adverse drug reactionsgenome wide association studiespersonalized medicinepharmacogenomics;
A medication with proven efficacy and safety in a series of rigorous clinical trials could still fail to work in some patients or cause serious adverse drug reactions (ADRs).1, 2 The interindividual responses to drugs are most likely affected by genetic variations, which can be divided into two types: (1) inherited variants (i.e., germ-line genetic variants); and (2) acquired variants (i.e., somatic mutation). Germ-line variants of genes encoding drug-metabolizing enzymes, drug transporters, drug targets, and human-leukocyte antigen (HLA) can affect individual response to medications. Somatic variants of genes are frequently associated with the development or progression of cancer, and affect the drug response of tumors that carry specific mutations, so called target therapy. Because of the impact of genetic variants on medication responses, how to give the “right drug” at the “right dose” for the “right patient” is a major goal in the era of precision medicine.3, 4
Pharmacogenomics is the application of current technology for the precise determination of genetic variants that influence drug response, and to develop personalized strategies that maximize therapeutic efficacy and assure drug safety. Large-scale genome-wide association studies and smaller-scale studies with a candidate-gene approach, that are used to study genes involved in drug metabolism enzymes, drug transporters, drug receptors, and HLA, have helped advance our understanding of the underlying mechanisms of ADRs and drug efficacy (Figure 1). Next-generation whole-genome sequencing, which provides a diverse genome map of multiple populations, will be useful for the future of pharmacogenetic studies.5 These studies take into account ancestral genetic structure, complex haplotypes, gene–gene interactions, and rare variants. They aim to detect and replicate novel pharmacogenetic loci of clinical significance. Based on these studies, the United States Food and Drug Administrations (FDA) have relabeled over 100 approved drugs with genetic information. A list of valid genomic biomarkers for clinical guidance can be found on the FDA website “Table of Pharmacogenomic Biomarkers in Drug Labels” (http://www.fda.gov/Drugs/ScienceResearch/ResearchAreas/Pharmacogenetics/ucm083378.htm). Some of these biomarkers have been implemented in medical practice.6, 7, 8
Download full-size image
3. Pain medication
Most clinical pain management options involve pharmacological interventions. Pain therapy has evolved over the years into a large specialty field. Ideal pain management approaches must provide adequate analgesia without excessive adverse effects. However, there are large interindividual differences in response to pain medications, concerning efficacy and the development of severe ADRs.9, 10, 11, 12, 13 Current pain management strategies are devised using the World Health Organization pain ladder, which begins with nonopioid medications, such as nonsteroidal anti-inflammatory drugs (NSAIDs), progressing to weak opioids, and culminating with strong opioids.9 Other pain medications include anticonvulsant drugs for neuralgia. Additionally, adjuvant therapies using antidepressant medications can aid in reducing chronic pain-associated anxiety.14, 15 In Taiwan, based on the National Health Insurance Research Database, we found the drugs that are most frequently used for pain management (Table 1, Table 2).
Download full-size image
Download full-size image
The field of pain medications is a valuable one to study germ-line variants using a pharmacogenomic approach, because of the considerable repercussions of these medications on the biological, psychological, sociological, and economical welfare of patients. Genetic studies have identified several polymorphic loci that govern the pharmacodynamics and kinetics of analgesic drugs.16, 17, 18 The aim of this review is to highlight recent advances in the pharmacogenomics of pain medicine.
3.1. Drug metabolism enzymes
The metabolism of drugs and other xenobiotics is often divided into multiple phases. Phase I enzymes are responsible for chemical modifications of the drugs. They include cytochrome P450 (CYP), cytochrome b5, and nicotinamide adenine dinucleotide phosphate-cytochrome P450 reductase. Phase II enzymes are involved in further conjugation of the active drug metabolites, some examples are glutathione S-transferases, aryl sulfatase, and uridine-glucuronosyltransferase. The hepatic CYPs are a multigene family of enzymes that play a critical role in the metabolism of many drugs, with each cytochrome isozyme displaying unique substrate specificities and susceptibility to induction and inhibition by exogenous chemicals. One of the most common CYPs involved in drug metabolism is cytochrome P450, family 2, subfamily D, polypeptide 6 (CYP2D6), whose metabolic rate can fluctuate by over 100-fold between the allelic variants expressed in different ethnic groups.19 For example, approximately 10% of the Caucasian population carries an autosomal recessive trait that yields a nonfunctional variant.20 Patients homozygous for this variant, known as poor metabolizers (PM), may have either a higher risk of adverse side effects because of drug overdose (i.e., tricyclic antidepressants or antiarrhythmics) or no drug efficacy due to poor transformation of the prodrug into its active metabolite. Conversely, in ultra-rapid metabolizers (UM), amplification of the CYP2D6 gene is correlated with enhanced enzyme activity and ultrarapid metabolism, which could decrease the optimal dosage for UM (Figure 1).21, 22, 23
3.2. Opioid analgesics-codeine: CYP2D6
Opioids are the pillar of pharmacological pain management.24 They have been used in therapeutics for centuries, and there are > 30 opioids currently marketed worldwide. However, opioid over dosage is highly correlated with respiratory depression, a harmful and life-threatening adverse effect.25 Careful therapeutic drug monitoring of the plasma concentrations can help provide effective pain relief while minimizing adverse effects. Morphine and diamorphine have been shown to have a wider therapeutic range or “safety margin” than some other opioids. Since it is not possible to tell which patients will require low or high doses, commencement dose is always low, unless it is a therapy change from another strong opioid.26 Thus, opioid treatment requires an individualized approach to ensure safety. Ample studies have demonstrated that this individualized treatment might rely on pharmacogenomics. The future promise of pharmacogenetics involves a rational drug regimen that maximizes efficacy and minimizes adverse events. However, variation in responses to pain medication is partially explained or predicted by patient's genetics, and pharmacogenetic testing for opioids is not widely applied in current clinical practice.
Codeine is widely administered after operations and is used in certain drug combinations for acute and chronic pain management.27, 28 Because codeine is a less potent μ-opioid receptor agonist than morphine, it has been classified as a weak opioid.29, 30 Codeine is a prodrug with a low affinity for the opioid receptor and low intrinsic activity at the μ-opioid receptor. The initial metabolism of codeine is through glucuronidation with O-demethylation,32resulting in morphine formation. The O-demethylation of codeine to morphine is mediated by CYP2D6.33 UMs (CYP2D6*1/*1 and *1/*2) should avoid codeine due to its potential for toxicity (Table 3). The Clinical Pharmacogenetics Implementation Consortium guidelines have provided information for guiding the dosing of codeine based on CYP2D6 genotypes.34It is suggested to consider alternative analgesics for UMs such as morphine or a nonopioi
Download full-size image
3.3. Opioid analgesics: drug receptors and transporters
In addition to the association between CYP2D6 (metabolism) and opioid response, the major candidate genetic contributors to opioid efficacy and adverse effects are ABCB1 (drug transporter) and OPRM1 (receptor interaction).23, 26, 38, 39, 40, 41 The μ-opioid receptor (OPRM1) is the primary binding site of action for many opioid drugs and for binding of beta-endorphins. The μ-opioid receptor is a transmembrane protein which belongs to the rhodopsin family of G-protein coupled receptors. The downstream signaling of μ-opioid receptors is through interactions with heterotrimeric G proteins.42 To date, over 100 genetic polymorphisms have been found in the OPRM1 gene. The most common studied nonsynonymous single nucleotide polymorphism is rs1799971, which is located in exon 1, and is a nucleotide change from adenosine (A) to guanosine (G) (A118G). Depending on the ethnicity, the polymorphism can be found in a proportion from 2% to nearly 50%. The μ-opioid receptor polymorphism was also found to be associated with opioid effects43. The functional study of OPRM1 revealed that the G allele (A118G) creates a novel CpG-methylation site. This methylation would suppress the increase of the OPRM1 expression level that takes place after a prolonged opioid administration. Patients with an inactive OPRMI allele might benefit from a k-agonist such as buprenorphine instead of a μ-agonist such as morphine.
3.4. Opioid analgesics-methadone: CYP3A4, CYP2B6, and CYP2D6
Methadone is an old opioid drug, having been discovered and developed during World War II.44 It is an opioid with a completely different chemical structure to morphine and oxycodone. Currently, methadone (including R- and S-methadone) are used as second-line opioids for the treatment of patients suffering from cancer pain.45 Methadone is a useful alternative in individuals who have had no responses to other opioids or suffering from adverse side effects (ADRs).46, 47, 48 It has demonstrated its efficacy in neuropathic pain conditions,49 with several case reports highlighting this specific indication owing to methadone's N-methyl-D-aspartate-receptor agonistic properties.50, 51 Both methadone enantiomers (R- and S-methadone) bind to the noncompetitive site of the N-methyl-D-aspartate receptor. In addition, R-methadone proved to be a stronger μ-opioid receptor agonist than S-methadone. However, because the pharmacokinetics of methadone was still unclear, the methadone has a narrow therapeutic index.52Studies on the metabolism of methadone have mainly been carried out on cohorts of opioid addicts under methadone maintenance treatment. In vivo, CYP3A4 and CYP2B6 are the major CYP isoforms involved in methadone metabolism, with CYP2D6 contributing to a minor extent, preferentially in metabolism of the R-enantiomer (Table 3).53
3.5. NSAIDs as nonopioid analgesics: CYP2C9
NSAIDs are a group of nonopioid analgesics commonly used for the treatment of acute pain after surgery or chronic pain. Although multimodal pain management by NSAIDS and opioids is important to enhance the efficacy of treatment and reduce the adverse effects, some adverse events including gastrointestinal side effects and cardiovascular side effects have been reported after long-term treatment.54 It implies that an interindividual variation limits the clinical utility and safety of NSAIDS. Based on the previous reports, CYP2C9 polymorphisms might play a significant role in the analgesic efficacy and toxicity of traditional NSAIDs (Table 3), such as flurbiprofen and celecoxib (Table 3).54, 55 For NSAIDs used as analgesics, treatment should start at half the lowest recommended dose in poor metabolizers (CYP2C9*3/*3 genotype) to avoid adverse cardiovascular and gastrointestinal events (Table 3). For traditional NSAIDs or selective cyclooxygenase-2 inhibitors, few data are currently available on the possible impact of these single nucleotide polymorphisms on analgesic efficacy.56
3.6. Cyclic antidepressant drugs as co-analgesics: CYP2D6 and CYP2C19
Cyclic antidepressants tackle depression by regulating the levels of neurotransmitters in the brain.57, 58 Currently, a major issue of the treatment with tricyclic antidepressants is their narrow therapeutic index, which results in their replacement by other antidepressants such as selective serotonin reuptake inhibitors.59, 60 However, because of their significant efficacy for pain management, cyclic antidepressants are frequently used as “co-analgesics” in the treatment of chronic pain, especially neuropathic pain61; they are also used for migraine prophylaxis. Some of the most commonly prescribed co-analgesics include amitriptyline, desipramine, imipramine, nortriptyline, doxepin, clomipramine, and protriptyline. Maprotiline (a tetracyclic compound) and amoxapine (a tricyclic dibenzoxazepine) are newer compounds that have a slightly different structure and toxicological profile.
Cyclic antidepressants are activated in the liver, after hydroxylation by CYP2D6 and demethylation by CYP2C19.62 CYP2D6 PMs have higher plasma concentrations of cyclic antidepressants than extensive metabolizers.63 PMs with CYP2D6*3/*3 genotypes should have the dose reduced by 60% to avoid arrhythmia and myelosuppression (Table 3). Therefore, the PMs have a higher risk of ADRs. By contrast, carriers of the CYP2D6 gene duplications would have ultrarapid metabolism of cyclic antidepressants, which may result in lower drug concentrations and poorer therapeutic responses.64 Thus, genetic variants may affect plasma concentrations of antidepressants, and dose adjustments before treatment are recommended.
4. Pharmacogenomics for pain medicine: immune-mediated drug hypersensitivity
4.1. Antiepileptic drugs for neuralgia: HLA association of drug-induced cutaneous ADRs
Antiepileptic drugs (AEDs) typically are used to control epilepsy or to treat some psychiatric disorders, e.g., bipolar disorder. These drugs, particularly carbamazepine, trileptal, lamotrigine, gabapentin, and topiramate, may also be used to treat some painful conditions, such as postherpetic neuralgia and fibromyalgia. However, anticonvulsants are a common cause of cutaneous ADRs (cADRs).65 The spectrum of cADRs ranges from a mild maculo–papular exanthema to life-threatening severe cutaneous reactions (SCARs) including drug rash with eosinophilia and systemic signs, Stevens–Johnson syndrome (SJS), and toxic epidermal necrolysis (TEN). More than 90% of SCARs occur within 2 months of AED use.66 The AEDs with risk of SCARs include carbamazepine, phenobarbital, phenytoin, and lamotrigine.66 cADRs represent a significant burden on the healthcare system, and SCAR with its most severe form, TEN, still carries a mortality rate as high as 35%.67 These life-threatening conditions could not be predicted and prevented until the recent advancement of pharmacogenomics in discovering of a strong association between a HLA allele (HLA-B*15:02) and carbamazepine-induced SJS/TEN in Han Chinese residing in Taiwan; these results were subsequently replicated in different populations (Figure 2).13, 68, 69, 70, 71, 72 The strong genetic association suggests a direct involvement of HLA in the pathogenesis of drug hypersensitivity. It has been shown that the HLA molecule presents an antigenic drug and results in clonal expansion and activation of CD8+cytotoxic T cells. A pharmacogenomic study also identified an unusual form of granulysin secreted by these cytotoxic T lymphocytes and natural killer cells responsible for the rapid and disseminated keratinocyte death in SJS/TEN.73 The high sensitivity and specificity of genetic markers provides a plausible basis for developing tests to identify individuals at risk for drug hypersensitivity. A large prospective study has shown that HLA-B*15:02screening before carbamazepine treatment can effectively reduce the incidence of carbamazepine-induced SJS/TEN.74, 75 Translational research demonstrates that preventing drug toxicity by screening people at risk before prescription of a drug is a clinical reality.75 Application of HLA-B*15:02genotyping as a screening tool for high risk populations taking carbamazepine is now recommended by many regulatory agencies worldwide, including Taiwan, USA, Thailand, Hong-Kong, and Singapore.
Download full-size image
This genetic association, albeit weaker, has also been found with phenytoin and lamotrigine (Table 3).76, 77 In Caucasian and Japanese populations the risk of developing hypersensitivity reactions to carbamazepine was found to be associated with another allele, HLA-A*31:01.78, 79 In addition to HLA, carbamazepine-induced SJS/TEN, a common type of AED-induced SCARs, was found to be associated with a specific T-cell receptor (Figure 2) clonotypes,80which could be another supportive biomarker to enhance the positive predictive value.81
4.2. Clozapine: HLA association of clozapine-induced agranulocytosis
Clozapine is a second-generation antipsychotic agent that belongs to the benzodiazepine group. It is indicated for the treatment of resistant schizophrenia or schizoaffective disorder and other psychotic conditions. Although clozapine has several advantages over conventional antipsychotics, the risk of agranulocytosis restricts its use. Based on a retrospective study, clozapine-induced agranulocytosis is associated with the rare HLA-DQB1(126Q) and HLA-B (158T) alleles. The pharmacogenetic test using these two polymorphisms has a sensitivity of 36% and specificity of 89% (Table 3).82 In addition, clozapine FDA label recommends for some CYP2D6 genotype-carrying patients a lower dose of clozapine (Table 3). However, a prospective study demonstrating whether a pharmacogenetic test based on this genetic association could have clinical utility and prevent clozapine-induced agranulocytosis has not yet been done.
4.3. Moving pharmacogenomic findings into medical practice for pain therapy
Before developing pharmacogenetic tests, a trial-and-error approach for analgesics was used for decades in analgesic treatment. Available pharmacogenomics studies indicated that some of the genetic markers have potential to be used clinically to guide the physicians to choose a right analgesic drug with a right dose to a right person. Some of the examples are listed in These are based on well-defined genetic variants which have been replicated by others and have high predictive values (either high negative predictive value or high positive predictive value or both). These tests also have high sensitivity and specificity. Some of the tests have also been validated by a prospective clinical trial and already in routine medical practice; for example, HLA-B*15:02 screening to prevent carbamazepine-induced SJS.75Many of the tests also have clinical guidelines for dose adjustment and alternative medications assembled by The Clinical Pharmacogenomics Implementation Consortium (e.g., codeine and carbamazepine). Although there have been clear advances in clinical implementation of pharmacogenetics, some barriers slow down the speed of pharmacogenetic testing widespread.83 For example, long turnaround time of genetic tests, less value of genomic technologies after health economic evaluation, uncertain interpretation for the pharmacogenetic genetic test results, population restriction, and lack of prospective studies documenting superiority of the pharmacogenetic-guided treatment approach have to be overcome in the future of personalized medicine of pain.
5. Special consideration for pain medication: drug–drug interaction
“Drug–drug interaction,” defined as an interaction between two or more drugs, could alter the effectiveness, toxicity, or half-life of the interacting drugs. Moreover, it may increase the risk for serious ADR. Drug–drug interaction has been estimated to contribute to approximately 6–10% of ADRs.84 For pain management, more than one analgesic drug may be administered. In addition, for treating other diseases at the same time, such as coronary artery disease, hypertension, diabetes mellitus, and infections, multiple drugs may be also administered.85 The CYP450 enzyme pathway has been thought to be primarily involved in the metabolism of drugs, especially certain opioids used for the management of severe chronic pain. Therefore, the events underlying drug–drug interactions in pain management using multiple drugs could explain why pharmacogenetic tests for the CYP gene are helpful for identification of the optimal drug dose and prevention of ADRs. Furthermore, some information regarding drug–drug interactions could be referred to on drug labels before administration of analgesic drugs. For example, the interaction between quetiapine (for the relief of pain or depression) and ritonavir (for the treatment of human immunodeficiency virus infection) may lead to rapid and severe weight gain.86
6. Summary and future directions
Despite the pharmacogenomic studies that show that genetic variants may account for individual differences in drug responses, few genetic markers can be used clinically to guide the dosage for optimal efficacy of pain therapies.8There are several reasons for the slow progress in this regard. Firstly, the pain sensitivity varies for individual patients and underlying causes of pain are not well characterized. The heterogeneous conditions could interfere with the evaluation of drug efficacy. More precise phenotypes are needed.87 Secondly, the prior usage of opioid and tolerance may affect the drug response. Thirdly, multiple genes may affect drug pharmacokinetics and pharmacodynamics and there are also drug–drug interactions, which may complicate the genetic studies. Fourthly, because the drugs used in pain management alter multiple signaling pathways, and the efficacy of drugs could not be easily measured due to the lack of key indicators or outcome. Therefore, pain management with drugs such as morphine using a “titration” method is currently still widely used in the clinic to guide dosage decisions.88, 89 By contrast, clinically useful genetic variants for ADRs are easier to identify based on pharmacogenomic studies, because some ADR conditions have clear phenotypes (i.e., SJS or TEN), critical pathways (i.e., HLA–T-cell receptor; Table 3 and Figure 2),80 and specific indicators (i.e., granulysin).73
In conclusion, the clinically useful pharmacogenomic tests currently available for pain therapy are more direct toward at predicting drug toxicity or dose adjustment. More studies are needed to identify genetic variants that determine drug efficacy of pain medications. This area of research is likely rapidly accelerating with the reducing cost of next-generation sequencing and well-established biobanking system worldwide.
We gratefully acknowledge the members of the Translational Resource Center and the National Center for Genome Medicine at Academia Sinica for their support. This study was supported by the Academia Sinica Genomic Medicine Multicenter Study (40-05-GMM). The funders had no role in study design, data collection, or analysis, the decision to publish or preparation of the manuscript.