The complete description of the human genome was published in 2002.1,2 Although application of genomics to medicine was initially embraced as the potential future for drug discovery3 and management,4 it has proven to be more difficult than was originally thought.5-7 Nevertheless, scientists have continued to unravel the role that genetic variability may play in altering the response to drug therapy, including variations in drug disposition.8,9 As a result, the clinical application of genetic information to guide clinical decision making and drug therapy is emerging.10 Genetic variability in response to drugs occurs as a result of molecular alterations of drug-metabolizing enzymes, drug targets, and transport proteins.11 Such insights have led to application of genetics in a variety of scenarios, including treatment of breast cancer,12 stroke,13 coronary artery disease,14 perioperative adverse events,15 and development of adverse drug reactions (ADRs).16

Rather than the traditional “one size fits all” approach to drug therapy, the application of targeted drug therapy might allow a more personalized approach, thereby minimizing the occurrence of therapeutic failures or adverse effects.17,18 Pharmacogenomic research seeks to characterize any particular individual’s susceptibility to a disease or drug effect based on their known genetic composition.19 When possible, it could revolutionize how drugs are discovered - by identifying those individuals most likely to respond to the experimental therapy - and eventually prescribed.20 Such an approach has been shown to be cost-effective.21 Compared to traditional dose-driven therapeutic drug monitoring - where pharmacological effect is associated with a given drug concentration and cannot be determined prior to the drug being used - pharmacogenomics would not depend on steady-state pharmacokinetics or compliance with drug administration for interpretation. Furthermore, it might be less invasive, could provide mechanistic information and predictive value for a number of drugs (e.g., metabolism by CYP3A4/5), and is stable throughout a patient’s life-span.22 Table 1 lists factors necessary for pharmacogenomics to result in a clinically important change in drug effect.

Table 1 Factors necessary for pharmacogenomics to result in a clinically important change in drug effect
Full size table

Despite the forgoing, the application of genetic variation to predict altered responses to drug therapy in patients in the intensive care unit (ICU) has received relatively little study.23 In part, it may be because diagnostic tests for genomic variability, until recently, have not been available within the short time frame necessary for decision making in the ICU.17,24,25 In addition, alterations in drug pharmacokinetics due to changes in bioavailability, volume of distribution, and drug- and disease-induced changes in drug metabolism and clearance that occur because of the nature of critical illness26 make it difficult to attribute observed differences in drug effects to alterations in genomic variability alone. Herein, we expand on the observations of Allen and Gelot23 and use the concepts of pharmacogenetics (i.e., application of a single genetic variant to describe an alteration in drug effect) and pharmacogenomics (i.e., the broader application of an individual’s genome to predict his or her response to medications) to explain some of the variability in clinical response commonly seen in a patient population. Our objective is to provide examples of the potential for the clinically relevant application of pharmacogenetics and pharmacogenomics to conditions and drugs commonly encountered in the ICU.

Concepts and definitions

Pharmacogenetics is the study of differences among individuals with regard to the clinical response to a particular drug (i.e., “one drug, many genomes”), whereas pharmacogenomics is the study of differences among compounds with regard to the gene expression response in a single normative genome (i.e., “many drugs, one genome”).27 Alterations in the human genetic code are common. When they occur in more than 1% of the population, they are described as genetic polymorphisms.28 Table 2 provides a glossary of terms frequently encountered when reading the literature on genetic alterations in drug effects. For further understanding of commonly used genetic terms and concepts, the reader is referred to other reviews.29-33 Examples of polymorphisms include variations in skin, hair, and eye color. Much of the genetic variation that occurs among individuals occurs because of single variations in the nucleotide sequence of the four base pairs of deoxyribonucleic acid - called single nucleotide polymorphisms (SNPs).34 By examining alterations in SNPs, geneticists have been able to identify the genetic variations associated with a particular condition, such as susceptibility to an ADR.35

Table 2 Glossary of commonly encountered terms in the genetic literature
Full size table

Many factors affect response to drugs, including sex,36 age,37 critical illness,38 and concomitant drug therapy.39 Increasingly, however, it is recognized that genetic variation may play a significant role in the body’s handling of drugs.28,40 Differences in drug response identify groups of individuals with altered genetic profiles. Examples include familial cholinesterase deficiency and prolonged neuromuscular blockade following succinylcholine administration,28 peripheral neuropathy following isoniazide administration in patients with N-acetyltransferase deficiency,41 differential drug effects for treatment of hypertension in individuals of Black African origin,42 and alterations in p-glycoprotein function identified in twin studies (monozygotic vs dizygotic).43

Variation in genes encoding drug-metabolizing enzymes may lead to an absence of, reduction in, or increase in drug effect. For example, a high percentage of ADRs occur as a result of a genetic variation in one of the common cytochrome P (CYP)-450 drug-metabolizing enzymes.44 Individuals with two homozygous alleles coding for enzyme function are termed “normal” or “extensive metabolizers” (EMs) when more than one copy of the gene is present. Alternatively, individuals with two alleles resulting in inactive or absent enzyme function are termed “poor metabolizers” (PMs). Individuals with one functioning and one dysfunctional allele are termed “intermediate metabolizers” (IMs).28,45,46 In simplistic terms, individuals who are EMs have a normal response to a standard drug dose, whereas PMs would be unable to metabolize the drug to the same extent and are thus subject to a potentially excessive drug effect (including toxicity). Drug-metabolizing enzymes may also be responsible for conversion of prodrugs to their active moiety. Therefore, EMs in this scenario could potentially experience drug toxicity.47 This situation may be clinically relevant, however, only if other aspects that affect the variability of a particular drug’s response (e.g., age, sex) are minimal and the therapeutic index for the agent is narrow (e.g., warfarin). Furthermore, few drugs are metabolized by one enzyme system alone, and alterations in one metabolic pathway may not be reflected in an altered drug response if the other metabolic pathways can overcome the lack of metabolism by the deficient pathway.48

Likewise, variation in genes encoding drug transport proteins may alter absorption, distribution, and excretion of medications, resulting in an absence of, reduction in, or increase in drug effect. Patients with decreased expression of drug transport proteins in the duodenum may have increased drug bioavailability and therefore potentially be at risk of toxic effects.48 Decreased expression in the blood-brain barrier may increase drug access to the central nervous system sites of action.49

Variability in drug response can also be explained by genetic variability in drug targets. SNPs have been associated with genes that encode receptors, resulting in altered function, substrate binding affinity, expression, and both up- and down-regulation of receptors.48 Increased expression of receptors in tissues where the pharmacological effect is not desired can result in untoward effects. Conversely, down-regulation in response to agonism may result in lack of response, or tachyphylaxis.34,50

Ultimately, the consequences of variations in drug-metabolizing enzymes, transport proteins, and targets can be ineffective, even potentially toxic, drug therapy. This situation may lead to delay in the therapeutic response or ADRs. The latter may be a reason for the ICU admission itself,51 and either may be potentially devastating if experienced by the critically ill patient with already compromised physiological reserve. Later, we describe examples of genetic variation that have the potential to result in, or complicate, an ICU admission by categorizing the consequences of genetic variation as a resultant increase or decrease in drug effect.

There has recently been consensus on the nomenclature applied to pharmacogenomics. If adopted, it would help standardize the description of the altered genetic component.52 The following link provides comprehensive information on drug and genetic alterations that is updated regularly (https://www.pharmgkb.org/).53,54 Built on a knowledge pyramid, it provides information on a variety of relevant material, ranging from primary literature citations to clinical interpretation and implementation to dosing guidelines. Included also are interpretations and explanations of drug labels from the US Food and Drug Administration; European Medicines Agency; the Pharmaceuticals and Medical Devices Agency, Japan; and Health Canada (Santé Canada). There is cross-linking between clinically important pharmacogenomics, drug labels, and dosing guidelines. Under drug labels, lists of drugs are provided and categorized by (1) the genetic testing required or recommended, or (2) actionable or informative pharmacogenomics. For well-known pharmacogenomic associations, information is organized by gene/drug association and linked to dosing guidelines and drug label information. For example, phenytoin is listed in the table next to the three genes that affect its disposition (CYP2C19, CYP2C9, HLA-B), and for each of these genes there is a link to important pharmacogenomic information, which includes background on the function of the proteins encoded, gene frequencies, nomenclature, testing information, and drug and disease associations for the gene in question. Armed with this information, practitioners should be able to develop alternative drug strategies as necessary.

Application

Two main approaches have been employed for the clinical application of pharmacogenomics. The traditional approach has been to identify variability in drug response among individuals, confirm that the variation is not attributable to evident patient parameters (e.g., altered volume of distribution), and then seek an explanation in the SNPs in genes encoding proteins involved in the disposition of that particular drug (candidate gene approach).55 Once an SNP is found, gene frequencies - stratified by population - are necessary to determine the utility of testing for the allele in a given patient. If the frequency of the particular allele is determined to be substantial, an attempt is made to determine its clinical validity.56 Clinical validity is the genotype’s ability to predict a particular phenotype.57 Therefore, the allele in question could be considered clinically significant if (1) the frequency is considered significant, (2) the SNP can be detected with adequate sensitivity and specificity, and (3) the SNP reliably predicts response (efficacy or toxicity). In this “reactive” scenario, specific genetic testing can be performed prior to initiating use of the drug.58

For this genetic testing to be applicable to the ICU, point-of-care testing with rapid response rates would need to be readily available to guide decision making during the short time frame necessary for management of a critically ill patient. Additionally, variability in response to drugs in the critically ill may often be explained by the patient’s disease status or concomitant drugs, making identification of genomic variability difficult and potentially clinically irrelevant.59 Point-of-care testing, for example, has been successfully employed in the setting of clopidogrel administration following percutaneous coronary intervention. In a prospective randomized, proof-of-concept trial, Roberts et al. tested for the CYP2C19*2 genotype by way of a buccal swab.24 The results of the genetic test were available within 60 min, and patients who were carriers of this allele in the intervention arm received prasugrel instead of clopidogrel.

An alternative approach is to perform a genome-wide association study wherein the entire genomic profile of groups of patients with and without the disease or condition are compared, with genes indicative of the condition identified. The subsequent likely candidate gene(s) can then be studied in a confirmatory patient population.32,60,61 Genetic information gained can be used in a preemptive fashion, with genome identification using polymorphisms that are known and available at the time of prescribing and accompanying clinical decision support incorporated into the electronic health record.57 Investigators who prospectively performed genetic testing on patients scheduled for coronary arteriography recently reported using this method. The genetic information was stored and linked to the patient. At the time of prescribing clopidogrel, if the patient had a polymorphism such as CYP2C19*2, a drug genome interaction was displayed, as were recommendations for prescribing based on this interaction.62 Unfortunately, there is no reliable way to predict admission to the ICU for most critically ill patients. There is opportunity, however, to stratify patients that may be at increased risk for ICU admission (e.g., acute coronary syndrome), or certain surgical populations (e.g., coronary artery bypass graft surgery) that will likely require ICU care.

Both of these approaches, if applied to the ICU setting, could theoretically result in improved patient outcomes. They could (1) minimize the inefficient, potentially harmful use of trial and error for dosing that could result in sub- or supra-therapeutic drug levels and (2) decrease the time to achieve therapeutic drug levels in patients with diminished physiological reserve. Increased efficiency in drug utilization may result in improved outcomes and reduced costs of care related to adverse events.

Genetics and drug metabolism

Polymorphisms leading to altered drug effect that may have important clinical applications have been identified.63 Some of the common genetic variations for drug-metabolizing enzymes and potential substrates of interest to critical care practitioners are described below. Table 3 lists some examples of genetic alterations for drugs administered in the ICU.

Table 3 Examples of genetic alterations of metabolic enzymes and effects on drugs commonly administered in the ICU
Full size table

CYP-450 isoenzymes

CYP-450 isoenzymes are responsible for oxidative metabolism of many drugs and endogenous substances. Although primarily located in hepatocytes, they are also found in enterocytes and embedded in cells of the kidney, lungs, and brain.64

CYP2C9

Drugs of interest metabolized by CYP2C9 and encountered in the ICU include warfarin, phenytoin, tolbutamide, glyburide, losartan, candesartan, irbesartan, and some of the non-steroidal anti-inflammatory agents, including celecoxib, diclofenac, ibuprofen, and flurbiprofen.33

Increased effect

The main genetic variants are CYP2C9*2/*3, which may lead to poor metabolism of such substrates as warfarin and non-steroidal anti-inflammatory agents, with associated increased bleeding risk.46,65 Phenytoin is mainly metabolized by CYP2C9.66 Patients with the CYP2C9*2/*3 alleles who are administered phenytoin (frequently administered in the ICU for seizure control) may be at increased risk for drug toxicity if administered standard doses.67 Monitoring drug levels is recommended.

CYP2C19

The CYP2C19 enzyme is responsible for metabolism of drugs such as the proton pump inhibitors omeprazole and lansoprazole, the benzodiazepine diazepam, the antifungal agent voriconazole, and the antiplatelet agent clopidogrel. Because the phenotypes of reduced (thrombosis) or exaggerated (bleeding) platelet inhibition are readily recognized, a large body of information concerning pharmacogenomic drug interactions with commonly prescribed medications and clopidogrel has been described.33,56

Increased effect

It is estimated that 15% of Caucasians and Africans and 30% of the Asian population have the CYP2C19*2 allele, which is most often implicated in loss of function and poor metabolism.23 In the case of the proton pump inhibitors, the presence of this allele may lead to excess acid suppression, which in the ICU has been associated with an increased risk of pneumonia.68 In contrast, omeprazole concentration was significantly elevated in EMs in the presence of clopidogrel - likely through clopidogrel-induced inhibition of CYP2C19.69

Decreased effect

Clopidogrel is widely used as an anti-platelet agent in the management of acute coronary syndromes,70 stroke,71 and other forms of peripheral vascular disease.72 It is a prodrug and requires metabolism, primarily by CYP2C19, to emerge in its active form. Failure to adequately metabolize clopidogrel to its active component may lead to adverse cardiac events73 and recurrent stroke,74 which are potentially devastating complications of inadequate anti-platelet therapy.74,75 Because the phenotype (poor inhibition of platelet function) can be readily identified, the correlation with the genetic contribution to adverse outcomes has received scrutiny.13,76,77 Although not all studies are supportive,78 meta-analyses showed that PMs are prone to early stent occlusion and adverse cardiac events.79,80 Tests are now available to identify PMs (failure to prevent platelet aggregation).24 These tests have been used in randomized trials to guide clopidogrel therapy with some success,81 making the requirement for more costly genotyping less necessary. They can be performed within a reasonable time frame, making the results clinically relevant for application in the ICU. When identified as a PM, the patient should be considered for alternative anti-platelet therapies (e.g., prasugrel).70 Algorithms have been developed using genetic testing to guide optimal drug management.82

CYP2D6

Also known as debrisoquine 4-hydroxylase, CYP2D6 is responsible for the metabolism of ~25% of prescribed drugs.33 The polymorphisms of the CYP2D6 gene have been categorized into four distinct phenotypes: (1) poor metabolizers, which have two defective alleles, thus rendering them incapable of metabolizing the gene substrates; (2) intermediate metabolizers, which have one defective and one normal gene; (3) extensive metabolizers, which have two functional alleles, and (4) ultra metabolizers (UMs), which have multiple copies of the functional gene.57,83,84 This categorization scheme has been adopted in many metabolic studies to describe genetic influences related to allelic distribution. The clinical consequences of drugs eliminated by CYP2D6 vary from toxic drug effects in PMs to inadequate drug effects in UMs.84 Although CYP2D6 appears to be highly polymorphic, testing for its variants CYP2D6*3/*4/*5/*6 has identified approximately 95% of PMs and IMs in a Caucasian population.85 Drugs affected by CYP2D6 polymorphisms commonly administered in the ICU include amitryptyline, codeine, hydrocodone, oxycodone, procainamide, citalopram, venlafaxine, paroxetine, haloperidol, respiridone, ondansetron, metoprolol, carvedilol, alprenolol, timolol, and tramadol.28,83 Dosing guidelines based on genetic testing have been published.86 Ethnic differences have also been observed, with patients of Asian ancestry having a higher proportion of PMs, as observed with haloperidol administration.87

Increased effect

Case reports of severe life-threatening and fatal adverse events have been reported in EMs given standard doses of codeine.84 Codeine is commonly administered to patients in the ICU as part of a multimodal pain regimen, often in combination with acetaminophen. Approximately 10% of codeine is metabolized via CYP2D6 to morphine (the metabolite responsible for analgesia). Therefore, enhanced analgesic effects are evident in EMs who may metabolize codeine more extensively, leading to a potential for increased morphine effect,84 including death in a breast-fed infant.88 Similar observations can be expected when tramadol is employed as it too is metabolized by CYP2D6, and its metabolite confers most of the analgesic effect.84

Decreased effect

Reduced effects of codeine may be seen in PMs of CYP2D6, resulting in diminished concentrations of morphine.89 Patients experiencing nausea and vomiting in the ICU are at increased risk of aspiration pneumonia and may be prescribed 5-hydroxytryptamine type 3 receptor antagonists (e.g., ondansetron) in an attempt to relieve these symptoms. The genetically determined ability to metabolize the drug may be associated with failure to relieve the nausea and vomiting (highest failure rate is in UMs).90 Similar observations have been made for ganisetron and dolasetron.91 Blood pressure control and management of heart failure utilizing carvedilol or metoprolol may be affected by CYP2D6 polymorphisms.92

CYP3A4/5

The genes for the metabolizing enzymes CYP3A4 (liver and upper intestine primarily) and CYP3A5 (liver, intestine, extrahepatic sites)93 are responsible for the genetic coding of enzymes involved in the metabolism of more than 50% of clinically utilized drugs and are subject to genetic variability.33 This includes metabolism of many of the highly utilized drugs employed for therapy in the ICU, including midazolam,94 fentanyl,95 and the calcineurin inhibitors cyclosporine and tacrolimus.28

Increased effect

Midazolam, which is exclusively metabolized by CYP3A4/5, serves as a probe for the function of this drug-metabolizing enzyme.59,96 Variability in the metabolism of midazolam occurs in the ICU97 and may, in part, be mediated by genetic differences in the CYP3A alleles.98,99 The CYP3A5*3 allele is associated with reduced midazolam metabolism and is differentially distributed among ethnic groups.100 In a recent study of ICU patients receiving midazolam, however, the presence of the allele CYP3A5*3/*3 was equally distributed as a percentage of the population at risk and not thought to be contributory to the development of coma or delirium.101

Opioids belonging to the phenypiperidine class (e.g., pethidine, fentanyl, sufentanil, alfentanil) are metabolized by the CYP450 enzyme system and are subject to genetic alterations in drug metabolism. Fentanyl, in particular, is metabolized by CYP3A4/5 and subject to variability in response.102 Alterations in p-glycoprotein (PGP) expression may also play a significant role in fentanyl disposition.103

The calcineurin inhibitors cyclosporine and tacrolimus are utilized to provide immunosuppression in patients undergoing solid organ transplant and are frequently initiated in the ICU following organ transplantation. A large body of studies has examined the effects of CYP3A4/5 polymorphisms on the metabolism of these drugs with variable results.28 For example, in a study of kidney transplant patients, genotyping revealed a significant effect of CYP3A5 polymorphisms on the plasma concentrations of tacrolimus in patients having the CYP3A5*3/*3 allele, thereby requiring less drug for effective immunity.104 No such relation was observed for cyclosporine. Dai et al. also observed a similar differential metabolism rate for patients with CYP3A5*1/*3 vs CYP3A5*3/*3 polymorphisms.105

Certain HMG-CoA reductase inhibitors (also referred to as statins), including lovastatin, atorvastatin, and simvastatin, are primarily metabolized by the CYP3A4/5 enzyme system.28 These agents are frequently encountered in the setting of the cardiovascular ICU, where their ongoing use has been associated with reduced delirium106 and a possible anti-inflammatory effect.107 Although pharmacogenetic differences can be detected between patients expressing CYP3A5*1 (expressors) vs CYP3A5*3 (non-expressors) and their ability to lower lipids,108 the clinical consequences are uncertain.

Uridine diphosphate-glucuronosyltransferase 2B7 (UGT2B7)

Morphine is metabolized by UGT2B7 to morphine-3-glucuronide, which is thought to produce adverse central nervous system (CNS) effects, and morphine-6-glucuronide, a more potent analgesic than the parent compound.49 Differences in genes encoding this enzyme result in alterations in morphine metabolism.

Increased effect

The UGT2B7*2 allele is associated with increased nausea after morphine administration.109 Comparatively, the 840GG and GA alleles have resulted in reduced morphine metabolism (compared with the AA allele).110

Catechol-O-methyltransferase

Decreased effect

Catecholamines, such as epinephrine, are frequently administered in the ICU. They may modulate the effects of epidural analgesics administered for pain management.111 Catecholamines are metabolized by catechol-O-methyltransferase, which is subject to genomic variation. Substitution of methionine for valine at codon 158 of the enzyme (Met/Met) resulted in a three- to four-fold reduction in enzyme activity, which has been shown to result in enhanced morphine analgesia compared with the Val/Val or Val/Met alleles.112

Butrylcholinesterase

Butrylcholinesterase (i.e., pseudocholinesterase) is responsible for the hydrolytic degradation of several drugs administered in the ICU, including succinylcholine,113 the benzylisoquinoline muscle relaxants atracurium and mivacurium,26 and the opioid remifentanil.114

Increased effect

Genetic alterations in enzyme function may lead to a prolonged drug effect. The genetic alterations occur as a result of a point mutation on chromosome 3.113 Approximately 4% of the population may be affected, and point mutations are more commonly observed in patients of European ancestry.46 In addition to the usual (U) variant with normal enzyme activity, there are several variants with altered hydrolyzing effect including the atypical (A), fluoride-resistant (F), and silent (S) types - the latter associated with complete absence of an enzyme effect. There are also variants with a normal enzyme effect but reduced quantity, including the H, J, and K variants.115 The presence of an enzyme effect can be detected by measuring the degree of hydrolysis that occurs in the presence of the enzyme inhibitors dibucaine and fluoride.113 A normal (homozygous typical) dibucaine number is 70-80, whereas a heterozygous atypical number is 50-60, and a homozygous atypical number is less than 30.46 Formal genetic sequencing provides more specific details of the genetic defect, and it is not uncommon for individuals to have more than one genetic variant.115

N-acetyltransferase 2

N-acetyltransferase 2 is responsible for the metabolism of agents such as hydralazine (often used for hypertensive emergencies during the perioperative period) and procainamide (an antiarrythmic agent now seldom used in the ICU).27

Increased effect

Six polymorphisms are associated with a slow-metabolizer phenotype characterized by a prolonged drug effect.116 When used over prolonged periods of time, this phenotype has been characterized by the development of antinuclear antibodies and systemic lupus erythematosus, which is a compelling reason to limit the use of hydralazine.117

Genetics and transport proteins

P-glycoprotein

P-glycoprotein (PGP) is an efflux transporter with excretory and protective functions that is found in barrier and elimination organs, including the choroid plexus, blood-brain barrier, hepatocytes, renal tubular cells, and small intestine. SNPs in the ABCB1 gene have been associated with altered expression and function of PGP, leading to clinically significant changes in drug response.48,103,118,119 Substrates of PGP utilized in the ICU include amitriptyline, clopidogrel, corticosteroids, digoxin, diltiazem, domperidone, azole antifungal agents, lansoprazole, levetiracetam, levofloxacin, loperamide, morphine, pantoprazole, phenobarbital, phenytoin, propranolol, ranitidine, rifampin, HMG-CoA reductase inhibitors, calcineurin inhibitors, and verapamil.118

Increased effect

Many opioids (including morphine, methadone, and fentanyl) are subject to PGP transport regulation, limiting accumulation in cells. P-glycoprotein may limit the transport of morphine into the CNS, and variants in its genetic composition may have important clinical effects. Individuals homozygous for the 3435C>T allele (TT carriers) experienced greater pain relief than the heterozygous (CT) type or homozygous wild type (CC).120 Absorption of the PGP substrate digoxin was significantly increased in patients with the 3435C>T and TT alleles compared with the native CC allele.121

Decreased effect

The effect of clopidogrel may, in part, be dependent on PGP. Stakonovic et al. demonstrated significantly reduced plasma concentrations of clopidogrel relative to the active metabolite in patients with the ABCB1 3435TT allele compared with patients having the CC or CT allele.122 This observation has been reported in reviews by other investigators.75,123,124

Genetics and drug targets

Adrenergic receptor

Increased effect

The β-adrenergic receptor (B1AR) blocking agents are commonly used to manage blood pressure and heart rate rhythm disturbances in the ICU. Parvez et al. examined the role that loss of function polymorphism of the β1AR (G389R) played in the response to B1AR blocking agents and determined that the presence of this allele was associated with enhanced response to β1AR blocking agents in patients being treated for atrial fibrillation.125

Decreased effect

Catecholamines are frequently used in the ICU to support patients with hemodynamic deterioration due to a variety of shock states.126 Polymorphisms of the β2-adrenergic receptor (β2AR) have been described.34 Altered responses to β2 agonists used in the management of asthma127,128 or hypotension during regional anesthesia129 have been observed in patients homozygous for the B16 ARG/ARG genotype. Alterations in receptor density in (and hence function of) the heart during cardiac surgery have been shown.130 Functional polymorphisms in the β1AR have been shown to affect the response to β1AR blocking agents employed for the management of heart failure.131 Other studies have found an association between β1-receptor polymorphisms and reduced ability of β-adrenergic receptor blocking agents to lower blood pressure.132,133

Dopamine receptor

Decreased effect

Management of delirium in the ICU frequently requires antipsychotic agents with activity at the dopamine receptor.59,134 Review of studies suggests that subjects with the Ins/Ins allele of the D2 receptor have a better response to antipsychotic agents than those who have the Ins/Del allele.135,136 Patients homozygous for the D2 receptor Taq 1 gene polymorphism had a lower response to haloperidol than heterozygous patients,137 but this observation was not consistent and did not prove to be significant in a subsequent meta-analysis.135

μ Opioid receptor

Decreased effect

Pain management is an important component of compassionate care in the ICU. Morphine, commonly employed in the ICU for pain management, interacts with the μ opioid receptor. Alteration of SNPs (A118 > G118) in the gene OPRM1 has led to genetic variation in receptor function and has been associated with alterations in pain relief.40,120 The polymorphism 118A>G results in less-effective analgesia. Individuals with this variant (GG) require more morphine than individuals with the wild-type variant (AA) and experience more postoperative nausea and vomiting.13,40,138-140 As many as 40% of patients may have a poor response to opioid administration.141 Because the phenotype (pain) is readily observed, investigation of the genetic component to patient variability in pain perception can be determined. It is seldom necessary, however, as most opioids are titrated to effect. In practical terms, when a patient has inadequate pain relief from an opioid, the contribution of genetics to the problem is likely to be small.142 Switching to an alternative analgesic class (e.g., non-steroidal analgesics) may prove beneficial.

Genetics and disease conditions

Sepsis

The heritable characteristic of sepsis is under-appreciated. Heredity may play a greater role in determining the outcome of an episode of sepsis than in outcomes in patients with cardiovascular disease32 or cancer.143 It is commonly observed that the clinical presentation of sepsis (its phenotype characterized by fever, hemodynamic instability, organ dysfunction) is associated with widely variable responses to treatment. In part, it may be due to the wide variability in inciting organisms (viral, bacterial, fungal) and the response to the infection itself.144 Failure of experimental therapies may in part be due to failure to recognize the complexity of the sepsis disease process and our inability to study it correctly.145,146 However, variation in genetic composition may also have a significant role to play.143,144,147-149 For example, innate immunity (the first line of defense against invading organisms) involves three major mechanisms: pathogen recognition, phagocytosis, development of a procoagulant inflammatory response.150 Ongoing genetic research has identified SNPs associated with variability in the innate immune response, including pathogen detection,151-153 inflammation, and coagulation.154,155 Table 4 lists some gene products that have been associated with altered responses to a septic episode.

Table 4 Reported gene association studies and sepsis
Full size table

Decreased effect

Genome-wide association studies suggest that there is alteration in the innate immune system in patients with septic shock.156 Genetic variants in components of the innate immune system - including the pattern recognition receptor toll-like receptor 1 (TLR1),157 toll-like receptor 4 (TLR4),158 toll-like receptor 2 (TLR2),159 migration inhibition factor,160 bacterial permeability increasing protein,161 CD14,153 and plasminogen activator inhibitor162 genes - are associated with differential organ dysfunction and inflammation during a sepsis episode.

Failure to recognize the diversity of response to sepsis as reflected by levels of the inflammatory cytokine tumour necrosis α (TNFα) may be one explanation for the lack of demonstrable efficacy in randomized trials.145,163 For example, in a study of 1057 critically ill patients, mortality was 48.7% in patients homozygous for the TNF-308AA allele compared with those having the TNF-308GA (30.5%) or TNF-308GG (29.5%) allele.164 When response (as reflected by high levels of TNFα) is taken into account, efficacy can be demonstrated.145 The TNF2 polymorphism of the TNFA gene (increased TNFα levels) may be associated with up to a 3.7-fold increased risk of death from septic shock.165 In theory, if specific patient polymorphisms resulting in a variation of TNF production are known to be present at the time of presentation, prescribing an anti-TNF strategy might be targeted at this patient population with a good likelihood of efficacy.

Genome-wide association studies have shown alterations in the adaptive immune response as well.166,167

Hypothetically, improvements in diagnostic precision and recognition of septic patients with distinct biomarkers for inclusion into clinical trials would minimize the likelihood of a type II statistical error.168 In adults, use of organ dysfunction in combination with a requirement for a vasopressor and an elevated lactate level identifies a population with septic shock.169 To date, however, no single biomarker or combination of biomarkers has the requisite sensitivity and specificity to fulfill this criterion for the pediatric population.170 The role that epigenetics may play in altering the host response to sepsis is still a focus of ongoing research.171

Malignant hyperthermia

Increased effect

Although rare, malignant hyperthermia may occur as a result of drug administration in the ICU. Succinylcholine113 and volatile anesthetic agents (used for ICU sedation)172 may precipitate this life-threatening condition, which is characterized by fever, myocardial necrosis, fatal arrhythmias, and acute renal failure. Malignant hyperthermia is inherited as an autosomal dominant condition and manifests as abnormal calcium metabolism at the cellular level.113 The genetic basis for the condition lies in mutations in the RYR1 (ryanodine) and CACNA1S genes.113 Genetic testing for susceptibility is available, and guidelines have been published.173 Management includes use of the ryanodine receptor antagonist dantrolene.174

Drug-induced hypersensitivity reactions

Increased effect

Adverse drug reactions have classically been categorized into type A (those that typically represent a dose response, are predictable based on their pharmacology, and frequently have a genetic component) or type B (idiosyncratic - not dose related or predictable; and the genetic contribution, although often not well characterized, may be important).175 Drug-induced hypersensitivity reactions can range from mild maculopapular exanthema to severe, life-threatening manifestations that require admission to the ICU, such as Stevens-Johnson syndrome (SJS), toxic epidermal necrolysis (TEN), and drug reaction with eosinophilia and systemic symptoms (DRESS).176,177

Although possible with other classes of medications, antiepilepsy drugs (AEDs) are the most commonly implicated, with an estimated incidence of severe hypersensitivity reactions of approximately one in 1000 to one in 10,000 patients exposed to AEDs. Associated AEDs are phenytoin, carbamazepine, lamotrigine, and phenobarbital.176,177 Several human leukocyte antigen (HLA) alleles have been associated with hypersensitivity reactions.178 The HLA-B*1502 allele is associated with SJS and TEN following carbamazepine exposure. The prevalence of HLA-B*1502 is highest in the Han Chinese, and that of carbamazepine-induced SJS and TEN is highest in this population as well.179 The HLA-B*1502 allele has also been implicated in phenytoin-induced SJS/TEN in Han Chinese and Thai populations.178,180 Carbamazepine hypersensitivity reactions (fever, eosinophilia) are more frequent in individuals who have at least one of three SNPs localized at the heat shock protein locus (two in HSP70-1 and one in HSP-Hom).175 Antiepilepsy drugs are commonly initiated in the ICU, particularly in neurosurgical and trauma ICUs. Genotyping in such high-risk populations could provide guidance on the appropriate antiepilepsy drug selected for a given ICU patient.

Patients infected with the human immunodeficiency virus (HIV) may be started on the non-nucleoside reverse transcriptase inhibitor (NNRTI) abacavir as part of an antiretroviral therapy drug cocktail. Approximately 5% of patients initiated on abacavir develop a hypersensitivity reaction characterized by fever, malaise, and a rash.181 Re-exposure may be life threatening. The allele HLA-B*5701 has been linked to abacavir hypersensitivity.182 Patients who have HLA-B*5701 in combination with another allele from the heat shock protein family (Hsp70 Hom M4973T) have a greater than 90% chance of developing abacavir hypersensitivity.183 In a randomized clinical trial, subjects randomized to a pre-prescription determination of HLA-B*5701 status and consequent management without abacavir if they screened positive had a 3.4% incidence of hypersensitivity reactions compared with a 7.8% incidence in the group randomized to abacavir as part of the standard-of-care protocol, thus demonstrating the utility of pre-prescription genomic screening.184 Based on these results, many HIV clinics are performing screening for HLA-B*5701 prior to prescribing abacavir. Patients treated with nevirapine (another NNRTI) are also subject to hypersensitivity reactions linked to an HLA class II allele HLA-DRB1*0101, but the severity of the reaction may be dependent on the CD4 count at the time of drug administration.185 The diagnostic possibility of a hypersensitivity reaction should be considered in any patient with HIV who presents with fever and a rash to the ICU, particularly if there has been a recent change in drug therapy.

Genotyping might reduce the incidence of ICU admission of patients with serious cutaneous hypersensitivity reactions. Additionally, genotyping could be utilized to guide AED selection among patients admitted with SJS/TEN. Point-of-care platforms unfortunately do not currently have the ability to perform HLA genotyping.57 A novel approach has been incorporation of an alert to the possible adverse effects and suggestions for genetic testing incorporated into the medical record.186

Cardiac channelopathies

Genetic mutations in the genes encoding for potassium and sodium channels in the heart (so-called channelopathies) may be associated with a predisposition to life-threatening arrhythmias. These arrhythmias may be a reason for ICU admission (sudden cardiac arrest in a previously healthy young individual) or occur as a consequence of interaction between a predisposed individual and drug administration while in the ICU (torsade de pointes).187

There are three major clinical conditions associated with genetic alterations in ion channel function in the heart: long QTc syndrome, catecholaminergic polymorphic ventricular tachycardia (CPVC), and Brugada syndrome.188 For the long QTc syndrome, although 16 genes have been implicated, only three genetic mutations - in the genes KCNQ1(LQT1), KCNHQ (LQT2), SNA5A(LQT3) - are responsible for approximately 75% of cases.187,189,190 The phenotype is sudden cardiac arrest or syncope associated with a prolonged QTc on the electrocardiogram. Specific electrocardiographic abnormalities may help classify the type of mutation.191 A scoring system (Schwartz score)188 has been developed to assess risk for arrhythmias that require further genetic assessment. Treatment is, to a large extent, dependent on the type of mutation - β-blockers for LQT1 and LQT2 and sodium channel blockers (mexiletine) for LQT3 - that should be initiated only under monitored conditions.188,191 Once a proband has been identified, family screening should occur and prophylactic therapy initiated as indicated.

Abnormalities in the RYR2 gene (responsible for intracellular calcium regulation) account for ~50% of the cases of CPVC identified.187,190 Typically, this arrhythmia is precipitated by exercise, but whether the stress-induced tachycardias (e.g., sepsis or postoperatively) often seen in the ICU can precipitate the arrhythmia has not (to our knowledge) been examined.

Brugada syndrome is characterized by a distinctive, coved-type, ST-segment elevation in leads V1 to V3, right ventricular conduction abnormalities, and life-threatening ventricular arrhythmias.190 The syndrome typically occurs in men during the fifth decade of life, at night, and with a family history of sudden cardiac death. Most genetic abnormalities are related to the sodium channel, with SCN5A responsible for ~20% of observed cases. More than 200 SCN5A mutations have been identified to date.

A number of drugs commonly employed in the ICU have been associated with the development of a prolonged QTc and, in some cases, torsade de pointes.192 Examples include amiodarone, erythromycin, clarithromycin, ketoconazole, and haloperidol.

An algorithm for identifying and managing these arrhythmias has been proposed.190

Thromboembolic disease

Decreased effect

The most common inherited thrombophilias include deficiencies in three natural anticoagulant proteins (antithrombin, protein C, protein S) and specific mutations in genes for factor V and prothrombin.193 The first presentation of an inherited defect in the coagulation system may be an acute thrombotic event that necessitates ICU admission. For example, about 2-3% of cases of portal vein thrombosis are due to an inherited coagulation disorder, but the odds of having such an event in the presence of an inherited coagulation defect is more than eight-fold that of normal individuals.194 Up to 30% of known patients with a thrombotic disorder have a recurrence.195 The factor V Leiden mutation involves a single amino acid substitution of glutamine for arginine at position 506 of the factor V gene, leading to activated protein C resistance.196 The prothrombin G20210A mutation represents a single point mutation (G®A at position 20210) in the factor II gene, leading to increased prothrombin concentration.197 There are racial differences in the presence of inherited thrombophilias, with factor V Leiden and prothrombin G20210A mutations more common in Caucasians and protein C and S deficiency more common in Asian populations.198 For example, the most common inherited thrombophilia in the Japanese population is 586A>G (also known as p.Lys196Glu), which accounts for up to 30% of cases, whereas in the Chinese population the most prevalent inherited thrombophilia is due to a protein C mutation (565C>T, p.Arg189Trp).198 Predisposing factors for the development of venous thrombotic events (VTEs) in susceptible patients include immobility, surgery, trauma, cancer, female sex, hormonal therapy, and pregnancy,193,199-201 which are prevalent conditions seen in the ICU and are addressed by routine use of VTE prophylaxis.202 Guidelines have been published concerning testing of patients and their families following an unprovoked VTE episode.203

Increased effect

Warfarin metabolism is subject to modification by genetic abnormalities in the CYP2C9 gene (*2 and *3 alleles vs *1), which is responsible for metabolism of S warfarin, and the VKORC1 gene (GA and AA alleles vs GG alleles), which is responsible for its molecular target.204 A meta-analysis of studies indicated that bleeding risk is increased in subjects with one of these mutations, and dosage adjustment is advised.205 However, a meta-analysis of studies of the utility of dosing based on genetic information to reduce the bleeding risk have produced conflicting results.204,206-210

Because many patients present with a de novo mutation, and there is a large variability in the phenotypic presentation of bleeding, the first presentation of patients with congenital bleeding disorders may be a traumatic event and bleeding of sufficient magnitude to warrant ICU admission.211 Diagnosis under these circumstances can be difficult. If a bleeding disorder is suspected, however, a schema for the diagnostic workup based on commonly available laboratory tests is available.211 Hemophilia A (factor VIII deficiency - F8 gene) and hemophilia B (factor IX deficiency - F9 gene) are common inherited X-linked recessive bleeding disorders resulting in deficient coagulation. There is a large variability in the number of genetic alterations leading to factor VIII and factor IX deficiency.212 Although most commonly observed early in life, the phenotype of bleeding may not manifest until the patient undergoes a surgical procedure or experiences a minor traumatic event that precipitates bleeding of sufficient magnitude to warrant treatment and observation in the ICU.211 Treatment in both instances of factor deficiency is replacement of the missing factor, although the possibility of using gene transfer therapy is becoming a reality.211,213

Altered platelet function due to von Willebrand’s disease is the most common inherited defective platelet function leading to increased bleeding. Von Willebrand’s disease is subclassified into three subtypes: type 1, partial quantitative deficiency; type 2, qualitative deficiency (further subclassified into four disorders); type 3, total deficiency.214 Types 1 and 2 are inherited in an autosomal dominant pattern, whereas type 3 is inherited as an autosomal recessive trait.215 Type 1 accounts for about 70% of cases.211 Management is with DDAVP, factor replacement, and antifibrinolytic agents.211,214

Reviews of the genetics of other rare bleeding disorders are available.216,217

Congestive heart failure

Congestive heart failure with or without arrhythmia is a common reason for ICU admission and is a significant burden on the health care system.218 The genetic contribution to the development of congestive heart failure as a result of cardiomyopathies is increasingly recognized.219 Hypertrophic cardiomyopathies have a genetic basis, and more than 13 genes with over 900 mutations in sarcomere expression have been identified.220 Similarly, the genetic contribution to dilated cardiomyopathies is being unraveled.219 The American Heart Association and the European Society of Cardiology both recognize the genetic component to many of these diseases in their classification schemes.221-223 Genetic testing for the presence of an inherited cardiomyopathy is available224 and may prove useful when making management decisions, such as early insertion of an automated implantable cardioverter-defibrillator.225

Response to heart failure therapies may be altered by genetic variations.226 Reductions in response to β-adrenergic receptor blocking agents was associated with alterations in the BAR ARG389ARG polymorphism in the Beta-Blocker Evaluation Survival Trial (BEST) trial of patients with severe heart failure. Patients with the ARG389GLY polymorphism had a 38% reduction in survival benefit compared with the ARG389ARG group.227 These investigators also showed that a deletion polymorphism in the α-2 receptor (α 2c DEL 322-325) resulted in subjects having a much greater sympatholytic effect to the β-adrenergic receptor blocking agent bucindolol with no effect on mortality compared with wild-type responders, who showed a 30% reduction in mortality.228 In a substudy - the Genetic Risk Assessment of Heart Failure in African Americans (GRAHF) study - of the African American Heart Failure Trial, the effects of subjects carrying the aldosterone synthase gene (CYP11B2) promoter polymorphisms C-344TT/TC or CC were examined.229 Patients with the C-344CC polymorphism (associated with increased aldosterone synthase activity) had significantly poorer event-free survival. The above studies serve as reminders that significant variation may exist in response to heart failure therapies. When an unexpected result does occur, the possibility that the variance has a genetic component should be considered.

Conclusion

Pharmacogenomics and pharmacogenetics in regard to drug therapy is beginning to be applied to patients in the ICU. As technology for rapid discovery of genetic mutations advances and becomes more available, it is possible that in the not too distant future clinicians will have rapidly available, patient-specific, genetic information at the bedside. Already, such application has resulted in improvements in determining the optimal antiplatelet therapy for patients with coronary artery disease. Improvements have also been made in the areas of warfarin dosing, pain management, and immunosuppressive therapy. Furthermore, malignant hyperthermia, inherited cardiac channelopathies, and coagulation disorders of relevance to critical care are being addressed with pharmacogenomics and pharmacogenetics. Given our current inability to predict efficacy and untoward adverse drug effects in a population that lacks the luxury of time for trial and error and has little physiological reserve to compensate for adverse effects, we predict that the application of pharmacogenomics in the ICU will increase and translate into clinically meaningful outcomes. As point-of-care genetic testing becomes more readily available, the genetic contribution of multiple drug therapies will become more predictable, and the combination of medications that is most likely to result in the desired effects for the individual could be chosen.