Impact of CYP2C19 and CYP3A4 Inhibitor Use on Clopidogrel Clinical Effectiveness in CYP2C19 Genotyped Patients Undergoing Percutaneous Coronary Intervention

Abstract

CYP2C19 and CYP3A4 contribute to clopidogrel bioactivation. CYP2C19 no‐function alleles diminish clopidogrel's antiplatelet effects and clinical effectiveness. Coadministration of either a CYP2C19 or a CYP3A4 inhibitor may also reduce clopidogrel's antiplatelet effects and lead to phenoconversion in patients without a CYP2C19 no‐function allele (normal/rapid/ultrarapid metabolizers: NM/RM/UMs). However, the impact of CYP2C19 or CYP3A4 inhibitor use on clopidogrel clinical effectiveness remains unclear. Rates of major atherothrombotic events (MAE) over 12 months after percutaneous coronary intervention (PCI) were evaluated in 3,242 patients across three sites who underwent CYP2C19 genotype testing and received P2Y₁₂ inhibitor therapy. Overall, 6.8% of patients were co‐prescribed a moderate or strong inhibitor of either CYP2C19 or CYP3A4, as defined by the U.S. Food and Drug Administration. In CYP2C19 genotype‐predicted NM/RM/UMs treated with clopidogrel (n = 1,624), the MAE rates were numerically higher, but not significantly different in patients receiving either a CYP2C19 or a CYP3A4 inhibitor vs. no inhibitor (18.4 vs. 12.8 events/100 patient‐years, adjusted hazard ratio (HR) 1.51, 95% confidence interval (CI), 0.85–2.68, P = 0.155). When evaluating CYP2C19 inhibitor and CYP3A4 inhibitor use separately, MAE rates were higher in clopidogrel‐treated NM/RM/UMs receiving a CYP2C19 inhibitor compared to no inhibitor (adjusted HR 2.22, 95% CI 1.01–4.91, P = 0.048), but no significant difference was observed in those receiving a CYP3A4 inhibitor compared to no inhibitor (adjusted HR 1.26, 95% CI 0.57–2.75, P = 0.569). These results suggest that concomitant use of a CYP2C19 inhibitor, but not a CYP3A4 inhibitor, may contribute to phenoconversion and decreased clopidogrel clinical effectiveness after PCI in CYP2C19 genotype‐predicted NM/RM/UMs.

Study Highlights.

• WHAT IS THE CURRENT KNOWLEDGE ON THE TOPIC?

CYP2C19 no‐function alleles diminish clopidogrel's antiplatelet effects and clinical effectiveness. Moderate and strong inhibitors of CYP2C19 or CYP3A4 can also diminish clopidogrel's antiplatelet effects and may lead to phenoconversion. However, it remains unclear whether the use of a CYP2C19 or CYP3A4 inhibitor impacts clinical outcomes in clopidogrel‐treated patients with or without a CYP2C19 no‐function allele.

• WHAT QUESTION DID THIS STUDY ADDRESS?

This multicenter pragmatic study evaluated the impact of CYP2C19 or CYP3A4 moderate and strong inhibitor use on major atherothrombotic events in 3,242 patients who underwent percutaneous coronary intervention and CYP2C19 genotyping.

• WHAT DOES THIS STUDY ADD TO OUR KNOWLEDGE?

Concomitant use of a strong CYP2C19 inhibitor, but not a CYP3A4 inhibitor, with clopidogrel was associated with diminished clopidogrel clinical effectiveness after percutaneous coronary intervention in patients without a CYP2C19 no‐function allele.

• HOW MIGHT THIS CHANGE CLINICAL PHARMACOLOGY OR TRANSLATIONAL SCIENCE?

These findings suggest that concomitant use of a strong CYP2C19 inhibitor may be associated with phenoconversion in patients without a CYP2C19 no‐function allele and may be an important factor in addition to CYP2C19 genotype to guide P2Y₁₂ inhibitor therapy selection in a real‐world clinical setting.

Dual antiplatelet therapy (DAPT) with aspirin and an oral P2Y₁₂ inhibitor (clopidogrel, prasugrel, or ticagrelor) is the standard of care after percutaneous coronary intervention (PCI) to prevent stent thrombosis and reduce the risk of major atherothrombotic events (MAE). ¹ In patients with acute coronary syndrome (ACS), prasugrel and ticagrelor demonstrated superiority in reducing recurrent MAE compared to clopidogrel in randomized clinical trials that did not consider CYP2C19 genotyping. ² , ³ However, prasugrel and ticagrelor are associated with higher bleeding risk, lower adherence rates, and higher cost compared to clopidogrel, which limit their widespread clinical utilization. ² , ³ , ⁴ Thus, clopidogrel remains commonly prescribed. ⁴

Clopidogrel is a prodrug that requires cytochrome P450 (CYP) enzymes for bioactivation to its thiol metabolite responsible for the drug's antiplatelet effects. Although multiple CYP isoforms contribute to this process in vitro, CYP2C19 is the predominant isoform. ⁵ , ⁶ It is well‐established that the CYP2C19 gene is highly polymorphic, and approximately 30% of the US population carry one or two CYP2C19 no‐function alleles (most commonly CYP2C19*2). ⁷ CYP2C19 no‐function allele carriers are genotype‐predicted as either CYP2C19 intermediate metabolizers (IMs) or poor metabolizers (PMs) (carrying one or two CYP2C19 no‐function allele copies, respectively); IM/PMs have reduced clopidogrel active metabolite formation and subsequently diminished antiplatelet effects and clinical effectiveness compared to normal (NM), rapid (RM), and ultrarapid (UM) metabolizers without a CYP2C19 no‐function allele. ⁸ , ⁹ In contrast, genetic variation in other CYP isoforms has not been consistently associated with clopidogrel pharmacokinetics and pharmacodynamics. ⁹ Use of CYP2C19 genotype testing to guide the selection of prasugrel or ticagrelor in genotype‐predicted CYP2C19 IM/PMs after PCI is associated with improved outcomes. ⁹ , ¹⁰ Thus, a growing number of institutions have implemented CYP2C19 genotyping and the use of genotype‐predicted metabolizer phenotype to guide P2Y₁₂ inhibitor selection in clinical settings. ¹¹

Beyond CYP2C19 no‐function alleles, drug–drug interactions (DDIs) studies have shown that CYP2C19 inhibitors can diminish clopidogrel's antiplatelet effects. Notably, coadministration of fluoxetine (a strong CYP2C19 inhibitor) inhibited clopidogrel active metabolite formation and impaired clopidogrel platelet inhibition. ¹² Additionally, the CYP2C19‐inhibiting proton pump inhibitors (PPIs), omeprazole and esomeprazole, can reduce active metabolite formation and antiplatelet activity of clopidogrel. ¹³ , ¹⁴ However, the inhibitor potency classification (weak vs. moderate) of omeprazole and esomeprazole is conflicting across drug‐interaction resources. ¹⁵ In addition to CYP2C19, CYP3A4 also contributes to clopidogrel active metabolite formation in vitro and in vivo. ⁵ , ¹⁶ Strong CYP3A4 inhibitors, such as ketoconazole and ritonavir, can also reduce clopidogrel bioactivation and impair its antiplatelet effects. ¹⁷ , ¹⁸ On the other hand, while CYP3A4 inhibitors have minimal effect on prasugrel active metabolite formation, they can increase ticagrelor plasma concentrations. ¹⁹ However, P2Y₁₂ inhibitor DDI studies that evaluate clinical outcomes and integrate the impact of CYP2C19 genotype are limited. Therefore, the impact of CYP2C19 and CYP3A4 inhibitors on P2Y₁₂ inhibitor clinical effectiveness in patients with and without a CYP2C19 no‐function allele remains unclear.

It is becoming increasingly recognized that DDIs impacting CYP enzyme activity can lead to “phenoconversion” (a mismatch between the genotype‐predicted vs. observed metabolic phenotype). For example, moderate and strong inhibitors of a CYP enzyme may convert a genotype‐predicted NM into an IM or PM and alter genotype associations with clinical outcomes. ²⁰ , ²¹ Emerging evidence has shown that certain actionable genotypes, most notably CYP2D6, are susceptible to phenoconversion in the presence of CYP enzyme inhibitors. ²² , ²³ , ²⁴ Although less well‐studied than CYP2D6, growing evidence has indicated the potential for CYP2C19 phenoconversion. ²¹ , ²⁵ , ²⁶ , ²⁷ However, CYP2C19 inhibitors and drug‐induced phenoconversion were not considered in the recent clopidogrel Clinical Pharmacogenetics Implementation Consortium (CPIC) guideline update because of limited evidence, and therefore CYP inhibitors are not currently considered when assigning CYP2C19 genotype‐predicted metabolizer phenotype to an individual patient and selecting P2Y₁₂ inhibitor therapy in clinical practice. ⁹

Therefore, to elucidate the potential impact of DDI‐induced phenoconversion on clopidogrel clinical effectiveness, the objectives of this study were to (1) evaluate the real‐world frequency of concomitant CYP2C19 or CYP3A4 inhibitor and oral P2Y₁₂ inhibitor use in patients who underwent clinical CYP2C19 genotype testing to guide P2Y₁₂ inhibitor therapy selection after PCI, (2) investigate the impact of CYP2C19 and CYP3A4 inhibitor use on MAE risk following PCI after stratifying by P2Y₁₂ inhibitor (clopidogrel or prasugrel/ticagrelor) and CYP2C19 genotype‐predicted metabolizer status (IM/PM or NM/RM/UM).

METHODS

Study population

The study population included adult patients who underwent PCI, were genotyped clinically for CYP2C19, and were treated with a P2Y₁₂ inhibitor, as previously described. ²⁸ Three institutions (University of Florida, Gainesville; University of Florida, Jacksonville; University of North Carolina, Chapel Hill) contributed data from 3,242 patients who met these inclusion criteria. Data collection was approved by the institutional review board at each institution.

CYP2C19 genotyping and P2Y₁₂ inhibitor prescribing

Clinical CYP2C19 genotyping was performed in a Clinical Laboratory Improvement Amendments (CLIA)‐licensed laboratory and reported in the electronic health record (EHR) at each site, as previously described. ²⁹ , ³⁰ , ³¹ The CYP2C19*2, *3, and *17 alleles were tested at each site; additional alleles (CYP2C19*4, *6, *8, and *10) were tested at the University of Florida, Gainesville. Genotype‐predicted metabolizer phenotypes (PM, IM, NM, RM, UM) were assigned according to CPIC recommendations. ⁹ In accordance with CPIC recommendations, ⁹ alternative P2Y₁₂ inhibitor therapy (prasugrel or ticagrelor) was recommended for CYP2C19 IMs and PMs, in the absence of contraindications; no prescribing recommendations were made for CYP2C19 NMs, RMs, and UMs. ²⁹ , ³⁰ , ³¹ The ultimate P2Y₁₂ inhibitor prescribing decision was left to the clinician's discretion.

Data abstraction

Demographic, clinical, genotype, and medication data were manually abstracted from the EHR at each site using a common data collection tool, as previously described. ²⁸ , ³² Baseline data were collected from the index PCI hospitalization (defined as the PCI performed in association with CYP2C19 genotyping). Longitudinal data were collected from subsequent inpatient and outpatient encounters at each site, which were supplemented by telephone interviews at the University of Florida, Jacksonville and EHR extracted prescription start and end dates at the University of North Carolina, Chapel Hill, for up to 12 months post‐PCI. Patients were followed up until the first occurrence of the clinical endpoint of interest or the last documented use of a P2Y₁₂ inhibitor within 12 months of PCI, whichever occurred first. Patients were included regardless of the length of follow‐up available, and thus, no missing data were identified, and no imputation was required in our dataset.

CYP2C19 and CYP3A4 inhibitor use

The use of moderate and strong inhibitors of CYP2C19 or CYP3A4, as determined by the FDA's Examples of Drugs that Interact with CYP Enzymes and Transporter Systems (November 2024), ³³ was assessed. Inhibitors that met these criteria are listed in Table S1 . Though the PPIs omeprazole and esomeprazole are classified as weak CYP2C19 inhibitors by the FDA, ¹⁵ PPI use status was also collected due to a warning in the clopidogrel prescribing information to avoid concomitant use of omeprazole/esomeprazole, ³⁴ conflicting inhibitor potency classification (weak vs. moderate) of omeprazole and esomeprazole across drug‐interaction resources, ¹⁵ and conflicting studies suggesting a potential impact of omeprazole/esomeprazole on clopidogrel clinical effectiveness. ³⁵ , ³⁶

Clinical outcomes

The primary endpoint was major atherothrombotic events (MAE), defined as the composite of the first occurrence of all‐cause death, myocardial infarction (MI), ischemic stroke, stent thrombosis, or unstable angina requiring revascularization. The secondary endpoint was major adverse cardiovascular events (MACE), defined as the composite of the first occurrence of cardiovascular death, MI, ischemic stroke, or stent thrombosis. All events were independently verified by an interventional cardiologist or reviewed by a clinical pharmacist, as described. ²⁸

Data analysis

Data were verified and cleaned at each site and then curated and aggregated at the University of Florida, Gainesville. Oral P2Y₁₂ inhibitor therapy (clopidogrel or prasugrel/ticagrelor) and the use of a CYP2C19 or CYP3A4 moderate/strong inhibitor (yes or no) were assigned to each patient at the time of the clinical event or last follow‐up in which P2Y₁₂ inhibitor treatment was documented. The frequency of CYP2C19/CYP3A4 inhibitor use, overall and by each inhibitor, was evaluated using descriptive statistics. Baseline characteristics were compared among patients receiving vs. not receiving an inhibitor using ${\chi }^2$, Fisher's exact test, Student's t‐test, or the Wilcoxon Rank‐sum test as appropriate.

The primary analysis evaluated the impact of concomitant use of either a CYP2C19 inhibitor or a CYP3A4 inhibitor vs. no inhibitor on clinical outcomes by CYP2C19 genotype‐predicted phenotype and P2Y₁₂ inhibitor therapy. Patients were stratified into three CYP2C19‐P2Y₁₂ inhibitor therapy groups: clopidogrel‐treated CYP2C19 IM/PMs, clopidogrel‐treated CYP2C19 NM/RM/UMs, and prasugrel or ticagrelor users regardless of CYP2C19 genotype‐predicted phenotype (alternative therapy) because it is well‐established that CYP2C19 genotype does not impact prasugrel or ticagrelor clinical effectiveness. ⁹ , ¹⁰ The time to the first occurrence of MAE (primary endpoint) and MACE (secondary endpoint) within 12 months post‐PCI was calculated for each patient. Patients who did not experience an event were censored at the date of the last follow‐up. Event rates in CYP2C19 or CYP3A4 inhibitor use (yes vs. no) groups were reported as the number of events per 100 patient‐years. Hazard ratios (HRs) and 95% confidence intervals (CIs) were estimated using Cox proportional hazards models. Each comparison model was stratified by site and ACS indication for PCI and was adjusted for key baseline demographic and clinical factors that included age, sex, self‐reported Black or African American race, and other baseline variables that differed across those co‐prescribed an inhibitor (yes vs. no) with a P‐value cutoff of <0.10. Variance inflation factors (VIF) and scaled Schoenfeld residuals were evaluated to assess for collinearity and for violations to the proportional hazards' assumptions, respectively. For correlated features (VIF > 5), variable selection was performed based on clinical relevance. Kaplan–Meier analyses were used to determine the cumulative incidence of MAE or MACE during the 12‐month follow‐up period across CYP2C19 or CYP3A4 inhibitor use (yes vs. no) groups.

A secondary analysis was conducted exclusively within clopidogrel‐treated CYP2C19 NM/RM/UMs (n = 1,624) to further investigate the impact of CYP2C19 inhibitor or CYP3A4 inhibitor use (separately) on MAE risk. In this analysis, MAE rates were compared across three groups: CYP2C19 moderate/strong inhibitor vs. CYP3A4 moderate/strong inhibitor vs. No inhibitor (reference). To evaluate the potential impact of omeprazole and esomeprazole (weak CYP2C19 inhibitors per FDA, ³³ with conflicting evidence on the CYP2C19 inhibition potential ¹⁵ ) on clopidogrel clinical effectiveness, an additional analysis was performed within clopidogrel‐treated NM/RM/UMs. MAE rates were compared across patients receiving omeprazole/esomeprazole (weak CYP2C19 inhibitors ³³ ) vs. any other PPI (non‐inhibitors of CYP2C19) vs. no PPI.

All analyses were performed in R statistical software (version 4.3.2 https://www.r‐project.org/). P values <0.05 were considered significant. Data and analytic codes supporting the findings of this study are available from the authors on reasonable request.

RESULTS

Study population

Baseline characteristics of the 3,242 patients in the study population are summarized in Table 1 . The mean age was 63 ± 12 years; 33% of patients were female, 21% self‐reported as Black/African American, and 76% had an ACS indication for PCI. Among the study population, 2.6%, 28%, 39%, 26%, and 4.4% were classified as genotype‐predicted CYP2C19 PM, IM, NM, RM, and UM, respectively.

Table 1.

Study population characteristics at the time of index PCI in the full cohort and stratified by inhibitor use status

	All patients, N = 3,242	CYP2C19 or CYP3A4 inhibitor, N = 219	No inhibitor, N = 3,023	P valueb
Demographics
Age (years)	63 ± 12	63 ± 12	63 ± 12	0.789
Age > 75	472 (15)	32 (15)	440 (15)	0.982
Female	1,070 (33)	105 (48)	965 (32)	<0.001
Black or African American	672 (21)	39 (18)	633 (21)	0.270
Hispanic	89 (2.7)	4 (1.8)	85 (2.8)	0.389
BMI (kg/m2)	30 ± 7	32 ± 8	30 ± 6	<0.001
Obesity (BMI $\ge$ 30 kg/m2)	1,486 (46)	122 (56)	1,364 (45)	0.002
Current smoker	967 (30)	41 (19)	926 (31)	<0.001
PCI indication
ACS	2,472 (76)	152 (69)	2,320 (77)	0.014
STEMI	760 (23)	35 (16)	725 (24)	0.007
Non‐STEMI	976 (30)	53 (24)	923 (31)	0.049
Unstable angina	736 (23)	64 (29)	672 (22)	0.017
Stent placed at index PCI	3,103 (96)	209 (95)	2,894 (96)	0.833
Medical History
Hypertension	2,623 (81)	190 (87)	2,433 (80)	0.023
Dyslipidemia	2,048 (63)	162 (74)	1,886 (62)	<0.001
Diabetes	1,251 (39)	104 (47)	1,147 (38)	0.005
Prior MI	791 (24)	57 (26)	734 (24)	0.561
Prior revascularization	1,259 (39)	92 (42)	1,167 (39)	0.318
Prior stent	886 (27)	66 (30)	820 (27)	0.334
Prior CABG	472 (15)	30 (14)	442 (15)	0.709
Stroke/TIA	376 (12)	23 (11)	353 (12)	0.600
PVD	265 (8.2)	17 (7.8)	248 (8.2)	0.818
Heart failure	501 (15)	38 (17)	463 (15)	0.421
Atrial fibrillation	265 (8.2)	31 (14)	234 (7.7)	<0.001
History of GI bleeding or ICH	94 (2.9)	3 (1.4)	91 (3.0)	0.162
Active cancer	191 (5.9)	19 (8.7)	172 (5.7)	0.070
Chronic kidney diseasea	715 (22)	58 (26)	657 (22)	0.102
Medications at discharge
Aspirin	3,171 (98)	215 (98)	2,956 (98)	>0.999
Anticoagulant	288 (8.9)	27 (12)	261 (8.6)	0.063
ACE inhibitor or ARB	2,250 (69)	151 (69)	2,099 (69)	0.881
$\beta$‐blocker	2,792 (86)	163 (74)	2,629 (87)	<0.001
Statin	3,059 (94)	199 (91)	2,860 (95)	0.021
Proton pump inhibitor	1,035 (32)	76 (35)	959 (32)	0.361

Data are presented as n (%) or mean ± standard deviation.

ACE, angiotensin converting enzyme; ACS, acute coronary syndrome; ARB, angiotensin receptor blocker; BMI, body mass index; CABG, coronary artery bypass grafting; CYP2C19, cytochrome P450 2C19 enzyme; CYP3A4, cytochrome P450 3A4 enzyme; GI, gastrointestinal; ICH, intracranial hemorrhage; MI, myocardial infarction; PCI, percutaneous coronary intervention; PVD, peripheral vascular disease; STEMI, ST‐segment elevation MI; TIA, transient ischemic attack.

^a Chronic kidney disease was defined as estimated glomerular filtration rate < 60 mL/min/1.73 m².

^b P value for comparison between CYP2C19 or CYP3A4 Inhibitor group and No inhibitor group.

CYP2C19 or CYP3A4 moderate/strong inhibitor and P2Y₁₂ inhibitor co‐prescribing

Overall, 219 (6.8%) patients were co‐prescribed a CYP2C19 or CYP3A4 moderate/strong inhibitor and a P2Y₁₂ inhibitor. Among them, 205 (93.6%) were co‐prescribed one inhibitor, 13 (5.9%) were co‐prescribed two inhibitors, and one (0.5%) was co‐prescribed three inhibitors. The proportion of patients co‐prescribed a CYP2C19 or CYP3A4 inhibitor was similar in each of the CYP2C19‐P2Y₁₂ inhibitor therapy groups (Figure 1 a ). Inhibitor users were less likely to have an ACS indication for PCI or be a current smoker, and were more likely to be female, obese, and have comorbid hypertension, dyslipidemia, diabetes, and atrial fibrillation (Table 1 ).

Figure 1.

Summary of CYP2C19 or CYP3A4 inhibitor use in the study population. (a) The study population was stratified by P2Y₁₂ inhibitor antiplatelet therapy and CYP2C19 genotype‐predicted phenotype. Patients in each subgroup were further categorized based on inhibitor use (yes vs. no) status. CYP2C19, cytochrome P450 2C19 enzyme. IM, intermediate metabolizer; NM, normal metabolizer; PM, poor metabolizer; RM, rapid metabolizer; UM, ultrarapid metabolizer. (b) Frequency of CYP2C19 or CYP3A4 moderate/strong inhibitor co‐prescribing with a P2Y₁₂ inhibitor. Less commonly prescribed inhibitors (frequency <0.5%) aggregated in Other* include erythromycin (0.4%), verapamil (0.3%), imatinib (0.1%), clarithromycin (0.09%), ritonavir (0.09%), dronedarone (0.06%), fluvoxamine (0.06%), cobicistat (0.03%), and nefazodone (0.03%). The enzyme inhibited by each is summarized in Table S2 .

Thirteen distinct CYP2C19 or CYP3A4 inhibitors were co‐prescribed (Table S2 ). The most common inhibitors are listed in Figure 1 b . Fluoxetine (strong CYP2C19 inhibitor, 2.1%) was the most commonly co‐prescribed inhibitor, followed by diltiazem (moderate CYP3A4 inhibitor, 2.0%), ciprofloxacin (moderate CYP3A4 inhibitor, 1.3%), and fluconazole (strong CYP2C19 and moderate CYP3A4 inhibitor, 0.6%).

Impact of CYP2C19 or CYP3A4 moderate/strong inhibitor use on clinical outcomes

Among the study population, 1,624 (50%) were clopidogrel‐treated NM/RM/UMs, 422 (13%) were clopidogrel‐treated IM/PMs, and 1,196 (37%) were treated with alternative therapy (Figure 1 a ). Patient characteristics across patients receiving vs. not receiving a CYP2C19 or CYP3A4 moderate/strong inhibitor within each of the three groups are summarized in Table S3 .

Overall, 274 patients (8.5%) experienced MAE. The median (interquartile range (IQR)) time to MAE or last follow‐up was 292 (IQR, 116–357) days post‐PCI for the overall population (289 (IQR, 104–357) and 302 (IQR, 159–356) days for patients with an ACS and non‐ACS indication for PCI, respectively). MAE risk across patients receiving vs. not receiving a CYP2C19 or CYP3A4 moderate/strong inhibitor within the three CYP2C19‐P2Y₁₂ inhibitor therapy groups is summarized in Table 2 . In clopidogrel‐treated NM/RM/UMs, MAE rates were numerically higher in patients receiving vs. not receiving a CYP2C19 or CYP3A4 inhibitor (18.4 vs. 12.8 events per 100 patient‐years, adjusted HR 1.51, 95% CI 0.85–2.68, P = 0.155; Figure 2 a ); however, the observed difference was not statistically significant. Similar results were observed with the secondary outcome MACE (13.2 vs. 8.3 events per 100 patient‐years, adjusted HR 1.76, 95% CI 0.89–3.46, P = 0.111; Figure 2 b ).

Table 2.

Major Atherothrombotic Events (MAE) and Major Adverse Cardiovascular Event (MACE) rates across patients receiving vs. not receiving an inhibitor

	CYP2C19 or CYP3A4 inhibitor			No inhibitor			Inhibitor vs. No inhibitor (reference)
	N	Events no (%)	Event rates (per 100‐pt‐years)	N	Events no (%)	Event rates (per 100‐pt‐years)	Unadjusted HR (95% CI)	P value	Adjusted HRa (95% CI)	P value
MAE
Clop – NM/RM/UM	118	14 (11.9)	18.4	1,506	122 (8.1)	12.8	1.46 (0.84–2.53)	0.183	1.51 (0.85–2.68)	0.155
Clop – IM/PM	27	2 (7.4)	12.5	395	42 (10.6)	18.4	0.74 (0.18–3.10)	0.681	0.76 (0.18–3.30)	0.719
Alternative therapy	74	9 (12.2)	16.6	1,122	85 (7.6)	11.3	1.49 (0.75–2.97)	0.253	1.31 (0.64–2.69)	0.465
MACE
Clop – NM/RM/UM	118	10 (8.5)	13.2	1,506	79 (5.3)	8.3	1.61 (0.83–3.12)	0.155	1.76 (0.89–3.46)	0.111
Clop – IM/PM	27	1 (3.7)	6.2	395	29 (7.3)	12.7	0.53 (0.07–3.92)	0.537	0.64 (0.08–4.87)	0.668
Alternative therapy	74	2 (2.7)	3.7	1,122	54 (4.8)	7.2	0.53 (0.13–2.15)	0.371	0.48 (0.11–2.00)	0.310

CI, confidence interval; Clop, clopidogrel; CYP2C19, cytochrome P450 2C19 enzyme; CYP3A4, cytochrome P450 3A4 enzyme; HR, hazard ratio; IM, intermediate metabolizer; MACE, major adverse cardiovascular events; MAE, major atherothrombotic events; NM, normal metabolizer; PM, poor metabolizer; RM, rapid metabolizer; UM, ultrarapid metabolizer.

^a All multivariable models were stratified by site and ACS indication for PCI and were adjusted for age, sex, self‐reported Black or African American race, and additional baseline characteristics that differed across inhibitor vs. no inhibitor groups within each strata: Clop – NM/RM/UM: obesity, current smoker, dyslipidemia, diabetes, and $\beta$‐blocker; Clop – IM/PM: atrial fibrillation and $\beta$‐blocker; Alternative therapy: obesity, current smoker, hypertension, dyslipidemia, diabetes, atrial fibrillation, and $\beta$‐blocker. Baseline characteristics are described in Table S3 .

Figure 2.

Time‐to‐cardiovascular clinical outcomes by CYP2C19/CYP3A4 Inhibitor Use Status. Kaplan–Meier curve for (a) major atherothrombotic events (MAE) and (b) major adverse cardiovascular events (MACE) over 12 months after the index percutaneous coronary intervention (PCI) among clopidogrel‐treated CYP2C19 normal, rapid, and ultrarapid metabolizers (NM/RM/UM) by inhibitor use status: CYP2C19 or CYP3A4 Inhibitor vs. No Inhibitor. The unadjusted log‐rank P‐value for each curve is presented. The tails of the Kaplan–Meier curves in panels (a) and (b) were truncated at 360 days after the index PCI. The Inhibitor group (red) had no events that occurred after day 360, when 19 patients were still in follow‐up. The No Inhibitor group (black) had n = 4 events that occurred after day 360, when 264 patients were still in follow‐up.

In clopidogrel‐treated IM/PMs, no difference in either MAE or MACE rate was observed in patients receiving vs. not receiving a CYP2C19 or CYP3A4 inhibitor (Table 2 ). Likewise, in patients receiving alternative therapy, neither MAE nor MACE significantly differed in patients receiving vs. not receiving a CYP2C19 or CYP3A4 inhibitor (Table 2 ).

When evaluating the impact of CYP2C19 inhibitor and CYP3A4 inhibitor use separately, clopidogrel‐treated NM/RM/UMs receiving a CYP2C19 inhibitor exhibited a higher MAE risk vs. those receiving no inhibitor (23.3 vs. 12.8 events per 100 patient‐years, adjusted HR 2.22, 95% CI 1.01–4.91, P = 0.048; Figure 3 ). The observed MAE rates in clopidogrel‐treated NM/RM/UMs receiving a CYP2C19 inhibitor (23.3 events per 100 patient‐years) were similar to MAE rates in genotype‐predicted IM/PMs treated with clopidogrel (18.0 events per 100 patient‐years). In contrast, there was no significant difference in MAE risk between patients receiving a CYP3A4 inhibitor vs. receiving no inhibitor (15.2 vs. 12.8 events per 100 patient‐years, adjusted HR 1.26, 95% CI 0.57–2.75, P = 0.569; Figure 3 ).

Figure 3.

Time‐to‐major atherothrombotic events (MAE) by CYP2C19 inhibitor or CYP3A4 inhibitor use status. Kaplan–Meier curve for MAE over 12 months after the index percutaneous coronary intervention (PCI) among clopidogrel‐treated CYP2C19 normal, rapid, and ultrarapid metabolizers (NM/RM/UMs) by inhibitor use status: CYP2C19 Inhibitor vs. CYP3A4 Inhibitor vs. No Inhibitor. The tails of the Kaplan–Meier curves were truncated at 360 days after the index PCI. The CYP2C19 Inhibitor and CYP3A4 Inhibitor groups had no events that occurred after day 360, with 4 and 15 patients still in follow‐up, respectively. In the No inhibitor group, n = 4 events occurred after day 360, when 264 patients were still in follow‐up. The multivariable model was stratified by site and ACS indication for PCI, and were adjusted for age, sex, self‐reported Black or African American race, obesity, dyslipidemia, stroke/TIA, and $\beta$‐blocker use. Baseline characteristics are described in Table S4 .

CYP2C19‐inhibiting PPI impact on clinical outcomes in clopidogrel‐treated NM/RM/UMs

In clopidogrel‐treated NM/RM/UMs, 579 (35.7%) patients were co‐prescribed a PPI. Among them, 382 (66%) were treated with omeprazole or esomeprazole (weak CYP2C19 inhibitors ³³ ), and 197 (34%) were treated with other non‐CYP2C19‐inhibiting PPIs, including pantoprazole: n = 176, lansoprazole: n = 17, dexlansoprazole: n = 2, or rabeprazole: n = 2. MAE rates did not differ between those receiving omeprazole or esomeprazole, those receiving a non‐CYP2C19‐inhibiting PPI, and those receiving no PPI (Figure 4 ). Also, no difference was observed in MAE risk between patients receiving omeprazole or esomeprazole vs. a non‐CYP2C19‐inhibiting PPI (15.9 vs. 14.7 events per 100 patient‐years, adjusted HR 0.82, 95% CI 0.39–1.75, P = 0.616).

Figure 4.

Time‐to‐major atherothrombotic events (MAE) by Proton Pump Inhibitor (PPI) Use Status. Kaplan–Meier curve for MAE over 12 months after the index percutaneous coronary intervention (PCI) among clopidogrel and PPI‐treated CYP2C19 normal, rapid, and ultrarapid metabolizers (NM/RM/UMs) by PPI use status: Omeprazole/esomeprazole vs. Other PPIs vs. No PPI. The tails of the Kaplan–Meier curves were truncated at 360 days after the index PCI. The Omeprazole/esomeprazole and Other PPIs groups had no events that occurred after day 360, when 49 and 50 patients were still in follow‐up, respectively. In the No PPI group, n = 4 events occurred after day 360, when 184 patients were still in follow‐up. The multivariable model was stratified by site and ACS indication for PCI, and was adjusted for age, sex, self‐reported Black or African American race, current smoker, Hispanic ethnicity, dyslipidemia, diabetes, prior MI, prior revascularization, stroke/TIA, heart failure, atrial fibrillation, history of GI bleeding or ICH, and CKD. Baseline characteristics were described in Table S5 .

DISCUSSION

CYP2C19, and to a lesser extent, CYP3A4, contribute to clopidogrel bioactivation. ⁵ It is well‐established that CYP2C19 no‐function alleles diminish clopidogrel's antiplatelet effects and clinical effectiveness. ⁹ , ¹⁰ While CYP2C19 and CYP3A4 inhibitors can also decrease clopidogrel active metabolite formation and antiplatelet effects, ¹² , ¹⁸ the impact of these DDIs on clinical outcomes remains unclear. The current study evaluated the prevalence of co‐prescribing an oral P2Y₁₂ inhibitor and an FDA‐defined CYP2C19 or CYP3A4 moderate/strong inhibitor and investigated the impact of inhibitor use on MAE risk in PCI patients genotyped for CYP2C19. The results showed that a CYP2C19 or CYP3A4 moderate/strong inhibitor was co‐prescribed with an oral P2Y₁₂ inhibitor in 6.8% of PCI patients in a real‐world clinical setting, and the prevalence of co‐prescription was similar in clopidogrel‐treated NM/RM/UMs, clopidogrel‐treated IM/PMs, and patients treated with alternative therapy. In clopidogrel‐treated NM/RM/UMs, MAE and MACE rates were numerically, but not significantly, higher in patients receiving vs. not receiving either a CYP2C19 or CYP3A4 moderate/strong inhibitor. In clopidogrel‐treated IM/PMs and patients receiving alternative therapy, no difference in MAE or MACE risk was observed in patients receiving vs. not receiving either a CYP2C19 or CYP3A4 inhibitor. When evaluating CYP2C19 and CYP3A4 inhibitors separately, clopidogrel‐treated NM/RM/UMs receiving a CYP2C19 inhibitor, but not a CYP3A4 inhibitor, exhibited higher MAE rates compared to patients receiving no inhibitor. Collectively, these data suggest that coadministration of a CYP2C19 inhibitor, but not a CYP3A4 inhibitor, may contribute to decreased clopidogrel clinical effectiveness after PCI in CYP2C19 genotype‐predicted NM/RM/UMs.

Randomized DDI crossover studies have shown that the selective serotonin reuptake inhibitors (SSRIs) fluoxetine and fluvoxamine (strong CYP2C19 inhibitors ³³ ) reduce plasma clopidogrel active metabolite concentrations and clopidogrel's antiplatelet effects. ¹² , ³⁷ In a retrospective study of 54,357 clopidogrel‐treated patients (22% with recent PCI) that were not genotyped for CYP2C19, patients co‐prescribed a CYP2C19‐inhibiting SSRI (fluoxetine or fluvoxamine) demonstrated an increased risk of ischemic events compared to patients receiving a non‐CYP2C19‐inhibiting SSRI (adjusted HR 1.12, 95% CI 1.01–1.24). ³⁸ In our analysis focused on clopidogrel‐treated NM/RM/UMs, MAE risk was also higher in those receiving a CYP2C19 inhibitor vs. no inhibitor after PCI. Of note, the proportion of patients prescribed a CYP2C19 inhibitor in our PCI patient population was relatively small (3.0%), and the agents most commonly used included fluoxetine (2.5%) and fluconazole (0.5%); both are classified as strong CYP2C19 inhibitors by the FDA. ¹⁵ , ³³ Fluconazole also moderately inhibits CYP3A4. ³³ Larger studies are needed to more confidently evaluate the influence of CYP2C19 inhibitor use and discern the relative impact of individual CYP2C19 inhibitors on clopidogrel clinical effectiveness after PCI in patients who are CYP2C19 genotype‐predicted NM/RM/UMs.

Similar to CYP2C19 inhibitors, randomized DDI crossover studies have shown that strong CYP3A4 inhibitors, such as ketoconazole and ritonavir, reduce clopidogrel's antiplatelet effects. ¹⁷ , ¹⁸ One observational study without considering CYP2C19 genotype demonstrated that clopidogrel‐treated patients receiving a CYP3A4 inhibitor were associated with higher overall mortality (adjusted HR 1.30, 95% CI 1.00–1.69), but lower thrombosis complications (adjusted HR 0.62, 95% CI 0.53–0.72), compared to those not receiving a CYP3A4 inhibitor. ³⁹ Thus, evidence regarding the impact of CYP3A4 inhibitors on clopidogrel clinical effectiveness is unclear. We observed no significant difference in MAE risk between clopidogrel‐treated NM/RM/UMs receiving vs. not receiving a CYP3A4 inhibitor. The most commonly prescribed CYP3A4 inhibitor in our study population was diltiazem (2.3%), a moderate CYP3A4 inhibitor. ³³ Although not all calcium channel blockers (CCBs) are CYP3A4 inhibitors, studies have shown that CCBs, as a drug class, reduce clopidogrel's antiplatelet effects. ⁴⁰ , ⁴¹ However, clinical outcome studies collectively demonstrate that CCBs do not appear to diminish clopidogrel's clinical effectiveness. ⁴² , ⁴³ , ⁴⁴ Notably, amlodipine was the most common CCB in this study, and the use of diltiazem and verapamil (a moderate CYP3A4 inhibitor ³³ ) was limited. ⁴² Other studies did not report the specific CCBs used. ⁴³ , ⁴⁴

Additionally, it should be noted that all CYP2C19 inhibitor users in our study were exposed to a strong CYP2C19 inhibitor, while most CYP3A4 inhibitor users were exposed to a moderate CYP3A4 inhibitor. Also, multiple CYP3A4 inhibitors such as ciprofloxacin and erythromycin were likely prescribed transiently rather than chronically. It remains unclear whether the distinction in inhibition potency and duration of inhibitor use contributed to the lack of an observed effect of CYP3A4 inhibitor use on clinical outcomes in clopidogrel‐treated NM/RM/UMs. Therefore, future studies are warranted to evaluate the impact of strong CYP3A4 inhibitor use on clopidogrel clinical effectiveness and discern the relative contribution of CYP2C19 and CYP3A4 activity to clopidogrel bioactivation.

Accumulating evidence demonstrates that CYP2C19 inhibitor use may lead to CYP2C19 phenoconversion, ¹⁵ , ²¹ , ²⁶ , ²⁷ which could impair the accuracy of genotype‐based CYP2C19 metabolizer phenotype prediction and influence outcomes of CYP2C19 genotype‐guided P2Y₁₂ inhibitor selection after PCI. Relative to CYP2C19 NM/RM/UMs not receiving a CYP2C19 inhibitor, MAE risk appeared to be similarly elevated in CYP2C19 NM/RM/UMs receiving a CYP2C19 inhibitor and CYP2C19 genotype‐predicted IM/PMs in our study population. Although the number of events in the CYP2C19 inhibitor group was small and should be interpreted with caution, these results suggest that the use of a strong CYP2C19 inhibitor might contribute to diminished clopidogrel clinical effectiveness after PCI through CYP2C19 phenoconversion in CYP2C19 genotype‐predicted NM/RM/UMs.

PPIs are widely prescribed to patients using a P2Y₁₂ inhibitor for its gastric protection effect. Omeprazole and esomeprazole demonstrate greater CYP2C19 enzyme inhibition in vitro and in vivo compared to lansoprazole, dexlansoprazole, pantoprazole, and rabeprazole. ¹³ , ¹⁴ , ⁴⁵ , ⁴⁶ The FDA classifies omeprazole and esomeprazole as weak CYP2C19 inhibitors, while the other PPIs are not listed as CYP2C19 inhibitors, ³³ and recent ACS clinical practice guidelines de‐emphasize the effect of PPIs on clopidogrel clinical effectiveness. ⁴⁷ Therefore, omeprazole and esomeprazole were not considered in our primary analysis. Nevertheless, the FDA has published a warning against the coadministration of omeprazole/esomeprazole and clopidogrel because of evidence that these PPIs can reduce clopidogrel's antiplatelet effects through CYP2C19 inhibition, ³⁴ and whether omeprazole/esomeprazole can diminish clopidogrel clinical effectiveness has remained controversial. ³⁵ , ³⁶ , ⁴⁸ In our study, MAE risk did not significantly differ in clopidogrel‐treated NM/RM/UM patients receiving omeprazole/esomeprazole compared to those receiving no PPI or compared to those receiving other PPIs (pantoprazole, lansoprazole, dexlansoprazole, and rabeprazole). Our results are consistent with a European registry that included 2,353 clopidogrel‐treated MI patients and observed no difference in cardiovascular events (composite of death, MI, or stroke) in patients without a CYP2C19 no‐function allele receiving PPI (mostly omeprazole/esomeprazole) vs. no PPI (adjusted HR 1.13, 95% CI 0.74–1.74). ³⁵ In alignment with FDA's current classification of omeprazole and esomeprazole as weak CYP2C19 inhibitors and recent ACS guidelines, ³³ , ⁴⁷ our results also suggest that omeprazole and esomeprazole use does not warrant clinical consideration when prescribing clopidogrel after PCI.

It is important to acknowledge several limitations. First, in this real‐world observational study, randomization to the use of a CYP2C19 or CYP3A4 inhibitor did not occur. Therefore, outcomes could have been impacted by the baseline differences between inhibitor use groups. Although baseline characteristics were carefully evaluated and covariate‐adjusted and stratified analyses were performed for each comparison group, residual and unmeasured confounding effects could remain. For example, although depression could be a risk factor for cardiovascular events, ⁴⁹ baseline data on depression diagnosis or severity was not available and considered in this study. Second, longitudinal clinical events were identified and validated based on EHR review. Events documented in healthcare systems other than the original institution where the index PCI was performed may have been missed. In addition, bleeding events were beyond the scope of the current study and were not included in the analysis. Thus, future studies evaluating the impact of CYP enzyme inhibitors and inducers on bleeding outcomes in CYP2C19 genotyped patients treated with clopidogrel are warranted. Third, pharmacokinetics and pharmacodynamics data, such as plasma clopidogrel active metabolite concentrations, platelet function, or CYP2C19 and CYP3A4 probe substrate metabolism, were not collected to directly assess phenoconversion of metabolic activity. Fourth, data on over‐the‐counter (OTC) medications and dietary supplements that inhibit CYP enzymes were not collected and evaluated, which may result in the underestimation of actual CYP inhibitor co‐prescribing within our cohort. Finally, the overall number of patients on a CYP2C19 or CYP3A4 inhibitor and the number of events observed in inhibitor users were small and confidence intervals were wide. Thus, the observed association between CYP2C19 inhibitor use and a higher MAE risk in clopidogrel‐treated NM/RM/UMs needs to be interpreted cautiously, and larger studies are needed to validate our findings.

In summary, results from this real‐world population who underwent PCI and received CYP2C19 genotype to guide P2Y₁₂ inhibitor therapy selection suggests that concomitant use of strong CYP2C19 inhibitors and clopidogrel may be associated with CYP2C19 phenoconversion and a higher MAE risk in CYP2C19 genotype‐predicted NM/RM/UMs. These findings support the potential value of evaluating both DDIs and pharmacogenetic test results to advance the practice of precision medicine. Future studies are needed to discern the effects of individual CYP2C19 inhibitors and CYP3A4 inhibitors on clopidogrel clinical effectiveness, and to determine whether integrating DDI and CYP2C19 genotype results to optimize P2Y₁₂ inhibitor prescribing improves outcomes.

FUNDING

This work was supported by grants from the National Institutes of Health (NIH) National Heart, Lung, and Blood Institute (R01HL149752), National Human Genome Research Institute (U01HG007269), and National Center for Advancing Translational Sciences (UM1TR004406 and UM1TR005128). Dr. Danwei Shao is supported by NIH grant T32GM086330 from the National Institute of General Medical Sciences. Dr. Jean G. Malavé is supported by NIH grant T32HG008958 from the National Human Genome Research Institute. The content is solely the responsibility of the authors and does not necessarily represent the official views of NIH.

CONFLICT OF INTEREST

Dr. Franchi has received consulting fees or honoraria from Werfen, outside the present work. He also has received institutional research grants from PLx Pharma and the Scott R. MacKenzie Foundation. Dr. Angiolillo declares that he has received consulting fees or honoraria from Anthos, Bayer, Boehringer Ingelheim, Bristol‐Myers Squibb, Chiesi, Faraday, Idorsia, Johnson & Johnson, Novartis, Novo Nordisk, PLx Pharma, Sanofi, SFJ Pharmaceuticals, Vectura, and Werfen. His institution has received research grants from Abbott, Amgen, AstraZeneca, Bayer, Chiesi, CSL Behring, DalCor Pharmaceuticals, Daiichi‐Sankyo, Edwards, Eli Lilly, Faraday, Janssen, Hikari DX, Novartis, Prolocor, and Vertex. All other authors declared no competing interests for this work.

AUTHOR CONTRIBUTIONS

D.S., and C.R.L. wrote the manuscript; D.S., J.G.M., L.H.C., and C.R.L. designed the research; D.S., J.G.M., J.S.R., F.F., E.C.K., D.J.A., G.A.S., L.H.C., and C.R.L. performed the research; D.S., J.G.M., L.H.C., and C.R.L. analyzed the data.

Supporting information

Table S1.