Pharmacogenetics of CYP2C19: functional and clinical implications of a new variant CYP2C19*17

Abstract

AIMS

Cytochrome P450 2C19 metabolizes many important drugs. In 2006, a variant allele (CYP2C19*17) associated with increased activity was discovered, but its likely clinical significance is controversial. Investigators disagree about the phenotype to be assigned to the two CYP2C19*17 genotypes. The aim of this study was to provide a critical summary, helpful to prescribers.

METHODS

We searched MEDLINE for papers on the allele from 2006 and then undertook historical searches through the reference lists of papers retrieved. The relevant information was critically assessed and summarized.

RESULTS

CYP2C19*17 was associated with increased enzymic activity. Substrates studied were omeprazole, pantoprazole, escitalopram, sertraline, voriconazole, tamoxifen and clopidogrel. Most studies used pharmacokinetic variables as outcome measure. For clopidogrel, activated by CYP2C19, pharmacodynamic consequences focused on platelet aggregation. While for most pharmacokinetic parameters of the substrates studied the average value was altered, the range of values showed mostly complete overlap for CYP2C19*1/*17 heterozygotes and wild-type homozygotes. Even for CYP2C19*17 homozygotes, the absolute effect was modest compared with the effect of previously identified loss-of-function alleles. In Helicobacter pylori eradication CYP2C19*2 carriage was associated with an altered eradication rate (odds ratio 4.20, 95% confidence interval 1.23, 16.44) relative to the wild-type, but CYP2C19*17 homozygosity was not. Prevalence of the variant allele was typically <5% in Asians and about four times higher in White and African populations.

CONCLUSIONS

Assignment of CYP2C19*17 homozygotes as extensive metabolizers rather than ultrarapid metabolizers is adequate. CYP2C19*17 genotyping is unlikely to have clinical utility except for drugs with very narrow therapeutic indices.

WHAT IS ALREADY KNOWN ABOUT THIS SUBJECT

• CYP2C19 polymorphisms may have clinical consequences in relation to drugs extensively metabolized by the enzyme.

• Recently a new variant CYP2C19*17 was discovered.

• Unlike other commonly studied variant alleles, notably CYP2C19*2 and CYP2C19*3, which are defective, CYP2C19*17 is associated with enhanced enzyme activity and three CYP2C19 phenotypes have been suggested: ultrarapid metabolizers (UM), extensive metabolizers (EM) and poor metabolizers (PM).

• There appears to be interethnic variability in the prevalence of the CYP2C19*17 allele.

WHAT THIS STUDY ADDS

• We provide the first critical quantitative review of this rapidly developing area of research to serve as a basis for subsequent research and overviews.

• We summarize data on population prevalence and functional effects of CYP2C19*17.

• We argue on the basis of current evidence that potentially significant clinical effects are unlikely except for drugs with very narrow therapeutic windows. Of studied substrates, only clopidogrel may fall into this category. Even then, only homozygotes of the variant allele are likely to be at significantly increased risk.

• The assignment of CYP2C19*17 homozygotes as EM, rather than UM, is adequate as the metabolic ratios of all probe drugs studied so far overlap completely the range of values seen in wild-type homozygotes.

• The implications of CYP2C19*17 on the clinical effects of tamoxifen require further study.

Introduction

The cytochrome P450 (CYP) genes and pseudogenes are grouped into 18 families and 44 subfamilies according to their sequence similarity. Of these, the CYP1, CYP2 and CYP3 families are particularly important in the metabolism of drugs, collectively accounting for most of their Phase I biotransformation [1]. A recent rare example of participation of a member of another CYP family, CYP4, in the metabolism of a commonly used drug is that of warfarin [2]. Various single nucleotide polymorphisms (SNPs) of CYP enzymes are known to be functionally important and alleles with SNPs in complete disequilibrium are known. For better communication of the results of studies of CYP polymorphisms, what is known as the star nomenclature was suggested [3, 4]. This nomenclature is now almost universally adopted, not only for the CYP genes but also for other genes. The wild-type gene is normally referred to as the *1 allele. For example, the CYP2C19*1 allele refers to the wild-type allele in the CYP2C19 subfamily (http://www.cypalleles.ki.se/).

The enzyme cytochrome P450 2C19 (CYP2C19) participates in the metabolism of a wide array of therapeutic drugs, across many drug classes (Table 1) [5]. The CYP2C19 gene maps onto chromosome 10 (10q24.1-q24.3), and encodes a 490-amino-acid protein. Several polymorphisms of the gene are known to be associated with reduced enzyme activity. Notable among these are CYP2C19*2, characterized by a 681G→A substitution in exon 5 leading to a splice-defective site, and CYP2C19*3, which has a point mutation in exon 4, leading to a premature stop codon. The presence of these alleles is generally considered to be sufficiently predictive for the phenotypes to be inferred from them. Thus individuals homozygous for the CYP2C19*2 and CYP2C19*3 alleles are considered to be poor metabolizers (PM), while subjects with at least one CYP2C19*1 allele are classified as extensive metabolizers (EM). More recent alleles identified include CYP2C19*17-21, few of which have been functionally characterized [6–8]. While the population prevalence of a number of CYP2C19 alleles, and their functional consequences, are well documented [9–11], this is less so for CYP2C19*17. The focus of this critical appraisal is on CYP2C19*17, for which a number of studies have now been reported since its discovery in 2006. Early data suggest that its functional effect may be clinically important. Two specific aims of this study were to summarize data on population prevalence of this allele and to assess critically its functional significance. This is important, as different authors have assigned discordant phenotypes to the CYP2C19*17 homozygotes and heterozygotes. For example, Wang et al. consider those with the CYP2C19*1/*17 heterozygotes as ultrarapid metabolizers (UM), whereas Sugimoto et al. consider them as EMs within the same group as the wild-type CYP2C19*1 homozygotes.

Table 1.

Examples of CYP2C19 drug substrates

Drug	Drug class and therapeutic effect
Clopidogrel	Antiplatelet
Escitalopram	Antidepressant
Nelfinavir	Antiviral
Mephenytoin	Anticonvulsant (used as probe drug)
Omeprazole	Proton pump inhibitor; antacid
Lansoprazole
Cyclophosphamide	Cytotoxic agent
Teniposide
Amitriptyline Citalopram Clomipramine	Antidepressant
Moclobemide sertraline
Tamoxifen	Anti-oestrogen
Voriconazole	Antifungal
Proguanil	Antimalarial
Propranolol	β-Blocker
Diazepam	Anxiolytic agent

Other enzymes may be more important than CYP2C19 in the metabolism of these substrates (e.g. CYP2D6 for tamoxifen).

Methods

We undertook a systematic electronic search of the literature using MEDLINE through the PubMed portal. Given that the CYP allele we were searching for was discovered only in 2006, we aimed for maximum sensitivity and searched using CYP2C19*17 as keyword. We then supplemented this search with careful historical searches through each of the CYP2C19*17-specific papers retrieved and through the citation lists appended to each electronic journal source identified, where available. We also read major reviews of human cytochrome P450 published since 2006 irrespective of whether CYP2C19*17 was identified in the main search fields, to identify citations specific to this allele.

We focused on the quantitative aspects of the relevant studies identified, irrespective of the designs used, as we did not anticipate many randomized controlled trials focusing on CYP2C19*17. In any case, during this phase of translational research studies using other study designs are highly informative on likely clinical validity [12]. Where suitable data were available we obtained pooled estimates using conventional methods of meta-analysis for proportions or odds ratios (ORs) [13]. Plots were generated from both reported mean and confidence intervals (CIs), or from our own calculations using the reported mean values, sample sizes and error estimates. CIs were calculated using either exact methods or asymptotic approximations depending on sample size. Medians and ranges were obtained from tabular data from the relevant publications.

Results and discussion

The CYP2C19*17 allele

The CYP2C19*17 allele is characterized by two SNPs in the 5′-flanking region (g.-3402C > T and g.-806C > T) of the gene. The two polymorphisms are in complete linkage disequilibrium with each other. The apparently faster activity of the encoded enzyme was ascribed to the recruitment of transcription factors to the mutated g.-806C > T site [6].

Functional consequences

Pharmacokinetic end-points

Omeprazole metabolism

Omeprazole is a widely-used probe drug for CYP2C19. In their report identifying CYP2C19*17 for the first time, Sim et al. [6] investigated its effects on omeprazole metabolism and reported that the metabolic ratio (omeprazole/5-OH-omeprazole) was significantly different for homozygotes with the variant allele when compared with homozygotes with the wild-type allele. In Swedes the median metabolic ratio was 0.50 [interquartile range (IQR) 0.37–0.73] and 0.20 (IQR 0.12–0.37) in Ethiopians. Their theoretical predictions using linear interpolation suggest a 35–40% lower omeprazole area under the plasma concentration–time curve (AUC) in CYP2C19*17 homozygotes than in CYP2C19*1 homozygotes. CYP2C19*1/*17 heterozygotes behaved essentially like the wild-type homozygotes except that the spread of metabolic ratios was broader than in the CYP2C19*17 homozygotes (Figure 1). Thus, although the metabolic ratios of the variant and wild-type homozygotes were significantly different, the range of values overlaps virtually completely for all genotypes in both Swedes and Ethiopians (Figure 1).

Figure 1.

Metabolic ratio of omeprazole and mephenytoin in Swedes and Ethiopians of different genotypes

A more recent small study [14] produced data consistent with those of Sim et al. Twenty-four-hour AUCs were no different between CYP2C19*1 homozygotes (n= 6) and CYP2C19*1/*17 heterozygotes (n= 6) for both omeprazole and pantoprazole with both single dosing and multiple dosing. The CIs for the wild-type homozygotes completely envelop those corresponding to the CYP2C19*17 heterozygotes (Table 2). Although those authors also studied lansoprazole, the study was not informative as there was only one subject with the CYP2C19*1/*17 genotype for this comparison.

Table 2.

Area under the 24-h concentration–time curve (AUC₂₄) for omeprazole and pantoprazole according to subject genotype and dosing regimen

	AUC24 mg h−1 l−1
	CYP2C19*1/*1 genotype	CYP2C19*1/*17 genotype
Drug and regimen	(Mean and 95% CI)	(Mean and 95% CI)
Omeprazole 20 mg day 1	0.64 (0.37, 0.91)	0.49 (0.31, 0.67)
Omeprazole 20 mg daily day 6	1.11 (0.69, 1.53)	0.86 (0.41, 1.31)
Pantoprazole 40 mg day 1	4.56 (3.28, 5.84)	3.42 (1.74, 5.10)
Pantoprazole 40 mg daily day 6	4.21 (2.68, 5.74)	3.32 (2.26, 4.38)

Results calculated using data reported by Hunfeld et al. [14].

In another small study of healthy White volunteers, Baldwin et al. [15] compared the single-dose pharmacokinetics of omeprazole in CYP2C19*17 homozygotes (n= 5) and subjects with the wild-type CYP2C19*1/*1 genotype (n= 11). Although there was a significantly lower AUC_∞ of both omeprazole and of the sulphone metabolite between the two genotype groups, both sets of AUC_∞s overlapped. There was no difference in the AUC_∞ of 5-hydroxyomeprazole between the two genotype groups (Table 3). Therefore, even between CYP2C19*17 and CYP2C19*1 homozygotes, identifiable by an analytically reliable test, this would have poor clinical validity [16].

Table 3.

Pharmacokinetic parameter estimates of omeprazole and its two main metabolites after a single oral 40-mg dose of omeprazole to two groups of healthy volunteers with distinct CYP2C19 homozygous genotypes [15]

	Mean (95% CI) CYP2C19*1/*1	Mean (95% CI) CYP2C19*17/*17
Parameter compound (nmol h−1 l−1)	n= 11	n= 5
AUC∞ omeprazole (OME)	4151 (2084–6218)	1973 (1476–2469)
AUC∞ 5-hydroxyomeprazole (5-OH)	3359 (2803–3915)	2989 (2273–3706)
AUC∞ omeprazole sulphone	3343 (1301–5384)	1083 (690–1476)
AUC∞ OME/ AUC∞ 5-OH (dimensionless)	1.2 (0.70–1.6)	0.66 (0.54–0.79)

The omeprazole metabolic ratio was also not discriminative, an observation consistent with the data first reported with the discovery of the CYP2C19*17 allele (Figure 2).

Figure 2.

Metabolic ratio of omeprazole in subjects with different CYP2C19 genotypes

Sertraline metabolism

The antidepressant sertraline is metabolized by CYP2C19 [17]. Subjects with defective CYP2C19 alleles (CYP2C19*2 and CYP2C19*3) showed higher blood levels of sertraline than those with the wild-type allele (CYP2C9*1) after administration of the same dose [17]. In contrast, the CYP2C19*17 allele did not influence dose-adjusted serum concentrations of either sertraline or N-desmethyl sertraline, its metabolite [18].

Escitalopram metabolism

Two recent studies have investigated the effect of CYP2C19*17 on escitalopram metabolism in healthy volunteers [19] and in psychiatric patients [20]. In their comparison of volunteers, homozygous for CYP2C19*17 and homozygous for CYP2C19*1, reported by Rosenborg et al., subjects were given 5 mg escitalopram twice daily for a week. The 12-h AUC of escitalopram was 21% lower in the former group with a probability significance level of 0.08. Based on this difference, which did not reach the conventional level of significance, the authors concluded that a clinically significant difference was seen in escitalopram kinetics. Given the considerable overlap in the AUC values observed in the two groups (mean 379; 95% CI 274, 484 vs. 300; 95% CI 228, 371 nm h⁻¹), this claim appears unjustifiable. The 12-h AUC of desmethylescitalopram, the principal metabolite, was also no different (184; 95% CI 115, 224 vs. 179; 164, 204 nM h⁻¹; P= 0.26).

In their study, Rudberg et al. [19] reported that the ratio of dose-adjusted escitalopram geometric mean concentration in CYP2C19*17 homozygotes relative to that of CYP2C19*1*1 subjects was significantly different to 1 (P= <0.01), as was the corresponding ratio for escitalopram (P < 0.01), after adjustment for gender and CYP2D6 defective alleles. However, the range of concentrations (nM mg⁻¹ day⁻¹) seen in the two genotype groups showed considerable overlap for both escitalopram (1.59; 95% 1.12, 2.25 in CYP2C19*17 homozygotes vs. 2.72; 95% CI 2.42, 3.07 in CYP2C19*1 homozygotes), and desmethylescitalopram (1.41; 95% CI 1.12, 1.78 vs. 2.02; 95% CI 1.86, 2.18) despite the significant difference in ratio of geometric means (Figure 3).

Figure 3.

Range of concentrations of escitalopram seen in subjects with different CYP2C19 genotypes. Reprinted by permission from Macmillan Publishers Ltd: [Clin Pharmacol Ther 2008; 83: 322, copyright (2008)

On the basis of their results, Rudberg et al. suggest that CYP2C19*17 subjects may have an ‘increased risk of therapeutic failure’ when given conventional doses of escitalopram [19]. The robustness of this inference is compromised by the fact that concentration measurements were made over a wide sampling-time window of 10–30 h after the last dose. While inclusion of ‘practice’ or ‘real-world’ patients improves generalizability of their results, this makes estimates of effect less precise and the development of a robust statistical predictor more difficult.

Voriconazole metabolism

Voriconazole, one of the newer antifungal agents, is extensively metabolized by CYP2C19 and to a lesser extent by CYP3A4 and CYP2C9 [21, 22]. In a study of the metabolism of voriconazole in healthy Chinese male volunteers, Wang et al. [22] classified those with the CYP2C19*1/*17 genotype as UMs. Despite screening 315 male volunteers, none was found to be a CYP2C19*17 homozygote, suitable for inclusion in their pharmacokinetic study. In common with other researchers, they classified CYP2C19*1/*1 subjects as EMs and CYP2C19*2/*2 subjects as PMs. They found that the 24-h area under the voriconazole concentration–time curve (AUC₂₄) was significantly lower in CYP2C19*1/*17 heterozygotes (3.39; 95% CI 2.88, 3.90 µg h⁻¹ ml⁻¹) than in the wild-type homozygotes (6.18; 95% CI 4.81, 7.55 µg h⁻¹ ml⁻¹) and in the latter than in the CYP2C19*2 homozygotes (16.30; 95% CI 14.75, 17.85 µg h⁻¹ ml⁻¹). These significant differences held when the AUC₂₄ was extrapolated to infinity (AUC_∞).

Pharmacodynamic and clinical end-points

Helicobacter pylori eradication

Several studies have now shown that the eradication of Helicobacter pylori with proton pump inhibitor (PPI)-containing triple therapy is influenced by polymorphisms of the CYP2C19 gene. In particular, eradication of the bacterium is more effective in CYP2C19*2 and CYP2C19*3 homozygotes and CYP2C19*2/*3 compound heterozygotes than in patients who are CYP2C19*1 homozygotes when treated with triple therapy including omeprazole and other PPIs extensively metabolized by the enzyme [23]. Therefore, it is possible that with the same dose of PPI, eradication may be less effective in CYP2C19*17 than in CYP2C19*1 carriers. Kurzawski tested this hypothesis and showed that the eradication rate was not associated with CYP2C19*17 carrier status. Figure 4 shows interestingly that while CYP2C19*2 carrier status was associated with an increased eradication rate, even CYP2C19*17 homozygotes showed no effect [24]. However, it is important to note that pantoprazole is not the most specific PPI substrate for the effect of CYP2C19*17 and a more significant effect may be seen with other agents within the same class.

Figure 4.

Odds ratio of successful eradication of Helicobacter pylori according to CYP2C19 genotype [24]

Clopidogrel

Clopidogrel is an antiplatelet prodrug that undergoes metabolic activation by CYP3A4 and CYP2C19 [25, 26]. Therefore, it is not surprising that carriage of the loss-of-function allele CYP2C19*2 has been associated with reduced effectiveness of clopidogrel, measured by adenosine diphosphate-induced platelet aggregation, in both healthy volunteers [27] and patients with coronary artery disease [28, 29]. However, CYP2C19*17 genotype was not associated with responsiveness to a loading dose of clopidogrel [30]. Generally, any poorer response is seen with the PM phenotype due to less extensive prodrug activation [28, 31]. Adverse consequences due to more rapid metabolism are unlikely to be picked up by the underpowered study in this respect. In a more recent study of the pharmacogenetics of clopidogrel in the Amish population, CYP2C19*2 genotype was associated with diminished platelet response but CYP2C19*17 was not [29].

Tamoxifen

Tamoxifen is an anti-oestrogenic compound, widely used in oestrogen receptor-positive breast cancer. Although the parent molecule is active, most of its activity is thought to be due to two of its metabolites, 4-hydroxytamoxifen and 4-hydroxytamoxifen (endoxifen). CYP2D6 is generally thought to be the key enzyme for the formation of the two metabolites, although other cytochrome P450 enzymes, including CYP2C19, are also known to be involved. In a recent case–control study, archival tissues were screened for CYP2D6, CYP2B6, CYP2C9, CYP3A4 and CYP2C19 variant alleles. CYP2C19*17 was surprisingly identified as a predictor of better tamoxifen response relative to carriers of alleles (CYPC19*1, *2 and *3) associated with impaired enzymic activity. The comparative hazard ratio for relapse was 0.45 (95% CI 0.21, 0.92). The hazard ratio relative to the wild-type allele carriers was 0.58 (95% CI 0.32, 1.01) [32].

Population distribution

In their seminal paper, Sim et al. [6] reported on the low frequency of the CYP2C19*17 allele in Chinese subjects (4%) relative to Ethiopians and Swedes, which had the same distribution (18% in both). This wide interethnic variability in allele frequency has now been confirmed by several studies, as shown in Figure 5[5, 18, 19, 22, 24, 31, 33–37]. The figure also shows pooled estimates of frequencies where more than one study is available on the same ethnic population. The Japanese, Chinese and Koreans have similarly low frequencies compared with the White and Black populations surveyed.

Figure 5.

Frequency of CYP2C19*17 allele in different populations

Relative importance

The CYP2C19*2 and CYP2C19*3 alleles are associated with loss of function and therefore it can be expected that they would be associated with increased activity of drugs, such as omeprazole, which are primarily deactivated via this pathway. In contrast, prodrugs such as clopidogrel, which are activated significantly via this pathway, would be expected to show loss of activity. CYP2C19*17 can be expected to show opposite effects to those seen with the null alleles. The evidence reviewed here suggests that, while this is the case, the magnitude of effect of the CYP2C19*17 allele is considerably smaller than that of CYP2C19*2 and CYP2C19*3. Even in homozygotes, any observed ultrarapid metabolic profiles have been within the range seen in the wild-type homozygotes. Figure 6 illustrates the comparative magnitude of effect. For tamoxifen, the clinical implications require urgent prospective validation.

Figure 6.

Metabolic ratio for omeprazole as a function of CYP2C19 genotype

CYP2C19*17 and breast cancer risk

A possible pharmacogenetic effect, leading to an observed association between CYP2C19*17 and breast cancer risk, was suggested by Justenhoven et al. [38]. In their study of 1015 cases and 1021 controls, single-factor analysis suggested a reduced risk of breast cancer for CYP2C19*17 heterozygotes (OR 0.80, 95% CI 0.66, 0.97; P= 0.021) and homozygotes (0.64, 95% CI 0.44, 0.94; P= 0.024). However, the associations became nonsignificant after adjusting for multiple testing. Further subgroup analysis suggested that for women using hormone therapy for at least 10 years, the decreased risk of breast cancer was significantly associated with carriage of at least one CYP2C19*17 allele (OR 0.57, 95% CI 0.39, 0.83; P= 0.004). The authors rationalized the reduced cancer risk to the CYP2C19*17 genotype increasing the catabolism of oestrogens. The reported association is therefore biologically plausible, as an association between excessive oestrogen exposure and breast cancer risk is well established. However, given the observational nature of the study, the weak strength of associations seen, the complex multifactorial breast cancer trait, the subgroup analyses necessary, and the known high susceptibility of such studies to bias leading to false-positive associations [39], robust validation of this association in different populations is required.

Conclusion

The CYP2C19*17 allele is associated with increased enzymic activity. However, the magnitude of effects is considerably smaller than has been reported with CYP2C19*2 and CYP2C19*3, albeit in opposite directions. The functional effects of CYP2C19*17 are unlikely to be clinically significant except for drugs with very narrow therapeutic windows. Of the drugs reviewed, only clopidogrel and tamoxifen may be of concern. For clopidogrel, only CYP2C19*17 homozygotes are likely to be at significantly increased risk of experiencing the effects of excessive inhibition of platelet aggregation. The assignment of CYP2C19*17 heterozygotes as EM, rather than UM, is adequate as the metabolic ratios of all probe drugs studied so far overlap completely the range of values seen in wild-type homozygotes. For the other CYP2C19 substrates commented on, genotyping for CYP2C19*17 is unlikely to have clinical utility.

Competing interests

None to declare.