Platelet inhibition during percutaneous catheter intervention (PCI) reduces the risk of early complications, in particular, peri-interventional myocardial infarction (MI) and the need for emergency reintervention. This has been shown consistently by studies on peri-interventional administration of glycoprotein (GP) IIb/IIIa antagonists (1- 2) and, more recently, by studies on pre-treatment with thienopyridines (3- 4). The benefit from peri-interventional GP IIb/IIIa receptor blockade is critically dependent on the level of platelet inhibition achieved: In the GOLD (AU—Assessing Ultegra) study, the quartile of patients with the lowest level of peri-interventional platelet inhibition (i.e., <95% inhibition of platelet aggregation), had a more than 2-fold increase in early (≤7 days) major adverse cardiac events (MACE) (composite of death, myocardial infarction, and urgent target vessel revascularization) (5).

The ISAR-REACT (Intracoronary Stenting and Antithrombotic Regimen—Rapid Early Action for Coronary Treatment) and ISAR-SWEET (Intracoronary Stenting and Antithrombotic Regimen: Is Abciximab a Superior Way to Eliminate Elevated Thrombotic Risk in Diabetics) trials demonstrated that pre-treatment with a 600-mg loading dose of clopidogrel can replace peri-interventional GP IIb/IIIa receptor blockade in low- to intermediate-risk PCI (6- 7). The platelet response to this loading dose is, however, highly variable (8- 10), and there is considerable concern that patients with a low or absent response to clopidogrel might be at increased risk of complications. Several criteria for such nonresponder status have been suggested (11- 12), but none of them have been validated clinically.

To explore this issue, we conducted the EXCELSIOR (Impact of Extent of Clopidogrel-Induced Platelet Inhibition During Elective Stent Implantation on Clinical Event Rate) study. This prospective observational study investigated the impact of the variability in peri-interventional platelet function on the clinical outcome of patients undergoing low to intermediate-risk PCI. Similarly to a previous study (13), we focused on absolute values of platelet reactivity rather than change from baseline. We hypothesized that a low level of platelet reactivity at the time of PCI was beneficial, irrespective of whether this was due to pharmacologic inhibition or to low baseline platelet aggregation. Specifically, we tested the hypothesis that the composite 30-day incidence of MACE differs by quartiles of the level of adenosine diphosphate (ADP)-induced platelet aggregation after loading with 600 mg clopidogrel. The reasons to administer a loading dose of 600 mg instead of the previously widely accepted dose of 300 mg (4,14) were that this has been: 1) used in large-scale clinical studies (6- 7,15); 2) shown to be safe (6- 7,15- 16); 3) shown to be more effective on platelet inhibition (9,17); and 4) associated with better clinical outcomes (18).

Patients undergoing elective coronary stent placement after pre-treatment with 600 mg clopidogrel and aspirin (≥100 mg per day for at least 5 days) were eligible for the study. We screened patients scheduled for cardiac catheterization as potential candidates for PCI. Major exclusion criteria were acute MI according to the American Heart Association/American College of Cardiology (AHA/ACC) criteria, chronic oral anticoagulation, thienopyridine treatment within the last 2 weeks before admission, contraindications to aspirin, clopidogrel, or heparin, cancer, hemodialysis, and hemodynamic instability. All patients gave written informed consent. The study was approved by the ethics committee of the medical faculty of the University of Freiburg, Germany.

Before the intervention, all patients were on aspirin and received a loading dose of 600 mg clopidogrel. Catheterization was timed according to the routine schedule of the catheterization laboratory. A delay of at least 2 h between clopidogrel loading and catheterization was strongly recommended. Noncompliance with this recommendation, however, did not preclude participation in the study. The choice of stent type was left to the operator’s discretion. All patients received an intra-arterial dose of 100 to 140 U/kg heparin; GP IIb/IIIa inhibitors were not allowed except for bail-out in case of extensive dissection or thrombus formation with deterioration of coronary flow to Thrombolysis In Myocardial Infarction (TIMI) flow grade <2. After PCI, all patients received aspirin (≥100 mg/day) and 75 mg/day clopidogrel for the duration of the study.

Immediately after inclusion in the study, blood was drawn for platelet function assays using tubes containing 3.8% sodium citrate (Sarstedt, Nümbrecht, Germany). We obtained the second blood sample at the time of catheterization before administration of heparin or contrast medium.

Platelet aggregation was assessed by turbidimetric aggregometry using a 4-channel Bio/Data PAP4 aggregometer (Mölab, Langenfield, Germany), as described previously (8). We prepared platelet-rich plasma by centrifugation of citrated venous blood at 750 g for 2 min and adjusted to 275 to 325 × 10⁹ thrombocytes/l by dilution with platelet-poor plasma from the same patient. We used ADP (Sigma, Munich, Germany) to induce aggregation. Light transmission in platelet-rich plasma was determined 5 min after addition of ADP at a final concentration of 5 μmol/l. Results were expressed as percentage of maximal light transmission using platelet-poor plasma from the same patient as reference (100% aggregation). Compared with maximal light transmission used in some of the previous studies (8- 10), light transmission at 5 min comprises both formation and stability of aggregates but yields a lower read out. The coefficient of variation of our optical aggregometry assay is 6.1% (8).

The primary end point of the study was the cumulative incidence of death from any cause, MI, or urgent target lesion revascularization (aortocoronary bypass surgery or PCI) due to myocardial ischemia within 30 days after stent placement. The diagnosis of myocardial infarction was based on either the development of pathologic Q waves in 2 or more contiguous electrocardiographic leads or an elevation of creatine kinase or its MB isoenzyme to at least 3 times the upper limit of normal in at least 2 blood samples. As a secondary end point, we monitored the plasma concentrations of troponin T. We obtained blood samples for determination of creatinine kinase and troponin T before PCI, at 8, 16, and 24 h after PCI (with a 2-h tolerance), and before discharge (36 to 48 h after PCI). At the same time points, we obtained 12-lead electrocardiographic recordings.

In addition, we assessed bleeding complications. According to the TIMI definitions (19), a bleeding event was defined as major if it was intracranial or if there were clinically significant overt signs of hemorrhage associated with a drop in hemoglobin of >5 g/dl. To account for transfusion, an increase in hemoglobin of 1 g/dl was assumed for each unit of blood.

We performed a phone interview at 30 days. For patients reporting cardiac symptoms, at least one clinical and electrocardiographic examination was performed in the outpatient clinic or by the referring physician. Thirty-day follow-up was completed in all patients. All information derived from contingent hospital readmission records or provided by the referring physician or by the outpatient clinic was entered into the computer database. All events were classified and adjudicated by a physician not involved in the follow-up process.

Sample size calculation was based on ISAR-REACT (6), which comprised a cohort with similar selection criteria and treatment strategy. Thus, we assumed an incidence of the primary end point of 4.2%. We designed our study to test the hypothesis that the incidence of the primary end point differed by quartiles of ADP-induced platelet aggregation. We intended to have a power of 0.80 to detect an effect size of 0.015 (e.g., 3-fold risk in the fourth quartile) with a 2-sided p value <0.05. With these assumptions, we obtained a sample size of at least 748 (nQuery Advisor, version 5.0, Statistical Solutions, Cork, Ireland) and aimed for a cohort of 800. For all statistical analyses, we used the SPSS software package, version 11.5 (SPSS, Chicago, Illinois). In general, discrete variables are reported as counts (percentages) and continuous variables as mean ± SD. We tested differences between groups with the chi-square test for discrete variables and with 1-way analysis of variance for continuous variables. In case of a highly skewed distribution of a continuous variable (skewness ≥2), we tested differences between groups by the Kruskal-Wallis test and report this variable as median (interquartile range). Event-free survival was analyzed by the Kaplan-Meier method, and differences in event-free survival were assessed by means of the log rank test. We calculated various univariable and multivariable logistic regression models for the primary end point as dependent variable. To control for potential confounders, the multivariable models included demographic, clinical and angiographic variables (Table 1) with a difference between quartiles of aggregation at p ≤ 0.20. In addition, we added time from clopidogrel loading and baseline platelet aggregation to the model. To identify factors that correlate with platelet aggregation immediately before intervention, we performed univariable and multivariable analyses with platelet aggregation as a continuous variable using the general linear model. Apart from baseline platelet aggregation and percentage inhibition of platelet aggregation, the multivariable model included as factors and covariables any demographic, clinical, and angiographic variable (Table 1) that in univariable models showed an association with platelet aggregation immediately before PCI at p ≤ 0.20. In the 2-tailed test, a p value of <0.05 was regarded as significant.

(Figure 1) shows the trial profile. The study cohort comprised 802 patients. (Table 1) shows the demographic and baseline clinical characteristics of the study cohort as well as the procedural and angiographic variables.

Thirty-day follow-up was completed in all patients. As shown in (Table 2), 15 patients incurred 19 major adverse cardiac events within 30 days. After PCI, troponin T concentrations above the detection limit (>0.03 μg/l) were found in 34.4% of the patients, but in only 7.5% troponin T increased to more than 10 times the detection limit (Table 2). Two patients received abciximab for bail-out.

Immediately before PCI, median ADP-induced (5 μmol/l) platelet aggregation was 14% with an interquartile range of 4% to 32%. As shown in (Figure 2) and (Table 2), the incidence of MACE during 30-day follow-up increased with quartiles of ADP-induced platelet aggregation. It was 0.5% in the 2 quartiles with the lowest platelet aggregation, but 3.1% in the third quartile and 3.5% in the highest quartile (p = 0.034). Thus, platelet aggregation above median was associated with a relative risk for 30-day MACE of 6.71 (95% confidence interval [CI] 1.52 to 29.41) (p = 0.003).

As shown in (Table 2), quartiles of platelet aggregation were not significantly associated with postprocedural elevations in troponin T or the composite 30-day incidence of MACE and post-procedural elevations in troponin T. Likewise, bleeding and vascular access site complications as well as the composite incidence of cardiac and noncardiac complications did not vary significantly with the level of ADP-induced platelet aggregation (Table 2).

The majority of the demographic, baseline clinical, procedural, and angiographic variables did not differ significantly between quartiles of platelet aggregation. Patients in the higher quartiles were, however, significantly older and had a significantly higher body mass index. Likewise, the proportion of patients with diabetes mellitus and impaired left ventricular function increased significantly with increasing quartiles of platelet aggregation, whereas the count of drug-eluting stents implanted decreased significantly. There were trends toward a higher proportion of active smokers and of patients with hypertension or with non-left anterior descending PCI and a trend toward a lower percentage diameter stenosis before PCI in the higher quartiles. After adjustment for these variables in a multivariable logistic regression model, platelet aggregation above median was associated with an adjusted odds ratio for 30-day MACE of 7.2 (95% CI 1.6 to 33.8) (p = 0.011) (Figure 3A).

As shown in (Table 3), patients belonging to the strata with the higher ADP-induced platelet aggregation before intervention more often underwent PCI within 2 h after loading with clopidogrel than patients in the lower strata. Conversely, the proportion of patients with an ADP-induced platelet aggregation above 14%, the median of the entire cohort, varied significantly with time after clopidogrel loading (Figure 4). To adjust the odds ratio for 30-day MACE for time from clopidogrel loading, we included this variable in the multivariable logistic regression model, which yielded an adjusted odds ratio for 30-day MACE of 9.6 (95% CI 2.1 to 44.3) (p = 0.004) (Figure 3A).

Patients belonging to the strata with the lower ADP-induced platelet aggregation before intervention also had a significantly lower platelet aggregation at baseline (Table 3). We therefore calculated a third multivariable logistic regression model which also included baseline platelet aggregation. In this model, the strong association between platelet aggregation at the time of intervention and 30-day MACE prevailed (Figure 3A).

In a 2-variable model, the variability in baseline platelet aggregation accounted for 32% and that of the inhibition of platelet aggregation from baseline to immediately before PCI for 56% of the variability in ADP-induced platelet aggregation before intervention. Time from clopidogrel loading had a strong impact on the change in platelet aggregation. In patients with a time from clopidogrel loading of <2 h, the percentage inhibition of ADP-induced platelet aggregation from baseline to immediately before PCI was 39 ± 54%, whereas it was 62 ± 53% with a delay of more than 2 h (p < 0.001). Apart from baseline platelet aggregation, percentage inhibition of platelet aggregation immediately before PCI and time from clopidogrel loading (PCI within the first 2 h after loading dose), age, hypertension, diabetes, body mass index, and number of drug-eluting stents were significantly (p < 0.05) related to platelet aggregation in univariable models. In addition, total cholesterol, reduced left ventricular function, stent in left anterior descending coronary artery, and balloon pressure during PCI were correlated with platelet aggregation by trend (p ≤ 0.20). These variables were entered in the multivariable general linear model (Table 4). In this model, baseline platelet aggregation and percentage inhibition of platelet aggregation from baseline to immediately before PCI prevailed as the strongest predictor of platelet aggregation, accounting for 30% and 53%, respectively, of the variability in ADP-induced platelet aggregation immediately before PCI.

To corroborate our analyses based on strata of ADP-induced platelet aggregation before intervention, we calculated multivariable logistic regression models with ADP-induced platelet aggregation before intervention entered as a continuous variable and performed stepwise adjustment for its determinants. As shown in (Figure 3)B, these models confirmed the primary analyses. After adjustments for demographic, clinical and angiographic variables as well as for time from clopidogrel loading and baseline platelet aggregation, a 10% increase in ADP-induced platelet aggregation before intervention was associated with an adjusted odds ratio for 30-day MACE of 1.32 (95% CI 1.04 to 1.61) (p = 0.026). This was similar to the corresponding unadjusted odds ratio of 1.31 (95% CI 1.03 to 1.67) (p = 0.026).

This prospective study demonstrated the clinical relevance of the variability of platelet aggregation after the 600-mg loading dose of clopidogrel in patients undergoing elective PCI. The 30-day incidence of MACE after elective PCI increased significantly with quartiles of ADP-induced platelet aggregation. It was 0.5% in the 2 quartiles with the lowest platelet aggregation, but 3.1% and 3.5% in the third and fourth quartiles, respectively. We, thus, found a substantial difference in the 30-day incidences of MACE between patients with a platelet aggregation above or below median, but very similar incidences within the 2 lowest and the 2 highest quartiles. Although event rates were low irrespective of the level of platelet aggregation, an ADP-induced platelet aggregation above median was associated with a more than 6-fold increase in risk of early MACE compared with a platelet aggregation below median. Our prospective study identified platelet function assessed before intervention as a strong independent prognostic factor in patients who receive pre-treatment 600 mg clopidogrel.

The observed variability in residual platelet aggregation immediately before PCI depends by 30% on the interindividual variability in platelet function even before the initiation of clopidogrel and by just above 50% on the variability in platelet responses to clopidogrel loading. An adequate delay from loading is critical to the clopidogrel effect at the time of PCI: the proportion of patients with suboptimal inhibition of platelet aggregation during PCI (i.e., platelet aggregation above median of entire cohort) varied considerably with time after loading, from 66.5% in those undergoing catheterization within 2 h after loading to 41.8% in those with a longer delay. This finding is consistent with earlier studies (8) demonstrating that after a 600-mg loading dose of clopidogrel 2 h are needed before a full (or near-full) effect is achieved. It is noteworthy that the proportion of patients with suboptimal platelet inhibition is still large even after a sufficient (i.e., >2 h) wait and that, consistent with our earlier study (8), an additional delay does not noticeably reduce this proportion.

The question arises whether the relation between MACE and platelet aggregation at the time of PCI can be attributed to the variability of time from clopidogrel loading or to baseline aggregation and whether it is independent of confounding baseline variables. To this end we calculated various multivariable logistic regression models. In the first step we adjusted for demographic, clinical, and angiographic covariables, which demonstrated an independent relation between MACE and platelet aggregation at the time of PCI. This confirmed that the variability in platelet aggregation at the time of PCI, irrespective of its cause, is relevant to clinical outcome. In a second and third step, we also adjusted for time from clopidogrel loading and baseline aggregation. In these models, the significant independent relation between MACE and platelet aggregation at the time of PCI prevailed. In the aggregate, the multivariable analyses demonstrate that the underlying biologic differences in response to clopidogrel are relevant to clinical outcome, rather than simply the differences in baseline characteristics or differences in time of exposure to clopidogrel before PCI.

In addition to the findings of the GOLD study on GP IIb/IIIa blockade (5), several clinical observations suggested that the peri-interventional level of platelet inhibition is crucial to the prevention of early complications after PCI. In TARGET (Do Tirofiban and ReoPro Give Similar Efficacy Outcomes Trial), differences in 30-day outcome after PCI between tirofiban and abciximab could be attributed to differences in the level of platelet inhibition at the time of PCI (20). Other studies reported consistent results for thienopyridines by showing that the early outcome of PCI is superior if the thienopyridine is effective at the time of PCI than if it is started immediately after PCI (3- 4). More recently, the ARMYDA (Antiplatelet Therapy for Reduction of Myocardial Damage During Angioplasty) study reported a more favorable 30-day outcome after pre-treatment with 600 mg than after pre-treatment with 300 mg clopidogrel (18).

Pursuing the concept that adequate peri-interventional platelet inhibition is critical, the present study focused on the impact of the peri-interventional level of platelet aggregation. To our knowledge, it provides the first prospective data that the peri-interventional level of platelet aggregation after loading with 600 mg clopidogrel is related to the outcome of elective PCI in low- to intermediate-risk patients. Most of the events that could be related to the level of platelet aggregation were detected within 48 h after PCI, but 4 events occurred between days 3 and 11. These subacute events may be interpreted as a late sequel of insufficient peri-interventional platelet inhibition or as a consequence of an inadequate suppression of platelet function that continues during chronic therapy. The latter interpretation is supported by recent retrospective studies that reported low levels of platelet inhibition under chronic treatment with clopidogrel in patients presenting with late stent thrombosis (21- 22).

Compared with previous studies (1- 2,6,23- 24), the incidence of MACE in the current study, in particular that of MI, was remarkably low. For most previous studies (1- 2,23- 24), a large extent of this difference can be attributed to technologic advances and to differences in patient selection, with the exclusion of higher-risk patients in the present study. We used, however, the same interventional approach and the same selection criteria as ISAR-REACT (6), the study which our power calculation was based on. Consequently, most of the baseline characteristics were similar in our study and, notably, two-thirds of the lesions treated in both studies were complex (AHA/ACC lesion type B2 or C). The proportion of patients with class III/IV angina was, however, higher in ISAR-REACT (40% vs. 25% in the present study), which may have contributed to the lower event rate in our cohort. The fact that the event rate was lower than projected reduces the power of the present study. The limited number of events prevents us from defining the exact threshold for the level of platelet aggregation above which patients need to be considered at high risk.

Similar to other studies (25- 26), about one-third of our patients incurred minor myocardial necroses, as evidenced by peri-interventional increases in cardiac troponin T. We could not detect any significant relation between minor myocardial necroses and any of the platelet function variables that we assessed in this study, although the power to detect such relation would have been high. One potential explanation for insensitivity could be that minor myocardial necroses are not a platelet-related event. As alternative mechanisms, one may consider distal embolization of plaque material or inadvertent obstruction of small side branches.

Most of the patients included in this study had a low risk profile. In patients with higher risk, suboptimal platelet inhibition may become even more important. Specifically, this appears to be true in patients with acute coronary syndromes, as suggested by a recent small study (27). This study showed poor outcome of PCI in acute myocardial infarction if platelet inhibition by clopidogrel was inadequate. Consistently, ISAR-REACT-2 revealed that in high-risk acute coronary syndromes, clopidogrel alone does not afford sufficient protection against post-interventional ischemic events and that abciximab is still needed to optimize outcome in this setting (15). Moreover, with the more widespread use of drug-eluting stents an increasing proportion of patients with high-risk angiographic features will undergo PCI. The risk of such interventions may become prohibitive if platelet inhibition is suboptimal.

The results of the present study, therefore, encourage the search for improved treatment regimens that give a more robust platelet inhibition or for new P2Y₁₂ receptor antagonists that offer a rapid efficacious and predictable platelet inhibition. In this respect, our study strongly endorses the use of adequate loading dosages of clopidogrel with adequate delays before PCI. This will solve only part of problem, however, because around 40% of the patients will still undergo PCI without optimal platelet inhibition. More recently, the use of loading dosages of clopidogrel even higher than 600 mg has been tested but showed no significant improvement (28- 29). The newer P2Y₁₂ receptor antagonists appear promising, but the ongoing phase 3 trials have to be awaited before they can enter clinical practice (30). While clopidogrel is still in use, our study suggests that, at least in high-risk patients, screening and correction of inadequate platelet inhibition can improve outcome.

We thank Katharina Andris, Devine Frundi, and Philipp Blanke for their committed and skillful help in the laboratory tests and in the inclusion of study patients, Kristhild Peitz and the catheterization laboratory staff for their enthusiasm in supporting the study, and Petra Enderle for excellent technical support.