Author + information
- Received January 28, 2016
- Revision received May 9, 2016
- Accepted May 18, 2016
- Published online August 22, 2016.
- Marianna Adamo, MDa,b,
- Sara Ariotti, MDa,c,
- Francesco Costa, MDa,d,
- Salvatore Curello, MDb,
- Aris Moschovitis, MDc,
- Ton de Vries, MAe,
- Harvey D. White, DScf,
- Stephan Windecker, MD, PhDc and
- Marco Valgimigli, MD, PhDa,c,∗ ()
- aThoraxcenter, Erasmus Medical Center, Rotterdam, the Netherlands
- bCatheterization Laboratory, Spedali Civili, Brescia, Italy
- cDepartment of Cardiology, Bern University Hospital, Bern, Switzerland
- dDepartment of Clinical and Experimental Medicine, Policlinico “G. Martino,” Messina, Italy
- eCardialysis BV, Rotterdam, the Netherlands
- fGreen Lane Cardiovascular Service, Auckland City Hospital, Auckland, New Zealand
- ↵∗Reprint requests and correspondence:
Dr. Marco Valgimigli, Bern University Hospital, Department of Cardiology, CH-3010 Bern, Switzerland.
Objectives The aim of this study was to investigate whether the 2 tirofiban formulations tested in the early and late phases of the PRISM (Platelet Receptor Inhibitor in Ischemic Syndrome Management) trial might differ with respect to risk for thrombocytopenia and clinical outcomes compared with unfractionated heparin (UFH).
Background Citrate-buffered tirofiban is currently marketed as brand-name drug. However, tirofiban has recently been promoted in some countries as a generic drug with different formulations, such as phosphate-buffered product.
Methods In the PRISM trial 3,232 patients were randomly assigned to receive tirofiban or UFH. In the early phase, 879 patients were allocated to phosphate-buffered tirofiban and 874 patients to UFH group. After a protocol amendment due to a study drug instability report, citrate-buffered tirofiban replaced the phosphate-buffered formulation. Therefore, in the late phase, 737 and 742 patients were treated with citrate-buffered tirofiban and UFH, respectively.
Results The relative risk for thrombocytopenia (nadir <90,000/mm3 or <100,000/mm3) was increased in patients treated with phosphate-buffered tirofiban in the early phase (odds ratio [OR]: 3.51; 95% confidence interval [CI]: 1.15 to 10.73; p = 0.027; and OR: 2.83; 95% CI: 1.11 to 7.22; p = 0.029, respectively) but not in patients treated with citrate-buffered tirofiban in the late phase (OR: 1.01; 95% CI: 0.20 to 5.05; p = 0.987; and OR: 0.99; 95% CI: 0.26 to 3.45; p = 0.991, respectively). Using a combined definition of thrombocytopenia (nadir <150,000/mm3 or a decrease ≥50%), the randomization period significantly modified the effect of the treatment (tirofiban vs. UFH) on platelet decrease (p for interaction = 0.024). Thrombocytopenia was associated with a 5- to 10-fold increased risk for TIMI (Thrombolysis In Myocardial Infarction) bleeding and a 2-fold increased risk for net adverse cardiovascular events.
Conclusions Phosphate-buffered tirofiban, currently marketed as a generic drug, is associated with a higher rate of thrombocytopenia with a potentially increased risk for adverse clinical outcomes compared with citrate-buffered tirofiban.
Patients with acute coronary syndromes (ACS) are frequently treated with intravenous glycoprotein IIb/IIIa inhibitors (GPI) (1,2). Among these agents, tirofiban was first approved by the U.S. Food and Drug Administration in 1998. Since initial approval, the dose has been revised, and tirofiban given as a high-dose bolus is currently the most frequently used GPI (3,4).
PRISM (Platelet Receptor Inhibitor in Ischemic Syndrome Management) was the first randomized clinical study investigating the safety and efficacy of tirofiban, and it demonstrated a clinical benefit of this GPI compared with unfractionated heparin (UFH) with respect to acute ischemic events and 30-day mortality in the absence of an increased risk for bleeding. At variance with all other placebo-controlled studies, PRISM reported a significant increase in the rate of thrombocytopenia in the tirofiban compared with UFH group (5).
Two different formulations of tirofiban were used in the PRISM trial in a sequential manner. After a protocol amendment due to a drug instability report, the phosphate-buffered product, which was used as the study drug during the early phase of the study, was replaced by the citrate-buffered formulation, which is currently marketed as a brand-name drug (Aggrastat; Correvio Ltd. in the United Kingdom and Medicure Pharma in the United States) (6). However, tirofiban has been recently promoted as a generic drug in several European countries with different formulations, such as phosphate-buffered tirofiban.
In this post hoc analysis of the PRISM trial, we sought to investigate whether the 2 tirofiban formulations used during the early and late phases of the study and currently marketed as generic and brand-name drugs might differ with respect to rates of thrombocytopenia and clinical outcomes compared with UFH.
The design and the main findings of the PRISM trial were previously reported (5).
Briefly, PRISM was a randomized, controlled, multicenter, double-blind trial including patients with non–ST-segment elevation ACS. Patients were randomly assigned to receive tirofiban (bolus of 0.6 μg/kg/min over 30 min followed by 0.15 μg/kg/min infusion for 48 h) or UFH (bolus of 5,000 IU followed by infusion of 1,000 IU/h for 48 h, adjusted for activated partial thromboplastin time at 6 and 24 h).
During the early recruitment phase of the trial, tirofiban was administered as a phosphate-buffered product that ranged in concentration from 0.17 to 0.5 mg/ml; sodium chloride was used to render the product iso-osmotic. During the late recruitment phase, this composition was abandoned and substituted by a citrate-buffered product (10 mmol/l) containing sodium chloride. The change in composition was deemed necessary because of instability report of the phosphate-buffered composition and the finding of precipitates in vials stored for 24 months or more (6). Sodium porcine heparin was provided as 1,000 U/ml (10-ml fill) or as 10,000 U/ml (5-ml fill) without differences through the early and late recruitment phases.
To maintain the randomization scheme, we primarily aimed to compare outcomes in patients treated with phosphate-buffered tirofiban versus UFH during the early phase and those treated with citrate-buffered tirofiban versus UFH during the late phase. As sensitivity analyses, we also compared patients receiving the 2 tirofiban formulations throughout the 2 different time periods.
Thrombocytopenia was defined as platelet nadir <90,000/mm3 (used in the PRISM trial ), as platelet count <100,000/mm3 (the most frequent cutoff used in previous studies [7–9]), and as a combination of nadir value <150,000/mm3 and decrease of platelet count ≥50% (used in a previous large registry ). Severe thrombocytopenia was defined as platelet count <50,000/mm3.
We also investigated the 30-day ischemic endpoints reported in the PRISM trial (2): death, myocardial infarction (MI), refractory ischemia; readmission for unstable angina, a composite of major adverse cardiovascular events including all single endpoints previously mentioned and a composite of death and MI.
Bleeding events were defined according to the TIMI (Thrombolysis In Myocardial Infarction) classification (11).
Finally, a composite endpoint of net adverse cardiovascular events (NACEs) including major adverse cardiovascular events and major or minor TIMI bleeding was assessed.
Continuous variables were expressed as mean ± SD and were compared using the Student t test. Categorical variables were expressed as counts and percentages and were compared using the chi-square or Fisher exact test, as appropriate.
The proportionality assumptions were checked by visual estimation after plotting the log cumulative hazard versus (log) time at follow-up after the index procedure and by applying a test for nonproportional hazards using Schoenfeld residuals, which failed to reject the null hypothesis that bleeding and ischemic event rate was affected by time. Therefore, a multivariate Cox regression analysis stratified by center, including baseline variables differently distributed at an alpha level of 0.10, was performed to calculate the relative risk of these endpoints and to evaluate whether thrombocytopenia was an independent predictor of outcomes. Each result was expressed as hazard ratio (HR) and corresponding 95% confidence interval (CI).
A stepwise logistic regression model was used to calculate the relative risk for thrombocytopenia according to the 3 different definitions and adjusted for baseline imbalances. Each result was reported as odds ratio (OR) and corresponding 95% CI.
Interaction tests between randomization period (early vs. late) and treatment (tirofiban vs. UFH) were done with likelihood ratio tests of the null hypothesis that the interaction coefficient was zero.
The Kaplan-Meier method was used to plot the cumulative incidence of bleeding events according to the presence of thrombocytopenia.
For all analyses, a 2-sided alpha value <0.05 was considered to indicate statistical significance. All statistical analyses were performed using SPSS version 21 (SPSS, Inc., Chicago, Illinois).
Baseline characteristics and duration of study drug infusion
In the PRISM trial, 3,232 patients were randomly assigned to receive tirofiban or UFH treatment. Before a formal study protocol amendment, during the early recruitment phase, 1,753 patients were enrolled, of whom 879 (50.1%) were allocated to the phosphate-buffered tirofiban group and 874 (49.9%) to the UFH arm. During the late recruitment phase, 737 of 1,479 patients (49.8%) were treated with citrate-buffered tirofiban and 742 (50.2%) with UFH.
The baseline features within the randomized arms during both the early and late recruitment periods are shown in Table 1. During the early phase, patients treated with UFH more frequently had prior coronary artery bypass graft surgery compared with those receiving phosphate-buffered tirofiban (20.4% vs. 16.5%, p = 0.037), whereas during the late phase, patients treated with UFH more commonly had non–ST-segment elevation MI compared with those receiving citrate-buffered tirofiban (27.4% vs. 21.4%, p = 0.008) (Table 1).
Among tirofiban-treated patients, those receiving citrate-buffered tirofiban more commonly had hypercholesterolemia (52% vs. 43.7%, p < 0.001) or previous heart failure (14.8 vs. 10.4%, p = 0.007), whereas patients treated with phosphate-buffered tirofiban were more frequently admitted for non–ST-segment elevation MI (25.4% vs. 21.4%, p = 0.018) (Table 1).
Approximately 60% of patients underwent medical management, and 70% of patients undergoing coronary angiography had multivessel coronary artery disease, which did not differ across groups (Table 1).
The duration of study drug infusion was well matched across groups. Phosphate-buffered (early phase) and citrate-buffered (late phase) tirofiban formulations were infused for a mean of 45.1 ± 10.6 h and 44.8 ± 10.5 h, respectively, whereas UFH was administered for a mean of 45.5 ± 9.5 h and 45.1 ± 10.1 h during the early and late phases, respectively.
Baseline platelet counts did not differ across groups. The rate of thrombocytopenia was significantly higher during the early recruitment phase, when phosphate-buffered tirofiban was compared with UFH, using a cutoff of 90,000/mm3 (1.7% vs. 0.5%, respectively, p = 0.030) or 100,000/mm3 (2.0% vs. 0.7%, respectively, p = 0.034) but not during the late recruitment phase, when citrate-buffered tirofiban was compared with UFH (0.3% vs. 0.1% [p = 0.621] and 0.7% vs. 0.7% [p = 0.971]) (Figure 1A). The rate of thrombocytopenia did not differ across groups on the basis of platelet nadir value <150,000/mm3 or decrease in platelet count ≥50% (Figure 1A).
Severe thrombocytopenia was observed in 4 patients treated with phosphate-buffered tirofiban versus none treated with UFH during the early recruitment phase (0.5% vs. 0%, p = 0.046) and in 1 patient treated with citrate-buffered tirofiban versus no patients allocated to the UFH group during the late recruitment phase (0.1% vs. 0%, p = 0.49).
Among tirofiban-treated patients, the rate of thrombocytopenia (nadir <90,000/mm3 or <100,000/mm3) was significantly higher in those treated with phosphate-buffered compared with those treated with citrate-buffered tirofiban, and a trend toward a more frequent platelet nadir <150,000/mm3 or a decrease in platelet count ≥50% in patients receiving phosphate-buffer tirofiban was observed (Figure 1A).
On multivariate-adjusted analysis, the relative risk for thrombocytopenia (nadir <90,000/mm3 or <100,000/mm3) was approximately 3-fold higher in patients treated with tirofiban during the early (OR: 3.51; 95% CI: 1.15 to 10.73; p = 0.027; and OR: 2.83; 95% CI: 1.11 to 7.22; p = 0.029, respectively) but not during the late recruitment phase, although formal statistical testing for interaction did not reach conventional thresholds of significance (Figure 1B).
The randomization period, however, emerged as a possible treatment modifier with respect to the risk for thrombocytopenia under the combined definition of platelet nadir <150,000/mm3 or decrease in platelet count ≥50% (p for interaction = 0.024) (Figure 1B).
Among patients receiving tirofiban, the phosphate-buffered formulation was associated with an increased risk for a platelet decrease <90,000/mm3 (OR: 3.99; 95% CI: 1.13 to 14.05; p = 0.031) or 100,000/mm3 (OR: 2.78; 95% CI: 1.02 to 7.63; p = 0.047) compared with citrate-buffered tirofiban.
Clinical endpoints at 30 days
After adjustment for baseline differences, citrate-buffered tirofiban was associated with a reduced risk for 30-day mortality compared with UFH (HR: 0.49; 95% CI: 0.27 to 0.89; p = 0.019), whereas no difference was observed between phosphate-buffered tirofiban and UFH (HR: 0.87; 95% CI: 0.49 to 1.54; p = 0.629). However, no significant interaction was observed between the randomization period and treatment with respect to the mortality endpoint (p for interaction = 0.151) (Figure 2). There were no differences between tirofiban and UFH with regard to other explored ischemic or bleeding endpoints (Figure 2).
Among tirofiban treated patients, no significant differences were noted between the 2 tirofiban formulations with respect to 30-day mortality and other clinical endpoints. A trend toward a higher rate of refractory ischemia in the citrate-buffered group was observed (Online Table 1).
Thrombocytopenia effect on ischemic and bleeding outcomes
Premature study drug discontinuation was deemed necessary in approximately 50% of patients among those who developed thrombocytopenia (platelet nadir <90,000/mm3 or 100,000/mm3), whereas this was infrequent in patients without thrombocytopenia (45.8% vs. 1.7% and 42.2% vs. 1.6%, respectively, p < 0.001 for all).
After adjustment for differences in baseline characteristics, thrombocytopenia was associated with a 5- to 10-fold increased risk for TIMI minor or major bleeding (Figure 3) and major bleeding (Table 3, Online Table 3). Patients who experienced a platelet nadir <100,000/mm3 also had a 2-fold increased risk for NACEs (HR: 2.36; 95% CI: 1.31 to 4.23; p = 0.004) (Table 3). Under the combined definition, a trend toward a higher risk for refractory ischemia was observed in patients with thrombocytopenia (Table 3).
The main findings of the present study can be summarized as follows. 1) Patients treated with phosphate-buffered tirofiban during the early recruitment phase of the PRISM trial more frequently experienced thrombocytopenia, with a 3-fold increased risk for platelet decrease <90,000/mm3 or 100,000/mm3 compared with those receiving UFH during the same randomization period. No difference was observed between citrate-buffered tirofiban and UFH during the late recruitment phase. 2) The randomization period (early vs. late) significantly modified the effect of treatment (tirofiban vs. UFH) with respect to the risk for thrombocytopenia (platelet nadir <150,000/mm3 or decrease in platelet count ≥50%). 3) Outside the randomization scheme, phosphate-buffered tirofiban was associated with a significantly higher risk for thrombocytopenia (platelet nadir <90,000/mm3 or 100,000/mm3) compared with the corresponding citrate-buffered product. 4) Thrombocytopenia was confirmed to be an independent predictor of adverse outcomes, with a 5- to 10-fold increased risk for TIMI minor or major bleeding and a 2-fold increased risk for NACEs. 5) After adjustment for baseline differences, citrate-buffered tirofiban was associated with a reduced risk for death compared with UFH, whereas no difference in mortality was observed between phosphate-buffered tirofiban and UFH.
Tirofiban use and formulations
American College of Cardiology and American Heart Association guidelines suggest the administration of GPIs in high-risk patients with non–ST-segment elevation ACS undergoing percutaneous coronary intervention irrespective of pre-treatment status with P2Y12 inhibitors (12,13). European Society of Cardiology guidelines recommend GPIs for patients with ACS in bailout situations or with thrombotic complications (14,15). As a result, these drugs are frequently used in clinical practice in patients with ACS undergoing invasive management, and tirofiban is currently the most frequently used GPI worldwide.
Citrate-buffered tirofiban is marketed as brand-name product (Aggrastat, Correvio [UK] Ltd. and Medicure Pharma [US]). In several European countries, tirofiban is marketed as a generic drug with formulations different from the citrate-buffered, including phosphate-buffered tirofiban (Hexal, Hikma, and MEDAC). No generic tirofiban products are currently marketed in the United States. Seven generic versions are available in India; at least 1 of these is a citrate-buffered product (Gland Pharma). No information regarding the excipients is in the public domain for the remaining 6 products.
Thrombocytopenia and outcomes
The association between thrombocytopenia and adverse clinical outcomes in the setting of ACS is well established, and patients receiving GPIs should be carefully scrutinized for changes in platelet count during and immediately after drug administration. Patients who experience significant platelet decreases after GPI administration are at increased risk for bleeding, recurrent MI, and death (7–10). In the Global Registry of Acute Coronary Events (GRACE) population, thrombocytopenia after GPI treatment was associated with a 4-fold increased risk for in-hospital mortality and a 2- to 3-fold increased risk for stroke and recurrent infarction (16). In the CRUSADE (Can Rapid Risk Stratification of Unstable Angina Patients Suppress Adverse Outcomes With Early Implementation of the ACC/AHA Guidelines) registry, patients who developed thrombocytopenia had a 3- and 4-fold increased risk for mortality and bleeding, respectively (10). In the PRISM population, thrombocytopenia emerged as an independent predictor of both bleeding and ischemic clinical outcomes, with a 5- to 10-fold increased risk for TIMI bleeding, a 2-fold increased risk for NACEs, and a trend toward an increased risk for refractory ischemia at 30-day follow-up. The lack of an association between thrombocytopenia and mortality in our cohort of patients may reflect a type II error.
Mechanism of thrombocytopenia
Drug-induced immune thrombocytopenia is a common hematologic problem. More than 200 drugs have been reported to cause immune thrombocytopenia (17); these include commonly used drugs such as antibiotics, anticonvulsants, and arginine-glycine-aspartic acid (RGD) mimetic agents such as tirofiban and eptifibatide. Immune thrombocytopenia can occur on first exposure to an RGD mimetic agent, and the platelet count usually drops abruptly within hours of commencement of drug administration (18), suggesting the presence of a naturally occurring antiplatelet antibody. Binding of fibrinogen, RGD peptides, or RGD mimetic drugs to the RGD recognition site of αIIbβ3 induces conformational changes and the emergence of cryptic epitopes previously “unseen” by the immune system (18). In the majority of patients with eptifibatide- and tirofiban-induced thrombocytopenia, antibody binding was found to be drug specific (19,20).
Extrapolating these findings to our analysis, it may be speculated that conformational changes in αIIbβ3 induced by phosphate-buffered tirofiban are more frequently recognized by naturally occurring antiplatelet antibodies compared with citrate-buffered tirofiban, thereby being more frequently associated with thrombocytopenia.
Our findings confirm previous observations that thrombocytopenia appears to be more a drug- than a class-specific side effect (21–23). The current observation that 2 distinct tirofiban formulations, which exert comparable antiplatelet effects, are associated with a remarkable difference in terms of drug safety, notably thrombocytopenia, reinforces the notion that the chemical structure of a given drug more than its anticipated target effect is associated with the risk for thrombocytopenia.
Generic drugs and side effects
Our current observation should also raise a word of caution with respect to current regulations for generic drug approval. The main principle underpinning the safe and effective use of generic drugs is the concept of bioequivalence (24).
The purpose of establishing bioequivalence is to demonstrate equivalence between a generic medicine and the original medicine in order to allow extrapolation of the pre-clinical and clinical testing performed with the original drug.
Although the active pharmaceutical ingredient does not differ between original and generic medicines, other (supposedly inactive) ingredients, known as excipients, may differ, and a number of pharmaceutical excipients are known to cause adverse effects or result in contraindications (25). As excipients may differ between originator medicines and generic preparations, which have been shown to be bioequivalent and therefore substitutable, there needs to be an awareness in the medical and health care community of differences in excipients and thus, the potential for generic formulations to induce safety issues. This may be particularly relevant when treating life-threatening disease such as ACS. Evidence has been published that differences in excipients between originator medications and their generic counterparts can cause problems (26,27); our present findings expand on previous observations (28).
First, the analysis was not pre-specified, and the small number of events, in particular with regard to thrombocytopenia, may have affected the results. However, sensitivity analyses performed using different definitions of thrombocytopenia and comparing patients treated with tirofiban versus UFH during the early versus late recruitment phases and patients receiving the 2 tirofiban formulations throughout the 2 different periods provided consistent results.
Second, in the PRISM trial, tirofiban was used as upstream treatment, which is discouraged in current guidelines, and was administered at a bolus dose of 0.6 μg/kg/min for 30 min followed by 0.15 μg/kg/min infusion for 48 h. This treatment modality differs from the approved high-dose bolus regimen of 25 μg/kg over 3 min, followed by 0.15 μg/kg/min infusion up to 48 h, which is currently used in clinical practice. However, the difference in the total dose of tirofiban administered according to the PRISM regimen versus the one currently in use is negligible. In a patient weighing 70 kg with normal renal function undergoing 48-h post-bolus infusion, the total drug exposure is 31.5 mg in 48.5 h for the PRISM scheme and 32 mg in 48.05 h for current use.
Third, the bleeding events were classified using the TIMI criteria because at the time of study recruitment, the Bleeding Academic Research Consortium classification was not available.
Phosphate-buffered tirofiban, currently marketed as a generic drug in several countries, is associated with a higher risk for thrombocytopenia and potentially increased risk for adverse clinical outcomes compared with a citrate-buffered tirofiban formulation. Careful post-marketing surveillance of both the brand-name and generic formulations of tirofiban is warranted.
WHAT IS KNOWN? Citrate-buffered tirofiban is currently marketed as a brand-name drug, whereas a phosphate-buffered product has recently been promoted as a generic drug in several European countries. In the PRISM trial, the 2 tirofiban formulations were used in the late and early phases of the study, respectively.
WHAT IS NEW? Phosphate-buffered tirofiban is associated with a higher rate of thrombocytopenia with potentially increased risk for adverse outcomes compared with citrate-buffered formulation.
WHAT IS NEXT? Post-marketing surveillance of the different tirofiban formulations is needed to ascertain whether brand-name tirofiban has a better safety profile compared with the generic products.
For supplemental tables, please see the online version of this article.
The authors have reported they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- acute coronary syndrome
- confidence interval
- glycoprotein IIb/IIIa inhibitor
- hazard ratio
- myocardial infarction
- net adverse cardiovascular event
- odds ratio
- arginine-glycine-aspartic acid
- unfractionated heparin
- Received January 28, 2016.
- Revision received May 9, 2016.
- Accepted May 18, 2016.
- American College of Cardiology Foundation
- Batchelor W.B.,
- Tolleson T.R.,
- Huang Y.,
- et al.
- Valgimigli M.,
- Percoco G.,
- Barbieri D.,
- et al.
- ↵U.S. Food and Drug Administration. Clinical Study Report MK-0383. A randomized, parallel, double-blind study to investigate the safety and clinical efficacy of MK-0383 vs. heparin in patients with unstable angina/non-Q-wave myocardial infarction (PRISM). Protocol No. 011–016. September 1997.
- Eikelboom J.W.,
- Anand S.S.,
- Mehta S.R.,
- Weitz J.I.,
- Yi C.,
- Yusuf S.
- McClure M.W.,
- Berkowitz S.D.,
- Sparapani R.,
- et al.
- Wang T.Y.,
- Ou F.S.,
- Roe M.T.,
- et al.
- Jneid H.,
- Anderson J.L.,
- Wright R.S.,
- et al.
- Amsterdam E.A.,
- Wenger N.K.,
- Brindis R.G.,
- et al.
- Windecker S.,
- Kolh P.,
- Alfonso F.,
- et al.
- Roffi M.,
- Patrono C.,
- Collet J.P.,
- et al.
- Reese J.A.,
- Li X.,
- Hauben M.,
- et al.
- Bougie D.W.,
- Wilker P.R.,
- Wuitschick E.D.,
- et al.
- Bougie D.W.,
- Rasmussen M.,
- Zhu J.,
- Aster R.H.
- Valgimigli M.,
- Biondi-Zoccai G.,
- Tebaldi M.,
- et al.
- Dunne S.,
- Shannon B.,
- Dunne C.,
- Cullen W.
- Mumoli N.,
- Cei M.,
- Luschi R.,
- Carmignani G.,
- Camaiti A.
- Patel A.,
- Jones S.A.,
- Ferro A.,
- Patel N.