Review Article. Pharmacogenetics and anaesthesia: the value of genetic profiling. Summary. The history of pharmacogenetics related to anaesthesia

Transcription

1 doi: /j x Review Article Pharmacogenetics and anaesthesia: the value of genetic profiling R. Landau, 1 L. A. Bollag 2 and J. C. Kraft 3 1 Professor of Anaesthesiology, 2 Assistant Professor of Anaesthesiology, 3 Research Study Assistant, University of Washington, Seattle, WA, USA Summary Approximately 50 years ago, pharmacogenetics was described as a new field of medicine that may explain human drug action. Anaesthesia played a key role in the early investigations. An understanding of how a person s DNA influences drug metabolism and effectiveness may allow individually tailored prescriptions, improving outcomes and safety. The ultimate goal of pharmacogenetic research is to offer tailored personalised medicine to improve both the efficacy of medication and patient safety by helping to predict risk of adverse outcomes. In this review, we present a selection of historical landmarks where anaesthesia has been a catalyst for pharmacogenetic development. We examine the level of evidence and cite examples of candidate genes and common polymorphisms known to alter the response to peri-operative medication. Finally, we set forth current views and potential exciting perspectives that may arise from the application of pharmacogenetics to the daily practice of anaesthesia and pain medicine.... Correspondence to: Dr. R. Landau rulandau@uw.edu Accepted: 23 August 2011 The history of pharmacogenetics related to anaesthesia Medical genetics began with the 20th century rediscovery of Gregor Mendel s original 19th century work on plant genetics [1]. In 1949, the landmark paper in Science by Linus Pauling and colleagues linked sickle cell anaemia to a derangement in a specific protein [2], and was the first proof that a genetic change alters the structure and function of a protein and results in a human disease. This set the stage for the birth of pharmacogenetics, a field first described by Arno Motulsky in 1957 [3], named by Friedrich Vogel in 1959 [4] and established by Werner Kalow s monograph in 1962 [5]. These scientists defined pharmacogenetics as the study of biochemical mechanisms that underlie individual differences in drug metabolism, efficacy and side effects. In the early 1950s, prolonged apnoea after suxamethonium was one of the iconic drug responses that provided a starting point for the launch of the new field of pharmacogenetics. In 1956, The Lancet published a paper that was the first to suggest a genetic basis for prolonged apnoea after suxamethonium [6]. Shortly thereafter, Kalow and Gunn described how an inherited variation in drug metabolism involving the enzyme butyrylcholinesterase affects the response to suxamethonium [7]. Since the first reports, now 10 years ago, describing initial findings from the Human Genome Project [8, 9], and its completion in 2003 [10], promises that these discoveries would translate into tangible clinical tests that may change drug prescriptions have been somewhat unfulfilled. Working towards this translation, the pharmacogenetics research network has established a pharmacogenomics knowledge base (PharmGKB) with the goal to collect, encode, and disseminate knowledge about the impact of human genetic variations on drug response, curate primary genotype and phenotype data, annotate gene variants and gene-drugdisease relationships via literature review, and summarise important pharmacogenetic genes and drug pathways [11]. We provide a summarised glossary of commonly used genetic terminology and recent genomic approaches (Table 1). Although numerous hurdles have limited the creation and implementation of pharmacogenetic testing, several Anaesthesia ª 2011 The Association of Anaesthetists of Great Britain and Ireland 165

2 Landau et al. Pharmacogenetics and anaesthesia Table 1. Glossary of common terms and technological advances. Common terms Allele Codon Epigenetics Genetics Genomics Linkage disequilibrium Genotype Mitochondrial DNA Nucleotide Pharmacogenetics Pharmacogenomics Phenotype Population stratification Polymorphism (poly = several; morph = form) Single Nucleotide Polymorphism (SNP) Variant (or mutant) Wild-type Description of different methods Chromatin immunoprecipitation (CHIP) Exome capture One of the forms (wild-type or variant) of a genetic locus on a chromosome. An individual inherits two alleles for each gene, one from each parent. Different alleles produce variation in inherited characteristics such as hair colour or blood type. Both alleles need to be known to determine the genotype. A homozygote person has two identical alleles for a gene; a heterozygote inherited a different allele from each parent A trinucleotide sequence of DNA or RNA that corresponds to a specific amino acid. Among 64 codons, 61 code for amino acids ( sense ) and 3 are used as stop signals ( non-sense ). Due to the degeneration of the genetic code, 1 6 codons can code for a single amino acid (hence only 20 amino acids) The study of heritable changes caused by the activation and deactivation of genes without any change in the underlying DNA sequence. Environmental stimuli such as ischaemia, hypotension or shock can alter cascades of gene expression. Epigenetic changes can be highly relevant in determining peri-operative outcomes after surgery and can also play an important role in disease susceptibility and drug response The study of the function of a single gene encoded by DNA, transcribed to RNA, and translated into protein The study of the entire genome all of the genetic instructions found in a cell, which includes 23 pairs of chromosomes in the nucleus and a small mitochondrial chromosome. The power of genomics is to see multiple pathways of genes and proteins involved in a given set of stimuli or diseases The non-random association of alleles at two or more loci, not necessarily on the same chromosome, it describes the situation in which the haplotype frequencies in a population deviate from the values they would have if the genes at each locus were combined at random An individual s collection of genes or the two alleles inherited for a particular gene. There are three possible genotypes for a polymorphism: wild-type homozygote (wt wt), heterozygote (wt v) or variant homozygote (v v) A small circular chromosome of single stranded DNA found inside mitochondria and passed from mother to offspring. Standard DNA isolation methods tend to only isolate nucleic DNA and special efforts are required to isolate mitochondrial DNA. Mitochondrial genetics is of special interest for the anaesthetist since mitochondria are key cellular organelles relevant to many aspects of anaesthesia and intensive care The basic building block of nucleic acids, which polymerise to make DNA and RNA. A nucleotide consists of a sugar molecule (either deoxyribose in DNA or ribose in RNA) attached to a phosphate group and a nitrogen-containing base (DNA: adenine (A), cytosine (C), guanine (G), thymine (T); RNA: uracil (U) takes the place of thymine) The study of drug response due to genetic variability Similar to pharmacogenetics, but incorporates sophisticated genomic tests and the whole genome to define the variability of drug response Observable traits or characteristics of an individual (e.g. hair colour, presence or absence of disease) that may or may not be genetic. It is usually the product of the genotype and all environmental effects Refers to differences in allele frequencies between cases and controls due to systematic differences in ancestry rather than a true association of genes with a disease One of two or more variants of a particular DNA sequence, present in more than 1% of a population. Most polymorphisms involve variation at a single base pair (SNP). Polymorphisms can also be much larger in size and involve long stretches of DNA A type of polymorphism involving variation of a single base pair when one nucleotide (A, T, C, or G) in the DNA sequence is altered. The average frequency is 1 variation for 1000 nucleotides. It is common for a given SNP to be inherited in a consistent haplotype block of DNA (called linkage disequilibrium) The other genetic sequence The first genetic variant cloned (usually the form that occurs most frequently) Technique allowing massive sequencing of DNA fragments (up to nucleotides), on a silicone plate Cutting-edge approach to selectively sequence coding regions of the human genome Protein coding genes constitute only approximately 1% of the human genome, but harbour 85% of the mutations with large effects on disease-related traits. Therefore, efficient strategies for selectively sequencing complete coding regions (whole exome) have the potential to contribute to the understanding of rare and common human diseases. Although still a subset of the genome, exome capture allows the investigation of a more complete set of human genes with the cost and time advantages of genome capture 166 Anaesthesia ª 2011 The Association of Anaesthetists of Great Britain and Ireland

3 Landau et al. Pharmacogenetics and anaesthesia Anaesthesia 2012, 67, Table 1. (Continued). Genome wide association studies (GWAS) Massively parallel sequencing (MPS) Based on the concept that a subset of markers (DNA sequences or SNPs) or haplotype blocks can identify regions of the human genome where genetic influences on diseases may reside. Once a given region of the genome has been identified and confirmed in a second population, genes within that chromosome region can be investigated to elucidate the causative gene and genetic variant, and therefore not all DNA from a given individual needs to be sequenced. However, as markers across the genome are currently widely spaced, the risk of such an approach is that important regions can be missed. In addition, rare SNPs that contribute greatly to disease in smaller populations of patients can be missed To date, GWAS have identified more than 250 common variants associated with risk alleles that contribute to a wide range of diseases. Most of these impart small effects on disease risk (e.g. odds ratio of 1 2); furthermore, even when extremely large studies have been performed, the vast majority of the genetic contribution to disease risk remains unexplained A technique that enables simultaneous screening of thousands of loci for disease-causing mutations, structural rearrangements, or epigenetic changes. On the RNA level, mutational analysis, posttranscriptional modifications and the profiling of abundant transcripts become possible in one experiment. This technique will most likely replace micro-arrays pharmacogenetic tests have been recently developed [12] and the US Food and Drug Administration (FDA) [13] and the European Medicines Agency have approved numerous drug label modifications to contain pharmacogenetic information [14]. Several conditions are required before a genetic test is approved and becomes commercially available: strong evidence is needed that the test is actually associated with clinical outcomes and that it can provide dose-guidance (predict the dose or the need for an alternate drug). Therefore, if no other drug is available or there is no algorithm to predict the dose, then genetic testing is unlikely to result in any useful information. Warfarin (testing of CYP2C9 and VKORC1), clopidogrel (testing of CYP2C19) and chemotherapy-based drugs such as tamoxifen (testing of CYP2D6) and trastuzumab (testing of HER2) are among the list of medicines for which pharmacogenetic testing is now available; none of these, however, is directly prescribed as anaesthetics or analgesic medications, but they may affect peri-operative outcomes, particularly warfarin, clopidogrel and b-blockers. Pharmacogenetics: relevance for the anaesthetist Numerous clinical trials and reviews have surfaced in recent years describing genetic associations with clinical outcomes in the field of anaesthesia, peri-operative outcomes and pain medicine [15 29]. Nonetheless, many clinicians remain sceptical and often wonder about the relevance of genetic research, as it is often considered that titration of drugs to the desired effect works well. The Clinical Pharmacogenetics Implementation Consortium (CPIC) was created in 2009 to establish a framework for understanding levels of evidence required for pharmacogenetics to be incorporated into clinical practice, and to address the need to provide very specific guidance to clinicians and laboratories so that pharmacogenetic tests are used wisely [30]. Indeed, one of the major obstacles to clinical implementation of pharmacogenetics has been the lack of clear peer-reviewed guidelines that translate laboratory test results into applicable clinical guides to prescribe specific drugs. A standard process for evaluating levels of evidence (linking gene variation to phenotypes) and strength of recommendations (linking genotypes to drug dosing recommendations) is proposed by the CPIC [30] (Table 2). The first guideline developed by the CPIC was published in January 2011, and provides dosing recommendations for azathioprine, mercaptopurine and thioguanine based on the thiopurine methyltransferase (TPMT) genotype with a strong classification of recommendation [31]. Thiopurines are most commonly used to treat non-malignant conditions and are also key anti-cancer agents, and prescribed as immunosuppressants in inflammatory bowel disease, rheumatoid arthritis and other immune conditions. An overview of all the drugs utilised in the perioperative period is beyond the scope of this review. We have selected six clinical examples for which pharmacogenetic tests have been evaluated with various levels of evidence for a genotype phenotype effect, and present their relevance for clinical practice and establishment of clinical recommendations. The first three relate to analgesia: codeine; oxycodone; and morphine. These are followed by three examples where there is stronger evidence for pharmacogenetic testing: warfarin; clopidogrel; and b-blockers. Codeine Cytochrome P450 (CYP450) is a super-family of liver enzymes that catalyse phase 1 drug metabolism. The D6 isozyme of the CYP2 family is particularly affected by genetic variability and currently has 80 identified CYP2D6 Anaesthesia ª 2011 The Association of Anaesthetists of Great Britain and Ireland 167

4 Landau et al. Pharmacogenetics and anaesthesia Table 2. Clinical Pharmacogenetics Implementation Consortium (CPIC) three-tier scheme for scoring of pharmacogenetic evidence. Quality of evidence linking drug-related phenotype to specific genetic variations Level 1 Includes consistent results from well-designed, well-conducted studies Level 2 Sufficient to determine the effects, but the number, quality, or consistency of the individual studies limit the strength of the evidence, by the inability to generalise to routine practice, or by the indirect nature of the evidence Level 3 Insufficient to assess the effects on health outcomes because of the limited number of studies, insufficient power of the studies, important flaws in their design or in the way they were conducted, gaps in the chain of evidence, or lack of information Level of evidence linking genotypes to drug dosing recommendations Level A Strong recommendation for the statement Level B Moderate recommendation for the statement Level C Optional recommendation for the statement alleles [32], resulting in a variable enzymatic activity ranging from 1% to 200%. As a result, each individual can be classified as having an ultra-rapid metabolism, an extensive metabolism, an intermediate metabolism or a poor metabolism, and microarray technology is available to classify individuals according to their metabolic phenotype. Furthermore, it is important to note that the distribution of CYP2D6 phenotypes varies according to ethnicity. Of note, approximately 7 10% of Caucasians have no CYP2D6 activity (poor metabolism) because of deletions, frameshift, or splice-site mutations of the gene. At the other end of the spectrum, 1 3% of Middle Europeans and up to 29% of Ethiopians have duplications of the CYP2D6 gene and are classified as ultra-rapid metabolisers [33]. Ultra-rapid metabolisers have up to 50% higher plasma concentrations compared with extensive metabolisers [34]. Codeine is a pro-drug, and requires O-demethylation catalysed by CYP2D6 to be converted into morphine and become analgesic; this metabolic pathway accounts for 10% of codeine clearance. The conversion of codeine into norcodeine by CYP3A4 and into codeine-6-glucuronide by glucuronidation represents approximately 80% of codeine clearance. Morphine is further metabolised into morphine- 6-glucuronide and morphine-3-glucuronide; both morphine and morphine-6-glucuronide display opioid activity. Codeine was initially prescribed because of the belief that being a weak opioid, it is safe and would not result in adverse outcomes. However, the death of a breastfed 13-day-old neonate through morphine overdose because his mother was taking codeine after childbirth resulted in a recent FDA warning on codeine use in nursing mothers [35]. Toxic blood levels of morphine or its active metabolite morphine- 6-glucuronide may arise in mothers and neonates who are CYP2D6 ultra-rapid or extensive metabolisers. It has been suggested that codeine be avoided in breastfeeding mothers with a CYP2D6 extensive or ultra-rapid metabolism genotype. Codeine and morphine clearance in breastfeeding mothers and their relation to CYP2D6 genotypes have since been evaluated [36 38]. Other life-threatening adverse events have been reported in patients with CYP2D6 ultrarapid metabolism [39, 40], and individuals in this subgroup are particularly at risk when prescribed pro-drugs with a narrow therapeutic range or concurrently with other drugs competing with critical metabolic pathways. Since 2007, the FDA requires manufacturers of prescription codeine products to state in the Precautions section of the drug label the known risks of prescribing codeine to breastfeeding mothers [41]. An FDA-approved genetic test (AmpliChip CYP450; Roche Diagnostics, Palo Alto, CA, USA) is commercially available to test genetic variants of CYP2D6 [42]. Overall, the level of evidence linking gene variation (CYP2D6) to phenotype (increased biotransformation of codeine into morphine) is strong; however, there is no randomised clinical trial assessing the benefits of genetic testing before codeine therapy. Currently, the only recommendation to avert risk is a cautionary insert to avoid codeine in breastfeeding mothers (or to apply genetic testing in mothers neonates if codeine is prescribed). Oxycodone Oxycodone is a semi-synthetic opioid agonist and is widely used as an analgesic for both acute and chronic pain. In the postoperative setting, oxycodone has been shown to be more potent than morphine for visceral pain relief [43]. Both oxycodone and oxymorphone, one of its metabolites, are potent analgesics that have been used in the postoperative setting [44]. Since the introduction of controlled-release oxycodone in 1995, annual prescriptions of oxycodone have increased several-fold [45]. Oxycodone undergoes metabolism in the liver through four different metabolic pathways catalysed by CYP3A4 and CYP2D6 (Fig. 1) [46]. N-demethylation of oxycodone by CYP3A4 into noroxycodone is quantitatively the most important metabolic route (in the order of 45%), while a smaller fraction (11%) of oxycodone is O-demethylated to 168 Anaesthesia ª 2011 The Association of Anaesthetists of Great Britain and Ireland

5 Landau et al. Pharmacogenetics and anaesthesia Anaesthesia 2012, 67, keto reduction (8%) α-/β-oxycodol UDPglucuronosyltransferase Oxymorphone-3- glucuronide Oxymorphone CYP2D6 (11%) CYP3A4 Figure 1 Oxycodone metabolism. Oxycodone Quinidine Paroxitine Itraconazole Ketoconazlole Grapefruit juice Noroxymorphone CYP3A4 (47%) CYP2D6 Noroxycodone oxymorphone by CYP2D6 [47]. The importance of an intact CYP3A4 pathway for oxycodone clearance has been emphasised in numerous pharmacokinetic studies; numerous drugs, as well as grapefruit juice intake, have been shown to interfere with this pathway [48 53]. Noroxymorphone is a metabolite of noroxycodone and oxymorphone that seems to be devoid of analgesic activity after systemic administration [54]. Early studies describing the impact of CYP2D6 metabolism (phenotype) on clinical outcomes of oxycodone (analgesia and side effects) have demonstrated a weaker effect in poor metabolisers [55, 56]. Anecdotal cases of adverse effects after oxycodone in patients who were genotyped for CYP2D6 genotype have been reported [57, 58]. Despite its widespread and increasing use for pain management and postoperative analgesia, evidence related to the pharmacogenetic influence of CYP3A4 and CYP2D6 for the clinical response (analgesic and side effect profile) of oxycodone or oxymorphone is particularly scarce. Oxycodone s analgesia and side effects were evaluated in 10 healthy Caucasian volunteers (all men) phenotyped and genotyped for CYP2D6 [59]. Experimental pain tests were performed in a 5-arm crossover, randomised, doubleblinded, placebo-controlled manner, with oral oxycodone 0.2 mg.kg )1. Differences in analgesic effect were found with increased analgesic effects in ultra-rapid metabolisers; conversely, poor metabolisers had a 2- to 20-fold reduction in the effects compared with extensive metabolisers. Notable differences in the incidence of spontaneously reported adverse reactions after oxycodone were reported by 2 2 ultra-rapid metabolisers, in comparison with only 1 6 extensive metabolisers and no toxicity reported in intermediate and poor metabolisers (0 2). Importantly, CYP3A4 blockade, such as that occurring with itraconazole, increased the analgesic efficacy of oxycodone as well as the toxicity of oxycodone, especially in CYP2D6 ultra-rapid metabolisers. These findings are consistent with reports of life-threatening events in ultra-rapid metabolisers receiving codeine [35, 39, 40]. The effects of itraconazole, an inhibitor of the CYP3A4- mediated N-demethylation of oxycodone, were evaluated after administration of oxycodone (0.1 mg iv and 10 mg orally) in 11 healthy Caucasian subjects [60]. Itraconazole affected the metabolism of oxycodone to a greater extent when oxycodone was given orally. As a result, dose adjustments of oral oxycodone may be necessary in CYP3A4 poor metabolisers to avoid opioid-related adverse effects. Overall, the level of evidence linking gene variation (CYP2D6 and CYP3A4) to phenotype (altered biotransformation of oxycodone into oxymorphone and overall clearance of oxycodone and oxymorphone) is strong; however, there is no randomised clinical trial on the benefits of genetic testing before oxycodone therapy. There is also no warning on the oxycodone label cautioning against prescription of oxycodone in CYP2D6 ultra-rapid metabolisers or in patients taking CYP3A4-inhibitors concomitantly. Given the widespread use and potential abuse of oxycodone prescription, further studies should certainly investigate this phenotype genotype association. Morphine Clinicians are well aware of the large and unpredictable inter-individual variability of response to morphine [61]. Genomic and pharmacogenetic research has considered numerous candidate genes as suitable targets for the study of pain and or analgesia [62]. Among the numerous candidate genes that have been considered important in opioid response, the l-opioid receptor gene (OPRM1, p.118a G), the catechol-o-methyltransferase gene (COMT, Val158Met), several variants of the ATP-binding cassette, sub-family B member 1 gene (ABCB1) and the CYP family of enzymes have been extensively reviewed [23, 25, 63]. A recent metaanalysis of all pain studies evaluating the impact of A118G polymorphism of OPRM1 on the response to opioids did not identify a strong association between this polymorphism and the response to opioids [64]. It is likely that the heterogeneity of the clinical situations (experimental pain, acute pain, labour pain, postoperative pain, chronic pain) and diversity of evaluated drugs and dosages precluded any significant findings. Overall, the level of evidence linking gene variation to morphine response is moderate, probably due to the inherent complexity of studying a heterogenous phenotype such as pain. Limitations that have prevented strong genotype phenotype associations from being identified in the context of pain studies include differences in pain modalities, sex differences, hurdles in extrapolating data Anaesthesia ª 2011 The Association of Anaesthetists of Great Britain and Ireland 169

6 Landau et al. Pharmacogenetics and anaesthesia from animal models to the response in humans, population stratification and environmental differences, in addition to the obvious polygenic nature of pain and analgesic response. The design and execution of large clinical studies analysing multiple haplotypes simultaneously remain the true challenge to date. One relatively small genome-wide association study in the context of acute postoperative pain was published [22], and recently, the possible impact of epigenetic-based strategies for pain therapy has been discussed [65]. Meanwhile, some interesting pre-clinical and even clinical trials have been exploring the value of gene therapies for management of chronic pain conditions [66 68]. Of interest as well are the new insights and developments brought by extensive research on the SCN9A gene, a gene involved in channelopathies that result in the inability to experience pain and potential targeted therapies [69]. Warfarin Balancing the risk of thrombosis against bleeding is a fundamental patient safety issue, and anaesthetists often assess therapeutic anticoagulation peri-operatively. Due to its narrow therapeutic window and significant variability in dose response, it is a leading cause of adverse drug reactions [70 72]. Traditionally, dosing has been performed by trialand-error, with the initial fixed dose adjusted based on the international normalised ratio (INR) until the INR value fell within a target range. Warfarin pharmacogenetics involves both the enzyme responsible for its metabolism, CYP2C9, and its target of action, Vitamin K epoxide reductase complex 1 (VKORC1), the key enzyme of the Vitamin K cycle and the molecular target of coumadins [73 76]. CYP2C9 is almost exclusively responsible for the metabolism (bio-inactivation) of the pharmacologically more active (S)-enantiomer of warfarin [77]. The CYP2C9 genotype explains approximately 10% of the observed variability in the therapeutic warfarin dose and VKORC1 polymorphisms account for approximately 30% of the variance in stabilised warfarin dose [78, 79]. In addition, VKORC1 contributes to a greater variability in dose response in Caucasians compared with Black individuals and Asians [80]. The CYP4F2 gene has also been associated with warfarin dosing [81 85]. A combination of CYP2C9, VKORC1 and CYP4F2 genotyping should allow to predict warfarin dosing (Table 3). In a cohort of 1053 Swedish subjects, a genomewide association study evaluating SNPs confirmed that VKORC1, CYP2C9 and CYP4F2 are the major genetic determinants of warfarin pharmacokinetics and dosing [86]. Other non-genetic factors including age, body mass index, sex, weight and INR are also known to play a role in warfarin response and collectively contribute to approximately 20% of variance in dose [87]. In 2007, the FDA approved pharmacogenetic information to be included in the warfarin product label [88]. The latest warfarin label reads: The patient s CYP2C9 and VKORC1 genotype information, when available, can assist in the selection of the starting dose [89]. The FDA proposed a relatively simple approach using genotype-stratified tables with the range of expected therapeutic warfarin doses (mg.day )1 ) based on CYP2C9 and VKORC1 genotypes to estimate warfarin dose; however, its accuracy has not been quantified. Another approach has focused on developing and validating comprehensive predictive algorithms to facilitate warfarin dosing that integrate both clinical and genetic factors. Numerous genotype-guided dosing algorithms for warfarin therapy in different cohorts have been proposed [90 98]. The algorithm with the best predictive power appears to be the one proposed by the International Warfarin Pharmacogenetics Consortium (IWPC) and is the largest study to date [93]. The greatest benefits of using an algorithm were observed in a subgroup of 46% of individuals who required 21 mg or less of warfarin per Table 3. Pharmacogenetics of warfarin dosing. Genotype Phenotype related to warfarin Dose CYP2C9 24 SNPs Metabolism of S-warfarin *1 Wild-type allele Normal Average *2, *3 Loss-of-function alleles Decreased (50%) Lower VKORC1 10 SNPs fi haplotypes (H) mrna expression AA H1, H2 Lowest Lower A B H1, H2, H7, H8, H9 Average Average BB H7, H8, H9 Highest Higher CYP4F2 SNP (rs ; V433M) Metabolism of VK1 V433 Highest capacity Lower V433M Average capacity Average M433 Reduced capacity Higher SNP, single nucleotide polymorphism; CYP2C9, Cytochrome P450 2C9; VKORC1, Vitamin K epoxide reductase complex subunit 1; VK1, Vitamin K1; CYP4F2, Cytochrome P450 4F2; V433M, Valine Methionine 433 polymorphism. 170 Anaesthesia ª 2011 The Association of Anaesthetists of Great Britain and Ireland

7 Landau et al. Pharmacogenetics and anaesthesia Anaesthesia 2012, 67, week or 49 mg or more per week for therapeutic anticoagulation. Finally, evaluation of cost-effectiveness of genotypebased warfarin dosing has resulted in variable findings [99 105]. A novel model has been proposed integrating genotyping based on the relative impact of each genotype on clinical outcomes [106]. While inevitable assumptions have to be made in all cost-effectiveness models, this model suggests that pharmacogenetic-guided dosing for warfarininitiation may improve health (quality-adjusted life-years), but at a high cost per quality-adjusted life-year gained. Overall, the level of evidence linking gene variation (CYP2C9 and VKORC1) to phenotype (bleeding thrombotic risk with warfarin dosing) is strong; however, there is, to date, no strong recommendation to apply pharmacogenetic testing before initiating warfarin therapy. Clopidogrel Clopidogrel is an adenosine diphosphate (ADP)-receptor antagonist, and like a pro-drug, it requires bioactivation into an active metabolite (R ) for inhibition of platelet aggregation. The range in response to inhibition of ADPinduced platelet aggregation with clopidogrel is extremely broad, with a wide distribution from < 10% to almost complete platelet anti-aggregation. A recent meta-analysis reviewing clinical outcomes after clopidogrel therapy emphasised that residual platelet reactivity despite clopidogrel treatment was significantly associated with an increased risk of death and or thrombotic recurrences [107]. Clopidogrel is metabolised in part by the enzyme CYP2C19 to achieve its antiplatelet activity. CYP2C19*1 is the wild-type genotype and results in a normal metabolic function. CYP2C19*2 and CYP2C19*3 are two common loss-of-function alleles that result in poor metabolism. Known ethnic differences in allele distribution account for differences in clinical outcomes with clopidogrel therapy. Although platelet response to clopidogrel is highly heritable, it is not entirely explained by CYP2C19, as one analysis showed that only 12% of the variation in response to clopidogrel was explained by the commonly studied CYP2C19*2 genotype [108]. Other genetic polymorphisms also associated with impaired CYP2C19 activity (CYP2C19*4, *5, *8) and adverse clinical events do exist although they are relatively uncommon. The effect of variant CYP2C19 alleles on clinical outcome in response to clopidogrel has been reviewed in several meta-analyses [109, 110]. Data from seven prospective studies with a total of 8043 patients with coronary artery disease treated with clopidogrel and followed up for a period of time ranging from 6 months to 8 years showed a significant association between the CYP2C19*2 polymorphism and an increased risk of major adverse cardiovascular events [109]. On 12 March 2010, the FDA approved a new label for clopidogrel with a boxed warning about the reduced effectiveness of clopidogrel in patients who are poor metabolisers with loss-of-function alleles CYP2C19*2 and *3, and suggested that carriers of these alleles receive a higher dose of clopidogrel or an alternative antiplatelet agent [111]. Although there is an expanding database on genetic polymorphisms that may affect clopidogrel metabolism and thus clinical outcomes, the American College of Cardiology Foundation (ACCF) and the American Heart Association (AHA) convened a task force to re-examine the evidence. The task force emphasised that while CYP2C19 variants have been shown in several studies to reduce clopidogrel metabolism and its pharmacodynamic effect and clinical effectiveness, there are no prospective studies demonstrating a clinical benefit to personalising antiplatelet therapy based on genotype analysis. They concluded that information regarding the predictive value of pharmacogenomic testing is very limited at this time, pending results of multiple ongoing studies. Overall, the level of evidence linking gene variation (CYP2C19) to phenotype (reduced effectiveness of clopidogrel resulting in increased myocardial infarctions and stent thrombosis) is strong; however, there is, to date, no strong recommendation to apply pharmacogenetic testing before adjusting clopidogrel dosing or switching to alternative antiplatelet therapies. Of interest and relevant to the pharmacogenetic relationship between CYP2C19 and clopidogrel response, proton pump inhibitors are often co-prescribed with clopidogrel for gastric protection. Co-administration of clopidogrel with proton pump inhibitors (CYP2C19-inhibiting drugs) decreases the antiplatelet effect of clopidogrel [112, 113] and might reduce clopidogrel efficacy [114]. Omeprazole is both a substrate and an inhibitor of CYP2C19; therefore, poor metabolisers will have not only impaired formation of the clopidogrel active metabolite but also the highest concentrations of omeprazole [114]. In a large retrospective cohort study of close to patients after coronary artery stenting, a significant increase in major adverse cardiac outcomes was found in patients concomitantly treated with clopidogrel and a proton pump inhibitors compared with those treated with clopidogrel alone [115]. However, a prospective trial in more than 2200 patients with acute myocardial infarction treated with clopidogrel confirmed that carriers of CYP2C19 loss-of-function alleles had a higher rate of subsequent cardiovascular events than those who were not carriers; concurrent use of proton pump inhibitors (70% of the cohort) and omeprazole specifically Anaesthesia ª 2011 The Association of Anaesthetists of Great Britain and Ireland 171

8 Landau et al. Pharmacogenetics and anaesthesia (50% of the cohort) did not impact on the response to clopidogrel [116]. Nonetheless, the FDA recommends that omeprazole (and not all proton pump inhibitors) should not be dosed concomitantly with clopidogrel. FDA states that not all proton pump inhibitors have the same inhibitory effect on CYP2C19 [117]. b-blockers Administration of b-blockers after an acute coronary event has become a quality measure by which hospitals are judged [118]. In the past decade, b-blockade in the operating room has become a standard of care in at-risk patients, with the goal of improving peri-operative outcomes. However, despite numerous in vitro studies defining extensively the biological effects of the multiple genetic variants of the b 1 - adrenergic receptor gene (ADRB1) and the b 2 -adrenergic receptor gene (ADRB2) (Table 4) [ ,144], the clinical relevance for most of the numerous pharmacogenetic results for b-blockers is still inconclusive. Detailed reviews are constantly updating the body of evidence on the pharmacogenetic effects of adrenergic receptors polymorphisms on the response to b-blockers in various cardiovascular conditions [ , , 152, 156] (Table 4). The Arg389 receptor of the b 1 -adrenergic receptor confers a gain-of-signalling effect (hyperfunctional state) and increases sensitivity to pharmacological b-blockade; conversely, the Gly389 receptor results in loss-of-signalling and a hypofunctional response as if it were already bblocked. Therefore, this apparent increased b 1 -adrenergic tone in individuals with Arg389 may represent an opportunity to use b-blockers with a greater impact in diseases responsive to anti-adrenergic therapy. Of interest, the Gly389 allele is more frequent in African Americans (42%) compared with Caucasians (28%) [127], and this has been suggested to in part explain the decreased sensitivity to b-blockade and altered clinical outcomes in African Americans treated with b-blockers [139, 157]. Individual genetic association studies have examined the association between the two common SNPs of ADRB1 and resting haemodynamics and hypertension as potential causal genes. The initial consensus was that these two polymorphisms of ADRB1 are most likely not to be associated with an increased risk for essential hypertension. A recent search from 163 genetic association studies involving seven genes and 37 distinct genetic variants was pooled into a web-based information system (the CUMAGAS-HYPERT database) along with data from genome-wide association studies involving the adrenoceptor family genes and provided some interesting findings [158]. Most of the published studies were underpowered to detect an association with a minor allele that would contribute to the risk for hypertension (a sample size of individuals would be needed to yield an odds ratio of , with a power of 80% [159]). Carriers of the Arg389 allele had a 16% reduced risk of hypertension but only in East Asians. Regarding the Ser49Gly polymorphism, carriers of the Ser49 had a 24% increased risk for hypertension that was not affected by ethnicity. This analysis also revealed a modest effect of the two common polymorphisms of ADRB2 (Arg16Gly and Gln27Glu), but only in Caucasians. There was no association recorded by any of the pooled genome wide association studies. Although b-blockers are widely used to treat essential hypertension, there have been too few large studies evaluating the effect of b-blockers in that clinical context based on genotype haplotypes of ADRB1 [119,131,134,135, ,145,146, ]; therefore, the level of evidence to conclude whether genetic variants of ADRB1 influence the response to b-blockers for antihypertensive therapy is overall weak. Whether the lack of consistency in findings between the various studies is due to differences in study design, population stratification (different ethnicities), the drug itself or the dose prescribed, or heterogeneity in outcome measures, a strong genetic effect of ADRB1 on blood pressure response to b-blockers remains to be determined. Clinical studies examining the effects of polymorphisms of ADRB1 and ADRB2 on b-blocker response and outcomes in individuals with cardiac diseases are numerous. Interestingly, while not all studies reported a positive association, studies with a positive association always showed Arg389 homozygosity to be a predictor of better clinical outcome in heart failure; therefore, this is considered as the SNP with the strongest level of evidence in the context of b-blocker response. We selected one clinical trial to illustrate the effect of ADRB1 polymorphisms on b-blocker response in patients with heart failure [139]. The study by Liggett et al. is unique in that it was a prospective, double-blinded, placebocontrolled longitudinal study that evaluated agonistmediated contractility in ex-vivo human ventricular studies and performed a genotype association study in a large cohort of healthy subjects versus individuals with heart failure (a sub-analysis of 1040 subjects who had participated in the BEST trial). Bucindolol-treated patients who were Arg389 homozygous had a significant reduction in mortality (38% reduction) and re-hospitalisation rates during the follow-up period of 5 years; conversely, individuals carrying the Gly389 allele were found not to benefit from bucindolol therapy. A small pharmaceutical company (ARCA Biopharma, Broomfield, CO, USA) issued a patent on methods of treating heart failure patients with bucindolol (Gencaro TM ) based on genetic targeting. Pending FDA approval of 172 Anaesthesia ª 2011 The Association of Anaesthetists of Great Britain and Ireland

9 Landau et al. Pharmacogenetics and anaesthesia Anaesthesia 2012, 67, Table 4. Polymorphisms of adrenergic receptors, and their relevance in b-blocker pharmacogenetics. Gene Polymorphism ADRB1 Arg389Gly Caucasians (24 34%) African Americans (39 46%) Hispanics (31 33%) Asians (20 30%) ADRB1 Ser49Gly Caucasians (12 16%) African Americans (23 28%) Hispanics (20 21%) Asians (14%) ADRB2 Arg16Gly Caucasians (39%) African Americans (49%) Hispanics (unknown) Asians (51%) ADRB2 Gln27Glu Caucasians (25%) African Americans (19%) Hispanics (unknown) Asians (9%) ADRB2 Thr164Ile Caucasians (1 2%) African Americans (0%) Hispanics (unknown) Asians (0%) ADRA2C Ins Del Caucasians (4%) African Americans (41%) Hispanics (unknown) Asians (unknown) Phenotype Condition ADRB1 Arg389Gly Healthy volunteers ADRB1 Arg389Gly Healthy volunteers Minor allele frequency according to ethnicity [120, 138] Functional consequences Ref Drug tested Exercise (no drug tested) Metoprolol Outcome parameter HR response to exercise Arg389 allele has higher basal and agonist-stimulated adenylyl cyclase activity Heart samples from heart failure patients with Arg389 allele have lower adenylyl cyclase activity after agonist stimulation than those with Gly389 allele Gly49 allele has greater agonist-promoted receptor down-regulation Gly49 is more sensitive to metoprolol inhibitory effects than Ser49 Gly16 allele has greater agonist-promoted receptor down-regulation Glu27 allele is resistant to receptor down-regulation Arg16-Gln27 haplotype causes almost complete receptor desensitisation Ile164 allele causes defective coupling with G protein Individuals carrying one Ile164 allele have a five-fold reduction in sensitivity to b 2 agonist-mediated vasodilation; vasoconstrictor sensitivity is increased Deletion allele causes loss of auto-inhibitory function of the variant receptor Attenuation of exercise-induced HR and SBP ADRB1 Arg389Gly Hypertension Atenolol Resting and exercise induced HR and BP ADRB1 Arg389Gly Hypertension Atenolol HR and BP response to chronic treatment ADRB1 Arg389Gly Hypertension Bisoprolol vs Atenolol (randomised) HR and BP response to chronic treatment ADRB1 Arg389Gly Hypertension Metoprolol Daytime DBP after 4 weeks treatment ADRB1 Arg389Gly Ser49Gly Hypertension Atenolol SBP and DBP response Sample size, Population n = 316 Caucasians n = 194 African American n = 221 Hispanics n = 142 Chinese n=16 Asian men n=34 Mixed ethnicity 18 men, 16 women n = 101 Caucasian (Sweden) 66 men, 35 women n = 147 Caucasian 86 men, 61 women n=40 Mixed ethnicity 24 men, 16 women n = 270 Caucasian (Italy) 149 men, 121 women Pharmacogenetic effect on outcome parameter Arg389 = Gly389 [128] Arg389 > Gly389 [130] [119] [120] [121] [122] [122] [123] [124] [125] [126] [127] [129] Arg389 > Gly389 [131] resting Arg389 = Gly398 exercise Arg389 = Gly389 [132] Arg389 = Gly389 [133] Arg389 > Gly389- carriers Arg389 = Gly389 Ser49 = Gly49 ADRB1 Arg389Gly Heart failure Carvedilol LVEF n = 224 Arg389-carriers > [119] Gly389 ADRB1 Arg389Gly Heart failure Metoprolol HR reduction, survival and adverse effects n = Caucasian 163 non-caucasian 507 men, 93 women Arg389 = Gly389 [136] [134] [135] Anaesthesia ª 2011 The Association of Anaesthetists of Great Britain and Ireland 173

10 Landau et al. Pharmacogenetics and anaesthesia Table 4. (Continued). Phenotype Outcome parameter Sample size, Population Condition Drug tested ADRB1 Arg389Gly Heart failure Carvedilol or LVEF and RVEF n = 199 bisoprolol Caucasian (non-randomised) 165 men, 34 women ADRB1 Arg389Gly Heart failure Metoprolol LVEF n = Caucasian 19 non-caucasian 37 men, 24 women ADRB1 Arg389Gly Heart failure Bucindolol (placebo-control) 5-year survival n = Caucasian 207 non-caucasian 825 men, 215 women ADRB1 Arg389Gly Heart failure Carvedilol LVEF (1.5 years) n = 135 Caucasian 106 men, 29 women ADRB1 Ser49Gly Heart failure Metoprolol LVEF n = 61 Mixed ethnicity 37 men, 24 women ADRB2 Gln27Glu Heart failure Carvedilol LVEF n = 80 Unknown ethnicity 68 men, 12 women ADRB1 ADRB2 ADRB1 ADRA2C ADRB1 ADRB2 ADRA2C ADRB1 ADRB2 ADRA2C Arg389Gly Ser49Gly Arg16Gly Gln27Glu Arg389Gly Ins Del Arg389Gly Ser49Gly Arg16Gly Gln27Glu Ins Del Arg389Gly Ser49Gly Arg16Gly Gln27Glu Ins Del Acute coronary syndrome Heart failure Heart failure Heart failure Non-specified b-blocker Metoprolol 5 6 months Non-specified b-blocker Metoprolol or carvedilol (non-randomised) 3-year mortality n = Caucasian 130 non-caucasian 381 men, 216 women LVEF n = 54 Mixed ethnicity 32 men, 22 women Death or heart transplantation Transplant-free survival n = 227 total n = 183 on b-blocker 188 Caucasian 39 non-caucasian 156 men, 71 women n = Caucasian 156 non-caucasian 471 men, 166 women Pharmacogenetic effect on outcome parameter Arg389 = Gly389- carriers Arg389 > Gly389- carriers Arg389 > Gly389- carriers Arg389-carriers > Gly389 Gly49-carriers > Ser49 [137] [138] [139] [140] [141] Glu27 > Gln27 [142] High risk: Arg16-Gln27 Intermediate risk: Gln27 carriers Low risk Gly16- Glu27 Arg389 > Gly389 Del > Ins Arg16-Gln27 > Gly16-Glu27 Arg389 = Gly389 Ser49 = Gly49 Arg16 = Gly16 Gln27 = Glu27 Ins = Del [143] [144] [145] [146] ADRB1, b 1 -adrenergic receptor gene; ADRB2, b 2 -adrenergic receptor gene; ADRA2C, a 2C -adrenergic receptor gene; Ins Del, Insertion Deletion; LVEF, left ventricular ejection fraction; SBP, systolic blood pressure; DBP, diastolic blood pressure; HR, heart rate. bucindolol with a specific indication for subgroups of the adrenoceptor genotypes, this would represent the first pharmacogenetic-guided therapy in the area of cardiovascular disease. Another potentially important study is that by Lanfear et al. that challenged the concept of b-blockers for all and evaluated whether benefits from b-blockade after an acute coronary event may be different based on individuals genetic profile [143]. Mortality rates were increased in individuals carrying certain variants of the ADRB2, raising the mortality rate to 20% at 3 years according to the haplotype combination of Arg16Gly and Gln27Glu. Patients with variants impairing b 2 AR down-regulation (Gly16- Glu27), indicating that receptor function does not undergo desensitisation, did benefit from b-blocker therapy. Conversely, patients with variants enhancing down-regulation (Arg16-Gln27) did not benefit from b-blockers, most likely because receptor density is lower at the cell surface, which mimics bar antagonist activity. In fact, the administration of b-blockers to such patients appeared to unmask negative effects as suggested by the increased mortality rate in comparison with non-treated patients (not on b-blockers). Pending replication in a larger cohort with a standardised treatment, this study provides convincing evidence that genetic variability of the b 2 AR has direct clinical relevance for b-blockers therapy. 174 Anaesthesia ª 2011 The Association of Anaesthetists of Great Britain and Ireland

11 Landau et al. Pharmacogenetics and anaesthesia Anaesthesia 2012, 67, Taken together, all these individual genetic association studies do not provide sufficiently strong evidence to change the way b-blockers are prescribed in a chronic or acute perioperative setting. Even if a handful of studies have suggested that individuals with a certain genotype haplotype do not respond optimally to b-blocker therapy, it is unlikely that b- blocker therapy will be withheld before additional randomised-controlled trials confirm these findings. If these initial findings are confirmed, then the subset of patients with specific genotypes that may at best not respond (Gly349 carriers of ADRB1) or actually even do worse (Arg16-Gln27 of ADRB2) will be identified before prescription of b-blocker therapy. The future of personalised medicine Perhaps the most exciting yet challenging development of personalised medicine emerged with the highly sophisticated technology that now allows whole genome sequencing at a cost that is no longer prohibitive. Considerable consideration is needed to decide how best to apply information gathered through whole genome sequencing to clinical practice [166]. Among these challenges, careful and comprehensive genetic counselling will be needed so that patients can decide whether they wish to undergo such genetic risk assessment. In addition, clinicians will need to identify effective ways of communicating meaningful information to patients about the many implications of their whole-genome sequences, which will have to include follow-up strategies when novel information is gathered that may change a specific risk assessment. For clinicians, databases with easy to access and well-validated information about genomic sequences and diseases need to be created, maintained and frequently updated to incorporate new findings for each disease risk and pharmacogenomic tests. The first integrated analysis of a complete human genome in a clinical context has been published, describing a 40-year-old male with a family history of coronary artery disease and sudden death, and addressing these issues in a fascinating article [167]. Disease and risk analysis of the genome for this individual study was focused on variants associated with genes for known Mendelian disease, novel mutations, variants known to have a pharmacogenetic effect, and SNPs previously associated with complex disease. The subject was found to have an increased genetic risk for myocardial infarction, type II diabetes and certain cancers. The patient was carrying the CYP2C19 variant with a loss of function, important for the metabolism of many drugs including clopidogrel. The rate of cardiovascular events is higher among patients taking clopidogrel with such a variant of CYP2C19, and therefore a higher dose of clopidogrel was recommended in the event of future use or consideration of newer agents with alternative metabolism. He also had three distinct genetic variations that suggested a lower maintenance dose of warfarin. He carried the single most important VKORC1 variant associated with a lower maintenance dose and was homozygous for a CYP4F2 SNP also associated with lower dosing, and a novel non-synonymous SNP in VKORC1 was found. Thus, warfarin loading, if ever indicated, could be managed in an individualised manner for this patient with lower expected doses. The patient has several variants associated with a good response to statins, including lower risk for myopathy, and one variant suggesting that he may need a higher dose to achieve a good response. With this report, the authors developed tools to integrate the subject s clinical characteristics, his family history and the results from whole genome sequencing including 2.6 million SNPs and 752 copy number variations to assist clinical decision-making. Large-scale implementation of such sophisticated methodology will require multidisciplinary approaches that include medical and genetic professionals, ethicists and regulatory agencies. Conclusions To improve peri-operative outcomes based on pharmacogenetic testing, one can foresee two medical areas that may soon provide some recommendations: cardiovascular and pain medicine. Pharmacogenetics of b-blockers and genetic testing to guide the dosing of clopidogrel could help tailor prescriptions with the aim of improving survival rates after acute coronary events. At the same time, algorithms that predict the dose of warfarin to prevent devastating haemorrhage as well as thrombotic events in the perioperative period are already available, and will, one hopes, be implemented with measured outcomes. As for pain therapies, genotyping of CYP isozymes to determine the pharmacokinetic profile and interactions of drugs with potentially devastating outcomes, such as those which occur with opioids relying on CYP2D6 and CYP3A4 for their metabolism, is likely to become strongly recommended. On the other hand, the pharmacogenetics of morphine seems more complex, and although there was some hope that simple genetic testing would predict the morphine dose required to provide optimal analgesia, this is unlikely to occur. Different polymorphisms seem to influence pain perception in numerous ways, and the response to analgesics differs depending on the pain modality, the opioid prescribed and even its route of administration. Interestingly, hurdles in establishing solid evidence to recommend the use of pharmacogenetic testing before drug prescription have been related to the lack of large wellconducted randomised studies with well-identified clinical Anaesthesia ª 2011 The Association of Anaesthetists of Great Britain and Ireland 175