Shashi Amur; Issam Zineh; Darrell R Abernethy; Shiew-Mei Huang; Lawrence J Lesko
Abstract and Introduction
Adverse drug reactions (ADRs) observed during drug development have been the cause for discontinuing development of many drugs. In addition, serious but rare ADRs observed after marketing have led to withdrawal of some drugs. A prioriidentification of individuals at risk of developing ADRs for a given drug will help develop strategies to reduce the risk for ADRs in these patients. US FDA initiatives and efforts at reducing ADRs to make drugs safer are described, including updating of drug labels to include genomic information intended to reduce ADRs. Pharmacogenomics can also be harnessed to identify individuals at risk of developing serious ADRs and to treat these individuals with alternative therapy, thus converting ADRs that are traditionally considered unavoidable to avoidable ADRs.
Adverse drug reactions (ADRs) have been reported to be the cause for withdrawal after marketing of 3.0–3.5% of the new molecular entities (NMEs). In addition, 2.4–12% of hospital admissions and 4.6% of deaths in hospitalized patients in the USA are attributed to ADRs,[2,3] and they have been reported to be the fourth leading cause of death. The costs associated with ADRs may exceed US$177 billion annually in the USA alone. This article will review the potential role of pharmacogenomics in reducing ADRs at the premarketing and postmarketing stages of drug development and will provide a perspective based on experiences at the US FDA.
Drug Development & ADRs
To address safety concerns during drug development, studies in animals are conducted beforehand and after the drug candidate is advanced to administration to humans. Clinical trials (Phases I, II and III) are conducted with several hundreds to thousands of participants. The results are evaluated to determine efficacy for the therapeutic indication being sought and safety in the context of achieving an acceptable benefit–risk ratio. The International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH) currently lists over a dozen guidelines governing the conduct of nonclinical and clinical safety assessments in drug development. Generally, two large, randomized, controlled Phase III clinical trials with prespecified statistical plans are conducted for demonstration of efficacy. However, even when a drug has demonstrated its benefit, as many as a third of all drugs fail during the clinical portion of drug development owing to major ADRs.[5,6]
Serious ADRs are sometimes not detected until after drug approval. Several factors are implicated for the inability to detect ADRs prior to marketing of a drug. They include:
- The exposure of a limited number (generally between 500–3000) of patients during clinical development and for a limited time;
- The randomized controlled trials may control for disease variability, but generally not for variability in individual response rates or adverse events that are difficult to anticipate;
- Many serious ADRs are infrequent or rare (<1 in 1000 in most cases and occasionally <1 in 10,000) and may not be detected against a background rate in the population receiving a placebo or active control.
To identify an ADR that occurs in one in 10,000 patients, at least 30,000 patients would need to be enrolled in the randomized clinical trial program to have any chance of detecting it. Thus, it is not possible to fully assess the safety of a NME until it has been taken by a large number of patients after marketing. As such, regulatory approval of a drug is based on an acceptable benefit–risk balance in the indicated population under conditions of use specified in the drug label.
Regulatory Agencies & Drug Safety
Patient safety is a top priority for all regulatory agencies and for healthcare professionals. To address the risk associated with drug exposure, the FDA works to reduce the risks of drugs to patients in the following ways:
- When the benefit–risk balance is determined to be unacceptable, the drug is not approved for marketing;
- At the time of approval, the FDA works with pharmaceutical companies to provide relevant efficacy and safety information in drug labels. Adverse events observed in clinical trials are described along with the information on management of risk wherever possible;
- The agency stipulates postmarketing requirements and postmarketing commitments to pharmaceutical companies in order to gain additional information about the drug product to address safety concerns of the drug;
- With the passage of FDA Amendments Act (FDAAA) of 2007, the FDA is now empowered to require pharmaceutical companies to carry out postmarketing research to assess signals of serious risk related to the drug or to identify an unexpected serious risk when available data indicates potential for a serious risk. The FDA also has authority to ask pharmaceutical companies to conduct additional studies if reports of ADRs become available in the postmarketing phase. Moreover, the FDA may require a risk evaluation and mitigation strategy (REMS), as specified in FDAAA, to help manage drug-associated risks;
- The Office of Surveillance and Epidemiology in the Center for Drug Evaluation and Research (CDER) examines reports of postmarketing reports of the adverse events;
- Several FDA initiatives are aimed at drug safety. Examples include the Safety First Initiative to strengthen and modernize CDER safety-related policies and procedures; the Safe Use initiative to partner with the healthcare delivery systems to ensure medicines are used safely; and appropriately, and the Sentinel Initiative to create an integrated electronic system to monitor medical product safety.
Challenges in Reducing ADR Risk
Understanding the mechanisms underpinning ADRs may help mitigate risk by identifying individuals at risk of developing ADRs for a given drug and to provide alternate treatment with a more desirable benefit–risk balance for those individuals. Multiple triggers that include intrinsic factors (e.g., genetics and race) and extrinsic factors (e.g., environment and co-medications) are responsible for the individuals' predisposition to ADRs. Even when Phase III clinical trials are large and have several hundred to thousands of patients enrolled, genetically diverse populations may not be studied in these trials. Also, knowledge of all the complex factors that affect drug safety is not always available at the premarketing stage. However, an understanding of this genetic diversity has been successfully employed in identifying individuals likely to experience ADRs in some cases and thus holds promise to improve drug safety.
Avoidable & Unavoidable Adverse Events
Adverse drug reactions can be grouped into 'avoidable' and 'unavoidable' categories. Avoidable ADRs are caused by product defects (e.g., particulate matter in intravenous fluids, color change of medicine), medication errors (e.g., prescribing and dispensing), and differences in drug exposure (pharmacokinetics) or by known factors related to the mechanism of action of the drugs (Figure 1). High exposure may also occur in patients receiving approved doses of the drug due factors such as genetic variations in drug-metabolizing enzymes (DMEs) or drug-transporter proteins (DTPs), or as a result of nongenetic factors (e.g., concomitant medications, dietary supplements and noncompliance). The ADRs due to pharmacokinetics represent a substantial proportion of all adverse events, with one report attributing 59% of the ADRs to pharmacokinetics and another as high as 90%.[9,10] Understanding of the factors responsible for the high exposure leads to determination of the appropriate dose for individuals identified as being at risk of developing the ADRs and taking action to decrease the probability of ADRs.
Figure 1. Avoidable and unavoidable adverse drug reactions. ADRs can be classified as avoidable and unavoidable. Some of the avoidable errors are caused by product defects and medication errors. Knowledge of the PK and PD mechanisms associated with the drug could lead to an understanding of the role of genetic polymorphisms in the DMEs and transporter protein(s) in the PK variations and in the ADRs. In conjunction with nongenetic factors, such as age, this information can be used to determine the appropriate dose needed by the patient. Unavoidable ADRs can be converted to avoidable ADRs either through understanding the mechanism of action of the drug or through findings of strong genetic associations to adverse events (for example, abacavir hypersensitivity and HLA-B*5701) which can be used in patient screening and selection.
ADR: Adverse drug reaction; DME: Drug-metabolizing enzyme; DTP: Drug-transporter protein; PD: Pharmacodynamic; PK: Pharmacokinetic.
It should be noted that off-target binding of the drug could lead to ADRs and may not be related to the same level of exposure that is responsible for on-target efficacy. Thus, an understanding of pharmacodynamics in these situations may assist in avoiding ADRs. Unavoidable ADRs are the idiosyncratic reactions for which the underlying mechanisms are not understood. These ADRs, which have been reported to contribute to 10% of the ADRs, may be acceptable risks in order to achieve efficacy of the drug in some life-threatening disease areas such as oncology. These unavoidable ADRs may become avoidable after gaining insight into the mechanisms underlying the ADRs or be considered acceptable based upon benefit–risk considerations. Strong genetic associations to ADRs can also lead to identification of individuals at risk of ADR even though the exact mechanisms are not understood. In these cases, the frequency of adverse events can be reduced by patient screening and selection.
Increased drug exposure is an important cause for toxicity and adverse events. If ADR(s) are observed at higher doses of a drug, patients with a deficiency or low expression of the drug-metabolizing enzyme(s) and/or transporter(s) that affect the disposition of the drug are more likely to have higher drug exposures and thus, are at a greater risk of developing toxicity. Pharmacokinetic studies can define individual drug exposure. However, it is not easy to identify patients at risk of developing adverse events due to high exposure before starting the treatment. Currently, dose adjustment is based on available information on patient factors such as body weight, organ dysfunctions and concomitant medications. In a few cases, therapeutic drug monitoring is employed. If patients at risk of excessive (or insufficient) exposure could be identified a priori, this could be of considerable use to optimize and individualize therapy. We will describe examples in which application of pharmacogenomics can provide important information to predict drug exposure in individual patients.
Genetic variations in DMEs and DTPs are reflected in the phenotypes (enzyme or transporter activities) that contribute to differences in drug metabolism and to the dynamics of drug–drug interactions. Thus, determining patient genotypes could identify individuals at risk of developing ADRs, provided the drug disposition pathway and the genotypes that predict higher drug exposure and associated ADRs are well-characterized. In addition, genetic variation in genes coding for enzymes, receptors, or other effector mechanisms involved in the mechanism of action of the drug can also influence the adverse events associated with the drug, although this area is not as well understood as variation in genes that code for DMEs. Understanding of the joint contribution of genetic variations in the metabolism, transport and mechanism of action of the drugs as well as nongenetic factors contributing to ADRs may be used to reduce their occurrence.
Pharmacogenomics: ADRs & Pharmacokinetics
Contribution of genetics to variability in drug metabolism is now well recognized. Interactions of environmental factors (e.g., smoking and diet) with pharmacogenetics of DMEs and DTPs are known to determine drug exposure in individuals. Approximately 60% of adverse events are believed to be associated with drugs metabolized by polymorphic phase 1 DMEs. Four major phenotypes can be typically identified by phenotyping or genotyping: poor metabolizer (PM), intermediate metabolizer (IM), extensive metabolizer (EM) and ultrarapid metabolizer (UM). For active drugs (where the administered drug has the pharmacological effect), PMs that have either no enzyme activity or reduced activity would be predicted to accumulate high drug concentrations due to decreased metabolism, relative to EMs receiving the same dose. The UMs, in contrast, are likely to have a less than expected response. However, for prodrugs (where the administered drug has little or no pharmacological activity and is metabolized in vivo to form an active metabolite as in the case of codeine metabolism to morphine), the UMs are expected to have high active metabolite exposure, and thus, more likely to develop ADRs than other phenotypes.
The majority (86%) of DMEs belong to the CYP gene superfamily.[9,15] So far, 107 CYP genes have been identified, 59 active genes and 48 pseudogenes. Polymorphisms in CYP2C9, CYP2C19 and CYP2D6 have been shown to cause clinically significant differences in exposure to several drugs. The FDA has evaluated the ADRs associated with the PMs in these enzymes in the context of several drugs, and has provided recommendations in the drug labels. Some clinically relevant examples are given below.
Two variants in CYP2C9 (CYP2C9*2 and *3) are known to be associated with reduced activity of the enzyme and patients who are homozygous or heterozygous cannot metabolize warfarin to the same degree as the wild-type CYP2C9 (*1/*1).Variation in the pharmacokinetics between patients makes it difficult to identify the appropriate starting dose and this variability has been linked to genetic variants of CYP2C9. These variants are also associated with reduced warfarin maintenance doses, longer time needed to achieve a stable dose, on average, and with a higher risk of serious bleeds.Recently, a pharmacogenetic algorithm was shown to estimate the therapeutic steady-state warfarin dose more accurately than one using clinical factors and international normalized ratio (INR) response alone. The FDA has discussed the impact of CYP2C9 polymorphisms on dosing in the warfarin label.
Two variants, CYP2C19*2 and *3, account for the majority of CYP2C19 PMs. Some of the less frequent variants associated with PM status are CYP2C19*4, *5, *6, *7 and *8. Individuals with PM status possess two loss-of-function alleles. PM status is associated with diminished response to clopidogrel. Patients with reduced CYP2C19 function have lower systemic exposure to the active metabolite of clopidogrel, diminished antiplatelet response, and generally exhibit higher cardiovascular event rates following myocardial infraction than patients with normal CYP2C19 function do. The FDA label informs healthcare providers in a boxed warning that "poor metabolizers of CYP2C19 treated with recommended doses of clopidogrel exhibit higher cardiovascular event rates following acute coronary syndrome than patients with normal CYP2C19 function".
Based on the genotype, individuals could be UMs, PMs with little or no activity, reduced activity or with normal activity.Individuals who are UMs due to a specific CYP2D6*2×2 genotype, convert codeine into its active metabolite, morphine more rapidly and completely than other people. This rapid conversion results in higher than expected serum morphine levels. In some instances, nursing mothers who are UMs of codeine achieve higher-than-expected serum levels of codeine's active metabolite, morphine, leading to higher-than-expected levels of morphine in breast milk, and potentially dangerously high serum morphine levels in their breastfed infants. Therefore, maternal use of codeine can potentially lead to serious ADRs, including death, in nursing infants. The FDA label for codeine carries the safety information in the label: "Advise patients that some people have a genetic variation that results in their liver changing codeine into morphine more rapidly and completely than other people. These people are more likely to have higher-than-normal levels of morphine in their blood after taking codeine, which can result in overdose symptoms such as extreme sleepiness, confusion or shallow breathing. In most cases, it is unknown if someone is an ultra-rapid codeine metabolizer. Nursing mothers taking codeine can also have higher morphine levels in their breast milk if they are ultra-rapid metabolizers. These higher levels of morphine in breast milk may lead to life-threatening or fatal side effects in nursing babies".
Polymorphisms in phase 2 DMEs are also known to influence the metabolism, enzyme activities and exposure-associated adverse events. An example is that of the association of a genetic polymorphism in the promoter region of UGT1A1*28 with irinotecan-associated toxicity. The wild-type allele (UGT1A1*1) has six TA repeats in the promoter region while the variant allele (UGT1A1*28) has seven TA repeats. Irinotecan is metabolized in vivo by carboxylesterases to the active metabolite, SN-38 that binds and blocks the topoisomerase I activity. The metabolite is eliminated by glucuronidation to SN-28 glucuronide, mediated primarily by UGT1A1. Patients homozygous for the UGT1A1*28 allele glucuronidate SN-38 less efficiently than patients who have one or two wild-type alleles (UGT1A1*1). Thus, the PMs are associated with higher plasma concentrations of SN-38 and are more likely to develop irinotecan toxicities, such as neutropenia. The label for irinotecan was updated in 2005 to add that UGT1A1*28 genotype is a risk factor for the development of severe neutropenia following initiation of irinotecan treatment. The label recommends that a reduced initial dose should be considered for patients known to be homozygous for the UGT1A*28 allele: "Individuals who are homozygous for the UGT1A1*28 allele are at increased risk for neutropenia following initiation of irinotecan treatment. A reduced initial dose should be considered for patients known to be homozygous for the UGT1A1*28. Heterozygous patients (carriers of one variant allele and one wild-type allele, which results in intermediate UGT1A1 activity) may be at increased risk for neutropenia; however, clinical results have been variable and such patients have been shown to tolerate normal starting doses".
Polymorphisms in transporters are also known to influence ADRs. The association of rs4149056 in the SLCO1B1 gene with simvastatin-induced myopathy has been described.[22,23] Among participants taking simvastatin 80 mg/day, CC homozygotes had a cumulative risk for myopathy of 18%, the CT genotype of 3% and 0.6% in TT homozygotes, and more than 60% of the myopathy cases could be attributable to the C variant in rs4149056 in the SLCO1B1 gene. Owing to recent advances in the sciences and technologies in the evaluation of transporters in drug disposition,[24–27] including studies in transporter genetics, understanding of the role of transporters in drug's safety and effectiveness will continue to improve.
Pharmacogenomics: ADRs & Pharmacodynamics
Genetic variation in proteins involved in the pharmacodynamics of the drug could contribute to ADRs. An example is that of warfarin, where genetic variation in VKORC1, the target of warfarin's mechanism of action, explains approximately 30% of the variability in warfarin dose. Since variations in CYP2C9 also influence variability in the maintenance warfarin dose, the warfarin label (revised in January 2010) provides recommendations on CYP2C9 and VKORC1 genotype-guided dose selection. Recently, a prospective study carried out by Medco and Mayo clinic demonstrated that dose modifications based on genetic testing for CYP2C9 and VKORC1 variants decreased hospitalization rates by approximately a third.However, opinions on whether pharmacogenetic-based dosing is ready for application in clinical practice currently differ among clinicians. Some believe that genetic testing should not be used to determine the starting dose of warfarin until data from randomized controlled trials support the testing and cost–effectiveness of the tests are determined.
Unavoidable ADRs: Can We Make Unavoidable ADRs Avoidable?
An adverse event is usually considered unavoidable when the risk factors leading to the ADR in the context of a particular drug are not known in advance. Establishing the risk factors associated with the ADR that leads to identifying individuals at risk can make an unavoidable ADR avoidable. For example, a hypersensitivity reaction was observed in 5% of the patients treated with abacavir in clinical trials. As a part of the postmarketing commitments, GlaxoSmithKline (GSK) carried out exploratory studies to identify the mechanism underlying abacavir hypersensitivity reaction (ABC-HSR). Initial studies by GSK and a report from Australia indicated association of HLA-B*5701 with ABC-HSR.[31,32] This association was confirmed by the large-scale Prospective Randomized Evaluation of DNA Screening in a Clinical Trial (PREDICT-1) study, which showed that clinically suspected ABC-HSR cases could be reduced from 7.8 to 3.4% by excluding HLA-B*5701-positive patients from abacavir treatment. Using a research tool, patch test, in addition to a clinical diagnosis of ABC-HSR, the immunology confirmed ABC-HSR cases were reduced from 2.7 to 0%. A decrease in ABC-HSR has since been reported in various parts of the world with the exclusion of HLA-B*5701 positive patients from abacavir treatment. The FDA updated the abacavir label in 2008 and recommended screening patients for HLA-B*5701 before prescribing the drug: "Patients who carry the HLA-B*5701 allele are at high risk for experiencing a hypersensitivity reaction to abacavir. Prior to initiating therapy with abacavir, screening for the HLA-B*5701 allele is recommended; this approach has been found to decrease the risk of hypersensitivity reaction".
Many ADRs are complex and multiple factors may be involved especially in different subsets of the population. In the case of abacavir, there are some HLA-B*5701-positive patients who are tolerant to the drug, suggesting that factors other than HLA-B*5701 are also needed for the ADR to occur. Conversely, when HLA-B*5701-positive patients are excluded from treatment, a few cases of ABC-HSR still occur. This information, even if it is incomplete, is extremely valuable since the test identifies the majority of patients at risk of developing ABC-HSR before prescribing abacavir. Depending on racial background, up to 10% of people are HLA-B*5701-positive and these individuals can be treated with an equivalent therapeutic alternative. This test is now widely used and is an example of successful application of genomics to avoid this ADR.
Some ADRs are life threatening and there is an urgent need for tests that identify patients at risk of developing these serious ADRs. To determine whether genetic factors are associated with these ADRs, candidate gene studies or genome-wide association analyses can pave the way to developing genetic tests that can help in converting unavoidable ADRs to avoidable ADRs. An example is that of carbamazepine (CBZ)-induced Stevens–Johnson syndrome (SJS)/toxic epidermal necrosis (TEN). The estimated risk for CBZ-associated SJS/TEN is 1.4 in 10,000 patients. HLA-B*1502 was found to have a strong association to CBZ-induced SJS/TEN in Han Chinese patients, where 44 out of 44 (100%) of the CBZ-induced SJS/TEN patients were found to be positive for the allele, whereas only 3% of the tolerant controls and 8.6% of the healthy controls carried the allele. This finding has been corroborated in other studies in Asians, particularly of Han Chinese ancestry.
A Perspective on Drug Labels
A multipronged surveillance and mechanistic approach is probably the best way to address complex problems such as ADRs. Attempts should be made to address avoidable ADRs during drug development. Rare but serious adverse events may be identified in Phase III clinical trials and occasionally in postmarketing phases. Some possible approaches to address unavoidable and rare adverse events are delineated below:
- The DME(s) and transporter(s) involved in the metabolism and transport of the NME should be identified in vitro and in vivo. If ADRs are observed during drug development, and the DME(s) and/or transporter(s) are polymorphic, evaluation of the association of the polymorphism(s) with safety biomarkers should be considered, especially if an exposure–response relationship exists;
- If ADRs appear to be unrelated to the pharmacokinetics of the drug, they could be related to its pharmacodynamics. Research into the pharmacodynamics of the drug, as in the case of warfarin, has led to identification of contribution of the VKORC1 polymorphism to dose selection. Thus, exploratory studies such as candidate gene studies, on the genes/proteins involved in the pharmacodynamics related to extensions of intended pharmacology or discrete adverse events could yield valuable information;
- If ADRs observed cannot be explained by the above strategies, another approach is to carry out genome-wide association studies that are unbiased and can generate a hypothesis, for example, about liver toxicity, based on the results. This approach might identify novel gene associations that need to be confirmed with additional studies;
- Tracking of ADRs is extremely important to identify safety signals as early as possible, particularly for those leading to rare adverse events. Ideally, an ADR database should have complete demographic and clinical information that can help in interpreting these safety signals to identify important event triggers. An example is that of the Organ Procurement and Transplantation Network (OPTN) database launched in 1999. This database requires submission of data in order to participate in the program. In addition to baseline data, the database also contains transplant recipient follow-up information on patient status, clinical and treatment information that includes adverse events, since post-transplant lymphoproliferative disorder is a rare ADR observed with transplant drugs, a separate post-malignancy form is generated after a malignancy has been reported in the follow-up;
- Identification and qualification of novel genomic/protein biomarkers associated with toxicities, such as hepatotoxicity and kidney toxicity, which may help to improve preclinical and clinical safety evaluation in clinical trials and lead to development of safer drugs.[39,40]
Drug development & adverse drug reactions
- Adverse drug reactions (ADRs) are reported to be the fourth leading cause of death.
- Approximately a third of the drugs fail in clinical trials during drug development owing to major ADRs.
- The US FDA works to reduce risk of ADRs in patients.
Potential role of pharmacogenomics in reducing ADRs
- Avoidable ADRs:
- Approximately 60% of ADRs are believed to be associated with drugs metabolized by polymorphic phase 1 drug-metabolizing enzymes.
- Knowledge of genetic polymorphisms in drug-metabolizing enzymes (DMEs) or drug-transporter proteins that predict higher drug exposure associated with ADRs, can help identify individuals at risk of developing ADRs.
- An understanding of the role of genetic variation in proteins involved in the pharmacodynamics of drugs may also lead to reducing ADRs.
- Unavoidable ADRs:
- ADRs are considered unavoidable when the risk factors contributing to the ADR are not known a priori.
- Establishing risk factors (e.g., genetic factors) that identify individuals at risk of developing ADRs can make unavoidable ADRs avoidable.
Pharmacogenomic information in drug labels
- In the labels of some FDA-approved drugs, genomic-based recommendations for dose selection or for not treating test-positive patients are provided.
- Mechanisms and selection criteria to update labels at the postmarketing stage with genomic information used at the FDA are described.
- Exploratory studies with new molecular entities should help address avoidable ADRs during drug development:
- Identification of the DME(s) and transporters in the metabolism of the new molecular entities. If ADRs are observed and if the DME(s) and/or transporter(s) are polymorphic, association with the ADR should be evaluated.
- If the ADRs are unrelated to pharmacokinetics, exploratory studies on the genes/proteins involved in the pharmacodynamics may yield valuable information.
- Novel technologies such as candidate gene studies and genome-wide association studies may be used for hypothesis generation.