The Pharmacogenomics Journal (2013) 13, 1–11;
published online 23 October 2012
Pharmacogenomics of drug-metabolizing enzymes: a recent update on clinical implications and endogenous effects
S C Sim1, M Kacevska1 and M Ingelman-Sundberg1
1Section of Pharmacogenetics, Department of Physiology and Pharmacology, Karolinska Institutet, Stockholm, Sweden
Correspondence: Dr SC Sim, Section of Pharmacogenetics, Department of Physiology and Pharmacology, Karolinska Institutet, SE-171 77 Stockholm, Sweden.
Interindividual differences in drug disposition are important causes for adverse drug reactions and lack of drug response. The majority of phase I and phase II drug-metabolizing enzymes (DMEs) are polymorphic and constitute essential factors for the outcome of drug therapy. Recently, both genome-wide association (GWA) studies with a focus on drug response, as well as more targeted studies of genes encoding DMEs have revealed in-depth information and provided additional information for variation in drug metabolism and drug response, resulting in increased knowledge that aids drug development and clinical practice. In addition, an increasing number of meta-analyses have been published based on several original and often conflicting pharmacogenetic studies. Here, we review data regarding the pharmacogenomics of DMEs, with particular emphasis on novelties. We conclude that recent studies have emphasized the importance of CYP2C19 polymorphism for the effects of clopidogrel, whereas the CYP2C9 polymorphism appears to have a role in anticoagulant treatment, although inferior to VKORC1. Furthermore, the analgesic and side effects of codeine in relation to CYP2D6 polymorphism are supported and the influence of CYP2D6 genotype on breast cancer recurrence during tamoxifen treatment appears relevant as based on three large studies. The influence of CYP2D6 polymorphism on the effect of antidepressants in a clinical setting is yet without any firm evidence, and the relation between CYP2D6 ultrarapid metabolizers and suicide behavior warrants further studies. There is evidence for the influence of CYP3A5 polymorphism on tacrolimus dose, although the influence on response is less studied. Recent large GWA studies support a link between CYP1A2 polymorphism and blood pressure as well as coffee consumption, and between CYP2A6 polymorphism and cigarette consumption, which in turn appears to influence the lung cancer incidence. Regarding phase II enzyme polymorphism, the anticancer treatment with mercaptopurines and irinotecan is still considered important in relation to the polymorphism of TPMT and UGT1A1, respectively. There is a need for further clarification of the clinical importance and use of all these findings, but the recent research in the field that encompasses larger studies and a whole genome perspective, improves the possibilities be able to make firm and cost-effective recommendations for drug treatment in the future.
Keywords: cardiovascular; conjugation; cytochrome P450; exogenous; hepatic; toxicity
The majority of phase I and phase II drug-metabolizing enzymes (DMEs) are polymorphic. This polymorphism causes important interindividual differences in drug and metabolite exposure and can determine drug response as well as the risk for adverse drug reactions. In recent years, we have increased our understanding regarding the clinical importance of such variation and several databases containing pharmacogenetic information regarding DMEs are readily available (reviewed in Sim and Ingelman-Sundberg1).
Genetic polymorphism of DMEs encompasses gene copy number variation including gene amplification and deletion, small insertions and deletions, as well as single-nucleotide polymorphisms (SNPs). A recent study examining global and local differentiation SNP profiles in 283 DME as well as transporter genes across 62 worldwide ethnic groups indicated that there is a positive selection on variation in DME genes and that this genetic differentiation contributes to population heterogeneity in drug response.2 The polymorphisms of DME genes are important determinants for drug response, and indeed the majority of pharmacogenomic drug labels refer to genes encoding phase I and phase II enzymes.3 In addition, both the US Food and Drug Administration (FDA) and the European Medicines Agency (EMA) highlight such variation in specific guidelines in clinical pharmacology and for drug development (http://www.ema.europa.eu/docs/en_GB/document_library/Scientific_guideline/
Guidances/UCM243702.pdf). In the present review, our objective is to focus on novel aspects of DME pharmacogenetics, both in terms of identifying recent studies that have provided additional evidence to or strengthened already known DME associations, as well as studies on new associations or mechanisms of DME polymorphism influencing phenotype. Herein, we present an update on the impact of DME genotypes on the metabolism and outcome of drugs, as well as other exogenous and endogenous substances.
SNPs affecting DME function
Associations of SNPs to drug treatment outcome is continuously being discovered, with a recent focus on genome-wide association (GWA) studies being conducted in many different ethnic groups. Although several SNPs identified in GWA studies are linked with SNPs causing functional alterations, determining the function of SNPs without apparent linkage or proven effects is a more laborious task that represents a bottleneck between the numerous DME polymorphisms identified and their functional role and molecular mechanism of action. Although DME SNPs with confirmed effects have been reported for all gene regions such as in 5′- or 3′-untranslated regions, non-synonymous SNPs causing amino-acid substitutions constitute the major basis for assigning new alleles among the Cytochrome P450 1-3 (CYP1-3) gene families, with close to 400 different unique alleles characterized to date (www.cypalleles.ki.se). Examples of alleles carrying SNPs interfering with the splicing machinery include CYP2D6*4, CYP2D6*41, CYP2C19*2, CYP2B6*6 and CYP3A5*3 (reviewed in Sadee et al.4). Short nucleotide sequences directing the spliceosome to the correct exon–intron boundary have limited tolerance to mutations and hence can lead to aberrantly spliced transcripts and abolished protein function, such as for CYP2C19*25, 6 and CYB2B6*6.7, 8, 9 An example of DME polymorphism interfering with transcriptional regulation is the UGT1A1*28 allele that harbors an extra promoter TA repeat leading to decreased gene expression,10, 11 whereas the CYP2C19*17 (g.-806C>T) allele results in increased gene expression.12 Increased transcription has also been reported for the g.-163C>A SNP variant in CYP1A2*1F, which can particularly enhance CYP1A2 inducibility (see for example, Djordjevic et al.13 and Han et al.14). A more recent functional intronic SNP was identified for CYP3A4 (g.15389C>T, rs35599367, CYP3A4*22), which was shown to reduce hepatic CYP3A4 mRNA levels potentially by disrupting the RNA elongation rate through changes in single-stranded DNA or RNA secondary structures.15
GWA studies meta-analyses and specific studies of DME polymorphism of clinical or endogenous importance
Recent GWA studies, and meta-analyses, providing weighted measures based on several original studies, have significantly improved our knowledge regarding genetic variation in DME genes that cause alterations in drug disposition. GWA data have, in particular, provided substantial support to previous findings, indicating the importance of DME variants in drug response (for example warfarin and CYP2C9, clopidogrel and CYP2C19), as well as providing evidence for other levels of associations, such as that of genetic variation in the aryl hydrocarbon receptor (AHR, a known regulator of CYP1A2 expression) and CYP1A2 (for which caffeine is used as a probe drug) with caffeine intake. In addition, GWA studies have established the relative role of the CYP2A6 locus in smoking behavior. Interpretation of GWA data, however, also has some associated uncertainties. Substantial variation between different GWA studies with respect to SNP coverage, population size studied and integration of replication cohorts, has sometimes made it difficult to cross compare and validate the various result outcomes and conclusions derived. Furthermore, deep sequencing with a higher coverage of genetic variation than GWA studies may prove more successful in identifying novel pharmacogenetic associations of DMEs. There is a GWA study database available (from National Human Genome Research Institute (http://www.genome.gov/gwastudies/), however, this database does neither record the population sizes with high accuracy, nor does it report whether the genome-wide significance stated for a specific gene is related to exploratory analyses, replication analyses (of sometimes non-significant results in the exploratory analysis), or whether significance was obtained after adjusting for subject characteristics or even other genetic variants known to influence the outcome.
As for meta-analyses, they provide a more analytical perspective on genotype–phenotype relationships by combining multiple studies that evaluate specific associations (for example clopidogrel and CYP2C19, tamoxifen and CYP2D6) and thus offer a more comprehensive deduction than individual studies.
In the proceeding sections we review GWA studies, recent meta-analyses as well as relevant specific DME studies carried out in the field of DMEs and their influence on drugs and endogenous phenotypes (the strongest findings are summarized in Table 1 and Figure 1). Furthermore, the set-up and outcome of recent GWA studies showing DME associations have been summarized in Table 2.
Phase I enzymes
CYP1A2, caffeine and blood pressure
Caffeine has long been known as one of the substrates and inducers of CYP1A2. In 2011, three independent GWA studies amounting to >72 000 subjects in total, found an association between CYP1A2 variation and the habitual consumption of coffee (see Table 2). In a genome-wide meta-analysis of 47 341 subjects, Cornelis et al.16 first presented an association between the behavioral caffeine intake trait and variations in both the CYP1A1-1A2 bidirectional promoter as well as a locus upstream the AHR that regulates the expression of CYP1A genes. These findings were later corroborated in another GW meta-analysis by Sulem et al.,17 whereas Amin et al.18 identified the CYP1A locus but not that of AHR. The effect of each individual CYP1A2 and AHR allele has been estimated to about 0.2 coffee cups per day.17 However, the identified polymorphisms at the two loci were together found to explain
CYP2A6 and smoking
CYP2A6 and smoking
CYP2A6 is the main enzyme involved in the metabolism of nicotine, and the influence of CYP2A6 polymorphism on smoking behavior has been extensively studied (see Gold and Lerman23). A recent GWA study by Thorgeirsson et al.24 found that SNP variation in both CYP2A6 (19q3) and nicotinic acetylcholine receptors (nAchR, 15q25) containing loci significantly affected cigarette consumption (European Network for Genetic and Genomic Epidemiology Consortium, n=31 266). Furthermore, the highly frequent CYP2A6*2 allele (rs1801272, allele frequency up to 10%) that gives rise to abolished CYP2A6 enzyme activity was found associated with a significantly lower number of cigarettes smoked per day (reduction of 0.68 cigarettes per day and CYP2A6*2 allele) (n=66 380, P=1.1 × 10−4), which was close to the effect observed for nAchR variation (0.80 cigarettes per day and allele, n=76 972, P=2.4 × 10−69).24 On the contrary, genome-wide significance for cigarette consumption was mainly found for the nicotinic receptor gene cluster and no apparent effect of CYP2A6 was observed in an analysis of 38 181 smokers (Tobacco and Genetics Consortium), although a follow-up analysis of 15 selected loci in 73 853 subjects identified a locus 40 kb downstream of CYP2A6 that was estimated to cause a 0.33 cigarette per day difference per allele.25 A smaller study (n=3441)26 found only nominally significant associations of CYP2A6 with cigarette consumption (0.27 cigarettes per day and allele for CYP2A6*2, P=2.88 × 10−5). A recent small and preliminary study examining functional magnetic resonance imaging of 31 subjects matched for cigarette consumption indicated that smokers with a reduced CYP2A6 phenotype or genotype (n=13–16) respond less to smoking cues in several regions of the brain compared with extensive metabolizers (EMs, n=15–18).27 However, this finding has to be reproduced in a larger set of individuals.
Thorgeirsson et al.24 also performed a genotype-based analysis of the CYP2A6 rs4105144 genotype, which exists in linkage disequilibrium with the defective CYP2A6*2 variant, in approximately 2000 lung cancer cases and 40 000 controls, where a small increased risk of lung cancer development was found for the major CYP2A6*1 allele (odds ratio, OR=1.09, P=0.04). Similarly, a reduced risk of lung cancer has been found for carriers of defective CYP2A6 alleles (for example CYP2A6*2 and *4) in Caucasian and Asian populations, and an enhanced CYP2A6 genotype effect was seen among smokers and for smoking-related lung cancer types (squamous cell lung carcinoma).28, 29, 30, 31 Thus, the link between CYP2A6 genotype and smoking has been emphasized recently, particularly with support from large GWA studies.
CYP2C9, CYP4F2 and warfarin
Coumarin-related drugs, used as anticoagulants for a variety of cardiovascular disease, are primarily metabolized by CYP2C9. They have a very narrow therapeutic index and a significant associated risk of bleeding and embolism. GWA studies on warfarin maintenance dosing have been performed in three different populations of Caucasians and Asians using univariate analyses.32, 33, 34 In all studies, VKORC1 emerged as the main contributor to determining the response to warfarin, whereas CYP2C9 showed a smaller but yet significant contribution in Swedish subjects (n=1053),33 a GW significant trend in Japanese subjects (n=1508),34 but being only moderately significant in a smaller cohort of European ancestry (n=181).32 After adjusting for VKORC1 and CYP2C9 genotype, age and gender by multiple regression analysis, CYP4F2 also reached GW significance in Swedes,33 whereas multiple regression did not lead to GW significance for CYP4F2 in Japanese subjects.34 CYP2C9 has also been found significant at a genome-wide level for acenocoumarol dosing in a study of 1450 subjects analyzed by a regression model adjusted for age, gender, BMI and the prothrombin international normalized ratio (INR) target measurement.35 Further adjusting for CYP2C9 and VKORC1 genotype identified CYP4F2 as GW significant in affecting acenocoumarol dosing.35 In support for the clinical relevance of genetic variants on warfarin dose variation, Epstein et al.36 found in a prospective study that the hospitalization rate for bleeding or thromboembolism caused by warfarin during the initial 6-month period is up to 43% lower in patients genotyped for VKORC1 and CYP2C9 polymorphisms to direct drug dosage (n=896) when compared with an age- and sex-matched non-genotyped control group from the same prescription benefit regime (n=2688). Furthermore, a study of 477 patients with VKORC1 and CYP2C9 genotype-guided warfarin dosing revealed that the guided method decreased the percentage of patients being out of range in INR and increased the time spent in therapeutic range to up to 3 months when compared with the 1866 patients treated in parallel by a standard scheme.37 Although both VKORC1 and CYP2C9 have been recently shown clinically relevant in warfarin guidance, further analyses of the pharmacological relevance is still necessary, especially for the specific impact of CYP2C9. At present, it appears that the genes together are able to predict about 35% of the variation in dosing; however, with VKORC1 being much more important than CYP2C9. Currently, drug regulatory agencies do not require genotyping before warfarin initiation; however, the warfarin drug label in the USA (Coumadin, FDA) presents three dosing subgroups based on the combined VKORC1 and CYP2C9 genotypes that should be considered if the patients’ genotype is known.
CYP3A4, statins and immunosuppressants
As mentioned, CYP3A4*22 (rs35599367), with an allele frequency of only a few percent, carries a mutation in intron 6 that causes reduced mRNA expression.15 Clinically, the CYP3A4*22 allele has been shown to influence the pharmacokinetics of cholesterol-lowering statin drugs15, 38 as well as the immunosuppressants tacrolimus39, 40 and cyclosporine.40 However, a follow-up study examining cyclosporine pharmacokinetics showed that the cyclosporine dose, blood concentration, rejection rate and delayed graft function were similar between cyclosporine-treated renal graft patients carrying the CYP3A4*22 allele (heterozygous, n=11) and non-carriers (n=161).41 Still, creatinine clearance was significantly reduced 3 months post transplant in CYP3A4*1/*22 patients as compared with those homozygous for the wild-type allele.41 In addition, when including co-variables in the analysis, the risk of delayed graft function was shown to be increased in CYP3A4*1/*22 patients, as 5 out of the 11 CYP3A4*22 carriers showed delayed graft function (45%) as compared with 26% in the total sample (n=39).41 The mechanism behind the potential increased risk for kidney impairment in transplantation patients carrying the CYP3A4*22 allele is presently unknown, and additional studies need to conform this finding.
CYP2C19 and clopidogrel
The antiplatelet agent Clopidogrel (Plavix) is extensively used worldwide for the prevention of ischemic events particularly in patients with coronary syndromes, percutaneous coronary intervention (PCI) and myocardial infarction. CYP2C19 has a major role in activating clopidogrel, and Hulot et al.42 were the first to demonstrate that the locus containing the CYP2C19 gene influenced its antiplatelet response. A proceeding GWA study on clopidogrel performed by Shuldiner et al.43 found that only the CYP2C19 polymorphism showed GW significance with respect to drug levels in healthy subjects, and the same study also showed that the defective CYP2C19*2 allele was associated with increased risk of cardiovascular events in patients. The clinical importance of CYP2C19 genotype on clopidogrel treatment has been extensively studied in recent years and a number of different meta-analyses, integrating between 7 and 32 studies with 8000–42 000 subjects per analysis, have been published.44, 45, 46, 47, 48, 49, 50, 51, 52 Most meta-analyses have addressed the effect of the defective CYP2C19 alleles (here called CYP2C19-def and mainly composed of CYP2C19*2 alleles) on clopidogrel treatment response. Some studies have also found that rapid CYP2C19 metabolism caused by the CYP2C19*17 allele is associated with reduced risk of cardiovascular events as well as increased risk of bleedings;45, 47, 48, 53 however, contradictory results have also been reported.46 The most recent and largest meta-analysis was published by Holmes et al.;44 it encompassed 32 studies and 42 016 patients and showed that the risk of bleeding was reduced in carriers of CYP2C19-def alleles (relative risk, RR=0.84; 95% confidence interval (CI), 0.75–0.94), with the risk of cardiovascular events also moderately increased (RR=1.18; 95% CI, 1.09–1.28), the significance of which however was lost when including only larger studies encompassing >200 events. Although Holmes et al.44 tend to smooth over the influences by CYP2C19 genotype on clopidogrel response, specific analyses of cardiovascular events such as myocardial infarction or stent thrombosis were all below 200 events per study where, for medium size studies (100–199 events per study), an increased risk for myocardial infarction as well as stent thrombosis was evident (RR=1.29, 95% CI, 1.06–1.58, and RR=1.54, 95% CI, 1.26–1.88, respectively). Strong criticism has been raised against the performance and interpretation of the meta-analysis, as studies that contained patients without a clear benefit of clopidogrel were included and emphasized, whereas the number of patients with PCI, in which clopidogrel has the highest indication and benefit, was low.54, 55, 56 Overall, the many meta-analyses have indicated the strongest effect of CYP2C19-def alleles on stent thrombosis, whereas CYP2C19*17 seems to have a similar effect on bleedings as well as cardiovascular events, although seemingly smaller in magnitude compared with CYP2C19-def.
Combined clinical variables such as age and diabetes have been suggested as more important in determining clopidogrel effects than CYP2C19-def genotypes.57 Interestingly, however, data available in Bouman et al.57 indicate that the defective CYP2C19*2 allele significantly impacts on platelet reactivity in response to clopidogrel more than any separate clinical variable. This is corroborated by Mega et al.58 who showed that increasing the dose of clopidogrel from 75 to 225 mg in CYP2C19-def heterozygote individuals leads to platelet reactivity measures similar to those of EMs (CYP2C19*1/*1), whereas however, sufficient inhibition of platelet reactivity could not be obtained in homozygous CYP2C19-def subjects with doses up to 300 mg. Furthermore, in a group of patients undergoing PCI, 91 subjects were subjected to a rapid genotyping scheme whereas 96 subjects went through standard clopidogrel treatment (75 mg per day).59 In the genotyping group, CYP2C19*2 carriers were allocated to prasugrel, with remaining subjects allocated to standard clopidogrel treatment.59 After 1 week of treatment, all 23 CYP2C19*2 carriers (100%) on prasugrel had sufficient inhibition of platelet reactivity, compared with 70% of CYP2C19*2 carriers (16 of 23) on standard treatment (P=0.009).59 In addition, platelet inhibition after 1 week of treatment among the CYP2C19*2 carriers was 73% in the genotype-guided group (prasugrel) but only 27% in the standard treatment group (clopidogrel;
Bouman et al.60 also found paraoxonase-1 to be a crucial enzyme in clopidogrel activation; however, not a single of many published studies have been able to confirm these findings, neither in gene-specific approaches (see for example, Sibbing et al.,61 Trenk et al.62) nor in GWA studies.43
Bouman et al.60 also found paraoxonase-1 to be a crucial enzyme in clopidogrel activation; however, not a single of many published studies have been able to confirm these findings, neither in gene-specific approaches (see for example, Sibbing et al.,61 Trenk et al.62) nor in GWA studies.43
In summary, despite the large number of studies published on this topic, guiding clopidogrel dosing based on CYP2C19 genotype is still a matter of debate. The significance of the CYP2C19 polymorphism is definitely stronger when one considers myocardial infarction and especially stent thrombosis. Whether the black box warning introduced by the FDA 2 years ago is relevant or of lower importance than initially thought is still being discussed. In addition, as the efficacy of the newer platelet aggregation inhibitors prasugrel and ticagrelor do not require activation of polymorphic enzymes, this issue might be less important in the future. On the basis of the initial prospective studies, however, it appears that genotype-guided treatment schemes is beneficiary for PCI patients in selecting the right antiplatelet regimen.
CYP2C19, CYP2D6 and tamoxifen treatment of breast cancer
Adjuvant antiestrogenic tamoxifen treatment has long been a standard for estrogen receptor-positive breast cancer. Conversion of tamoxifen into the high-affinity estrogen receptor antagonist endoxifen requires metabolism, which to a large extent is carried out by CYP2D6. There is a general consensus that the formation of endoxifen is highly linked to CYP2D6 polymorphism,63, 64, 65 whereas the effect of CYP2D6 genotype on the pharmacodynamic response to tamoxifen has not been clearly defined.65 The majority of clinical studies investigating CYP2D6 genotype in relation to tamoxifen response have included small patients cohorts and, in addition, have been limited in terms of coverage of CYP2D6 alleles. Two meta-analyses have been published in the field,66, 67 but both analyses were limited to the CYP2D6*4 and *10 alleles as well as covering mostly small studies, which resulted in a small to non-existent effect of CYP2D6 genotype on tamoxifen outcome. High CYP2D6 allele coverage is an important factor in accurate determination of CYP2D6 effects, as exemplified by a power increase from 8 to 63% and a concomitant decrease in the P-value of recurrence hazard ratio analyses by increasing the allele coverage from CYP2D6*4 only to 33 alleles.68, 69 Recently, larger studies on CYP2D6 genotype and tamoxifen response have been published, where six studies qualify for a ~500 participant selection of invasive non-metastatic breast cancer in either postmenopausal women or a mix of pre- and postmenopausal women where postmenopausal women were always composing at least 75%.68, 69, 70, 71, 72, 73 Median follow-up was at least 5 years for all studies and the treatment regimens were to a major extent non-chemotherapy-based (75–100%). Furthermore, a minimum of six CYP2D6 alleles were genotyped in all studies and breast cancer recurrence was the main outcome parameter in all but one study.73 Of utmost importance, however, is the serious criticism raised against two of the five studies on breast cancer recurrence that did not find any relationship between CYP2D6 genotype and tamoxifen response.71, 72 This is apparently due to a lack of Hardy–Weinberg equilibrium, which was caused by genotyping analyses of tumor instead of germline DNA samples, as tumoral chromosomal rearrangements can occur in the region containing CYP2D6.74 The remaining three large studies on breast cancer recurrence all found a significant effect of CYP2D6 polymorphism on tamoxifen treatment response (Table 3). Taking these large recurrence studies together, one might conclude an overall odds ratio of between 2 and 3 for poor metabolizers. However, it must be emphasized that two of the three studies have used both tumor and lymphocyte derived DNA for genotyping analyses.69, 70
Table 3 - Association of CYP2D6 genotype with tamoxifen (20 mg for 5 years) response in larger (around 500 patients or more) studies of invasive, non-metastatic breast cancer.
Recently, a GWA study presented associations of the C10orf11 gene with recurrence-free survival in 240 tamoxifen-treated breast cancer patients.75 Although CYP2D6 was not identified at the genome-wide level, it was shown that allelic variation in the C10orf11 gene together with variation in the two candidate genes CYP2D6 and ABCC2 had a cumulative effect on tamoxifen outcome with respect to the total number of risk alleles among all three genes.75 More studies are needed, in particular with respect to the relationship between CYP2D6 polymorphism and survival.
Polymorphism of CYP2C19 has also been implicated in the response to tamoxifen recently, however, the data thus far have been contradictive.76, 77, 78
CYP2B6 and non-nucleoside reverse transcriptase inhibitors (NNRTIs)
CYP2B6 metabolizes the two NNRTI's nevirapine and efavirenz that are used for the treatment of HIV infection (cf79). Nevirapine levels and clinical outcome has been associated with the CYP2B6 c.516G>T variant that causes a Q172H amino acid substitution commonly present in several different CYP2B6 alleles (see www.cypalleles.ki.se/cyp2b6.htm). In a study of 126 children, homozygosity for the 516T allele was associated with reduced nevirapine clearance and an enhanced increase in CD4+ T cells both at 12 and 24 weeks of treatment.80 Furthermore, in analyses of 175 subjects with cutaneous adverse events and 587 controls, the CYP2B6 516TT genotype appeared to modify the risk for nevirapine-induced and HLA-related (HLA-Cw*04) cutaneous adverse events, increasing the odds ratio from 2.4 (95% CI, 1.4–4.1) to 6.3 (95% CI, 2.5–15.7) when compared with the wild-type 516GG genotype.81 An enhanced frequency of early efavirenz treatment discontinuation (OR=2.6, 95% CI, 1.3–5.2) has also been associated with the CYP2B6 516TT genotype in a study of 373 efavirenz-treated patients,82 which however was not found in a smaller study of 105 subjects.83 In addition to NNRTI drug response, the CYP2B6 genotype at position 516 and 983 corresponding to CYP2B6*16 and *18 was shown to influence central nervous system (CNS) adverse events in Whites,84 thus supporting previously observed CNS-related effects (mood disorder, sleep disorder, fatigue).85, 86 In conclusion, the CYP2B6 polymorphism could be of significance for outcome of NNRTI treatment.
CYP3A5 and tacrolimus
Therapeutic drug monitoring is highly recommended for transplant patients receiving the immunosuppressive drug tacrolimus, as tacrolimus is subject to pharmacokinetic inter-patient variability and since its therapeutic window is small. Kidney transplantation is the most common indication for tacrolimus, and the drug is mainly metabolized by CYP3A enzymes where the common defective CYP3A5*3 allele is the most significant genetic determinant. In white populations, the frequency of carriers of the wild-type CYP3A5*1 allele (carriers are commonly referred to as CYP3A5 ‘expressors’) is only about 15%, whereas it is up to 50% and 90% in Asians and Blacks, respectively (http://www.ncbi.nlm.nih.gov/SNP/snp_ref.cgi?rs=776746). As reviewed by Barry and Levine,87 the weighted mean oral clearance of five studies is almost 50% lower (range 26–65%) in CYP3A5*3 homozygotes (CYP3A5 ‘non-expressors’) compared with CYP3A5*1 carriers. A recent meta-analysis by Tang et al.88 addressed dose requirements and rejection rates in 18 studies of renal transplant patients (n=1443) and five studies of liver transplants (n=336), and a clear effect of CYP3A5 genotype was claimed on tacrolimus dose at all treatment follow-up occasions (2 weeks to 12 months). However, an effect on rejection rates was concluded only after the first month of treatment, which is mirrored by the most prominent effect of CYP3A5 genotype on tacrolimus dose at this time point.88 Thus, CYP3A5 genotyping could be a useful tool to guide tacrolimus initiation doses in the prevention of early graft rejection.
CYP2C19, CYP2D6 and antidepressant treatment
CYP2D6 is involved in the metabolism of many antidepressants, and poor response of antidepressants has been associated with SNP polymorphism as well as the ultrarapid metabolizer (UM) phenotype caused by gene amplification,89, 90, 91, 92, 93 although negative findings have also been observed for CYP2D6 polymorphism.94, 95, 96 On the basis of the variable types of antidepressants, differences in outcomes measured and a relatively low sample size in most studies, no clear conclusion can at present be drawn on the impact of CYP2D6 genotype on response to antidepressants. As for duplicated CYP2D6 genes, suicide rates and suicide behavior has been shown to be enhanced in UMs.97, 98, 99, 100 An enhanced suicide effect could be due to increased antidepressant elimination and thus poor response, or to an endogenous effect not related to antidepressant treatment (see below).
CYP2C19 effectively metabolizes escitalopram and the serum levels vary according to CYP2C19 genotype.101, 102 The defective CYP2C19*2 allele has also been associated with depressive symptoms in older healthy subjects from the Swedish Twin Registry suggesting endogenous effects,103 and the rapid CYP2C19*17 allele has shown to reduce the remission rate within the group of subjects that tolerated escitalopram in the STAR*D study (OR=0.80, 95% CI, 0.63–1.00).104 Further studies are needed before conclusions can be drawn with respect to antidepressant response as well as endogenous functions for CYP2C19 and CYP2D6.
CYP2D6 and codeine
CYP2D6 activates codeine into the analgesic substance morphine. Although no analgesic effect is obtained in CYP2D6 poor metabolizers, UMs are at risk of excessive morphine levels causing, for example, sedation and respiratory depression both in adults and in infants of CYP2D6 UM breast-feeding mothers (see for example Sim and Ingelman-Sundberg1 and Supplements in Crews et al.105). Sistonen et al.106 retrospectively addressed infant CNS depression (sleepiness and lethargy) among 26 infant cases and 85 infant controls being breast-fed by mothers taking codeine. Although the main risk factor for infant CNS depression was maternal CNS depression (58% in infant cases vs 7% in infant controls, P=1.5 × 10−7), the maternal CYP2D6 genotype increased the risk of infant CNS depression (OR=17, P=0.043).106 FDA has included codeine drug label information on increased bioactivation in CYP2D6 UMs and the preference to choose lower doses for the shortest period of time in breast-feeding mothers as well as in the general population, to avoid overdose symptoms such as sleepiness, confusion or shallow breathing.
CYP2D6 and endogenous brain functions
CYP2D6 polymorphism has been suggested to influence personality traits, and allusions to an association with schizophrenia and Parkinson’s disease has been claimed (see Dorado et al.107). Recently, a number of studies have found a relationship between CYP2D6 genotype and suicidal behavior that is manifested in an overrepresentation of alleles with more than two CYP2D6 gene copies (UMs) in suicidal subjects.97, 98, 99, 100 Furthermore, this risk was suggested to lie solely or partly in a more severe type of suicidal behavior among UMs as found by Penas–Lledo et al.,97 thus potentially leading to a higher rate of deaths resulting from suicide attempts. It is known that CYP2D6 is able to metabolize CNS active substances and that CYP2D6 is expressed in the brain (see1). Recently, Kirchheiner et al.108 were able to show that brain blood perfusion levels were affected by CYP2D6 genotype. Poor metabolizers were found to have a 15% higher thalamic blood perfusion level than EMs at rest (P<0 .05="">0>
CYP2C9 and non-steroidal anti-inflammatory drugs (NSAIDs)
CYP2C9 and non-steroidal anti-inflammatory drugs (NSAIDs)
The effect of CYP2C9 genotype on the risk of gastrointestinal bleedings during non-steroidal anti-inflammatory drug treatment is at present ambiguous. A smaller meta-analysis of three studies111 indicated an increased OR of 1.8 (95% CI, 1.2–2.5) for variant CYP2C9 alleles (mainly CYP2C9*2 and *3). However, a few additional studies have rather added to the confusion by suggesting that the link between CYP2C9 genotype and bleeding risk is completely open.112 The non-steroidal anti-inflammatory drugs flurbiprofen and celecoxib have, however, without firm reasons, received FDA drug label information regarding CYP2C9, stating that known or suspected CYP2C9 poor metabolizers should administer the drugs with caution and in poor metabolizers (that is CYP2C9*3/*3) half the lowest recommended dose of celecoxib should be considered as the starting treatment.
Phase II enzymes
UGT1A1, irinotecan and endogenous functions
Irinotecan is a chemotherapeutic used in combination treatments of mainly colorectal cancer. Excessive levels of irinotecan’s bioactive metabolite, SN-38, can lead to severe neutropenia and this effect has been shown more pronounced in patients with the UGT1A1*28 allele, which carries an extended promoter repeat causing reduced UGT1A1 transcription and activity (see Sim and Ingelman-Sundberg1). The risk of neutropenia in UGT1A1*28/*28 subjects compared with carriers of none or one UGT1A1*28 allele was shown in a meta-analysis of a total of 1998 irinotecan-treated patients to be as significant in low dose (RR=2.4, 95% CI, 1.3–4.4) as in medium-dose patients (RR=2.0, 95% CI, 1.6–2.5), whereas the risk was significantly increased in high-dose patients (RR=7.2, 95% CI, 3.1–16.8).113 Owing to the increased risk of neutropenia in UGT1A1*28 carriers, the FDA has since 2005 recommended genotyping for UGT1A1*28 to select subjects benefiting from a lower initial irinotecan dose, although genetic testing is not required. Homozygosity for the UGT1A1*28 allele (>20% of subjects in certain populations) causes benign hyperbilirubinemia (Gilbert’s syndrome) due to a decreased rate of bilirubin conjugation by UGT1A1, and more detrimental UGT1A1 alleles can cause the more severe symptoms observed in Crigler–Najjar Syndrome.114 As expected, UGT1A1 was identified as the main component to influence serum bilirubin levels in GWA studies115, 116, 117 (Table 2). These findings have also been demonstrated in humanized mice carrying the UGT1A1*28 allele, which resemble subjects with Gilberts syndrome in terms of mild jaundice caused by increased bilirubin levels.118 Thus, the link between UGT1A1 polymorphisms and bilirubin levels is firmly established.
Thiopurine methyltransferase (TPMT) and thiopurine drugs
TPMT methylates the thiopurine drug 6-mercaptopurine (6-MP) that is used directly or administered as a prodrug (azathioprine) for the treatment of, for example, leukemia and chronic inflammatory disease such as Crohn’s disease. Excessive levels of 6-MP can cause myelosuppression and myelotoxicity and blood count is normally monitored during treatment. TPMT is highly involved in 6-MP metabolism and TMPT activity and TPMT genotype is known to affect the risk of toxicity. TPMT activity can be phenotyped or genotype-predicted, but the sensitivity for genotyping analyses in predicting reduced TPMT phenotypes has been ambiguous. Recently, heterozygosity or homozygosity for variant TPMT alleles were shown to yield odds ratios for leukopenia of 4.3 (95% CI, 2.7–6.9) and 20.8 (95% CI, 3.4–126. 9), based on 18 and 5 studies, respectively.119 In fact, TPMT genotyping or phenotyping (TPMT activity in red blood cells) is recommended by the FDA.
The field of DME polymorphism continuously develops and recently more substantial evidence has been obtained for the clinically most significant polymorphisms (see summarizing Figure 1). Indeed the recent studies have emphasized the importance of CYP2C19 polymorphism for the therapeutic effects of clopidogrel, and the role of CYP2D6 polymorphism for tamoxifen treatment appears to be relevant. The CYP2C9 polymorphism is relevant in particular, for predicting patients requiring low doses of warfarin, but is of lower importance than VKORC1 and demographic factors. The analgesic and side effects of codeine are influenced by CYP2D6 polymorphism, whereas the effect of CYP2D6 genotype on antidepressant response is without any firm evidence. It is, however, important to clarify the relation between CYP2D6 UMs and suicide behavior, especially in the light of recent reports on effects of CYP2D6 genotype on brain perfusion and brain glucose metabolism. Good evidence for the influence of CYP3A5 polymorphism on dosing of the immunosuppressive agent tacrolimus are at hand, and regarding phase II enzymes, anticancer treatment with mercaptopurines and irinotecan is important in relation to the polymorphism of TPMT and UGT1A1, respectively. Furthermore, interesting links between CYP1A2 variation and caffeine intake as well as blood pressure has been emphasized recently, as has also the previously debated association of CYP2A6 genotype with cigarette consumption.
The need for clarification of the cost-benefit of all these associations in the clinical setting remains an issue. At present, it is difficult to conclude which tests should be required, but recent research in the field encompassing much larger studies, where also a whole genome perspective is addressed, is the accurate way in order to be able to make firm and cost-effective recommendations for the drug treatment in the future.
1.Sim SC, Ingelman-Sundberg M. Pharmacogenomic biomarkers: new tools in current and future drug therapy. Trends Pharmacol Sci 2011; 32: 72–81.
2.Li J, Zhang L, Zhou H, Stoneking M, Tang K. Global patterns of genetic diversity and signals of natural selection for human ADME genes. Hum Mol Genet 2011; 20: 528–540.
3.Frueh FW, Amur S, Mummaneni P, Epstein RS, Aubert RE, DeLuca TM et al. Pharmacogenomic biomarker information in drug labels approved by the United States food and drug administration: prevalence of related drug use. Pharmacotherapy 2008; 28: 992–998.
4.Sadee W, Wang D, Papp AC, Pinsonneault JK, Smith RM, Moyer RA et al. Pharmacogenomics of the RNA world: structural RNA polymorphisms in drug therapy. Clin Pharmacol Ther 2011; 89: 355–365.
5.de Morais SM, Wilkinson GR, Blaisdell J, Nakamura K, Meyer UA, Goldstein JA. The major genetic defect responsible for the polymorphism of S-mephenytoin metabolism in humans. J Biol Chem 1994; 269: 15419–15422.
6.Ibeanu GC, Blaisdell J, Ghanayem BI, Beyeler C, Benhamou S, Bouchardy C et al. An additional defective allele, CYP2C19*5, contributes to the S-mephenytoin poor metabolizer phenotype in Caucasians. Pharmacogenetics 1998; 8: 129–135.
7.Lang T, Klein K, Fischer J, Nussler AK, Neuhaus P, Hofmann U et al. Extensive genetic polymorphism in the human CYP2B6 gene with impact on expression and function in human liver. Pharmacogenetics 2001; 11: 399–415.