Author + information
- Received February 19, 1998
- Revision received April 7, 1998
- Accepted April 23, 1998
- Published online August 1, 1998.
- Scott D. Berkowitz, MD, FACPa,* (, )
- David C. Sane, MD∗,
- Kristina N. Sigmon, MAa,
- Jane H. Shavender, BAa,
- Robert A. Harrington, MD, FACCa,
- James E. Tcheng, MD, FACCa,
- Eric J. Topol, MD, FACC†,
- Robert M. Califf, MD, FACCa,‡,
- for the Evaluation of c7E3 for the Prevention of Ischemic Complications (EPIC) Study Group
- ↵*Address for correspondence: Dr. Scott D. Berkowitz, Divisions of Hematology/Coagulation and Cardiology, Department of Medicine, Box 3471, Duke University Medical Center, Durham, North Carolina 27710
Objectives. This study sought to determine the frequency of thrombocytopenia and its relation with clinical outcomes in high risk patients undergoing percutaneous coronary revascularization who received either the platelet glycoprotein (GP) IIb/IIIa receptor antagonist abciximab (ReoPro, c7E3 Fab) or conventional therapy.
Background. The development of thrombocytopenia on exposure to GPIIb/IIIa antagonists threatens the utility and economic viability of this drug class for patients with vascular disease.
Methods. We analyzed data from the Evaluation of c7E3 for the Prevention of Ischemic Complications trial (EPIC), a 2,099-patient, randomized trial of placebo, abciximab bolus or abciximab bolus plus a 12-h infusion during high-risk coronary revascularization.
Results. Thrombocytopenia (nadir platelet count <100 × 109/liter) developed in 81 patients (3.9%) during their hospital stay, with 19 (0.9%) developing severe (<50 × 109/liter) thrombocytopenia. Both thrombocytopenia and severe thrombocytopenia were more frequent in the bolus-plus-infusion arm (5.2% and 1.6%, respectively) than in the bolus-only and placebo arms combined (p = 0.020 and p = 0.025, respectively). Acute profound thrombocytopenia developed in two patients in the bolus-plus-infusion arm. Patients with thrombocytopenia experienced more unfavorable clinical outcomes than those who did not develop thrombocytopenia, regardless of treatment assignment, but those with thrombocytopenia who received abciximab had fewer worse outcomes at 30 days. Multivariable logistic modeling revealed a lower baseline platelet count, older age and lighter weight to be important predictors of thrombocytopenia. In a logistic regression model, bolus-plus-infusion treatment was a significant predictor of thrombocytopenia (p = 0.016) and remained so after adjustment for procedures and baseline risk factors (p = 0.0077).
Conclusions. Thrombocytopenia was associated with adverse clinical outcomes and excessive bleeding, but patients receiving abciximab fared better than those receiving placebo.
Platelet- and thrombin-mediated thrombosis (1–3)plays a pivotal role in the development of abrupt closure and acute ischemic complications after coronary intervention. Pharmacologic interruption of this process reduces ischemic events after coronary intervention, but also leads to a higher risk of hemorrhagic complications (4–6). Much of the hemorrhagic risk is attributed to the effect of both thrombin inactivation and inhibition of platelet function, whereas the contribution of thrombocytopenia is less well understood. In the setting of the acute coronary syndromes, trials of thrombolytic therapy provide the only current source of data on thrombocytopenia.
In the Evaluation of c7E3 for Prevention of Ischemic Complications (EPIC) trial, the use of a monoclonal chimeric human-murine antibody Fab fragment to the platelet glycoprotein (GP) IIb/IIIa receptor (c7E3 Fab, abciximab, ReoPro), in combination with aspirin and intravenous heparin, produced a 35% reduction in the 30-day composite-event rate (death, myocardial infarction or urgent coronary revascularization) in high risk patients undergoing coronary interventions, as compared with aspirin and heparin therapy alone (5,7). Our purposes were to define in this study group the frequency of thrombocytopenia and to analyze its relation to the 30-day clinical outcomes and hemorrhagic complications.
The inclusion and exclusion criteria, therapies and primary analysis of the EPIC trial have been previously published (5). Briefly, 2,099 patients <80 years of age undergoing a percutaneous transluminal coronary intervention (balloon coronary angioplasty or directional coronary atherectomy), with clinical or angiographic attributes associated with a high risk of abrupt vessel closure, were enrolled.
Patients at risk for bleeding complications (history of hemorrhagic diathesis, major operation within 6 weeks of enrollment, gastrointestinal or genitourinary bleeding of clinical significance within the previous 6 weeks, stroke within 2 years of enrollment or any stroke with significant residual neurologic deficit or presumed or documented history of vasculitis) were excluded from participation in the study.
All patients received aspirin and intravenous heparin and were randomly assigned to one of three treatment groups: placebo bolus plus placebo infusion; abciximab bolus (0.25 mg/kg body weight) with placebo infusion; or abciximab bolus (0.25 mg/kg) plus abciximab infusion (10 μg/min). The bolus was administered within 10 to 60 min before the procedure. The abciximab infusions were given intravenously and continuously for 12 h unless clinically contraindicated.
Defining the group with thrombocytopenia
Platelet counts were obtained from complete blood count samples drawn into EDTA and were determined before the infusion, at 30 min and at 2, 12 and 24 h after initiation of the study agent, and then daily until discharge or day 7, whichever occurred first. The term thrombocytopeniawas assigned if there was a nadir platelet count <100 × 109/liter and an independent reviewer excluded pseudothrombocytopenia. The term severe thrombocytopeniawas assigned to those nadir platelet counts <50 × 109/liter. Acute thrombocytopeniawas defined as a platelet count <100 × 109/liter within 24 h of the initiation of therapy. Acute severe thrombocytopeniawas defined as a platelet count <50 × 109/liter within 24 h of the initiation of therapy. Acute profound thrombocytopeniawas defined as platelet count <20 × 109/liter within 24 h of the initiation therapy. Pseudothrombocytopeniawas defined as artifactual, in vitro thrombocytopenia detected by automated laboratory cell counters recognized by preservation of the 30-min platelet count in citrate anticoagulant, while a simultaneous platelet count in EDTA anticoagulant was low (<100 × 109/liter). Platelet clumps were identified on a peripheral blood smear in some cases, and this was used as an independent criterion for pseudothrombocytopenia. The protocol recommended that heparin and aspirin be discontinued if the platelet count decreased to ≤60 × 109/liter and that platelet transfusion be initiated if the platelet count decreased to <50 × 109/liter.
Bleeding events were classified as major or minor according to the Thrombolysis in Myocardial Infarction trial criteria (8), and net blood loss was computed using the criteria of Landefeld et al. (9). Details have been described previously (10).
Continuous measures are expressed as the median values with interquartile ranges (25th and 75th percentiles), whereas categoric factors are expressed as percentages. Multivariable logistic modeling was applied to examine the relation between baseline patient characteristics and thrombocytopenia. The effect of abciximab bolus-plus-infusion treatment on the development of thrombocytopenia was explored using multiple logistic regression analysis to account for procedures (coronary artery bypass graft surgery, repeat catheterizations, repeat percutaneous transluminal coronary intervention, balloon pump use, stent implantation) occurring before the development of thrombocytopenia. A second logistic regression model was developed to examine the effect of thrombocytopenia on a “major bleeding” outcome after adjustment for previously identified predictors of bleeding (10). Multiple linear regression techniques were used in parallel to explore the relation between thrombocytopenia and the bleeding index in the presence of known predictors. All clinical outcomes reported are at 30 days.
True thrombocytopenia developed in 81 (3.9%) of the 2,099 patients undergoing high-risk percutaneous coronary revascularization, with 19 (0.9%) developing severe thrombocytopenia at some point during the index hospital period (Table 1). Pseudothrombocytopenia was assigned by a masked reviewer (D.C.S.) to 8 patients (1.1%) in the bolus-plus-infusion arm, 18 patients (2.6%) in the bolus-only arm and 9 patients (1.3%) in the placebo arm. Both thrombocytopenia and severe thrombocytopenia were more frequent among patients treated with bolus plus infusion (p = 0.020 and p = 0.025, respectively) than among patients receiving bolus only or placebo. The absolute platelet decline (baseline platelet count minus nadir platelet count) was greater for all patients in the bolus-plus-infusion group than for all patients in the placebo group (Wilcoxon rank-sum test, p = 0.019). The median time to the development of thrombocytopenia tended to be shorter in patients receiving abciximab (0.50 vs. 1.68 days). There was no difference observed in the duration of thrombocytopenia between treatments.
Acute thrombocytopenia, acute severe thrombocytopenia and acute profound thrombocytopenia occurred more frequently among patients treated with bolus plus infusion than among patients receiving placebo (p = 0.002 for acute thrombocytopenia) (Table 1). Two cases of acute profound thrombocytopenia occurred in this trial. Both were in the bolus-plus-infusion treatment group. Patient 1 had a baseline platelet count of 205 × 109/liter. Two hours after the abciximab bolus a nadir count of 2 × 109/liter was discovered. Twelve units of platelets raised the count to 13 × 10/liters and another 12 U raised it to 77 × 109/liter. No bleeding occurred. Two weeks later the count was 409 × 109/liter. Patient 2 had a baseline platelet count of 147 × 109/liter. Two hours after the bolus the count was 19 × 109/liter (performed in citrate). The patient twice received 10 U of platelets and was begun on intravenous immunoglobulin and steroids. The patient’s count then rose to 103 × 109/liter 3.5 days after receiving abciximab. No bleeding occurred.
Baseline characteristics associated with thrombocytopenia
The distributions of the key baseline characteristics for patients who developed thrombocytopenia are displayed in Table 2. Univariably, thrombocytopenia occurred more often in patients who were older, of lighter weight, of African ancestry and had peripheral vascular disease or three-vessel coronary artery disease.
Unfavorable outcomes associated with thrombocytopenia
Patients with thrombocytopenia more commonly experienced unfavorable clinical outcomes (death, myocardial infarction, bypass surgery, additional percutaneous revascularization and balloon pump insertion) and longer median lengths of stay in the coronary care unit and hospital than those who did not develop thrombocytopenia, regardless of treatment assignment (Table 3). Patients who developed thrombocytopenia had significantly fewer undesirable outcomes (excluding death and additional percutaneous transluminal coronary intervention) if they were treated with abciximab (bolus only or bolus + infusion) versus placebo. A definitive causal relation between thrombocytopenia and worse clinical outcomes and could not be determined. The outcome events of myocardial infarction, percutaneous coronary revascularization and bypass grafting occurred before the development of thrombocytopenia in 75% to 80% of the patients with thrombocytopenia. The possibility that early discontinuation of the study medication for acute thrombocytopenia led to the development of an unfavorable clinical outcome did not appear to be the case: only five patients developed both thrombocytopenia and an undesirable clinical event before completion of the study medication infusion.
Hemorrhage associated with thrombocytopenia
The hematologic variables associated with the development of thrombocytopenia are shown in Table 4. Less hemorrhage, as judged by the bleeding index, occurred in patients with thrombocytopenia who were treated with abciximab as compared with placebo (p = 0.046), and they required fewer packed red cell transfusion (p = 0.018).
Bleeding events associated with thrombocytopenia are presented in Table 5. Among patients who received placebo, major bleeding was more common (69.6% vs. 4.5%) in those with thrombocytopenia than in those with normal platelet counts. Among patients who received abciximab, major bleeding was again more common in those with thrombocytopenia than in those without it (43.1% vs. 11.2%). Among patients with thrombocytopenia, major bleeding was more common if they received placebo instead of abciximab (69.6% vs. 43.1%, p = 0.030). Intracranial hemorrhage was not clinically apparent in patients with thrombocytopenia who received abciximab.
Predictors of developing thrombocytopenia
Multivariable logistic modeling, applied to examine the relation between selected baseline patient characteristics and thrombocytopenia, revealed three important predictors of developing thrombocytopenia: lower baseline platelet count, older age and lighter weight (Table 6A). The probability of developing thrombocytopenia over the observed range of baseline platelet count, age and weight is shown in Figure 1. A risk score developed to show the overall incidence of thrombocytopenia according to the number of baseline risk factors present highlights the importance of these predictors; if a patient >65 years old, weighing <80 kg, presented with a baseline platelet count <200 × 109/liter, the patient’s risk of developing thrombocytopenia in the hospital was nearly 10% (Table 6B). Although patients with advancing age and lower body weight were previously shown in the EPIC trial to be at higher risk of postprocedural ischemic events, no correlation was found between the number of risk factors present for developing thrombocytopenia and the rate of adverse events.
Effect of abciximab on development of thrombocytopenia
The effect of abciximab bolus-plus-infusion treatment on the development of thrombocytopenia was explored using multiple logistic regression analysis to account for procedures performed before the occurrence of thrombocytopenia. These procedures included cardiac catheterization, percutaneous transluminal coronary revascularization, stent or balloon pump placement and bypass surgery. Without adjustment for procedures and baseline risk factors (initial platelet count, age and weight), abciximab bolus plus infusion proved to be an important predictor of thrombocytopenia (chi-square statistic 5.8, p = 0.016). After adjusting for the predictive effect of the procedures and the presence of baseline risk factors (chi-square 172.5, p < 0.0001), the bolus-plus-infusion treatment remained a significant predictor of thrombocytopenia (chi-square 7.1, p = 0.0077).
Role of thrombocytopenia in major bleeding
Predictors of bleeding in the EPIC population were previously identified (10). In the current analysis, a logistic regression model was developed to examine the effect of thrombocytopenia on the “major bleeding” outcome after adjustment for these predictors of bleeding (chi-square 300.51, p < 0.0001), and it validated thrombocytopenia as a significant additional predictor of major bleeding (chi-square 13.04, p = 0.0003). Multiple linear regression techniques used in parallel to explore the relation between thrombocytopenia and the bleeding index in the presence of known predictors proved thrombocytopenia to be a very potent predictor of an increased bleeding index (chi-square 114.98, p < 0.0001).
Thrombocytopenia is an important and not infrequent occurrence in patients with medically intractable unstable or postinfarction angina undergoing revascularization without thrombolytic therapy. In the EPIC trial, thrombocytopenia was associated with undesirable ischemia-related outcomes, including death, and excessive hemorrhage. Important predictors were a lower baseline platelet count, older age and lighter weight. We noted an increased likelihood of thrombocytopenia in patients receiving abciximab, particularly as a bolus plus infusion. However, patients who developed thrombocytopenia but who also had received abciximab had fewer worse outcomes (re-infarction, need for intraaortic balloon pump or bypass surgery).
Thrombocytopenia is associated with excessive bleeding
Thrombocytopenia in the EPIC trial, regardless of whether the patient was treated with abciximab or not, was also a significant additional predictor of major bleeding and a very potent predictor of an increased bleeding index. Among the 2,018 patients who did not develop thrombocytopenia, morebleeding events were recorded in those who received abciximab than in those who received placebo, as would be expected with antagonism of the platelet GPIIb/IIIa receptor. Among the 81 patients who developed thrombocytopenia, treatment with abciximab was associated with lessbleeding as compared with thrombocytopenia without abciximab; these patients endured fewer major bleeding events and fewer transfusions of packed red blood cells than those who received placebo. This appears counter-intuitive in light of the published report of the bleeding complications in the entire EPIC population (10). However, that study did not separately analyze patients to evaluate the effect of thrombocytopenia and focused on major bleeding unrelated to bypass surgery. Owing to the small number of patients with thrombocytopenia (81 of 2,099), the effect of thrombocytopenia on bleeding within each treatment group may have been diluted in that analysis.
Thrombocytopenia associated with abciximab infusion is distinct
A distinct subgroup of patients with thrombocytopenia who had received abciximab was observed. Patients with thrombocytopenia who received placebo were a sicker group: they had worse clinical outcomes, had a higher bleeding index (p = 0.0462) and were more frequently transfused with packed red blood cells (p = 0.018). They developed thrombocytopenia later; 24 h after initiation of treatment the frequency of thrombocytopenia among patients in the placebo group was half that of those who received abciximab (Fig. 2). In the abciximab-treated group the observed frequency of thrombocytopenia was higher, but it was quantitatively less severe.
Acute thrombocytopenia in the EPIC trial
Acute thrombocytopenia was observed almost exclusively in abciximab-treated patients. The two patients who developed acute profound thrombocytopenia had received abciximab. Although rare, the entity of acute profound thrombocytopenia associated with abciximab has been described, and the evaluation and management has been reviewed (11,12). These two patients had platelet counts detected below 20 × 109/liter at the 2-h platelet count evaluation. This was found in previously reported cases and supports the package insert recommendation to obtain a platelet count at 2 to 4 h after abciximab initiation. Analysis of the time to thrombocytopenia revealed that one-third of patients had thrombocytopenia 2 h after the abciximab bolus and one-half of them had it 12 h after. Although it cannot be determined at that stage which cases of thrombocytopenia are related to abciximab therapy, evaluation of the platelet count soon after initiation of abciximab identifies the patients at risk, potentially permitting sufficient time to discontinue the infusion with effect. In the initial reports of profound thrombocytopenia (11–13), platelet transfusion was efficacious and was not associated with worsening thrombocytopenia or thrombosis. Transfusion is recommended if the platelet count is below 20 × 109/liter.
Potential mechanisms for abciximab-associated thrombocytopenia
The mechanism of the thrombocytopenia observed with abciximab infusion is not understood, although postulates exist. Antibodies to the GPIIb/IIIa receptor can be encountered in immune and drug-induced thrombocytopenias (14,15), but by design abciximab has no Fc portion, making untenable the hypothesis that platelets with abciximab on their GPIIb/IIIa receptors are removed from the circulation by the Fc receptors of the macrophages of the reticuloendothelial system. This does not, however, rule out the development of aggregates of platelets and c7E3 Fab and subsequent removal from the circulation. Another consideration is the potential presence of residual uncleaved c7E3 Fab2in the infused preparation, but this appears unlikely according to data from the manufacturer (personal communication of Robert Jordan, Centocor). Because the construction of the abciximab fragment requires cleavage of the chimeric antibody across the hinge region of the molecule, a human anti-chimeric antibody (HACA) reaction to this damaged human antibody structure might occur after abciximab infusion. The development of a HACA response suggests that the potential for immune-mediated reactions exists. In the EPIC trial, the development of HACA within the first 12 weeks of exposure to abciximab was 5.8% overall, with 6.5% developing in the bolus-plus-infusion group, 5.2% in the bolus-only group and none in the placebo group (16). Also detected was a low affinity pre-immune (pre-dose) HACA reactivity in 8% of the patients in EPIC, which appeared to be specific for the Fab cleavage site of human immunoglobulins. However, no correlation was observed between pre-dose or induced HACA and thrombocytopenia (personal communication of Carrie L. Wagner, Centocor).
Binding of abciximab to the GPIIb/IIIa receptor could conceivably produce a conformational change, resulting in the expression of new epitopes, termed ligand-induced binding sites (LIBS) (17). If patients have preformed antibodies that cross react with LIBS, the platelets may become coated with immunoglobulin and be removed from the circulation. Bednar et al. (18)reported the presence in human plasma samples of preexisting immunoglobulin G–type antibodies that selectively bound to platelets in the presence of their GPIIb/IIIa antagonist and led to thrombocytopenia.
Platelet modifications might also be created either by the presence of abciximab on the platelet surface in relation to other platelet antigens or by the ability of abciximab to bind to the vitronectin (αvβ3) receptor (where it can inhibit platelet-mediated thrombin generation) or the Mac-1 (CD11b/CD18) integrin receptors of activated monocytes (19). The unusual pharmacokinetic properties of the abciximab bolus plus infusion (13,20,21)might heighten the opportunity for the evolution of these scenarios; there is an extended platelet-bound half-life, with small amounts of the drug remaining on circulating platelets for 15 days or more (16). Any of the these aforementioned developments could culminate in platelet activation and platelet–platelet interactions with consumption or removal of the platelets from the circulation by macrophages of the reticuloendothelial system.
This analysis has certain limitations. A definitive delineation of the cause(s) for each of the 81 cases of thrombocytopenia was not possible. The thrombocytopenia protocol was primarily designed to detect the presence of pseudothrombocytopenia and HACA-induced thrombocytopenia; poor compliance with other aspects of the workup of thrombocytopenia led to incomplete investigation of other potential causes. Clean data are not available to evaluate the presence of heparin-induced thrombocytopenia. Data on heparin use were only obtained in the catheterization laboratory; once the patient left the laboratory, only the highest and lowest rates of heparin infused during the first 3 days were collected. Plasma samples to detect antibodies to heparin were collected in the extreme minority of cases. Also, the relation between thrombocytopenia and procedure could not be defined because only the day, and not the time, of the procedures was collected. Still, our analysis represents an important description of thrombocytopenia and augments the limited knowledge developed from previous studies of thrombocytopenia in the setting of acute coronary syndromes, most of which were derived from patients treated with thrombolytic therapy (8,22,23)
Thrombocytopenia was a frequent covariate among patients who developed ischemia-associated morbidity and mortality; however, among patients with thrombocytopenia, treatment with abciximab was associated with fewer ischemic events and less bleeding. Independent predictors of thrombocytopenia were lower baseline platelet count, increased age, lighter weight and treatment with bolus plus infusion of abciximab. The subgroup with thrombocytopenia with abciximab treatment appeared to be distinct from other types of thrombocytopenia and could be detected by a declining platelet count in the first 24 h after therapy initiation. A platelet count performed at 2 to 4 h after the abciximab infusion is begun will identify those patients at risk of developing thrombocytopenia, especially those who will develop acute severe and acute profound thrombocytopenia.
The authors express their appreciation for the critical review and suggestions made by Barry S. Coller, MD, and Robert E. Jordan, PhD; the evaluation of the human anti-chimeric data performed by Carrie L Wagner, PhD; and the expert editorial support provided by Penny K. Hodgson.
↵‡ A list of the EPIC investigators can be found in the N Engl J Med 1994;330:956–61.
☆ This study was supported by a grant from Centocor, Inc., Malvern, Pennsylvania.
This study was also presented in part at the 45th Annual Scientific Session of the American College of Cardiology, Orlando, Florida, March 24–27, 1996.
- Evaluation of c7E3 for the Prevention of Ischemic Complications trial
- human anti-chimeric antibody
- ligand-induced binding site
- Received February 19, 1998.
- Revision received April 7, 1998.
- Accepted April 23, 1998.
- American College of Cardiology
- Scharf R.E.,
- Tomer A.,
- Marzec U.M.,
- Teirstein P.S.,
- Ruggeri Z.M.,
- Harker L.A.
- Tschoepe D.,
- Schultheib H.P.,
- Kolarov P.,
- et al.
- Schwartz L.,
- Bourassa M.G.,
- Lesperance J.,
- et al.
- ↵Bovill EG, Terrin ML, Stump DC, et al., for the TIMI investigators. Hemorrhagic events during therapy with recombinant tissue-type plasminogen activator, heparin, and aspirin for acute myocardial infarction. Ann Intern Med 1991;115:256–65.
- ↵Aguirre FV, Topol EJ, Ferguson JJ, et al., for the EPIC Investigators. Bleeding complications with the chimeric antibody to platelet glycoprotein IIb/IIIa integrin in patients undergoing percutaneous coronary intervention. Circulation 1995;91:2882–90.
- Berkowitz S.D.,
- Harrington R.A.,
- Rund M.M.,
- Tcheng J.E.
- Tcheng J.E.,
- Ellis S.G.,
- George B.S.,
- et al.
- Curtis B.R.,
- McFarland J.G.,
- Wu G.G.,
- Visentin G.P.,
- Aster R.H.
- Jordan R.E.,
- Wagner C.L.,
- Mascelli M.A.,
- et al.
- Frelinger A.L. 3d.,
- Cohen I.,
- Plow E.F.,
- et al.
- Bednar B.,
- Bednar R.A.,
- Cook J.J.,
- et al.
- The European Cooperative Study Group,
- Simoons M.L.,
- de Boer M.J.,
- van den Brand M.J.,
- et al.
- Harrington R.A.,
- Sane D.C.,
- Califf R.M.,
- et al.
- Rao AK, Pratt C, Berke A, et al., for the TIMI Investigators. Thrombolysis in Myocardial Infarction (TIMI) trial phase I: hemorrhagic manifestations and changes in plasma fibrinogen and the fibrinolytic system in patients treated with recombinant tissue plasminogen activator and streptokinase. J Am Coll Cardiol 1988;11:1–11.