E, measurement at the sampling time; IPA, inhibition of platelet aggregation; k and A, coefficients of the equation; MPA, maximal platelet aggregation; R2, coefficient of determination. AUECpred is calculated from A (regression coefficient, slope) × E (the value of the measurement [MPA, ΔMPA, or IPA]) + k (intercept, constant).
LETTERS TO THE EDITOR
Dissecting pharmacodynamics to determine the optimal sampling time and measurement for assessing the antiplatelet effect of clopidogrel
Article first published online: 1 OCT 2012
© 2012 International Society on Thrombosis and Haemostasis
Journal of Thrombosis and Haemostasis
Volume 10, Issue 10, pages 2196–2199, October 2012
How to Cite
KIM, S.-D., JEONG, Y.-H., LEE, S.-W., SHIN, I. H., PARK, J.-Y., PARK, S.-W., YOON, Y.-R. and SONG, J.-K. (2012), Dissecting pharmacodynamics to determine the optimal sampling time and measurement for assessing the antiplatelet effect of clopidogrel. Journal of Thrombosis and Haemostasis, 10: 2196–2199. doi: 10.1111/j.1538-7836.2012.04870.x
- Issue published online: 1 OCT 2012
- Article first published online: 1 OCT 2012
- Accepted manuscript online: 27 JUL 2012 10:41AM EST
- Received 17 May 2012, accepted 12 July 2012
Clopidogrel has been used to treat and prevent a variety of atherothrombotic diseases, but recurrent cardiovascular events may nonetheless occur. Multiple studies have demonstrated substantial interindividual pharmacodynamic variability with clopidogrel therapy, and an impaired platelet response to clopidogrel or ‘high on-treatment platelet reactivity’ (HPR) has been associated with the occurrence of ischemic events . Even though significant efforts have been made to standardize methods for quantifying platelet reactivity (PR), there is still a lack of scientific consensus on the timing of blood sampling and optimal measurement . Previous studies have shown that the antiplatelet effect of clopidogrel varies with the time of day, even during maintenance therapy [2,3]. However, clinical studies that have examined the link between PR and clinical outcome during clopidogrel treatment have used different postdose sampling times [4–6], or failed to specified the timing of blood sampling . On-treatment PR, the absolute level of PR during clopidogrel treatment, has become a widely adopted indicator of clopidogrel response variability, and HPR is accepted as a major risk factor for ischemic events after percutaneous coronary intervention . However, there are limited data indicating that on-treatment PR provides a better indication than others, such as absolute or relative changes in PR from baseline .
The area under the concentration–time curve (AUC) is the most commonly used pharmacokinetic parameter used to characterize total exposure of a drug, but its routine clinical use is limited by the need for multiple blood samples. To reduce the number of blood samples, limited sampling strategies (LSSs) have been devised to determine the best sampling times for prediction of the AUC by the use of regression models . Platelet aggregation-inhibition–time curve (area under the effect curve [AUEC]) is pharmacodynamic counterpart of the AUC, and can be used as a surrogate marker for overall platelet aggregation-inhibition status during clopidogrel treatment [9–11]. In the present study, we developed LSS models to estimate the AUEC of clopidogrel in healthy subjects, in order to determine the optimal sampling time and measurement for assessing platelet response to clopidogrel.
Two datasets from prospective trials involving a total of 123 healthy subjects at independent institutions in Korea were retrospectively reviewed. The first dataset, used for the development of LSS models, was obtained from the pharmacodynamic data of 43 subjects after administration of clopidogrel bisulfate (Plavix) in a phase I trial that compared two clopidogrel salt preparations at the Kyungpook National University Hospital, Daegu, Korea . The second dataset, used for external validation, was derived from the pharmacodynamic data of 80 subjects after a 300-mg loading dose (LD) of clopidogrel. All subjects received clopidogrel as a 300-mg clopidogrel LD on day 1, and as a 75-mg daily maintenance dose (MD) on days 2–6. Healthy male subjects were between 20 and 55 years of age, and other characteristics of the subjects are shown in Data S1. Our protocols were approved by local institutional review boards. In the development dataset, blood samples were taken during the 300-mg LD period (day 1: predose , and 2, 4, 8, 12 and 24 h after the first dose) and the MD period (day 6 [+ 120 h]: predose , and 2, 4, 8, 12 and 24 h after the sixth dose). PR was quantified with the use of light transmittance aggregometry (LTA), with 5 μm ADP as the agonist. Detailed descriptions of the assay method and statistical analysis are given in Data S1.
Three measurements were used to define clopidogrel response. Maximal platelet aggregation (MPA) was defined as the maximal percentage change in OD relative to platelet-poor plasma at each time point. The absolute change in aggregation (ΔMPA) was calculated as baseline aggregation (MPAbaseline) minus postdose aggregation (MPApostdose) . The relative change in aggregation (inhibition of platelet aggregation [IPA]) was calculated from the observed MPA at each time point for each treatment, with the following formula :
Observed AUECs (AUECobs) during the LD (AUEC0–24) and MD (AUEC120–144) periods were separately calculated by non-compartmental analysis with WinNonlin Pro 5.2 (Pharsight Corporation, Mountain View, CA, USA). AUECobs was calculated from the MPA vs. time curve by use of the linear trapezoidal rule.
Simple linear regression was used to assess the ability of the single measurement–time points to predict AUEC0–24 after the clopidogrel LD. A measurement–time point indicated a particular measurement (MPA, ΔMPA, or IPA) at each time point (day 1: predose , or 2, 4, 8, 12 or 24 h after dosing). Each analysis gave the parameters of the following equation, allowing calculation of the predicted AUEC (AUECpred):
At and kt represent the coefficients of the equation at time t used to calculate predicted AUECEt, and Et is the value of the measurement (MPA, ΔMPA, or IPA) at time t. Correlations between AUEC120–144 and a single measurement–time point were also evaluated with the same procedure. Finally, the LSS models were validated for predictive power by both internal and external validations (Data S1).
The mean MPA profiles in the development and validation dataset are shown in Fig. S1. The AUEC was significantly higher during the MD period (1436.8 ± 225.9% h) than during the LD period (1247.9 ± 233.0% h, P < 0.001), and the AUEC values between two periods were correlated significantly (r = 0.78, P < 0.001) (Table S1). Repeated-measures one-factor anova indicated a significant heterogeneity in MPAs during both the LD and MD periods (P < 0.001). Table 1 shows the regression equations and their R2 for estimation of the AUEC in the development dataset. All three measurements at each time were significantly correlated with their AUECs (P < 0.05). MPA had higher R2 values than ΔMPA and IPA, and all of the measurements at 12 h postdose had the highest R2 values. The best correlations with AUEC0–24 and AUEC120–144 were for MPAs at 12 h postdose (MPA12 and MPA132, R2 = 0.76 and R2 = 0.77, respectively).
|A||k||R 2||A||k||R 2||A||k||R 2|
|Loading-dose period (0–24 h)|
|Maintenance-dose period (120–144 h)|
Leave-one-out cross-validation analysis on the predictive performance parameters of each regression equation showed that the models with the largest R2 values (Table 1) also had the best predictive performance (Table S2). MPA at 12 h postdose (MPA12 and MPA132) had the lowest bias (mean relative prediction errors [MPEs] of 0.96% and 0.74%, respectively) and the highest precision (root mean squared relative errors [RMSEs] of 9.87% and 8.23%, respectively). Bootstrapping analysis also showed that the optimal sampling time–measurement combination was MPA at 12 h postdose (MPA12 and MPA132), which appeared in 887 (88.7%) and 697 (69.7%) of the 1000 models (P < 0.001). The best correlations with the AUEC were consistently for MPA at 12 h postdose (R2 = 0.93, MPA = 1.43%, RMSE = 10.0%) in the independent external validation dataset. Subsequent bootstrapping in the validation dataset also indicated that the most commonly selected sampling time–measurement combination was MPA at 12 h postdose, which appeared in 557 (55.7%) of the 1000 models (P < 0.001).
This is the first study to validate the optimal sampling time and measurement for assessing the antiplatelet response to clopidogrel after LDs and MDs. The main findings are as follows: (i) there is high intraindividual variability in PR during clopidogrel loading and maintenance treatment; and (ii) MPA at 12 h postdose provides the least biased and most precise estimate of the clopidogrel AUEC. These observations strengthen our conclusion that the antiplatelet response to clopidogrel may be best indicated by ADP-induced maximum platelet aggregation at 12 h postdose measured by LTA.
On the basis of our results, we suggest that timed blood sampling should be used in studies that aim to assess PR during clopidogrel treatment, and that blood sampling at 12 h postdose is a reasonable choice for future studies that aim to determine PR during clopidogrel treatment. Furthermore, clopidogrel administration in the late evening could be considered, as the maximal pharmacodynamic effect of clopidogrel can cover the vulnerable morning period, in which there is an increased incidence of stent thrombosis, myocardial infarction, stroke, and sudden cardiac death .
The AUEC of clopidogrel has not been proven to reflect overall platelet aggregation-inhibition status during clopidogrel therapy. However, several studies have shown a relationship between MPA values at various sampling times (e.g. 6 h or 12–18 h after clopidogrel loading) and clinical outcome [4,5]. As individual MPA values are significantly related to clinical outcome, the summation of separate MPA values (i.e. the AUEC) could be considered as a representative measurement of the entire clinical effect of clopidogrel.
The present study has several limitations. First, the sample size was relatively small, subjects were all healthy East Asian subjects, and the data analysis was performed by reanalysis of prospective trials. Further large-scale prospective studies including coronary artery disease patients are needed to confirm our results, and the optimal sampling time should be selected on the basis on pharmacodynamic characteristics and other important clinical factors. Second, determination of MPA with LTA was the only method used to measure the antiplatelet response to clopidogrel. Prospective trials using more simplified methods, such as the VerifyNow P2Y12, assay are needed. Third, the optimal time and measurement after high-dose clopidogrel LD or MD cannot be suggested on the basis of this study. Finally, because there were no available data between 12 and 24 h postdose, a dedicated study to bridge this time gap is required.
Disclosure of Conflict of Interests
Y.-H. Jeong has received honoraria for lectures from Sanofi-Aventis, Daiichi Sankyo Inc., and Otsuka. The other authors state that they have no conflict of interest.
- 6Adjusted clopidogrel loading doses according to vasodilator-stimulated phosphoprotein phosphorylation index decrease rate of major adverse cardiovascular events in patients with clopidogrel resistance: a multicenter randomized prospective study. J Am Coll Cardiol 2008; 51: 1404–11., , , , , , , , , .
- 11Twenty-four-hour area under the concentration–time curve/MIC ratio as a generic predictor of fluoroquinolone antimicrobial effect by using three strains of Pseudomonas aeruginosa and an in vitro pharmacodynamic model. Antimicrob Agents Chemother 1996; 40: 627–32., , , .
Data S1. Methods
Figure S1. Maximal platelet aggregation profiles in (A) development dataset (n = 43) and (B) validation dataset (n = 80).
Table S1. Pharmacodynamic parameters of clopidogrel (5 μM ADP-induced platelet aggregation) after 300-mg loading (day 1) and 75-mg daily maintenance doses (days 2 to 6) (n = 43)
Table S2. Validation of regression models to predict platelet aggregation-inhibition-time curve (AUEC) during 300-mg loading (AUEC0–24) and 75-mg maintenance doses (AUEC120–144) (n = 43)
|JTH_4870_sm_FigS1.tif||147K||Supporting info item|
|JTH_4870_sm_SupplementaryMethods-TableS1-S2.doc||91K||Supporting info item|
Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.