Is CYP2C Haplotype Relevant for Efficacy and Bleeding Risk in Clopidogrel-Treated Patients?

A recently discovered haplotype—CYP2C:TG—determines the ultrarapid metabolism of several CYP2C19 substrates. The platelet inhibitor clopidogrel requires CYP2C19-mediated activation: the risk of ischemic events is increased in patients with a poor (PM) or intermediate (IM) CYP2C19 metabolizer phenotype (vs. normal, NM; rapid, RM; or ultrarapid, UM). We investigated whether the CYP2C:TG haplotype affected efficacy/bleeding risk in clopidogrel-treated patients. Adults (n = 283) treated with clopidogrel over 3–6 months were classified by CYP2C19 phenotype based on the CYP2C19*2*17 genotype, and based on the CYP2C19/CYP2C cluster genotype, and regarding carriage of the CYP2:TG haplotype, and were balanced on a number of covariates across the levels of phenotypes/haplotype carriage. Overall, 45 (15.9%) patients experienced ischemic events, and 49 (17.3%) experienced bleedings. By either classification, the incidence of ischemic events was similarly numerically higher in PM/IM patients (21.6%, 21.8%, respectively) than in mutually similar NM, RM, and UM patients (13.2–14.8%), whereas the incidence of bleeding events was numerically lower (13.1% vs. 16.6–20.5%). The incidence of ischemic events was similar in CYP2C:TG carries and non-carries (14.1% vs. 16.1%), whereas the incidence of bleedings appeared mildly lower in the former (14.9% vs. 20.1%). We observed no signal to suggest a major effect of the CYP2C19/CYP2C cluster genotype or CYP2C:TG haplotype on the clinical efficacy/safety of clopidogrel.

Clopidogrel is the oldest and most extensively studied of the second-generation thienopyridines-platelet aggregation inhibitors of the P2Y12 receptor antagonist class.It is indicated for the prevention of occlusive arterial incidents (ischemic incidents) in patients at increased risk.As a monotherapy, or combined with aspirin, clopidogrel is indicated in patients who underwent percutaneous coronary intervention (PCI), embolization of cerebrovascular aneurisms or arterio-venous malformations (AVM), cerebrovascular stenting, in patients who have sustained acute myocardial infarction/acute coronary syndromes, transitory ischemic attack (TIA) or ischemic stroke, and in those with verified peripheral artery diseases (PAD) [6][7][8].Regarding clopidogrel, CYP2C19 activity has a specific role: clopidogrel is a prodrug, and after oral administration, around 85% is inactivated in the first pass by the hepatic carboxylesterase 1 (CES1), whereas the remaining 15% requires enzymatic biotransformation into an active moiety, a reversible P2Y12 antagonist.While several hepatic CYP450 enzymes are involved, CYP2C19 is the most important one.This is a delicate role, since inadequate clopidogrel activation results in recurrent ischemic events, whereas excessive activation might, at least theoretically, increase the risk of bleeding [9][10][11][12][13][14].The situation is further complicated by the fact that antiplatelet-treated patients commonly receive gastroprotective treatments, typically proton pump inhibitors, all of which are CYP2C19 substrates, and some (omeprazole, esomeprazole) are rather strong enzyme inhibitors [15].The platelet response to clopidogrel shows considerable interindividual variability, and some 4-30% of patients fail to achieve adequate platelet inhibition [16,17].After more than a decade of research, PM and IM CYP2C19 phenotypes have been extensively shown to result in lower active metabolite levels, and higher (preserved) platelet reactivity than NM (and, likely, RM and UM) phenotype, and are associated with an increased risk of major ischemic events (reviewed in, e.g., [4,12,18,19]).On the other hand, RM and UM patients display the highest active metabolite levels, but it is unclear whether this means greater efficacy and/or greater bleeding risk vs. NM patients-some studies have suggested that both could be the case [20][21][22][23][24], whereas others have suggested no association between the *17 allele and clopidogrel pharmacokinetics, pharmacodynamics, and ischemic or bleeding outcomes (i.e., no relevant difference between NM, RM, and UM subjects) after accounting for the *2 allele [25][26][27][28][29].It has been postulated that the observed relationship between the *17 allele and clopidogrel antiplatelet effects may be due to absence of the no function *2 allele rather than presence of the increased function *17 allele [12,28].The clinical utility of CYP2C19 genotype-guided strategy for selection of P2Y12 inhibitors has been evaluated in multicenter randomized clinical trials [18]: PHARMCLO [30], POPular-Genetics [31], and TAILOR-PCI [32,33].Overall, genotyping for the CYP2C19 no function alleles and identification of PM and IM subjects to guide P2Y12-targeted antiplatelet therapy has a reasonable clinical utility, whereas the identification of RM/UM subjects is not considered relevant [34].
Recently, a novel haplotype composed of two non-coding variants in the CYP2C gene cluster, CYP2C18 NM_000772.3:c.*31T(rs2860840) and NM_000772.2:c.819+2182G(rs11188059), has been identified, and is reported to be associated with the ultrarapid metabolism of CYP2C19 substrates escitalopram and sertraline [35,36].The haplotype is referred to as "CYP2C:TG".It has been reported to be associated with exposure to and efficacy of a proton pump inhibitor omeprazole [37] and the azole antifungal voriconazole [38].Another study found no association between the CYP2C:TG haplotype and in vivo exposure to CYP2C19 substrates citalopram, sertraline, voriconazole, omeprazole, pantoprazole, and rabeprazole in healthy volunteers, and no association between this haplotype and CYP2C18/2C19 abundance in human liver sample or activity in vitro [39].
Considering that the role of RM/UM phenotypes in bioactivation of clopidogrel is still doubtful, we aimed to explore whether accounting for the CYP2C:TG might add information about the risk of ischemic and bleeding events in patients treated with clopidogrel over 3-6 months.For this purpose, the incidence of events was analyzed with respect to the "classically" defined CYP2C19 phenotypes, that is, based on the CYP2C19*2*17 genotypes, phenotypes defined with respect to the CYP2C cluster genotype, that is combined CYP2C19*2*17 and CYP2C rs11188059 and rs2860840, and with respect to the presence of the CYP2C:TG haplotype.

Study Outline
The present data are preliminary results from a larger prospective study ("Pharmacogenomics in Prediction of Cardiovascular Drugs Adverse Reactions") that started on 1 January 2022, and will last 60 months and include 1200 subjects.The study (Clinical-Trials.gov,NCT05307718) is conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committees of the School of Medicine, University of Zagreb, (reg.number 380-59-10106-20-111/125; class 641-01/20-02/01) and the University Hospital Centre Zagreb (class 8.1-20/142-2; number 02/21 AG), Croatia.All subjects provided written informed consent for genotyping the pharmacogenes of interest, and for publishing anonymized data for scientific purposes.The present analysis included consecutive adults (≥18 years of age), unrelated Caucasians of European (Slavic) descent, who started clopidogrel due to percutaneous coronary intervention (PCI), embolization of cerebrovascular aneurisms (AVM), or cerebrovascular stenting requiring >3 months of therapy, and were followed up over at least 3 and to a maximum of 6 months (all patients with no events were observed over 6 months) between 1 January 2022 and 1 March 2024.The decision to prescribe clopidogrel was at the discretion of attending cardiologist/stroke neurologist, and was in line with the approved prescribing information including recommendations to avoid it in patients who need to be treated with strong CYP2C19 inducers or inhibitors.Patients were genotyped for CPY2C19*2, CYP2C19*17, CYP2C rs11188059G>A, and rs2860840C>T.Phenotypes (PM, IM, NM, UM, RM) were determined based on the CYP2C19*2,*17 genotype, and on combined CYP2C19*2,*17 and CYP2C18 rs11188059 and CYP2C18 rs2860840 genotypes (cluster genotype).They were also classified as CYP2C:TG carriers and non-carriers.Phenotypes/haplotype carriage were considered as "exposures" indicating genetically determined enzyme activity, and incidence of ischemic and bleeding events were recorded.Levels of exposure were then balanced on a number of relevant covariates, and their effects on outcomes of interest were estimated.

Patients
HIV patients were not included in this study.Patients undergoing PCI were not included if older than 80 years of age, were treated with continuous post-interventional glycoprotein IIb/IIIa inhibitors, had thrombocytopenia (<150 × 10 9 /L), more pronounced renal failure (creatinine > 200 µmol/L), anemia (hematocrit < 30%), hemorrhagic diathesis, history of major surgery within previous 6 weeks, or hemorrhagic stroke within previous 6 months.No particular exclusion criteria were implemented for neurovascular patients (i.e., undergoing cerebrovascular stenting or embolization of cerebrovascular aneurisms/arterio-venous malformations).

Genotyping Procedures
Genomic DNA was extracted from whole blood samples using the QIAamp DNA Blood Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer's protocol.Genotyping was performed using TaqMan ® Drug Metabolism Genotyping Assays or TaqMan ® SNP Genotyping Assays (Applied Biosystems, Carlsbad, CA, USA) for CYP2C19*2 (c.681G>A, rs4244285; assay ID: C__25986767_70), CYP2C19*17 (c.-806C>T, rs12248560; assay ID: C____469857_10), CYP2C18 c.*31C>T (rs2860840; assay ID: C__11201742_10) and CYP2C18 c.819+2182G>A (rs11188059; C__31983321_10), and TaqMan ® Universal PCR Master Mix (Applied Biosystems, Carlsbad, CA, USA) by real-time PCR genotyping on the 7500 Real-Time PCR System (Applied Biosystems, Carlsbad, CA, USA), according to the manufacturer's instructions.Genotyping of CYP2C19*2 and *17 was performed with methods validated for routine pharmacogenetic testing in the clinical laboratory included in external quality assessment schemes, while rs11188059 and rs2860840 variants were validated by frequency for Caucasian population.Genotyping of CYP2C19*2 and *17 was performed with methods implemented and validated for routine pharmacogenetic testing in the clinical laboratory that regularly participates in external quality assessment schemes (European Molecular Genetics Quality Network-EMQN; Reference Institute for Bioanalytics-RfB).
The following bleeding events were recorded: (i) all intracranial bleedings (e.g., nontraumatic intraparenchymal hematoma, subarachnoid, intraventricular, or subdural hemorrhage [40]), except for microhemorrhages and minor asymptomatic sulcal subarachnoid hemorrhages; (ii) any extracranial bleeding that required at least non-surgical medical intervention; (iii) any intraspinal or intraocular bleeding that compromised vision; (iv) fatal bleeding of any location or cause.
No grading based on severity of either ischemic or bleeding events was employed.

Data Analysis
All ischemic events and all bleeding events were considered jointly, regardless of severity or location.To estimate the effects of CYP2C19 phenotypes and of CYP2C:TG haplotype, different phenotypes (levels of exposure) and CYP2C:TG carriers and noncarriers were mutually balanced on several covariates that could have affected the outcome of interest: age, sex, indication for clopidogrel treatment (3 levels: percutaneous coronary intervention, embolism of cerebral aneurism, cerebrovascular stenting), chronic kidney, heart or liver failure (jointly), hypertension, diabetes or dyslipidemia, concurrent use of aspirin, concurrent use of anticoagulants, concurrent use of gastroprotective treatments (yes or no), and concurrent use of proton pump inhibitors (yes or no).For this purpose, we used entropy balancing [41] implemented in package WeightIt (Version 1.1.0)[42] in R [43].Entropy balancing is a data preprocessing method that (where this is possible) guarantees covariate balance via a reweighting scheme that assigns a scalar weight to each sample unit such that the reweighted groups satisfy a set of balance constraints that are imposed on the sample moments of covariate distributions [41].Standardized mean differences (d) < 0.1 between pairs of exposure (phenotype, haplotype) levels were considered to indicate adequate balance.Balanced data were analyzed by fitting weighted log-binomial models with robust variance estimation to binary outcomes to generate adjusted proportions.Since both the sample and number of events were limited, confidence intervals around relative risks would have been rather wide.Instead, we present adjusted proportions, their confidence intervals, and Cochran-Armitage test of trend in adjusted (weighted) proportions, where appropriate.We used SAS 9.4 for Windows (SAS Inc., Cary, NC, USA).

Outcomes across CYP2C19 Phenotypes-Raw Data
Although some phenotypes were uncommon (e.g., PM, UM by *2,*17 classification, NM by cluster-based classification), a tendency of a numerically higher incidence of ischemic events and a lower incidence of bleeding events was apparent for PM/IM vs. NM, RM and UM phenotypes by either classification (Figure 1A,B).The tendency was more clearly visible when phenotype levels was collapsed to 3 (based on a biological rationale) (Figure 1C) for the combined PM/IM phenotype (n = 60), NM (n = 88 based on *2,*17, n = 23 based on the cluster-genotype), and RM/UM phenotypes (n = 135 based on *2,*17, n = 200 based on cluster-genotype) (Figure 1C).For the cluster-based phenotype, the tendency was apparent also when phenotypes were collapsed to PM/IM, combined NM and RM (n = 128), and UM (n = 95) (Figure 1C).This might be biologically more plausible, since a large prevalence of UM subjects in this phenotype classification is based predominantly on a "switch" of many of the NM and some of the RM subjects to the UM phenotype, indicating that there is a greater similarity between NM and RM subjects than between RM and UM subjects.

Outcomes across Phenotypes-Balanced Data
Patients across the collapsed levels of CYP2C19 phenotypes (PM/IM combined, NM, and RM/UM combined based on *2,*17; and PM/IM, NM, and RM combined, and UM based on the CYP2C19/CYP2C cluster) differed largely with respect to most of the covariates (Table 2).

Outcomes across Phenotypes-Balanced Data
Patients across the collapsed levels of CYP2C19 phenotypes (PM/IM combined, NM, and RM/UM combined based on *2,*17; and PM/IM, NM, and RM combined, and UM based on the CYP2C19/CYP2C cluster) differed largely with respect to most of the covariates (Table 2).After entropy balancing, all subsets were mutually closely similarly with an adequate balance (all d < 0.1) (Table 3).The adjusted proportions showed similar trends as the raw data: numerically higher proportion of ischemic events and lower proportion of bleeding events in PM/IM patients vs. NM and combined RM/UM patients, or vs. the combined NM/RM and UM patients (Figure 2).groups, with coefficients of variation of 0.470, 0.410, and 0.315, respectively. 2Effective sample sizes were 49.6 for PM/IM, 117 for NM/RM, and 85.5 for UM.Entropies by phenotype level were 0.102, 0.046, and 0.052, respectively.Weights ranged from 0.486 to 1.903 in all groups, with coefficients of variation of 0.462, 0.314, and 0.335, respectively.AVM-aneurisms of arteriovenous malformations; CAD-coronary artery disease; CHF-chronic heart failure; CKD-chronic kidney failure; CLDchronic liver failure; CV-cerebrovascular; CVI-cerebrovascular incident; PAD-peripheral artery disease; PCI-percutaneous coronary intervention; PPI-proton pump inhibitor.The sample was modest in size and the number of events was low, so formal tests of trend in proportions for all outcomes yielded relatively low z-values.However, despite this lack of precision of the estimates, numerical trends appeared obvious and were similar for both types of phenotype definitions (Figure 2).

Outcomes Regarding the CYP2C:TG Haplotype
The CYP2C:TG carriers (n = 162) and non-carriers (n = 121) mildly differed regarding the covariates, and the prevalence of ischemic events (14.8% vs. 17.4%) was generally similar between them (Table 4).The prevalence of bleeding events appeared to be slightly lower in haplotype carriers (14.8% vs. 20.7%,d = −0.153).After entropy balancing, virtually identical values for all covariates were achieved for haplotype carriers and non-carriers: the incidence of ischemic events remained closely similar (14.1% vs. 16.1%) in CYP2C:TG The sample was modest in size and the number of events was low, so formal tests of trend in proportions for all outcomes yielded relatively low z-values.However, despite this lack of precision of the estimates, numerical trends appeared obvious and were similar for both types of phenotype definitions (Figure 2).

Outcomes Regarding the CYP2C:TG Haplotype
The CYP2C:TG carriers (n = 162) and non-carriers (n = 121) mildly differed regarding the covariates, and the prevalence of ischemic events (14.8% vs. 17.4%) was generally similar between them (Table 4).The prevalence of bleeding events appeared to be slightly lower in haplotype carriers (14.8% vs. 20.7%,d = −0.153).After entropy balancing, virtually identical values for all covariates were achieved for haplotype carriers and non-carriers: the incidence of ischemic events remained closely similar (14.1% vs. 16.1%) in CYP2C:TG carriers and non-carriers, whereas the difference in bleeding events was minimal (14.9% vs. 20.1%)(Table 4).Table 4. Patient characteristics in CYP2C:TG carriers (TG-other, TG-TG) and non-carriers in the CYP2C cluster diplotype and ischemic and bleeding events before and after entropy balancing.Data are mean ± SD for age, and count (percent) for raw data, or percent in the entropy-balanced data 1 .Depicted are standardized mean differences (d; values < 0.1 indicate fully adequate balance).Shown are also outcomes of interest, i.e., incidence of ischemic and bleeding events: for raw data (before entropy balancing of covariates), simple proportions are provided; for adjusted (balanced) data, incidence is given with 95% confidence intervals.

Before Entropy Balancing
After Entropy Balancing

Discussion
As a result of extensive investigations-including genome-wide association studies and candidate-gene analyses of observational data and data from large randomized trialsthe current guidelines by the Clinical Pharmacogenetic Implementation Consortium on the CYP2C19 genotype (genotype-predicted phenotype) in clopidogrel-treated patients strongly recommend that in cardiovascular indications, clopidogrel should be avoided in PM/IM CYP2C19 metabolizers (preference towards the non-CYP2C19-dependent P2Y12 antagonists like prasugrel or ticagrelor) [4].A moderate recommendation for clopidogrel avoidance in PM/IM patients refers to cerebrovascular indications [4].No recommendation pertains to RM or UM patients in either indication, implying that differentiation between NM, RM, and UM subjects appears to be of no clinical utility [4].The considerable inter-subject variability in response to clopidogrel is considered largely genetically determined [44], and only partly explained by identification of the loss-of-function CYP2C19 alleles [45][46][47].
Polymorphisms in several other genes (e.g., CYP2B6, CES1, SCOS5P1, CDC42BPA, CTRAC1, ABCB1, G4GALT2, P2RY12, PON1) have been suggested as relevant for the antiplatelet effects of clopidogrel, although their impact on "hard" clinical outcomes is yet to be clarified [27,28,48,49].It is reasonable to envisage a risk-grading system encompassing multiple genetic variants and non-genetic factors that would be more predictive and reliable than any individual indicator alone, regarding the clinical efficacy/safety of clopidogrel in cardiovascular and neurovascular indications [27,50].
The newly discovered CYP2C:TG haplotype has been shown associated with accelerated metabolism (by around 25%) of CYP2C19 substrates escitalopram and sertraline [35,36], and it has been suggested that it could affect the exposure to/effects of other CYP2C19 substrates such as omeprazole [37] and voriconazole [38].However, opposing observations (not indicating effects of CYP2C:TG) have been reported, as well [39].Currently, it is considered that the CYP2C:TG haplotype provides useful additional information to predict enhanced CYP2C19 activity, an effect that might be ethnically specific, but further investigations are needed before it could be routinely used in clinical settings [51].
Considering that in the case of CYP2C19 phenotypes and clopidogrel efficacy/safety, "rapid" or "ultrarapid" designation has not been thus far considered practically relevant (i.e., no distinction has been made vs. normal metabolizers), we considered it reasonable to explore a concept that "additional information" about enhanced activity conveyed by the CYP2C:TG haplotype might also additionally inform about efficacy and bleeding risks in clopidogrel-treated patients.For this purpose, we assessed the incidence of ischemic and bleeding events with respect to PM/IM, NM, and RM/UM phenotype based on the "classical" criteria of the CYP2C19*2*17 genotype, PM/IM, NM/RM, and UM phenotype informed by the combined CYP2C19*2*17 genotype and rs2860840 and rs11188059 (cluster CYP2C19/CYP2C genotype), and with respect to the presence of the CYP2C:TG haplotype.Due to the limited sample size and limited number of events, the present risk estimates are imprecise (wide confidence intervals), but numerical trends are obvious: higher incidence of ischemic events in PM/IM patients vs. mutually similar NM/RM/UM patients by either classification; lower incidence of bleeding events in PM/IM patients vs. NM/RM/UM (also mutually similar) by either classification; a similar incidence of ischemic events in CYP2C:TG carriers and non-carriers, and a mildly lower incidence of bleeding events in the former than in the latter.Despite this imprecision, however, the patient subsets were not extremely small (between 60 and 135 across the collapsed phenotype categories, 162 vs. 121 regarding CYP2C:TG), and the observed proportions do not appear critically fragile: indeed, one or two events more or less could have occurred in any subset by shear chance, but it is unlikely that this would have substantially changed the observed numerical trends.It could be objected that mixing cardiovascular and neurovascular patients/indications was not appropriate, but we deemed it a feasible option since clopidogrel is indicated in both settings, where it is expected to convey similar effects on ischemic/bleeding events through the same mechanism of action.Also, some residual confounding should be attributed to the fact that we did not consider factors that could have influenced effects of the commonly co-administered aspirin, or other already mentioned genetic factors that have been suggested to affect the antiplatelet effect of clopidogrel.On the other hand, we did account for a number of factors known to have an impact on the risk of ischemic or bleeding events in clopidogrel-treated patients: (i) in line with the prescribing recommendations, clopidogrel was avoided in patients using strong CYP2C19 inducers or inhibitors, and in HIV-positive patients; (ii) exposed and control subjects were balanced according to age, chronic kidney disease, hypertension, dyslipidemia, diabetes, serious liver disease, known history of coronary, cerebral or peripheral artery disease (and, by virtue of this, also indirectly regarding treatments used in such patients); (iii) concomitant aspirin or anticoagulant use [27,50], the use of gastroprotection (which may directly reflect on incidence of gastro-intestinal bleeding), and, specifically, the use of proton pump inhibitors (PPIs).This latter factor is of two-fold relevance: (i) all PPIs are CYP2C19 substrates that could compete with clopidogrel or inhibit it [15] and thus affect clopidogrel bioactivation and its effects [52,53].Of note, the effects of pantoprazole and rabeprazole used in the present sample of patients are minimal, if any, both in vitro and in vivo [15,52,53]; (ii) CYP2C:TG could affect their clearance/exposure and, thus, efficacy in gastroprotection.In this regard, it is of note that no gastro-intestinal bleedings were recorded during the observed period.By using the entropy balancing method, all contrasted "exposures" (different phenotypes, haplotypes) were perfectly balanced with respect to all these factors.Finally, certain validity to the present estimates is conveyed by the fact that In PM/IM, NM, and RM/UM patients classified based on "classical" criteria of the CYP2C19*2*17 genotype, we observed the incidence of ischemic and bleeding events in line with the theoretical expectations [4], and the observations based on the CYP2C19/CYP2C cluster genotype-based phenotypes were closely similar.Under these circumstances, it appears justified to state that no clear-cut, strong signal has been detected in the present exploration that would indicate a major role of the CYP2C:TG haplotype for informing the risk of treatment failure or bleeding events in clopidogrel-treated patients.

Figure 1 .
Figure 1.Raw incidence of ischemic and bleeding events in clopidogrel-treated patients with different CYP2C19 phenotypes: poor (PM), intermediate (IM), normal (NM), rapid (RM), and ultrarapid (UM) metabolizer.Depicted are percentages and number of patients with events/total number in brackets.(A) Incidence of events across phenotypes defined based on the CPY2C19*2,*17 genotype.(B) Incidence of events across phenotypes defined based on the CYP2C19/CYP2C cluster genotype.(C) Incidence of events across collapsed phenotype categories.

Figure 1 .
Figure 1.Raw incidence of ischemic and bleeding events in clopidogrel-treated patients with different CYP2C19 phenotypes: poor (PM), intermediate (IM), normal (NM), rapid (RM), and ultrarapid (UM) metabolizer.Depicted are percentages and number of patients with events/total number in brackets.(A) Incidence of events across phenotypes defined based on the CPY2C19*2,*17 genotype.(B) Incidence of events across phenotypes defined based on the CYP2C19/CYP2C cluster genotype.(C) Incidence of events across collapsed phenotype categories.

Figure 2 .
Figure 2. Adjusted incidence (based on the fully balanced data in Table3) of ischemic and bleeding events in clopidogrel-treated patients with different CYP2C19 phenotypes: poor (PM), intermediate (IM), normal (NM), rapid (RM), or ultrarapid (UM) metabolizer.Depicted are percentages from the weighted pseudopopulations.

Figure 2 .
Figure 2. Adjusted incidence (based on the fully balanced data in Table3) of ischemic and bleeding events in clopidogrel-treated patients with different CYP2C19 phenotypes: poor (PM), intermediate (IM), normal (NM), rapid (RM), or ultrarapid (UM) metabolizer.Depicted are percentages from the weighted pseudopopulations.

Table 1 .
Patient characteristics.Values are n (percent) or mean ± SD (minimum-maximum) for age.

Table 2 .
Patient characteristics across the CYP2C19 phenotypes (poor, intermediate, normal, rapid, ultrarapid metabolizers [PM, IM, NM, RM, UM, respectively]) defined based on the CYP2C19*2/*17 genotype or based on the CPY2C19/CYP2C cluster genotype.Data are mean ± SD or count (percent).Shown are maximum standardize mean differences (Max d) between any two of the three phenotype levels.

Table 3 .
Patient characteristics across the CYP2C19 phenotypes (PM, IM, NM, RM, UM) defined based on the CYP2C19*2*17 genotype or on the CPY2C19/CYP2C cluster genotype after entropy balancing1.Data are mean ± SD or percent patients with a characteristic in the weighted pseudopopulation.Shown are maximum standardize mean differences (Max d) between any two of the three phenotype levels.