharmacogenetic testing has taken hold (although not yet in cardiovascular therapeutics) is in screening for HLA-B susceptibility alleles in the prevention of SJS/TEN. Initial studies with the antiretroviral abacavir strongly implicated HLA-B*57:01 as a risk allele,80 and an RCT in 1956 patients demonstrated that a genetically guided approach could eliminate drug-related SJS/TEN while the incidence was ≈3% in conventionally treated subjects.81 As a result, current guidelines for use of abacavir mandate HLA-B*57:01 testing. Studies in Southeast Asia have strongly associated HLA-B*15:02 with carbamazepine-related SJS/TEN.82,83 One national program that mandated preprescription testing resulted in a switch from carbamazepine to other anticonvulsants, with a resultant drop in carbamazepine-induced SJS/TEN but a rise in cases because of other anticonvulsants.84 Thus, simply mandating testing without an accompanying educational program is likely to be ineffective. In European ancestry subjects, there is a different risk allele, HLA-B*31:01, for carbamazepine-associated SJS/TEN.28

    Role of Genomics in the Drug Development Process

    An emerging body of evidence indicates that human genetics and phenome mining in EHRs can be used to improve the likelihood of successful drug development and possibly to direct repurposing (the identification of new indications for agents already approved or in development).

    GWAS and Drug Targets

    One early study testing incorporation of human genetics in drug development examined the relationship between genetic mechanisms of disease, identified by GWAS, and drugs targeting those mechanisms.85 The authors contended that when genetic evidence was present, confidence in the disease indication for that drug was enhanced. More importantly, an association between genetic mechanisms of disease inferred by GWAS could also identify repurposing opportunities. A follow-up study suggested that genetic evidence, most often in terms of variation in known disease pathways targeted by marketed drugs, was evident in 2% of drug candidates at the preclinical stage and 8.2% in approved drugs; this result strongly supports the idea that drugs targeting known or suspected genetic mechanisms for disease have a greater likelihood of reaching the market than all drugs entering the development pipeline.86

    One large study in rheumatoid arthritis used GWAS and subsequent advanced informatics methods (including functional annotation, known cis-acting expression quantitative trait loci, and pathway analyses) to identify 98 candidate genes in a set of 29 880 rheumatoid arthritis cases and 73  758 controls.87 Among the 98 biological candidate genes identified, many were targets of approved therapies in rheumatoid arthritis and the remainder, therefore, became candidates for drug targeting. Indeed, for some of these genes, drugs had previously been developed for other indications (notably malignancies), and the suggestion was, therefore, advanced that these drugs (palbociclib, capridine-β, flavopiridol, and alvocidib) could be similarly repurposed for rheumatoid arthritis.

    After a large trial showed no overall efficacy of the cholesteryl ester transfer protein inhibitor dalcetrapib, a GWAS in 5749 patients receiving the drug identified common SNPs in ADCY9 as a significant determinant of effective therapy88; this finding suggests that the drug could be retested in subjects with the favorable allele.89

    Rare Variants in Lipid Genes Identify New Targets

    The development of PCSK9 inhibitors highlights another way in which human genetics impacts the drug development process. The seminal observation that mutations in PCSK9 were a rare cause of familial hypercholesterolemia90 was followed by evidence that this likely represents gain of function.91 Sequencing PCSK9 in the Dallas Heart Study cohort then identified common truncating variants (ie, expected to be loss of function) associated with low low-density lipoprotein cholesterol in African ancestry subjects.92 A follow-up analysis in the Atherosclerosis Risk in Communities Study showed that the truncating variants were associated with not only lower low-density lipoprotein cholesterol (137±37 versus 116±33 mg/dL) but also a ≈90% reduction in coronary artery disease events.93 There were several other key observations that enabled the rapid development of PCSK9 inhibitors for coronary artery disease. First, individuals with truncating mutations did not seem to have any detrimental phenotype, and indeed, subsequent investigations have identified individuals homozygous for such truncating mutations, with extraordinarily low low-density lipoprotein cholesterol, and no other apparent clinical phenotype.94 The second key observation was that the analyses were conducted in African ancestry subjects, in whom truncating mutations in PCSK9 are especially common. This reinforces the critical need for studying populations of diverse ancestry in genomics and in pharmacogenomics.

    Other studies have identified rare loss-of-function variants, often in lipid pathways, associated with beneficial changes in lipid profiles and in decreased coronary artery disease events. This strategy has been used to validate the ezetimibe target NPC1L195 and to implicate ANGPTL396 and APOC97 as mediators of hypertriglyceridemia and of increased coronary artery disease. These genetic observations not only reinforce the role for triglycerides as a direct cause of coronary artery disease but also identify these molecules a potential drug targets. These approaches have also been used to implicate SLC30A8, encoding a zinc transporter in pancreatic beta cells, as protective against diabetes mellitus in obese subjects,98 suggesting that this transporter could be a druggable target. The approach of identifying loss-of-function variants that protect against important human phenotypes has not yet been applied widely in other cardiovascular diseases, such as hypertension, arrhythmias, or heart failure.

    Targeted Drug Therapy in Mendelian Disease

    Studies of Mendelian diseases are identifying molecular mechanisms of disease that are now being translated into new drug development or repurposing of existing drugs. One example is the role of the noninactivating late sodium current as the cause of type 3 long-QT syndrome. Case reports and multiple small studies have suggested that sodium channel–blocking drugs, including known agents, such as flecainide or mexiletine, and newer agents, such as ranolazine, can shorten the QT interval in these subjects.99,100 However, clinical trials have not yet answered the question of whether these therapies reduce cardiovascular events.

    In cystic fibrosis, specific mutations have been associated with normal cell surface expression of dysfunctional chloride transporters, and in these instances, the small molecule ivacaftor has been shown to improve chloride transport and can exert remarkable changes in functional status.101 This presumably reflects the idea that even small changes in chloride transport can ameliorate the underlying functional defect; and so chloride transport need not be normalized, but only improved a small amount to result in a large clinical effect. The cost of ivacaftor has been cited at >$300 000/yr, so the trade-off between functional improvement (including not only fewer hospitalizations but also improved quality of life) and cost has become controversial. One analysis in Great Britain suggested that the cost of the drug far exceeded the measurable cost savings it confers (eg, by reducing hospitalizations).102

    In Duchenne muscular dystrophy, eteplirsen, an RNA therapeutic designed to skip transcription of dystrophin exon 51 and, thus, generate a truncated but functional molecule, has been evaluated in 8 patients receiving 1 of 2 doses of active drug and 4 receiving placebo.103 Patients receiving the drug showed increased dystrophin expression and improvement in some functional indices (eg, longer walk time at the lower active dose but not at a higher dose). Although staff at the Food and Drug Administration and an advisory committee argued against approval in part because of the small numbers and lack of dose–response data, the drug was granted approval in 2017; the sponsor is to conduct a further trial of efficacy, although this will not be placebo controlled.104

    In Marfan syndrome, molecular dissection of underlying mechanisms has implicated abnormal TGF-β (transforming growth factor β) signaling, and in mouse models, there is a striking beneficial effect of angiotensin receptor blockers.105 In clinical trials, angiotensin receptor blockers perform as well as, but to date, not better than, standard β-blocker therapy.106 Response to β-blockers may also be subject to pharmacogenomic modification, as variation in ADRB1 is associated with differential response to atenolol in patients with Marfan syndrome.107

    These mechanism-based examples highlight the potential for genomic exploration of mechanisms of disease to advance therapy in specific individuals. Outstanding issues in the field include the cost of developing and delivering these new therapies and the evidence that regulatory authorities require to allow these new therapies on the market.

    The Future

    Rare Variants in Pharmacogenes

    A major focus in pharmacogenomics over the past 50 years has been on describing common variants with large effect sizes for certain drugs. This work has in turn led to experiments in implementing this information to improve the outcome of drug therapy. Indeed, review of genetic variant information in even small numbers of drug–gene pairs reveals that most subjects carry a common genetic variant that would be of clinical import when a target drug is prescribed.108 However, the large effect size/common variant scenario has, in some ways, limited pharmacogenomic science. Table 2 shows the numbers of nonsynonymous variants reported in the Genomic Aggregation Database in genes known to modulate actions of multiple drugs. Although these are individually rare, collectively they again reinforce the idea that most individuals will have a genetic variant in a pharmacogene that may become important to their therapy at some point in the future. Some of these are common but only in specific ancestries (eg, D36Y, an Ashkenazi-specific VKORC1 variant46). Thus, a major challenge facing genomic and pharmacogenomic science is the development of robust methods to identify all such variants across populations109111 and to predict and characterize function of these variants—individually and in combination—as they are described. These methods could be in silico; in vitro functional, high-throughput assays112,113; or could search for drug response phenotypes in large clinical populations such as EHRs.114 One interesting emerging technology is the use of induced pluripotent stem cells to derive specific cell types (eg, hepatocytes and cardiomyocytes) and examine variability in drug response,115 as discussed below. Further, application of CRISPR-Cas9 for rapid and reliable gene editing in these cells is now being deployed to assess a causative relationship between genetic variants and specific phenotypes.116,117 As methods evolve to drive these cells to a fully mature phenotype, this approach may become more common.

    Beyond Candidate Genes

    One conventional focus in pharmacogenomic discovery has been on candidate genes defined using pharmacokinetic or pharmacodynamic principles outlined above. This has served the field well, but new tools are now becoming available to understand further the complexities of human biology and the way in which this impacts variable drug responses. For example, not all patients with HLA-B risk alleles exposed to abacavir develop SJS/TEN.118 One hypothesis that has some experimental support is the idea that the interaction of drug with the HLA molecule alters the repertoire of peptides presented to tissue resident T cells.119

    Figure 5 highlights the complexities of analyzing drug response when multiple genetic variants contribute; systems biology approaches may be one way forward in identifying critical nodes modifying drug action.120 Unbiased approaches beyond genomics are increasingly applied to understand variable drug responses. For example, a transcriptomic approach in statin-exposed cell lines identified glycine amidinotransferase as a potential functional link between sterol signaling and myotoxicity.121 Induced pluripotent stem cells are also being developed to understand and predict drug actions. Here, 2 broad approaches can be envisioned. In the first, a battery of standard induced pluripotent stem cells would be used to derive specific cells, such as cardiomyocytes or vascular endothelial cells, and the response of these cells to drug challenges could then be established. This approach is being developed by a consortium of the Food and Drug Administration and industry to better predict drugs at risk for causing QT prolongation and torsades de pointes.122 A second approach is to study induced pluripotent stem cells from individuals with known drug exposures and a range of drug responses to then define pathways predisposing to toxicity. Using this approach, cardiomyocytes from individuals with anthracycline toxicity demonstrated impaired mitochondrial, metabolic, calcium handling, and antioxidant activity and decreased cell viability.115 Similarly, transcriptomic analysis of iPSC cardiomyocytes from normal subjects displaying extremes of QT prolongation with sotalol challenge identified altered expression of a series of genes potentially involved in repolarization as mediators of this effect.123

    Measuring Plasma Concentrations

    The field of pharmacogenomics had its start in studies that measured plasma drug concentrations and identified pharmacokinetically based outliers. With the development of newer (eg, mass spectrometry–based) rapid analytical methods, a return to incorporating measured plasma drug concentrations to identify outlier subjects can be envisioned. This approach, combined with genetic data, has been reported to avoid high concentrations of atorvastatin and rosuvastatin.124 In another study, leftover blood samples were combined with EHR records of fentanyl administration to develop a population-based pharmacokinetic model for fentanyl administration to neonates and children.125 A mass spectrometry–based multiplexed assay of multiple antihypertensive drugs was used to distinguish subjects with resistant hypertension from those with noncompliance.126

    Genetic Risk Scores—The QT Example

    Methods to combine multiple risk alleles into a genetic risk score may be one way of approaching analysis of multigenic drug responses (eg, bottom, Figure 5). A GWAS on QT duration involving >100 000 subjects identified 35 loci with signals at genome-wide significance.127 The locus with the most statistically significant signal was near NOS1AP, which encodes an ancillary protein for neuronal NO synthase; the P value was <10212 but (as is usual in GWAS of complex traits) a small effect size, 3.5 ms/allele. This locus has been recognized to influence the QT for >10 years128 and has been proposed to be a modulator of calcium or potassium channel function129 or to affect cell–cell coupling in heart.130 In large populations, variants in NOS1AP have been associated with slightly increased risk of sudden death in populations131 and in variability in the extent to which a congenital long-QT syndrome mutation produces a clinical phenotype, that is, NOS1AP variants have been implicated as modulator genes for this congenital arrhythmia syndrome.132,133 However, the complex genetic architecture of the QT interval suggests that single variants are unlikely to be highly predictive for any trait. Recently, Strauss et al134 have implemented a GRS that combines variant information at 61 GWAS QT loci, weighted by their effect sizes, and found that the GRS was a strong predictor (r2=0.3; P<0.02) of the extent to which the QT interval prolonged in response to challenge with the QT prolonging drug dofetilide. This result suggests that the genetic architecture of the QT interval is determined by multiple genetic variants and that drug challenge may have differential effects, depending on the specific variants contributing to QT interval in an individual. This is another manifestation of the idea of reduced repolarization reserve in which drug challenge produces a range of QT responses, depending on individual genetic makeup.135,136 The study by Strauss et al also included an analysis of the ability of the GRS to distinguish between cases of drug-induced torsades de pointes and controls exposed to drug and not developing marked QT prolongation. An initial GWAS of this phenotype in 216 cases and 771 controls revealed no single locus even approaching genome-wide significance,137 but the GRS could distinguish between cases and controls in the same set with a P<106. GRS are generally developed by using a weighted combination of GWAS-derived associations; newer methods that consider biological function (eg, a variant in a transcription factor and its target gene) may refine this approach.

    A Role for the EHR

    To take full advantage of emerging statistical methods and increasingly inexpensive genotyping and sequencing, the field will require not only accrual of large numbers of drug-exposed patients with well-curated drug response phenotypes but also better methods to study subjects with unusual drug responses and to study drug responses (and other phenotypes) in patients with unusual genotypes. EHRs represent an opportunity not only for implementing pharmacogenetic information but also to accumulate the large cohorts that are required to identify these drug response outliers.

    As discussed above, genetics will become an increasing part of the drug development process. One particularly appealing approach is to exploit the recently developed technology of the phenome-wide association study. A GWAS asks the question “With what genetic locus is a particular phenotype associated?”, the question in phenome-wide association study is “With what human phenotypes is a particular genetic variant (or set of variants) associated?” Phenome-wide association study is enabled by the development of methods to curate phenotypes in large EHRs in which genotyping data are also available.138 Phenome-wide association study has been used to define pleiotropic genetic effects, replicate and extend genetic association study results, define disease subtypes, and may also help repurpose medications.114,139

    Genomic Medicine

    Genomic medicine is the concept that an individual’s genomic information can be used to individualize care. Although the plummeting costs of genotyping and sequencing make this an appealing prospect in principle, the challenges in implementing this vision are considerable and include engaging multiple partners (including payers); improving information technology infrastructure; a need for educational initiatives; a need for improved data accumulation, storage, and analysis approaches; and a need for medical and economic outcomes. Pharmacogenetics has traditionally been viewed as an example of low hanging fruit in genomic medicine and yet, as this review makes clear, the field continues to evolve, this prospect is only now beginning to be approached, and the challenges are considerable. Opportunities raised by new technologies, inexpensive genotyping and sequencing, and aggregation of large numbers of subjects with genetic information and well-characterized drug response phenotypes should enable further discovery and allow the field to have a real impact on drug development, drug utilization, and individualized therapy.

    Nonstandard Abbreviations and Acronyms

    CYP

    cytochrome P450 drug oxidizing enzyme

    EHR

    electronic health record

    GRS

    genetic risk score

    GWAS

    genome-wide association study

    HLA

    human leukocyte antigen

    IGNITE

    implementing genomics in practice

    RCT

    randomized clinical trial

    SAE

    serious adverse event

    SJS/TEN

    Stevens–Johnson syndrome/toxic epidermal necrolysis

    SNP

    single nucleotide polymorphism

    TPMT

    thiopurine methyltransferase

    Footnotes

    Correspondence to Dan M. Roden, MD, Vanderbilt University Medical Center, 2215B Garland Ave, Room 1285B, Nashville, TN 37232. E-mail

    References

    • 1. Mozaffarian D, Benjamin EJ, Go AS, et al.. Heart disease and stroke statistics-2016 update: a report from the american heart association.Circulation. 2016; 133:e38–e360.LinkGoogle Scholar
    • 2. Pirmohamed M. Personalized pharmacogenomics: predicting efficacy and adverse drug reactions.Annu Rev Genomics Hum Genet. 2014; 15:349–370. doi: 10.1146/annurev-genom-090413-025419.CrossrefMedlineGoogle Scholar
    • 3. Lazarou J, Pomeranz BH, Corey PN. Incidence of adverse drug reactions in hospitalized patients: a meta-analysis of prospective studies.JAMA. 1998; 279:1200–1205.CrossrefMedlineGoogle Scholar
    • 4. Landrigan CP, Parry GJ, Bones CB, Hackbarth AD, Goldmann DA, Sharek PJ. Temporal trends in rates of patient harm resulting from medical care.N Engl J Med. 2010; 363:2124–2134. doi: 10.1056/NEJMsa1004404.CrossrefMedlineGoogle Scholar
    • 5. Pirmohamed M, James S, Meakin S, Green C, Scott AK, Walley TJ, Farrar K, Park BK, Breckenridge AM. Adverse drug reactions as cause of admission to hospital: prospective analysis of 18 820 patients.BMJ. 2004; 329:15–19. doi: 10.1136/bmj.329.7456.15.CrossrefMedlineGoogle Scholar
    • 6. Roden DM, Johnson JA, Kimmel SE, Krauss RM, Medina MW, Shuldiner A, Wilke RA. Cardiovascular pharmacogenomics.Circ Res. 2011; 109:807–820. doi: 10.1161/CIRCRESAHA.110.230995.LinkGoogle Scholar
    • 7. Zaiou M, El Amri H. Cardiovascular pharmacogenetics: a promise for genomically-guided therapy and personalized medicine.Clin Genet. 2017; 91:355–370. doi: 10.1111/cge.12881.CrossrefMedlineGoogle Scholar
    • 8. Caudle KE, Dunnenberger HM, Freimuth RR, Peterson JF, Burlison JD, Whirl-Carrillo M, Scott SA, Rehm HL, Williams MS, Klein TE, Relling MV, Hoffman JM. Standardizing terms for clinical pharmacogenetic test results: consensus terms from the Clinical Pharmacogenetics Implementation Consortium (CPIC).Genet Med. 2017; 19:215–223. doi: 10.1038/gim.2016.87.CrossrefMedlineGoogle Scholar
    • 9. Kalman LV, Agúndez J, Appell ML, et al.. Pharmacogenetic allele nomenclature: International workgroup recommendations for test result reporting.Clin Pharmacol Ther. 2016; 99:172–185. doi: 10.1002/cpt.280.CrossrefMedlineGoogle Scholar
    • 10. Van Driest SL, Wells QS, Stallings S, et al.. Association of arrhythmia-related genetic variants with phenotypes documented in electronic medical records.JAMA. 2016; 315:47–57. doi: 10.1001/jama.2015.17701.CrossrefMedlineGoogle Scholar
    • 11. Berg JS, Amendola LM, Eng C, et al.; Members of the CSER Actionability and Return of Results Working Group. Processes and preliminary outputs for identification of actionable genes as incidental findings in genomic sequence data in the Clinical Sequencing Exploratory Research Consortium.Genet Med. 2013; 15:860–867. doi: 10.1038/gim.2013.133.CrossrefMedlineGoogle Scholar
    • 12. Britt BA, Locher WG, Kalow W. Hereditary aspects of malignant hyperthermia.Can Anaesth Soc J. 1969; 16:89–98.CrossrefMedlineGoogle Scholar
    • 13. Kalow W. Familial incidence of low pseudocholinesterase level.Lancet. 1956; 268:576–577CrossrefGoogle Scholar
    • 14. O’reilly RA, Aggeler PM, Hoag MS, Leong LS, Kropatkin ML. Hereditary transmission of exceptional resistance to coumarin anticoagulant drugs. The first reported kindred.N Engl J Med. 1964; 271:809–815. doi: 10.1056/NEJM196410152711602.CrossrefMedlineGoogle Scholar
    • 15. Gaxiola B, Friedl W, Propping P. Epinephrine-induced platelet aggregation. A twin study.Clin Genet. 1984; 26:543–548.CrossrefMedlineGoogle Scholar
    • 16. Swan GE, Benowitz NL, Jacob P, Lessov CN, Tyndale RF, Wilhelmsen K, Krasnow RE, McElroy MR, Moore SE, Wambach M. Pharmacogenetics of nicotine metabolism in twins: methods and procedures.Twin Res. 2004; 7:435–448. doi: 10.1375/1369052042335269.CrossrefMedlineGoogle Scholar
    • 17. Birkenfeld AL, Jordan J, Hofmann U, Busjahn A, Franke G, Krüger N, Igel S, Mürdter T, Drescher S, Shi S, Engeli S, Schwab M, Eichelbaum M, Luft FC, Fromm MF. Genetic influences on the pharmacokinetics of orally and intravenously administered digoxin as exhibited by monozygotic twins.Clin Pharmacol Ther. 2009; 86:605–608. doi: 10.1038/clpt.2009.170.CrossrefMedlineGoogle Scholar
    • 18. Arnett DK, Davis BR, Ford CE, Boerwinkle E, Leiendecker-Foster C, Miller MB, Black H, Eckfeldt JH. Pharmacogenetic association of the angiotensin-converting enzyme insertion/deletion polymorphism on blood pressure and cardiovascular risk in relation to antihypertensive treatment: the Genetics of Hypertension-Associated Treatment (GenHAT) study.Circulation. 2005; 111:3374–3383. doi: 10.1161/CIRCULATIONAHA.104.504639.LinkGoogle Scholar
    • 19. Ioannidis JP, Trikalinos TA, Khoury MJ. Implications of small effect sizes of individual genetic variants on the design and interpretation of genetic association studies of complex diseases.Am J Epidemiol. 2006; 164:609–614. doi: 10.1093/aje/kwj259.CrossrefMedlineGoogle Scholar
    • 20. Hulot JS, Bura A, Villard E, Azizi M, Remones V, Goyenvalle C, Aiach M, Lechat P, Gaussem P. Cytochrome P450 2C19 loss-of-function polymorphism is a major determinant of clopidogrel responsiveness in healthy subjects.Blood. 2006; 108:2244–2247. doi: 10.1182/blood-2006-04-013052.CrossrefMedlineGoogle Scholar
    • 21. Ciszkowski C, Madadi P, Phillips MS, Lauwers AE, Koren G. Codeine, ultrarapid-metabolism genotype, and postoperative death.N Engl J Med. 2009; 361:827–828. doi: 10.1056/NEJMc0904266.CrossrefMedlineGoogle Scholar
    • 22. Sibbing D, Koch W, Gebhard D, Schuster T, Braun S, Stegherr J, Morath T, Schömig A, von Beckerath N, Kastrati A. Cytochrome 2C19*17 allelic variant, platelet aggregation, bleeding events, and stent thrombosis in clopidogrel-treated patients with coronary stent placement.Circulation. 2010; 121:512–518. doi: 10.1161/CIRCULATIONAHA.109.885194.LinkGoogle Scholar
    • 23. Roden DM, Stein CM. Clopidogrel and the concept of high-risk pharmacokinetics.Circulation. 2009; 119:2127–2130. doi: 10.1161/CIRCULATIONAHA.109.865907.LinkGoogle Scholar
    • 24. Evers J, Eichelbaum M, Kroemer HK. Unpredictability of flecainide plasma concentrations in patients with renal failure: relationship to side effects and sudden death?Ther Drug Monit. 1994; 16:349–351.CrossrefMedlineGoogle Scholar
    • 25. Motsinger-Reif AA, Jorgenson E, Relling MV, Kroetz DL, Weinshilboum R, Cox NJ, Roden DM. Genome-wide association studies in pharmacogenomics: successes and lessons.Pharmacogenet Genomics. 2013; 23:383–394. doi: 10.1097/FPC.0b013e32833d7b45.CrossrefMedlineGoogle Scholar
    • 26. Maranville JC, Cox NJ. Pharmacogenomic variants have larger effect sizes than genetic variants associated with other dichotomous complex traits.Pharmacogenomics J. 2016; 16:388–392. doi: 10.1038/tpj.2015.47.CrossrefMedlineGoogle Scholar
    • 27. Daly AK, Donaldson PT, Bhatnagar P, et al.; DILIGEN Study; International SAE Consortium. HLA-B*5701 genotype is a major determinant of drug-induced liver injury due to flucloxacillin.Nat Genet. 2009; 41:816–819. doi: 10.1038/ng.379.CrossrefMedlineGoogle Scholar
    • 28. McCormack M, Alfirevic A, Bourgeois S, et al.. HLA-A*3101 and carbamazepine-induced hypersensitivity reactions in Europeans.N Engl J Med. 2011; 364:1134–1143. doi: 10.1056/NEJMoa1013297.CrossrefMedlineGoogle Scholar
    • 29. Pasanen MK, Neuvonen M, Neuvonen PJ, Niemi M. SLCO1B1 polymorphism markedly affects the pharmacokinetics of simvastatin acid.Pharmacogenet Genomics. 2006; 16:873–879. doi: 10.1097/01.fpc.0000230416.82349.90.CrossrefMedlineGoogle Scholar
    • 30. Link E, Parish S, Armitage J, Bowman L, Heath S, Matsuda F, Gut I, Lathrop M, Collins R. Slco1b1 variants and statin-induced myopathy–a genomewide study.N Engl J Med. 2008; 359:789–799.CrossrefMedlineGoogle Scholar
    • 31. Brunham LR, Lansberg PJ, Zhang L, Miao F, Carter C, Hovingh GK, Visscher H, Jukema JW, Stalenhoef AF, Ross CJ, Carleton BC, Kastelein JJ, Hayden MR. Differential effect of the rs4149056 variant in SLCO1B1 on myopathy associated with simvastatin and atorvastatin.Pharmacogenomics J. 2012; 12:233–237. doi: 10.1038/tpj.2010.92.CrossrefMedlineGoogle Scholar
    • 32. Voora D, Shah SH, Spasojevic I, Ali S, Reed CR, Salisbury BA, Ginsburg GS. The SLCO1B1*5 genetic variant is associated with statin-induced side effects.J Am Coll Cardiol. 2009; 54:1609–1616. doi: 10.1016/j.jacc.2009.04.053.CrossrefMedlineGoogle Scholar
    • 33. Shuldiner AR, O’Connell JR, Bliden KP, et al.. Association of cytochrome P450 2C19 genotype with the antiplatelet effect and clinical efficacy of clopidogrel therapy.JAMA. 2009; 302:849–857. doi: 10.1001/jama.2009.1232.CrossrefMedlineGoogle Scholar
    • 34. Cooper GM, Johnson JA, Langaee TY, Feng H, Stanaway IB, Schwarz UI, Ritchie MD, Stein CM, Roden DM, Smith JD, Veenstra DL, Rettie AE, Rieder MJ. A genome-wide scan for common genetic variants with a large influence on warfarin maintenance dose.Blood. 2008; 112:1022–1027. doi: 10.1182/blood-2008-01-134247.CrossrefMedlineGoogle Scholar
    • 35. Takeuchi F, McGinnis R, Bourgeois S, Barnes C, Eriksson N, Soranzo N, Whittaker P, Ranganath V, Kumanduri V, McLaren W, Holm L, Lindh J, Rane A, Wadelius M, Deloukas P. A genome-wide association study confirms VKORC1, CYP2C9, and CYP4F2 as principal genetic determinants of warfarin dose.PLoS Genet. 2009; 5:e1000433. doi: 10.1371/journal.pgen.1000433.CrossrefMedlineGoogle Scholar
    • 36. Postmus I, Trompet S, Deshmukh HA, et al.; Welcome Trust Case Control Consortium. Pharmacogenetic meta-analysis of genome-wide association studies of LDL cholesterol response to statins.Nat Commun. 2014; 5:5068. doi: 10.1038/ncomms6068.CrossrefMedlineGoogle Scholar
    • 37. Wei W-Q, Li X, Feng Q, et al.. Abstract 15433: Lpa variants are associated with residual cardiovascular risk in patients receiving statins.Circulation. 2017; 136:A15433.Google Scholar
    • 38. Christophersen IE, Rienstra M, Roselli C, et al.; METASTROKE Consortium of the ISGC; Neurology Working Group of the CHARGE Consortium; AFGen Consortium. Large-scale analyses of common and rare variants identify 12 new loci associated with atrial fibrillation.Nat Genet. 2017; 49:946–952. doi: 10.1038/ng.3843.CrossrefMedlineGoogle Scholar
    • 39. Parvez B, Vaglio J, Rowan S, Muhammad R, Kucera G, Stubblefield T, Carter S, Roden D, Darbar D. Symptomatic response to antiarrhythmic drug therapy is modulated by a common single nucleotide polymorphism in atrial fibrillation.J Am Coll Cardiol. 2012; 60:539–545. doi: 10.1016/j.jacc.2012.01.070.CrossrefMedlineGoogle Scholar
    • 40. Parvez B, Shoemaker MB, Muhammad R, Richardson R, Jiang L, Blair MA, Roden DM, Darbar D. Common genetic polymorphism at 4q25 locus predicts atrial fibrillation recurrence after successful cardioversion.Heart Rhythm. 2013; 10:849–855. doi: 10.1016/j.hrthm.2013.02.018.CrossrefMedlineGoogle Scholar
    • 41. Shoemaker MB, Bollmann A, Lubitz SA, et al.. Common genetic variants and response to atrial fibrillation ablation.Circ Arrhythm Electrophysiol. 2015; 8:296–302. doi: 10.1161/CIRCEP.114.001909.LinkGoogle Scholar
    • 42. Rettie AE, Wienkers LC, Gonzalez FJ, Trager WF, Korzekwa KR. Impaired (S)-warfarin metabolism catalysed by the R144C allelic variant of CYP2C9.Pharmacogenetics. 1994; 4:39–42.CrossrefMedlineGoogle Scholar
    • 43. Aithal GP, Day CP, Kesteven PJ, Daly AK. Association of polymorphisms in the cytochrome P450 CYP2C9 with warfarin dose requirement and risk of bleeding complications.Lancet. 1999; 353:717–719. doi: 10.1016/S0140-6736(98)04474-2.CrossrefMedlineGoogle Scholar
    • 44. Rost S, Fregin A, Ivaskevicius V, Conzelmann E, Hörtnagel K, Pelz HJ, Lappegard K, Seifried E, Scharrer I, Tuddenham EG, Müller CR, Strom TM, Oldenburg J. Mutations in VKORC1 cause warfarin resistance and multiple coagulation factor deficiency type 2.Nature. 2004; 427:537–541. doi: 10.1038/nature02214.CrossrefMedlineGoogle Scholar
    • 45. Rieder MJ, Reiner AP, Gage BF, Nickerson DA, Eby CS, McLeod HL, Blough DK, Thummel KE, Veenstra DL, Rettie AE. Effect of VKORC1 haplotypes on transcriptional regulation and warfarin dose.N Engl J Med. 2005; 352:2285–2293. doi: 10.1056/NEJMoa044503.CrossrefMedlineGoogle Scholar
    • 46. Yang Y, Peter I, Scott SA. Pharmacogenetics in Jewish populations.Drug Metabol Drug Interact. 2014; 29:221–233. doi: 10.1515/dmdi-2013-0069.CrossrefMedlineGoogle Scholar
    • 47. Kawai VK, Cunningham A, Vear SI, Van Driest SL, Oginni A, Xu H, Jiang M, Li C, Denny JC, Shaffer C, Bowton E, Gage BF, Ray WA, Roden DM, Stein CM. Genotype and risk of major bleeding during warfarin treatment.Pharmacogenomics. 2014; 15:1973–1983. doi: 10.2217/pgs.14.153.CrossrefMedlineGoogle Scholar
    • 48. Roth JA, Boudreau D, Fujii MM, Farin FM, Rettie AE, Thummel KE, Veenstra DL. Genetic risk factors for major bleeding in patients treated with warfarin in a community setting.Clin Pharmacol Ther. 2014; 95:636–643. doi: 10.1038/clpt.2014.26.CrossrefMedlineGoogle Scholar
    • 49. Collet JP, Hulot JS, Pena A, Villard E, Esteve JB, Silvain J, Payot L, Brugier D, Cayla G, Beygui F, Bensimon G, Funck-Brentano C, Montalescot G. Cytochrome P450 2C19 polymorphism in young patients treated with clopidogrel after myocardial infarction: a cohort study.Lancet. 2009; 373:309–317. doi: 10.1016/S0140-6736(08)61845-0.CrossrefMedlineGoogle Scholar
    • 50. Simon T, Verstuyft C, Mary-Krause M, Quteineh L, Drouet E, Méneveau N, Steg PG, Ferrières J, Danchin N, Becquemont L; French Registry of Acute ST-Elevation and Non-ST-Elevation Myocardial Infarction (FAST-MI) Investigators. Genetic determinants of response to clopidogrel and cardiovascular events.N Engl J Med. 2009; 360:363–375. doi: 10.1056/NEJMoa0808227.CrossrefMedlineGoogle Scholar
    • 51. Mega JL, Close SL, Wiviott SD, Shen L, Hockett RD, Brandt JT, Walker JR, Antman EM, Macias W, Braunwald E, Sabatine MS. Cytochrome p-450 polymorphisms and response to clopidogrel.N Engl J Med. 2009; 360:354–362. doi: 10.1056/NEJMoa0809171.CrossrefMedlineGoogle Scholar
    • 52. Holmes MV, Perel P, Shah T, Hingorani AD, Casas JP. CYP2C19 genotype, clopidogrel metabolism, platelet function, and cardiovascular events: a systematic review and meta-analysis.JAMA. 2011; 306:2704–2714. doi: 10.1001/jama.2011.1880.CrossrefMedlineGoogle Scholar
    • 53. Johnson JA, Roden DM, Lesko LJ, Ashley E, Klein TE, Shuldiner AR. Clopidogrel: a case for indication-specific pharmacogenetics.Clin Pharmacol Ther. 2012; 91:774–776. doi: 10.1038/clpt.2012.21.CrossrefMedlineGoogle Scholar
    • 54. Wang Y, Zhao X, Lin J, Li H, Johnston SC, Lin Y, Pan Y, Liu L, Wang D, Wang C, Meng X, Xu J, Wang Y; CHANCE Investigators. Association between CYP2C19 loss-of-function allele status and efficacy of clopidogrel for risk reduction among patients with minor stroke or transient ischemic attack.JAMA. 2016; 316:70–78. doi: 10.1001/jama.2016.8662.CrossrefMedlineGoogle Scholar
    • 55. Bouman HJ, Schömig E, van Werkum JW, Velder J, Hackeng CM, Hirschhäuser C, Waldmann C, Schmalz HG, ten Berg JM, Taubert D. Paraoxonase-1 is a major determinant of clopidogrel efficacy.Nat Med. 2011; 17:110–116. doi: 10.1038/nm.2281.CrossrefMedlineGoogle Scholar
    • 56. Paré G, Ross S, Mehta SR, Yusuf S, Anand SS, Connolly SJ, Fox KA, Eikelboom JW. Effect of PON1 Q192R genetic polymorphism on clopidogrel efficacy and cardiovascular events in the clopidogrel in the unstable angina to prevent recurrent events trial and the atrial fibrillation clopidogrel trial with irbesartan for prevention of vascular events.Circ Cardiovasc Genet. 2012; 5:250–256. doi: 10.1161/CIRCGENETICS.111.961417.LinkGoogle Scholar
    • 57. Perera MA, Cavallari LH, Limdi NA, et al.. Genetic variants associated with warfarin dose in African-American individuals: a genome-wide association study.Lancet. 2013; 382:790–796. doi: 10.1016/S0140-6736(13)60681-9.CrossrefMedlineGoogle Scholar
    • 58. Ramos E, Doumatey A, Elkahloun AG, Shriner D, Huang H, Chen G, Zhou J, McLeod H, Adeyemo A, Rotimi CN. Pharmacogenomics, ancestry and clinical decision making for global populations.Pharmacogenomics J. 2014; 14:217–222. doi: 10.1038/tpj.2013.24.CrossrefMedlineGoogle Scholar
    • 59. Limdi NA, Brown TM, Yan Q, Thigpen JL, Shendre A, Liu N, Hill CE, Arnett DK, Beasley TM. Race influences warfarin dose changes associated with genetic factors.Blood. 2015; 126:539–545. doi: 10.1182/blood-2015-02-627042.CrossrefMedlineGoogle Scholar
    • 60. Arwood MJ, Deng J, Drozda K, Pugach O, Nutescu EA, Schmidt S, Duarte JD, Cavallari LH. Anticoagulation endpoints with clinical implementation of warfarin pharmacogenetic dosing in a real-world setting: a proposal for a new pharmacogenetic dosing approach.Clin Pharmacol Ther. 2017; 101:675–683. doi: 10.1002/cpt.558.CrossrefMedlineGoogle Scholar
    • 61. Kimmel SE, French B, Kasner SE, et al.; COAG Investigators. A pharmacogenetic versus a clinical algorithm for warfarin dosing.N Engl J Med. 2013; 369:2283–2293. doi: 10.1056/NEJMoa1310669.CrossrefMedlineGoogle Scholar
    • 62. Pirmohamed M, Burnside G, Eriksson N, et al.; EU-PACT Group. A randomized trial of genotype-guided dosing of warfarin.N Engl J Med. 2013; 369:2294–2303. doi: 10.1056/NEJMoa1311386.CrossrefMedlineGoogle Scholar
    • 63. Verhoef TI, Ragia G, de Boer A, et al.; EU-PACT Group. A randomized trial of genotype-guided dosing of acenocoumarol and phenprocoumon.N Engl J Med. 2013; 369:2304–2312. doi: 10.1056/NEJMoa1311388.CrossrefMedlineGoogle Scholar
    • 64. Gage BF, Bass AR, Lin H, et al.. Effect of genotype-guided warfarin dosing on clinical events and anticoagulation control among patients undergoing hip or knee arthroplasty: the GIFT randomized clinical trial.JAMA. 2017; 318:1115–1124. doi: 10.1001/jama.2017.11469.CrossrefMedlineGoogle Scholar
    • 65. Mega JL, Hochholzer W, Frelinger AL, Kluk MJ, Angiolillo DJ, Kereiakes DJ, Isserman S, Rogers WJ, Ruff CT, Contant C, Pencina MJ, Scirica BM, Longtine JA, Michelson AD, Sabatine MS. Dosing clopidogrel based on CYP2C19 genotype and the effect on platelet reactivity in patients with stable cardiovascular disease.JAMA. 2011; 306:2221–2228. doi: 10.1001/jama.2011.1703.CrossrefMedlineGoogle Scholar
    • 66. ClinicalTrials.gov. https://www.clinicaltrials.gov/ct2/show/NCT01742117. 2015.Google Scholar
    • 67. Coenen MJ, de Jong DJ, van Marrewijk CJ, Derijks LJ, Vermeulen SH, Wong DR, Klungel OH, Verbeek AL, Hooymans PM, Peters WH, te Morsche RH, Newman WG, Scheffer H, Guchelaar HJ, Franke B; TOPIC Recruitment Team. Identification of patients with variants in TPMT and dose reduction reduces hematologic events during thiopurine treatment of inflammatory bowel disease.Gastroenterology. 2015; 149:907.e7–917.e7. doi: 10.1053/j.gastro.2015.06.002.CrossrefGoogle Scholar
    • 68. ClinicalTrials.gov. https://clinicaltrials.gov/ct2/show/NCT03093818. 2017.Google Scholar
    • 69. Holmes DR, Dehmer GJ, Kaul S, Leifer D, O’Gara PT, Stein CM. ACCF/AHA clopidogrel clinical alert: approaches to the FDA “boxed warning”. A report of the American College of Cardiology Foundation Task Force on Clinical Expert Consensus Documents and the American Heart Association.Circulation. 2010; 122:537–557.Google Scholar
    • 70. Luzum JA, Pakyz RE, Elsey AR, et al.; Pharmacogenomics Research Network Translational Pharmacogenetics Program. The pharmacogenomics research network translational pharmacogenetics program: outcomes and metrics of pharmacogenetic implementations across diverse healthcare systems.Clin Pharmacol Ther. 2017; 102:502–510. doi: 10.1002/cpt.630.CrossrefMedlineGoogle Scholar
    • 71. Manolio TA, Abramowicz M, Al-Mulla F, et al.. Global implementation of genomic medicine: we are not alone.Sci Transl Med. 2015; 7:290ps13. doi: 10.1126/scitranslmed.aab0194.CrossrefMedlineGoogle Scholar
    • 72. Cavallari LH, Weitzel KW, Elsey AR, et al.. Institutional profile: University of Florida Health Personalized Medicine Program.Pharmacogenomics. 2017; 18:421–426. doi: 10.2217/pgs-2017-0028.CrossrefMedlineGoogle Scholar
    • 73. Pulley JM, Denny JC, Peterson JF, et al.. Operational implementation of prospective genotyping for personalized medicine: the design of the Vanderbilt PREDICT project.Clin Pharmacol Ther. 2012; 92:87–95. doi: 10.1038/clpt.2011.371.CrossrefMedlineGoogle Scholar
    • 74. Bell GC, Crews KR, Wilkinson MR, Haidar CE, Hicks JK, Baker DK, Kornegay NM, Yang W, Cross SJ, Howard SC, Freimuth RR, Evans WE, Broeckel U, Relling MV, Hoffman JM. Development and use of active clinical decision support for preemptive pharmacogenomics.J Am Med Inform Assoc. 2014; 21:e93–e99. doi: 10.1136/amiajnl-2013-001993.CrossrefMedlineGoogle Scholar
    • 75. Cavallari LH, Beitelshees AL, Blake KV, et al.. The IGNITE pharmacogenetics working group: an opportunity for building evidence with pharmacogenetic implementation in a real-world setting.Clin Transl Sci. 2017; 10:143–146. doi: 10.1111/cts.12456.CrossrefMedlineGoogle Scholar
    • 76. Cavallari LH, Lee CR, Beitelshees AL, et al.; IGNITE Network. Multisite investigation of outcomes with implementation of CYP2C19 genotype-guided antiplatelet therapy after percutaneous coronary intervention.JACC Cardiovasc Interv. 2018; 11:181–191. doi: 10.1016/j.jcin.2017.07.022.CrossrefMedlineGoogle Scholar
    • 77. Peterson JF, Field JR, Unertl KM, Schildcrout JS, Johnson DC, Shi Y, Danciu I, Cleator JH, Pulley JM, McPherson JA, Denny JC, Laposata M, Roden DM, Johnson KB. Physician response to implementation of genotype-tailored antiplatelet therapy.Clin Pharmacol Ther. 2016; 100:67–74. doi: 10.1002/cpt.331.CrossrefMedlineGoogle Scholar
    • 78. Relling MV, Klein TE. CPIC: clinical pharmacogenetics implementation consortium of the pharmacogenomics research network.Clin Pharmacol Ther. 2011; 89:464–467. doi: 10.1038/clpt.2010.279.CrossrefMedlineGoogle Scholar
    • 79. Bank P, Caudle KE, Swen JJ, Gammal RS, Whirl-Carrillo M, Klein TE, Relling MV, Guchelaar HJ. Comparison of the guidelines of the clinical pharmacogenetics implementation consortium and the Dutch pharmacogenetics working group.Clin Pharmacol Ther. 2018; 103:599–618. doi: 10.1002/cpt.762.CrossrefMedlineGoogle Scholar
    • 80. Mallal S, Nolan D, Witt C, Masel G, Martin AM, Moore C, Sayer D, Castley A, Mamotte C, Maxwell D, James I, Christiansen FT. Association between presence of HLA-B*5701, HLA-DR7, and HLA-DQ3 and hypersensitivity to HIV-1 reverse-transcriptase inhibitor abacavir.Lancet. 2002; 359:727–732.CrossrefMedlineGoogle Scholar
    • 81. Mallal S, Phillips E, Carosi G, et al.; PREDICT-1 Study Team. HLA-B*5701 screening for hypersensitivity to abacavir.N Engl J Med. 2008; 358:568–579. doi: 10.1056/NEJMoa0706135.CrossrefMedlineGoogle Scholar
    • 82. Chung WH, Hung SI, Hong HS, Hsih MS, Yang LC, Ho HC, Wu JY, Chen YT. Medical genetics: a marker for Stevens-Johnson syndrome.Nature. 2004; 428:486. doi: 10.1038/428486a.CrossrefMedlineGoogle Scholar
    • 83. Hung SI, Chung WH, Jee SH, et al.. Genetic susceptibility to carbamazepine-induced cutaneous adverse drug reactions.Pharmacogenet Genomics. 2006; 16:297–306. doi: 10.1097/01.fpc.0000199500.46842.4a.CrossrefMedlineGoogle Scholar
    • 84. Chen Z, Liew D, Kwan P. Effects of a HLA-B*15:02 screening policy on antiepileptic drug use and severe skin reactions.Neurology. 2014; 83:2077–2084. doi: 10.1212/WNL.0000000000001034.CrossrefMedlineGoogle Scholar
    • 85. Sanseau P, Agarwal P, Barnes MR, Pastinen T, Richards JB, Cardon LR, Mooser V. Use of genome-wide association studies for drug repositioning.Nat Biotechnol. 2012; 30:317–320. doi: 10.1038/nbt.2151.CrossrefMedlineGoogle Scholar
    • 86. Nelson MR, Tipney H, Painter JL, Shen J, Nicoletti P, Shen Y, Floratos A, Sham PC, Li MJ, Wang J, Cardon LR, Whittaker JC, Sanseau P. The support of human genetic evidence for approved drug indications.Nat Genet. 2015; 47:856–860. doi: 10.1038/ng.3314.CrossrefMedlineGoogle Scholar
    • 87. Okada Y, Wu D, Trynka G, et al.; RACI consortium; GARNET consortium. Genetics of rheumatoid arthritis contributes to biology and drug discovery.Nature. 2014; 506:376–381. doi: 10.1038/nature12873.CrossrefMedlineGoogle Scholar
    • 88. Tardif JC, Rhéaume E, Lemieux Perreault LP, et al.. Pharmacogenomic determinants of the cardiovascular effects of dalcetrapib.Circ Cardiovasc Genet. 2015; 8:372–382. doi: 10.1161/CIRCGENETICS.114.000663.LinkGoogle Scholar
    • 89. Dubé MP, de Denus S, Tardif JC. Pharmacogenomics to revive drug development in cardiovascular disease.Cardiovasc Drugs Ther. 2016; 30:59–64. doi: 10.1007/s10557-015-6637-y.CrossrefMedlineGoogle Scholar
    • 90. Abifadel M, Varret M, Rabès JP, et al.. Mutations in PCSK9 cause autosomal dominant hypercholesterolemia.Nat Genet. 2003; 34:154–156. doi: 10.1038/ng1161.CrossrefMedlineGoogle Scholar
    • 91. Maxwell KN, Breslow JL. Adenoviral-mediated expression of Pcsk9 in mice results in a low-density lipoprotein receptor knockout phenotype.Proc Natl Acad Sci USA. 2004; 101:7100–7105. doi: 10.1073/pnas.0402133101.CrossrefMedlineGoogle Scholar
    • 92. Cohen J, Pertsemlidis A, Kotowski IK, Graham R, Garcia CK, Hobbs HH. Low LDL cholesterol in individuals of African descent resulting from frequent nonsense mutations in PCSK9.Nat Genet. 2005; 37:161–165. doi: 10.1038/ng1509.CrossrefMedlineGoogle Scholar
    • 93. Cohen JC, Boerwinkle E, Mosley TH, Hobbs HH. Sequence variations in PCSK9, low LDL, and protection against coronary heart disease.N Engl J Med. 2006; 354:1264–1272. doi: 10.1056/NEJMoa054013.CrossrefMedlineGoogle Scholar
    • 94. Zhao Z, Tuakli-Wosornu Y, Lagace TA, Kinch L, Grishin NV, Horton JD, Cohen JC, Hobbs HH. Molecular characterization of loss-of-function mutations in PCSK9 and identification of a compound heterozygote.Am J Hum Genet. 2006; 79:514–523. doi: 10.1086/507488.CrossrefMedlineGoogle Scholar
    • 95. Stitziel NO, Won HH, Morrison AC, et al.. Inactivating mutations in NPC1L1 and protection from coronary heart disease.N Engl J Med. 2014; 371:2072–2082.CrossrefMedlineGoogle Scholar
    • 96. Dewey FE, Gusarova V, Dunbar RL, et al.. Genetic and pharmacologic inactivation of ANGPTL3 and cardiovascular disease.N Engl J Med. 2017; 377:211–221. doi: 10.1056/NEJMoa1612790.CrossrefMedlineGoogle Scholar
    • 97. Crosby J, Peloso GM, Auer PL, et al.. Loss-of-function mutations in apoc3, triglycerides, and coronary disease.N Engl J Med. 2014; 371:22–31CrossrefMedlineGoogle Scholar
    • 98. Flannick J, Thorleifsson G, Beer NL, et al.; Go-T2D Consortium; T2D-GENES Consortium. Loss-of-function mutations in SLC30A8 protect against type 2 diabetes.Nat Genet. 2014; 46:357–363. doi: 10.1038/ng.2915.CrossrefMedlineGoogle Scholar
    • 99. Benhorin J, Taub R, Goldmit M, Kerem B, Kass RS, Windman I, Medina A. Effects of flecainide in patients with new SCN5A mutation: mutation-specific therapy for long-QT syndrome?Circulation. 2000; 101:1698–1706.LinkGoogle Scholar
    • 100. Chorin E, Hu D, Antzelevitch C, Hochstadt A, Belardinelli L, Zeltser D, Barajas-Martinez H, Rozovski U, Rosso R, Adler A, Benhorin J, Viskin S. Ranolazine for congenital long-QT syndrome type III: Experimental and long-term clinical data.Circ Arrhythm Electrophysiol. 2016; 9:e004370.LinkGoogle Scholar
    • 101. Ramsey BW, Davies J, McElvaney NG, et al.; VX08-770-102 Study Group. A CFTR potentiator in patients with cystic fibrosis and the G551D mutation.N Engl J Med. 2011; 365:1663–1672. doi: 10.1056/NEJMoa1105185.CrossrefMedlineGoogle Scholar
    • 102. Whiting P, Al M, Burgers L, Westwood M, Ryder S, Hoogendoorn M, Armstrong N, Allen A, Severens H, Kleijnen J. Ivacaftor for the treatment of patients with cystic fibrosis and the G551D mutation: a systematic review and cost-effectiveness analysis.Health Technol Assess. 2014; 18:1–106. doi: 10.3310/hta18180.CrossrefGoogle Scholar
    • 103. Mendell JR, Rodino-Klapac LR, Sahenk Z, et al.; Eteplirsen Study Group. Eteplirsen for the treatment of Duchenne muscular dystrophy.Ann Neurol. 2013; 74:637–647. doi: 10.1002/ana.23982.CrossrefMedlineGoogle Scholar
    • 104. Kesselheim AS, Avorn J. Approving a problematic muscular dystrophy drug: implications for FDA policy.JAMA. 2016; 316:2357–2358. doi: 10.1001/jama.2016.16437.CrossrefMedlineGoogle Scholar
    • 105. Habashi JP, Judge DP, Holm TM, et al.. Losartan, an AT1 antagonist, prevents aortic aneurysm in a mouse model of Marfan syndrome.Science. 2006; 312:117–121. doi: 10.1126/science.1124287.CrossrefMedlineGoogle Scholar
    • 106. Lacro RV, Dietz HC, Sleeper LA, et al.; Pediatric Heart Network Investigators. Atenolol versus losartan in children and young adults with Marfan’s syndrome.N Engl J Med. 2014; 371:2061–2071. doi: 10.1056/NEJMoa1404731.CrossrefMedlineGoogle Scholar
    • 107. Van Driest SL, Sleeper LA, Dietz H, et al.. Impact of variants in ADRB1 and CYP2C9 on atenolol and losartan response in Marfan syndrome.Paper presented at: Pediatric Academic Societies 2017 Annual Meeting. San Francisco, CA; May 6–9, 2017.Google Scholar
    • 108. Van Driest SL, Shi Y, Bowton EA, Schildcrout JS, Peterson JF, Pulley J, Denny JC, Roden DM. Clinically actionable genotypes among 10,000 patients with preemptive pharmacogenomic testing.Clin Pharmacol Ther. 2014; 95:423–431. doi: 10.1038/clpt.2013.229.CrossrefMedlineGoogle Scholar
    • 109. Nelson MR, Wegmann D, Ehm MG, et al.. An abundance of rare functional variants in 202 drug target genes sequenced in 14,002 people.Science. 2012; 337:100–104. doi: 10.1126/science.1217876.CrossrefMedlineGoogle Scholar
    • 110. Zhou Y, Ingelman-Sundberg M, Lauschke VM. Worldwide distribution of cytochrome p450 alleles: a meta-analysis of population-scale sequencing projects.Clin Pharmacol Ther. 2017; 102:688–700.CrossrefMedlineGoogle Scholar
    • 111. Lek M, Karczewski KJ, Minikel EV, et al.; Exome Aggregation Consortium. Analysis of protein-coding genetic variation in 60,706 humans.Nature. 2016; 536:285–291. doi: 10.1038/nature19057.CrossrefMedlineGoogle Scholar
    • 112. Gray VE, Hause RJ, Fowler DM. Analysis of large-scale mutagenesis data to assess the impact of single amino acid substitutions.Genetics. 2017; 207:53–61. doi: 10.1534/genetics.117.300064.CrossrefMedlineGoogle Scholar
    • 113. Fowler DM, Stephany JJ, Fields S. Measuring the activity of protein variants on a large scale using deep mutational scanning.Nat Protoc. 2014; 9:2267–2284. doi: 10.1038/nprot.2014.153.CrossrefMedlineGoogle Scholar
    • 114. Denny JC, Bastarache L, Roden DM. Phenome-wide association studies as a tool to advance precision medicine.Annu Rev Genomics Hum Genet. 2016; 17:353–373. doi: 10.1146/annurev-genom-090314-024956.CrossrefMedlineGoogle Scholar
    • 115. Burridge PW, Li YF, Matsa E, et al.. Human induced pluripotent stem cell-derived cardiomyocytes recapitulate the predilection of breast cancer patients to doxorubicin-induced cardiotoxicity.Nat Med. 2016; 22:547–556. doi: 10.1038/nm.4087.CrossrefMedlineGoogle Scholar
    • 116. Bellin M, Casini S, Davis RP, D’Aniello C, Haas J, Ward-van Oostwaard D, Tertoolen LG, Jung CB, Elliott DA, Welling A, Laugwitz KL, Moretti A, Mummery CL. Isogenic human pluripotent stem cell pairs reveal the role of a KCNH2 mutation in long-QT syndrome.EMBO J. 2013; 32:3161–3175. doi: 10.1038/emboj.2013.240.CrossrefMedlineGoogle Scholar
    • 117. Bersell KB, Yang T, Hong CC, Petropoulou E, Jamshidi Y, Wells QS, Kannankeril PJ, Roden DM. Genomic editing in iPSCs establishes that a rare tbx5 variant causes brugada syndrome.Paper presented at: Selected Student Platform Presentation, 2016 ASCI-AAP Annual Meeting. Chicago, IL; April 15–17, 2016.Google Scholar
    • 118. Pavlos R, Mallal S, Ostrov D, Buus S, Metushi I, Peters B, Phillips E. T cell-mediated hypersensitivity reactions to drugs.Annu Rev Med. 2015; 66:439–454. doi: 10.1146/annurev-med-050913-022745.CrossrefMedlineGoogle Scholar
    • 119. Ostrov DA, Grant BJ, Pompeu YA, et al.. Drug hypersensitivity caused by alteration of the MHC-presented self-peptide repertoire.Proc Natl Acad Sci USA. 2012; 109:9959–9964. doi: 10.1073/pnas.1207934109.CrossrefMedlineGoogle Scholar
    • 120. MacRae CA, Roden DM, Loscalzo J. The future of cardiovascular therapeutics.Circulation. 2016; 133:2610–2617. doi: 10.1161/CIRCULATIONAHA.116.023555.LinkGoogle Scholar
    • 121. Mangravite LM, Engelhardt BE, Medina MW, et al.. A statin-dependent QTL for GATM expression is associated with statin-induced myopathy.Nature. 2013; 502:377–380. doi: 10.1038/nature12508.CrossrefMedlineGoogle Scholar
    • 122. Gintant G, Fermini B, Stockbridge N, Strauss D. The evolving roles of human iPSC-derived cardiomyocytes in drug safety and discovery.Cell Stem Cell. 2017; 21:14–17. doi: 10.1016/j.stem.2017.06.005.CrossrefMedlineGoogle Scholar
    • 123. Stillitano F, Hansen J, Kong CW, et al.. Modeling susceptibility to drug-induced long QT with a panel of subject-specific induced pluripotent stem cells.eLife. 2017; 6:e19406.CrossrefMedlineGoogle Scholar
    • 124. DeGorter MK, Tirona RG, Schwarz UI, Choi YH, Dresser GK, Suskin N, Myers K, Zou G, Iwuchukwu O, Wei WQ, Wilke RA, Hegele RA, Kim RB. Clinical and pharmacogenetic predictors of circulating atorvastatin and rosuvastatin concentrations in routine clinical care.Circ Cardiovasc Genet. 2013; 6:400–408. doi: 10.1161/CIRCGENETICS.113.000099.LinkGoogle Scholar
    • 125. Van Driest SL, Marshall MD, Hachey B, Beck C, Crum K, Owen J, Smith AH, Kannankeril PJ, Woodworth A, Caprioli RM, Choi L. Pragmatic pharmacology: population pharmacokinetic analysis of fentanyl using remnant samples from children after cardiac surgery.Br J Clin Pharmacol. 2016; 81:1165–1174. doi: 10.1111/bcp.12903.CrossrefMedlineGoogle Scholar
    • 126. McNaughton CD, Brown NJ, Rothman RL, Liu D, Kabagambe EK, Levy PD, Self WH, Storrow AB, Collins SP, Roumie CL. Systolic blood pressure and biochemical assessment of adherence: a cross-sectional analysis in the emergency department.Hypertension. 2017; 70:307–314. doi: 10.1161/HYPERTENSIONAHA.117.09659.LinkGoogle Scholar
    • 127. Arking DE, Pulit SL, Crotti L, et al.; CARe Consortium; COGENT Consortium; DCCT/EDIC; eMERGE Consortium; HRGEN Consortium. Genetic association study of QT interval highlights role for calcium signaling pathways in myocardial repolarization.Nat Genet. 2014; 46:826–836. doi: 10.1038/ng.3014.CrossrefMedlineGoogle Scholar
    • 128. Arking DE, Pfeufer A, Post W, et al.. A common genetic variant in the NOS1 regulator NOS1AP modulates cardiac repolarization.Nat Genet. 2006; 38:644–651. doi: 10.1038/ng1790.CrossrefMedlineGoogle Scholar
    • 129. Chang KC, Barth AS, Sasano T, Kizana E, Kashiwakura Y, Zhang Y, Foster DB, Marbán E. CAPON modulates cardiac repolarization via neuronal nitric oxide synthase signaling in the heart.Proc Natl Acad Sci USA. 2008; 105:4477–4482. doi: 10.1073/pnas.0709118105.CrossrefMedlineGoogle Scholar
    • 130. Kapoor A, Sekar RB, Hansen NF, et al.; QT Interval-International GWAS Consortium. An enhancer polymorphism at the cardiomyocyte intercalated disc protein NOS1AP locus is a major regulator of the QT interval.Am J Hum Genet. 2014; 94:854–869. doi: 10.1016/j.ajhg.2014.05.001.CrossrefMedlineGoogle Scholar
    • 131. Kao WH, Arking DE, Post W, et al.. Genetic variations in nitric oxide synthase 1 adaptor protein are associated with sudden cardiac death in US white community-based populations.Circulation. 2009; 119:940–951. doi: 10.1161/CIRCULATIONAHA.108.791723.LinkGoogle Scholar
    • 132. Crotti L, Monti MC, Insolia R, Peljto A, Goosen A, Brink PA, Greenberg DA, Schwartz PJ, George AL. NOS1AP is a genetic modifier of the long-QT syndrome.Circulation. 2009; 120:1657–1663. doi: 10.1161/CIRCULATIONAHA.109.879643.LinkGoogle Scholar
    • 133. Tomás M, Napolitano C, De Giuli L, Bloise R, Subirana I, Malovini A, Bellazzi R, Arking DE, Marban E, Chakravarti A, Spooner PM, Priori SG. Polymorphisms in the NOS1AP gene modulate QT interval duration and risk of arrhythmias in the long QT syndrome.J Am Coll Cardiol. 2010; 55:2745–2752. doi: 10.1016/j.jacc.2009.12.065.CrossrefMedlineGoogle Scholar
    • 134. Strauss DG, Vicente J, Johannesen L, Blinova K, Mason JW, Weeke P, Behr ER, Roden DM, Woosley R, Kosova G, Rosenberg MA, Newton-Cheh C. Common genetic variant risk score is associated with drug-induced QT prolongation and torsade de pointes risk: a pilot study.Circulation. 2017; 135:1300–1310. doi: 10.1161/CIRCULATIONAHA.116.023980.LinkGoogle Scholar
    • 135. Roden DM. Taking the “idio” out of “idiosyncratic”: predicting torsades de pointes.Pacing Clin Electrophysiol. 1998; 21:1029–1034.CrossrefMedlineGoogle Scholar
    • 136. Roden DM. Predicting drug-induced QT prolongation and torsades de pointes.J Physiol. 2016; 594:2459–2468.CrossrefMedlineGoogle Scholar
    • 137. Behr ER, Ritchie MD, Tanaka T, et al.. Genome wide analysis of drug-induced torsades de pointes: lack of common variants with large effect sizes.PLoS One. 2013; 8:e78511. doi: 10.1371/journal.pone.0078511.CrossrefMedlineGoogle Scholar
    • 138. Denny JC, Bastarache L, Ritchie MD, et al.. Systematic comparison of phenome-wide association study of electronic medical record data and genome-wide association study data.Nat Biotechnol. 2013; 31:1102–1110. doi: 10.1038/nbt.2749.CrossrefMedlineGoogle Scholar
    • 139. Rastegar-Mojarad M, Ye Z, Kolesar JM, Hebbring SJ, Lin SM. Opportunities for drug repositioning from phenome-wide association studies.Nat Biotechnol. 2015; 33:342–345. doi: 10.1038/nbt.3183.CrossrefMedlineGoogle Scholar
    • 140. Simon JA, Lin F, Hulley SB, Blanche PJ, Waters D, Shiboski S, Rotter JI, Nickerson DA, Yang H, Saad M, Krauss RM. Phenotypic predictors of response to simvastatin therapy among African-Americans and Caucasians: the Cholesterol and Pharmacogenetics (CAP) Study.Am J Cardiol. 2006; 97:843–850. doi: 10.1016/j.amjcard.2005.09.134.CrossrefMedlineGoogle Scholar
    • 141. Lamba J, Hebert JM, Schuetz EG, Klein TE, Altman RB. PharmGKB summary: very important pharmacogene information for CYP3A5.Pharmacogenet Genomics. 2012; 22:555–558. doi: 10.1097/FPC.0b013e328351d47f.CrossrefMedlineGoogle Scholar
    • 142. Birdwell KA, Grady B, Choi L, et al.. The use of a DNA biobank linked to electronic medical records to characterize pharmacogenomic predictors of tacrolimus dose requirement in kidney transplant recipients.Pharmacogenet Genomics. 2012; 22:32–42. doi: 10.1097/FPC.0b013e32834e1641.CrossrefMedlineGoogle Scholar