Polymorphisms in the drug-metabolizing enzymes represent the first recognized and, so far, the most documented examples of genetic variants with consequences in drug response and toxicity. The major phase I enzymes are the CYP superfamily of isoenzymes. N-acetyltransferase, uridine diphosphate glucuronosyltransferase (UGT), and glutathione S-transferase are examples of phase II metabolizing enzymes that exhibit genetic polymorphisms. Thiopurine S-methyltransferase (TPMT) and dihydropyrimidine dehydrogenase (DPD) are examples of nucleotide base-metabolizing enzymes. eTable 6-1 lists examples of polymorphic metabolizing enzymes and corresponding drug substrates whose plasma concentrations and pharmacologic effects may be altered as a consequence of genetic variation. Examples of such effects are discussed in the following sections.
Currently, 57 different CYP isoenzymes have been documented to be present in humans, with 42 involved in the metabolism of exogenous xenobiotics and endogenous substances such as steroids and prostaglandins.10 Fifteen of these isoenzymes are known to be involved in the metabolism of drugs, but significant interindividual variabilities in enzyme activity exist as a result of induction, inhibition, and genetic inheritance. Functional genetic polymorphism has been discovered for CYP2A6, CYP2B6, CYP2C9, CYP2C19, CYP2D6, and CYP3A4/5, and their impact on drug therapy are described below.
Polymorphisms in the CYP2D6 gene are the best characterized among all of the CYP variants. Over the years, at least 100 gene variants and 120 alleles have been identified in the CYP2D6 gene (www.cypalleles.ki.se/cyp2d6.htm). Despite the extensive number of alleles, Sistonen et al.11 showed genotyping 12 SNPs that represent 20 different haplotypes could provide 90% to 95% accuracy in predicting the real phenotype (outward expression of genotypes). More specifically, different studies showed that the CYP2D6 phenotypes of extensive metabolizer (EM) carrying two functional alleles and poor metabolizer (PM) carrying two nonfunctional alleles could be predicted with up to 99% confidence with six genotypic variants. CYP2D6*1 is considered the wild-type variant and exhibits normal enzyme activity. CYP2D6*2 has the same activity as CYP2D6*1 but is capable of duplication or amplification. Both these variants are present in EMs. The two null variants, CYP2D6*4 (c.1846G>A, defective splicing) and CYP2D6*5 (gene deletion), are predominantly found in white PMs (5% to 10% of population) and result in an inactive enzyme and absence of enzyme, respectively. The predominant variants in people of Asian and African heritage are CYP2D6*10 (c.100C>T, Pro34Ser) and CYP2D6*17 (c.1023C>T, Arg296Cys), respectively, both resulting in single-amino-acid substitution and consequent reduction in enzyme activity. Other than *10 and *17, *9, *29, *36, and *41 variants are also associated with lower enzyme activity in the intermediate metabolizers (IMs) phenotype (carriers of one nonfunctional allele and one allele with diminished activity).10 In addition to *2, gene duplication or amplification had been documented for *1, *4, *6, *10, *17, *29, *35, *41, *43, and *45 variants,12 with resultant higher enzyme activity in the ultrarapid metabolizer (UM) phenotype (carriers of multiple copies of functional alleles).
The presence of two defective alleles (CYP2D6*3, CYP2D6*4 [more common], CYP2D6*5, or CYP2D6*6) in PM results in significant impaired ability to metabolize CYP2D6-dependent substrates. Depending on the importance of the affected CYP2D6 pathway to overall drug metabolism and the drug’s therapeutic index, clinically significant side effects may occur in PMs as a result of elevated drug concentrations,13 for example, of perhexiline (neuropathy), perphenazine (sedation and parkinsonism), and propafenone (proarrhythmic events).
The therapeutic implication of CYP2D6 polymorphism is different if the substrate in question is a prodrug. In this case, PMs would not be able to convert the drug into the therapeutically active metabolite. Two examples of prodrugs dependent on CYP2D6-mediated conversion to active forms are codeine and tramadol. Codeine and tramadol are converted by CYP2D6 to morphine and O-desmethyltramadol, respectively, and CYP2D6 PMs would experience little analgesic relief after taking these drugs.14 Another example is CYP2D6-catalyzed conversion of tamoxifen to the more potent antiestrogen metabolite, endoxifen, in which case PMs have been shown to have shortened time to recurrence of breast cancer and worse relapse-free survival.15
Patients who are EMs have a wide range of CYP2D6 activity, with UMs possessing very high enzyme activity on one end of the spectrum and IMs possessing diminished activity on the other end. Both have clinical implications in terms of dosage adjustment for CYP2D6 substrates. For the CYP2D6 substrate nortriptyline, a patient with three copies of CYP2D6*2 was shown to require doses three- to fivefold higher than normally recommended to achieve therapeutic plasma concentrations (50 to 150 ng/mL [50 to 150 mcg/L; 190 to 570 nmol/L]).16,17 In the same report, another patient with duplicated CYP2D6*2 required twice the usual recommended daily dose (300 mg vs. 25 to 150 mg) to achieve adequate therapeutic response.17 There are similar reports of lower drug efficacy in UMs with antiemetics such as ondansetron.18 Conversely, UMs administered the usual therapeutic dose of codeine might exhibit symptoms of narcotic overdose associated with high morphine concentration. This toxicity potential had been reported in several case reports.19,20
Furthermore, the consequence of CYP2D6-mediated drug interactions can be different in patients with different metabolic phenotypes. The UM phenotype has been reported to affect the potential for drug interaction with paroxetine, a potent CYP2D6 inhibitor as well as a CYP2D6 substrate, whence a UM with three functional CYP2D6 gene copies had undetectable paroxetine concentration with standard dosing and showed no inhibitory effect at CYP2D6.21
In general, the magnitude of drug interactions involving inhibition of CYP2D6 is much greater in EMs versus PMs, who have either little or no enzyme activity. For example, Hamelin et al.22 showed that in EMs, but not PMs, hemodynamic responses to metoprolol (a CYP2D6 substrate) were pronounced and prolonged during concomitant diphenhydramine administration. Potent CYP2D6 inhibitors such as paroxetine and fluoxetine may reduce the metabolic capacity of EMs significantly so that they appear phenotypically as PMs.23 Given the abundance and greater antiestrogenic activity of endoxifen,15 the use of paroxetine or fluoxetine in tamoxifen-treated patients should best be avoided. When there is a need for concurrent antidepressant administration with tamoxifen, those with lesser extent of CYP2D6 inhibition, such as citalopram and venlafaxine,23 would be better alternatives.
The high prevalence of CYP2D6*10 (associated with lower enzyme activity) in the Asian population provides a biologic and molecular explanation for the higher drug concentrations and/or lower dosage requirements of neuroleptic medications and mianserin in people of Asian heritage.24,25 The widespread presence of the CYP2D6*17 variant among people of African heritage suggests that native African populations would metabolize CYP2D6 substrates at a slower rate than do other ethnic groups.26,27 However, there are no genotype- and phenotype-based data to document the need for prescribing lower doses of psychotropics and other CYP2D6 substrates in native African populations.
In addition to the therapeutic implications of genetic polymorphisms, one study showed that the CYP2D6 polymorphism also has an economic impact.28 The annual cost of treating UMs and PMs was $4,000 to $6,000 higher than the cost of treating EMs or IMs. The cost of genotyping can be considerably less than that incurred in a patient with a serious adverse drug reaction. In 2005, the FDA approved the AmpliChip® CYP450 Test (Roche Diagnostics) for analyzing 27 CYP2D6 alleles in addition to the CYP2C19*1, *2, and *3 alleles (discussed below) to assist clinicians in individualizing therapy with drugs metabolized through the CYP2D6 and 2C19 pathways.
Sidebar: Clinical Controversy…
The CYP2D6 poor metabolizer phenotype has been associated with poor outcomes with tamoxifen in postmenopausal women with breast cancer. However, a group of investigators genotyped tumor tissue from tamoxifen-treated women with breast cancer who were enrolled in a large clinical trial and found no relationship between CYP2D6 genotype and tamoxifen treatment response. More recently, several other well-known pharmacogenomics investigators called these findings into question, citing significant problems with distribution of the genotype frequencies.29 Thus, the role of CYP2D6 genotyping to predict tamoxifen response has yet to be resolved.
The principal defective alleles for the CYP2C19 genetic polymorphism are CYP2C19*2 (c.19154G>A, aberrant splice site) in exon 5 and CYP2C19*3 (c.17948G>A, premature stop codon) in exon 2 of CYP2C19, resulting in inactive enzyme and the PM phenotype. The clinical relevance of the CYP2C19 polymorphism has been demonstrated for proton pump inhibitors and clopidogrel.
PMs for the CYP2C19 polymorphism showed up to a 10-fold increase in the area under the curve (AUC) of omeprazole compared with EMs.30 The presence of a defective CYP2C19 allele has been associated with improved Helicobacter pylori cure rates after dual (omeprazole and amoxicillin)31 and triple (omeprazole or lansoprazole, clarithromycin, and amoxicillin) therapy.32 The cure rate achieved with dual therapy was 100% in PMs compared with 60% and 29% in heterozygous and homozygous EMs, respectively.31 In two studies included in the meta-analysis of 20 studies using triple therapy,32 EMs had H. pylori eradication rates of 74% to 83% versus 100% cure rates in all 15 PMs included in the two studies.
These differences likely reflect the higher achievable intragastric pH in the PM group.33 Interestingly, EMs who failed initial triple therapy (lansoprazole, clarithromycin, and amoxicillin) and were retreated with high-dose lansoprazole (30 mg four times daily) and amoxicillin achieved a 97% H. pylori eradication.34 A gene–dose effect in attainment of desirable intragastric pH ranges and H. pylori eradication rate, as well as the cost effectiveness of pharmacogenomic-guided dosing was shown for lansoprazole.35
Conversely, IMs and PMs of CYP2C19 may have reduced response to the antiplatelet agent clopidogrel and potentially require higher doses or alternative antiplatelet therapy (e.g., prasugrel) for better clinical outcomes. This is because clopidogrel is a prodrug that requires conversion via CYP2C19 to its active form. In IMs and PMs, usual doses of clopidogrel are less effective at inhibiting platelet aggregation and preventing cardiovascular events than in EMs.36 The recent report of a point-of-care device for detecting CYP2C19*2 with a turnaround time of approximately 1 hour37 could facilitate use of CYP2C19 genotyping in accordance to consensus-based guidelines.36
Similar to the CYP2D6 polymorphism, people of Asian heritage also metabolize most CYP2C19 substrates at a slower rate than do whites.38 This is a reflection of a higher prevalence of both PMs (13% to 22.5% vs 2% to 6% in whites) and heterozygotes for the defective CYP2C19 alleles (10% to 30%) in Asians.30 This genotypic difference may explain the practice of prescribing lower diazepam dosages for patients of Chinese heritage.39 Similar to CYP2D6 alleles with high enzyme activity, the allelic variant CYP2C19*17 is associated with a very rapid metabolism phenotype, and carriers of this allelic variant would likely require higher doses of proton pump inhibitors40,41 and other CYP2C19 substrates such as voriconazole.42
Another polymorphic isoenzyme of CYP2C is CYP2C9, which metabolizes narrow therapeutic index drugs such as warfarin, phenytoin, and tolbutamide. Warfarin is a racemic mixture, and the S-isomer, which possesses about three times the anticoagulant effects of the R-isomer, is metabolized by CYP2C9. CYP2C9*2 (c.19154G>A) and CYP2C9*3 (c.17948G>A) are the two most common CYP2C9 variants in whites, and both exhibit single-amino-acid substitutions at positions critical for enzyme activity.43 This could have clinically important consequences in warfarin-treated patients. For example, a 90% reduction in S-warfarin clearance was reported in CYP2C9*3 homozygotes compared with subjects homozygous for the wild-type (*1) variant,44 requiring dose reduction to 0.5 mg/day in a report of a CYP2C9*3 homozygote initially given usual doses of warfarin.45 The clinical relevance of CYP2C9 polymorphism in warfarin dosing was recently reviewed in a meta-analysis of 39 studies.46 In one study included in the meta-analysis,47 an overrepresentation of CYP2C9 variant alleles was observed in 81% of patients requiring low-dose warfarin therapy (≤1.5 mg/day). The low-dose group was reported to have more difficulty with warfarin induction, requiring longer hospital stays to stabilize the warfarin regimen and experiencing a higher incidence of bleeding complications.
Both CYP2C9*2 and CYP2C9*3 are more common in whites than in Asians and Africans. More recently, the CYP2C9*8 allele was shown to reduce warfarin clearance and dose requirements.48 The CYP2C9*8 allele occurs in approximately 10% of African Americans and may have important implications for metabolism of CYP2C9 substrates in this population. Ultrarapid CYP2C9-mediated metabolism has also been recently reported resulting in higher dosage requirements of warfarin and phenytoin,49,50 and the molecular mechanism is currently under investigation.
The CYP2C9 polymorphism, in conjunction with a polymorphism in the vitamin K oxidoreductase complex subunit 1 gene (VKORC1), provide extensive pharmacogenomic data that form the basis for a consensus-based guideline,51 and the potential of cost savings.52 The use of CYP2C9 and VKORC1 genotypes in dosing warfarin is discussed in section “Polymorphisms in Drug Target Genes” for more detail.
In addition to the wild type CYP2A6*1, several variants for the CYP2A6 polymorphism have been identified (www.cypalleles.ki.se): CYP2A6*2 (single amino acid substitution), CYP2A6*4 (gene deletion), CYP2A6*5 (gene conversion), and CYP2A6*20 (frameshift) are associated with abolished enzyme activity. Deletion of the CYP2A6 gene is very common in Asian patients,53 which likely accounts for the dramatic difference in the frequency of PMs in Asian (20%) versus white populations (≤1%). Nicotine is metabolized by CYP2A6, and the clinical relevance of the CYP2A6 polymorphism lies in management of tobacco abuse. Investigators reported that nonsmokers were more likely to carry the defective CYP2A6 allele than were smokers. Smokers who had the defective CYP2A6 allele smoked fewer cigarettes and were more likely to quit. The inability to metabolize nicotine, secondary to the presence of a defective CYP2A6 allele, likely leads to enhanced nicotine tolerance and increased adverse effects from nicotine. Based on these observations, CYP2A6 inhibition may have a role in the management of tobacco dependency.53
Although the role of CYP2B6 in the metabolism of anticancer drugs such as cyclophosphamide and ifosfamide has been studied, it is with the antiretroviral agents that its clinical relevance was revisited and highlighted. The nonnucleoside reverse transcriptase inhibitor efavirenz is metabolized by CYP2B6. Many patients receiving efavirenz experience central nervous system (CNS) adverse effects that are related to variable systemic exposure to the drug, which could be related to the lower metabolizing efficiency of the CYP2B6*6, *16, or *18 alleles.54 A prospective study demonstrated that dose reduction for 6 months in 12 patients with high efavirenz concentrations secondary to CYP2B6 polymorphism resulted in both effective anti-human immunodeficiency virus (HIV)-1 activity with HIV-1 load less than 50 copies/mL as well as lower incidence of CNS adverse effects.55
Within the CYP3A subfamily, at least three isoenzymes, namely, CYP3A4, CYP3A5, and CYP3A7, have been characterized. Despite as much as 40-fold interindividual variability in its expression, functional CYP3A4 is expressed in most adults, with intestinal expression playing a significant role in the first-pass metabolism of numerous drugs. Although several CYP3A4 variants (e.g., *6, *17, *20) have been associated with reduced activity, their low frequency suggest limited clinical relevance.
CYP3A5 is reported to be polymorphic in 60% of African Americans and 33% of whites, with CYP3A5*3 (c.6986 A>G, aberrant splice site) in intron 3 as the primary allele variant. In contrast to individuals with the CYP3A5*1 allele, subjects with CYP3A5*3 have no functional CYP3A5 enzyme.56 CYP3A4 and CYP3A5 mediate the metabolism of more than 50% of all clinically useful drugs. However, with overlapping substrate specificities, it remains unknown whether some drugs are substrates for CYP3A5 but not CYP3A4 and vice versa. Although variability exists between dose-adjusted concentration and CYP3A5 genotypes, studies have shown a correlation between trough concentrations of tacrolimus and CYP3A5 genetic constitution.57
Phase II and Nucleotide-Base Metabolizing Enzymes
The clinical relevance of genetic polymorphisms in TPMT, DPD, and UGT enzymes has been demonstrated in the treatment of cancer. The TPMT gene has four mutant alleles: TPMT*3A (the most common), TPMT*2, TPMT*3B, and TPMT*3C. Thiopurine drugs such as 6-thioguanine, 6-mercaptopurine, and its precursor, azathioprine, are inactivated by TPMT, and patients who are homozygous or heterozygous for the TPMT mutant alleles are at higher risk for developing serious hematological toxicities during treatment with the thiopurines.58 DPD mediates the metabolism of 5-fluorouracil and its precursor capecitabine, and patients with a defective allele of the DPYD gene encoding for DPD cannot metabolize 5-fluorouracil and thus may experience enhanced drug-related neurotoxicity.59 The camptothecin derivative irinotecan (CPT-11) is activated by carboxylesterase to SN-38, a potent topoisomerase I inhibitor. SN-38 is inactivated by glucuronidation via the polymorphic UGT1A1 enzyme, which may play a role in CPT-11–related toxicity. An extra thymine-adenine (TA) repeat within the TATA section of the UGT1A1 promoter results in the (TA)7TAA allele (also known as UGT1A1*28), which possesses lower enzyme activity than the wild-type (TA)6TAA allele. Impaired SN-38 glucuronidation secondary to the (TA)7TAA allele may result in abnormally high SN-38 concentrations. A prospective clinical trial demonstrated more severe diarrhea and neutropenia in irinotecan-treated patients who are homozygous or heterozygous carriers of the (TA)7TAA allele.60 A subsequent meta-analysis showed dose-related increases in the risk for severe neutropenia with irinotecan with the UGT1A1 (TA)7TAA allele.61 In 2005, the FDA approved the Invader® UGT1A1 Molecular Assay (Third Wave Technologies) to genotype for UGT1A1 alleles, and the labeling for irinotecan was revised to recommend dose adjustment for individuals who are homozygous for the (TA)7TAA allele.