OUP user menu

Oral anticoagulation with coumarin derivatives and antiplatelet effects of clopidogrel

Dirk Sibbing, Nicolas von Beckerath, Tanja Morath, Julia Stegherr, Julinda Mehilli, Nikolaus Sarafoff, Siegmund Braun, Stefanie Schulz, Albert Schömig, Adnan Kastrati
DOI: http://dx.doi.org/10.1093/eurheartj/ehq023 1205-1211 First published online: 16 February 2010

Abstract

Aims A relevant proportion of patients receiving aspirin and clopidogrel after percutaneous coronary intervention (PCI) also require oral anticoagulation with a coumarin derivative such as phenprocoumon. Both clopidogrel and phenprocoumon are metabolized by the hepatic cytochrome P450 system and a drug–drug interaction may exist at this level. The aim of this study was to investigate the impact of phenprocoumon on the antiplatelet effects of clopidogrel in patients with coronary artery disease.

Methods and results Patients (n = 1223) eligible for this study were under dual maintenance antiplatelet treatment with aspirin and clopidogrel. Adenosine diphosphate-induced platelet aggregation (in AU*min) was measured with multiple electrode platelet aggregometry on a Multiplate analyzer (Dynabyte, Munich, Germany). From the entire study population, 124 (10.1%) patients were under concomitant phenprocoumon treatment at the time point of platelet function testing. Platelet aggregation (median [interquartile range]) was significantly higher in patients with (n = 124) concomitant phenprocoumon treatment compared with patients without (n = 1099) phenprocoumon treatment (308 [190–493] AU*min vs. 224 [145–390] AU*min; P = 0.0001, adjusted P = 0.002).

Conclusion Phenprocoumon significantly attenuates the antiplatelet effects of clopidogrel. The impact of this interaction on the risk of thrombotic and bleeding events after PCI requires further investigations.

  • Clopidogrel
  • Phenprocoumon
  • Oral anticoagulation
  • Platelet aggregation

Introduction

For patients undergoing coronary stent placement, dual antiplatelet treatment with aspirin and clopidogrel is the therapy of choice to prevent thrombosis of the treated vessels and subsequent ischaemic events.1,2

Clopidogrel, an orally administered pro-drug, needs to be converted in vivo into its active thiol metabolite by the hepatic cytochrome P450 (CYP) system, which targets and inhibits the platelet adenosine diphosphate (ADP) P2Y12 receptor.3,4 A number of hepatic isoenzymes are involved in this two-step metabolization process of clopidogrel, including CYP2C19, CYP2C9, CYP3A4, CYP1A2, and CYP2B6.4,5 A large interindividual variability exists in platelet response to clopidogrel and clopidogrel low responsiveness, either following clopidogrel loading611 or in the setting of clopidogrel maintenance treatment,12,13 and has been associated with a higher risk of ischaemic events following percutaneous coronary intervention (PCI) including the occurrence of stent thrombosis.

Causative factors for clopidogrel low responsiveness are manifold.1419 Recent studies2027 have set the focus on specific genetic variants in genes encoding for isoenzymes of the hepatic CYP system, such as the CYP2C19 gene, that have been linked to clopidogrel low responsiveness and to higher rates of ischaemic events following PCI. In the context of hepatic CYP metabolization, a significant drug–drug interaction of proton pump inhibitors (PPI) and especially of the PPI omeprazole has been reported as well.2830 This drug–drug interaction may be explained by the circumstance that PPI treatment not only interferes with the clopidogrel metabolism at the level of the hepatic CYP-dependent metabolization process, but also at the level of intestinal absorption by changing the gastric pH value. Recently, however, the clinical impact of this well described drug–drug interaction has been questioned by the preliminary results of the Clopidogrel and the Optimization of GI Events (COGENT) Trial,31 showing no negative influence of concomitant PPI treatment on the incidence of ischaemic events in clopidogrel-treated patients. Besides PPIs, a number of other commonly prescribed drugs are also metabolized by the hepatic CYP system and do therefore have the potential to interact with clopidogrel metabolism.

Parts of patients with a history of coronary stent placement and stent-related antiplatelet treatment with aspirin and clopidogrel are also in need of concomitant oral anticoagulation with a coumarin derivative due to other comorbidities such as prior mechanic valve placement, atrial fibrillation, or a recent history of pulmonary embolism or deep vein thrombosis. Coumarin derivatives such as phenprocoumon are also metabolized by the hepatic CYP system and specifically by isoenzymes CYP3A4 and CYP2C9.32,33 Whether concomitant phenprocoumon treatment interferes with the antiplatelet effects of clopidogrel has never been investigated before.

The goal of this platelet function study was to investigate the impact of the coumarin derivative phenprocoumon on the antiplatelet effects of clopidogrel in patients with previous coronary stenting under maintenance treatment with aspirin and clopidogrel.

Methods

Patients

For this cross-sectional observational study, a total of 1223 consecutive patients were recruited at the Deutsches Herzzentrum München (Technische Universität München, Munich, Germany) between August 2007 and February 2009. Patients eligible for this study were in a stable condition and under dual antiplatelet treatment with aspirin and clopidogrel (75 mg/dL). Patients who were enrolled in this study were admitted for a control coronary angiography following coronary stenting which was scheduled as per institution protocol. All patients had undergone a PCI a median of 7 months [interquartile range (IQR) = 6–8] before study inclusion and had platelet function results available directly at hospital admission. The presence of an acute coronary syndrome and treatment with GP IIb/IIIa inhibitors during the 10 days before platelet function testing were exclusion criteria. The study of platelet function testing during angiography was approved by the institutional Ethics committee and complied with the Declaration of Helsinki and all patients gave written informed consent for it.

Blood sampling

For all patients, blood samples were taken directly on hospital admission in a fasting state and before any further in-hospital drug administration with a loose tourniquet through a short venous catheter inserted into a forearm vein. The first tube drawn was labelled as a discard, and was not used for platelet function testing. Blood was placed in 4.0 mL plastic tubes containing the anticoagulant lepirudin (25 µg/mL, Refludan, Hirudin blood collection tubes, Dynabyte). Blood samples were kept at room temperature for at least 30 min before platelet function testing.

Platelet function testing

ADP-induced platelet aggregation in whole blood was assessed with multiple electrode platelet aggregometry (MEA) on a Multiplate analyzer (Dynabyte) as previously described.34,35 The agonist ADP was used in a final concentration of 6.4 µmol/L, as it is recommended by the manufacturer. Aggregation measured with MEA is quantified as AU and area under the curve (AUC) of aggregation units (AU*min). All material used including ADP was obtained from the manufacturer (Dynabyte).

Study endpoints

The primary endpoint of this study was the ADP-induced platelet aggregation (in AU*min) in patients with concomitant phenprocoumon treatment vs. patients without phenprocoumon treatment. The secondary endpoint was the proportion of patients with a low responsiveness to clopidogrel in the group of patients with and without phenprocoumon treatment. For the present analysis, we defined a low response to clopidogrel treatment by setting a cutoff point at the upper quintile (20%) of patients according to MEA measurements.6

Statistical analysis

Variables are presented as mean ± SD, counts (%) or median with interquartile range [IQR]. For statistical analysis, categorical variables were compared using χ2 test. Kolmogorov–Smirnov test was used to test for normal distribution of continuous data. Normally distributed continuous data were compared between groups with two-sided t-test and non-normally distributed continuous data were compared between groups by two-sided Wilcoxon test. Platelet function data obtained with MEA were not normally distributed, are presented as median [IQR], and were compared between two groups with two-sided unpaired Wilcoxon test. For multivariate analysis, a multiple linear regression model was used with ADP-induced platelet aggregation as a continuous and dependent variable and as this variable was not normally distributed, it was log-transformed for inclusion into the model. Independent variables were concomitant treatment with phenprocoumon and a number of other clinically relevant variables considered to have a possible influence on the study results. Adjustment was therefore done by including all clinical variables into the model that have been shown to impact the antiplatelet effects of a clopidogrel maintenance treatment based on previous observations29 and also for age and all co-administered PPIs. Multivariate analysis was also performed by including platelet aggregation measurements as a dichotomous independent variable and by using a cutoff value defining the upper quintile of patients as clopidogrel low responders.6 The dichotomous model allowed the calculation of the odds ratio (OR) and the corresponding 95% confidence interval (CI). For all calculations a P-value <0.05 was considered statistically significant. Analyses were performed using the software package S-PLUS version 4.5 (Insightful Corp., Seattle, WA, USA).

Results

Patients

A total of 1223 patients were enrolled in this study. The mean age of patients was 67.6 ± 10.3 years and the proportion of women was 22% (n = 274). From the entire study population, 124 (10.1%) patients were under concomitant treatment with phenprocoumon at the time point of platelet function testing, whereas 1099 patients were not treated with a coumarin derivative. The mean blood international normalized ratio (INR) was 1.8 ± 0.6 in phenprocoumon treated patients vs. 0.95 ± 0.1 in patients not treated with a coumarin derivative. The different indications for the administration of phenprocoumon at the time point of platelet function testing were as follows (parts of patients had multiple indications): in 94 patients due to persistent or paroxysmal atrial fibrillation, in 5 patients due to persistent or paroxysmal atrial flutter, in 10 patients due to prior mechanic valve placement, in 7 patients due to a left ventricular thrombus, in 8 patients due to a recent history of pulmonary embolism, in 8 patients due to a recent history of deep vein thrombosis, and in 17 patients due to a history of a cerebral ischaemic insult. The baseline characteristics of the study population according to co-administration of phenprocoumon are shown in Table 1. Patients with concomitant phenprocoumon treatment were older, had higher serum creatinine levels, had less frequently a positive history of a myocardial infarction, and had more frequently a history of coronary bypass surgery. Comorbidities associated with the prescription of phenprocoumon were found more commonly in this group of patients. The other variables were well balanced between both groups, including the co-medication with relevant cardiovascular drugs. The proportion of patients with concomitant PPI use was also similar between the two groups.

View this table:
Table 1

Baseline characteristics of the study population

VariableWith phenprocoumon (n = 124)Without phenprocoumon (n = 1099)P-value
Age (years)71.4 ± 9.367.1 ± 10.4<0.001
Woman, n (%)26 (21.0)248 (22.6)0.69
Body mass index (kg/m2)27.6 [24.7–31.2]26.9 [24.6–29.8]0.17
Serum creatinine (mg/dL)1.12 ± 0.321.01 ± 0.350.001
Platelet count (×103/µL)210 ± 60216 ± 600.26
Diabetes mellitus, n (%)24 (19.4)291 (26.5)0.09
Active smokers, n (%)7 (5.7)118 (10.7)0.08
Arterial hypertension, n (%)82 (66.1)744 (67.7)0.72
Hypercholesterolaemia, n (%)91 (73.4)815 (74.2)0.85
Family history of CAD, n (%)50 (40.3)472 (42.9)0.58
Previous MI, n (%)36 (29.0)444 (40.4)0.01
Previous bypass surgery, n (%)29 (23.4)174 (15.8)0.03
Multivessel disease, n (%)108 (87.1)945 (86.0)0.73
Atrial fibrillation at admission, n (%)42 (33.9)16 (1.5)<0.001
Atrial flutter at admission, n (%)2 (1.6)3 (0.3)0.03
Mechanic valve placement, n (%)10 (8.1)0 (0.0)<0.001
Prior pulmonary embolism, n (%)8 (6.5)10 (0.9)<0.001
Prior deep vein thrombosis, n (%)8 (6.5)12 (1.1)<0.001
Prior ischaemic cerebral insult, n (%)17 (13.7)45 (4.1)<0.001
Prior LV thrombus, n (%)7 (5.7)9 (0.8)<0.001
Co-medication at admission, n (%)
 Aspirin124 (100.0)1099 (100.0)>0.99
 Beta-blocker114 (91.9)1041 (94.7)0.20
 ACE inhibitor104 (83.9)840 (76.4)0.06
 Calcium-channel antagonists18 (14.5)163 (14.8)0.92
 Nitrates3 (2.4)32 (2.9)0.76
Concomitant PPI treatment, n (%)
 Pantoprazole21 (16.9)183 (16.7)0.94
 Omeprazole10 (8.1)69 (6.3)0.44
 Esomeprazole5 (4.0)39 (3.6)0.78
 Lansoprazole0 (0.0)1 (0.9)0.73
Concomitant statin treatment, n (%)
 Simvastatin83 (66.9)755 (68.7)0.69
 Atorvastatin13 (10.5)176 (16.0)0.11
 Pravastatin7 (5.7)56 (5.1)0.79
 Fluvastatin3 (2.4)32 (2.9)0.76
  • Data presented are mean ± SD or number of patients (%). Body mass index is expressed as the median [interquartile range].

  • CAD, coronary artery disease; LV, left ventricular; PPI, proton pump inhibitor.

Platelet aggregation and phenprocoumon treatment

For the entire study population (n = 1223), ADP-induced platelet aggregation (median [IQR]) assessed with MEA was 229 [148–401] AU*min. ADP-induced platelet aggregation was significantly higher in patients with vs. patients without phenprocoumon treatment (308 [190–493] AU*min vs. 224 [145–390] AU*min, respectively; P = 0.0001). Figure 1 shows box-blot analyses of ADP-induced platelet aggregation values according to concomitant phenprocoumon treatment. We assessed ADP-induced platelet aggregation among patients on phenprocoumon treatment and compared the values between patients with an INR ≥2.0 (n = 48) with those with an INR <2.0 (n = 76). Despite the numerically higher values in patients with an INR ≥2.0 vs. patients with an INR <2.0 (351 [191–520] AU*min vs. 289 [190–420] AU*min, respectively; P = 0.19), the difference between both subgroups was not statistically significant.

Figure 1

Adenosine diphosphate-induced platelet aggregation and phenprocoumon. Box plot analyses (n = 1223 patients) of multiple electrode platelet aggregometry measurements for adenosine diphosphate-induced platelet aggregation according to phenprocoumon treatment. Boxes indicate 25th and 75th percentiles and whiskers denote 10th and 90th percentiles.

A cutoff value of 447 AU*min was observed to define the upper quintile (20%) of patients. These patients were considered as clopidogrel low-responders. Based on this cutoff value for MEA measurements under clopidogrel treatment, 245 (20%) patients of the present study population were found to be clopidogrel low responders. The remaining 978 patients (80%) were defined as normal responders. The proportion of patients with a low response to clopidogrel was significantly higher in patients with concomitant phenprocoumon treatment (n = 124) compared with patients without (n = 1099) phenprocoumon treatment [37 (29.8%) vs. 208 (18.9%); OR: 1.8, 95% CI: 1.2–2.8; P = 0.004].

Multivariate analysis

A multivariate analysis with ADP-induced platelet aggregation as the dependent and continuous variable demonstrated that concomitant treatment with phenprocoumon was independently associated with an attenuated platelet response to clopidogrel maintenance treatment (Table 2). Other variables that showed an independent association with the ADP-induced platelet aggregation were diabetes mellitus, body mass index, serum creatinine level, co-administration of omeprazole, active smoking, and a history of myocardial infarction. Detailed results of the multivariate analysis are demonstrated in Table 2. Multivariate analysis was also performed by including platelet aggregation measurements as a dichotomous independent variable and by using a cutoff value of 447 AU*min to define clopidogrel low responsiveness. In this model, concomitant treatment with phenprocoumon was found to be an independent predictor of low responsiveness to clopidogrel maintenance treatment (OR: 2.0, 95% CI: 1.3–3.1; P = 0.002).

View this table:
Table 2

Multivariable linear regression model

VariableRegression coefficientP-value
ValueStandard error
Phenprocoumon0.2680.0690.0001
Age0.0020.0020.33
Body mass index0.0120.0050.02
Serum creatinine0.1280.0620.04
Diabetes0.1800.0490.0002
Active smoker0.2170.0690.002
Previous myocardial infarction0.1650.0430.0001
Previous bypass surgery−0.0060.0560.92
Pantoprazole−0.0050.0560.93
Omeprazole0.2730.0850.001
Esomeprazole−0.0080.1110.94
Lansoprazole0.4710.7190.51
  • Multivariable linear regression model with adenosine diphosphate-induced platelet aggregation (log-transformed) assessed with multiple electrode platelet aggregometry as the dependent variable.

Discussion

To the best of our knowledge this is the first study investigating the impact of concomitant treatment with a coumarin derivative on the antiplatelet effects of clopidogrel in a large cohort of CAD patients with previous PCI under dual maintenance treatment with aspirin and clopidogrel. The key result of the present study is that concomitant treatment with phenprocoumon significantly attenuates the antiplatelet effects of clopidogrel. Patients under concomitant treatment with phenprocoumon exhibited approximately 35% higher values of ADP-induced platelet aggregation as compared with the remaining patients and about one-third of patients under phenprocoumon treatment were classified as clopidogrel low responders. In the multivariate analysis, we included a number of possible confounding variables. Results of this analysis support an independent association of phenprocoumon treatment with attenuated responsiveness to clopidogrel and results are strengthened by the fact that a number of well established predictors of clopidogrel responsiveness, such as diabetes mellitus,17,29,36,37 body mass index,19,29 renal insufficiency,17,29 or the intake of omeperazole,28,29 were also found to be independent predictors of clopidogrel responsiveness in the present study cohort.

For this platelet function study, we assessed the antiplatelet effects of clopidogrel in a steady-state situation in patients with clopidogrel maintenance treatment. This obviates the need to adjust results to a loading interval when platelet function is tested following the administration of a single clopidogrel loading dose. We used the MEA technique on a Multiplate analyzer for standardized assessment of ADP-induced platelet aggregation in whole blood.6,34,35,38 MEA is capable of detecting the amount of platelet inhibition achieved using different P2Y12 antagonists such as clopidogrel, cangrelor, and the active metabolites of clopidogrel and prasugrel in varying doses.39,40 The predictive value of MEA measurements for the occurrence of stent thrombosis and other ischaemic events following coronary stenting has been demonstrated recently in a prospectively designed clinical trial including 1608 clopidogrel-treated patients with coronary stent placement.6 The upper quintile of patients regarding platelet aggregation values to define clopidogrel low responsiveness that has been used in this trial was also used for the present analysis to study the association of clopidogrel low response and phenprocoumon intake. In terms of investigating possible drug–drug interactions in the setting of clopidogrel treatment, the reliability and validity of MEA measurements have been demonstrated in a number of previous studies investigating the impact of different PPI,29,30 unfractionated heparin or bivalirudin,41 and calcium-channel blockers38 on clopidogrel responsiveness.

Results of our study strongly suggest that a concomitant intake of phenprocoumon significantly alters the in vivo biotransformation of clopidogrel into its active thiol metabolite. Both drugs are metabolized by the hepatic CYP system and a drug–drug interaction at this level is therefore self-evident. The major isoenzymes for phenprocoumon metabolization are CYP2C9 and CYP3A4 and both isoenzymes are also involved in the metabolization process of clopidogrel.4,5,32,33 Phenprocoumon, when taken regularly, claims CYP2C9 and CYP3A4 pathway capacities,32,33 which may in turn lead to reduced biotransformation of clopidogrel resulting in higher ADP-induced platelet aggregation values as observed in our study. Concerning the CYP3A4 pathway, a negative impact on the antiplatelet effects of clopidogrel has also been demonstrated for other drugs that claim CYP3A4 pathway capacities such as calcium-channel blockers38 or ketoconazole.42 Concerning the CYP2C9 pathway, the same common loss-of-function polymorphisms in the CYP2C9 gene have been shown to influence the metabolization of both phenprocoumon32 and clopidogrel.5 Data, however, are in part conflicting since other studies24 have failed to show a relation between polymorphisms in the CYP2C9 or CYP3A4 gene and clopidogrel pharmacodynamics.

The clinical implications of the findings reported here remain to be determined. Available evidence from different clinical trials does not suggest that patients receiving concomitant treatment with a coumarine derivate are at higher risk for suffering coronary thrombotic events.43 This seems to contrast with the findings of our present study showing that phenprocoumon significantly attenuates the antiplatelet effects of clopidogrel. It may be speculated, however, that the negative impact of phenprocoumon on the antiplatelet effects of clopidogrel in patients is offset by the strong effect of phenprocoumon treatment on the coagulation system and in particular on the availability of thrombin that prevents an increased risk of thrombotic events. The impact of this drug–drug interaction on the risk of thrombotic events after PCI requires further analyses and results reported here may provide the rationale for such investigations.

Limitations

The present study has limitations that merit mention. We were only able to investigate the impact of the coumarin derivative phenprocoumon in the present study population, since this is the most widely prescribed coumarin derivative in Germany and most European countries. Further studies have to demonstrate whether the results reported here for phenprocoumon can be extrapolated to other coumarin derivatives such as warfarin or acenocoumarol. The number of patients (n = 124) with triple therapy (aspirin, clopidogrel, and phenprocoumon) is relatively small, and further studies with larger cohorts are certainly warranted. In addition, we only provide pharmacodynamic platelet aggregation data for clopidogrel treatment. Pharmacokinetic data on the metabolization of clopidogrel in the presence or absence of concomitant phenprocoumon treatment is not provided, but may be an important issue for further studies to corroborate present results. Furthermore, we only focused on measuring the platelet response to clopidogrel after stimulation with ADP. Measuring a number of markers of platelet responsiveness and activity simultaneously was not undertaken here, but would be an interesting issue for further studies. Finally, the present study is a cross-sectional observational study and therefore it is subject to limitations inherent to all such analyses. Specifically, the non-randomized nature of coumarin administration is likely to reflect both observed and unknown selection bias.

Conclusion

Phenprocoumon significantly attenuates the antiplatelet effects of clopidogrel. The impact of this interaction on the risk of thrombotic and bleeding events after PCI requires further investigations.

Funding

Material for platelet aggregation measurements were provided free of charge from Dynabyte (Munich, Germany).

Conflict of interest: D.S. reported receiving speaker fees from Dynabyte and fees for advisory board activities from Eli Lilly. N.B. reported receiving speaker fees from Eli Lilly and fees for advisory board activities from Eli Lilly and sanofi-aventis. A.D. reported receiving speaker fees from Eli Lilly, sanofi-aventis, and Bristol-Myers Squibb.

Acknowledgements

Material for platelet function analysis on the Multiplate device was provided free of charge from Dynabyte. The sponsors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; or preparation, review, or approval of the manuscript.

References

View Abstract