Comparative Effectiveness and Safety of a Catheterization Laboratory–Only Eptifibatide Dosing Strategy in Patients Undergoing Percutaneous Coronary Intervention

Adjunctive platelet glycoprotein IIb/IIIa inhibitors reduce the incidence of periprocedural ischemic complications in patients undergoing elective or urgent percutaneous coronary intervention (PCI).^1–3 Eptifibatide, a small-molecule glycoprotein IIb/IIIa inhibitor, is the most commonly used agent in this class. The drug label recommends that it be dosed as a double-bolus followed by an infusion lasting 18 to 24 hours or to hospital discharge and at least ≥12 hours. This recommendation is based on the dosing regimen that was tested in the pivotal trials that established the clinical efficacy of the drug.⁴ Whereas the bolus plus infusion regimen is the standard of care, the need for prolonged infusion has been questioned because of possible increased bleeding complications, greater cost, and prolonged hospitalization. Eptifibatide has rapid onset and offset of action, and the current dosing recommendations for eptifibatide were derived from pharmacokinetic studies assessing the need to obtain adequate (>80%) and persistent inhibition of platelet aggregation.^5–7 It is unclear, however, as to how long platelet inhibition must be maintained after contemporary PCI, and a small randomized study previously suggested that an eptifibatide bolus and a short infusion (<2 hours) were noninferior to the standard regimen of bolus plus infusion and might be associated with a reduction in bleeding.⁸

See Editorial by Sherwood and Tcheng

Because of these issues, some interventional cardiologists use a catheterization laboratory–only eptifibatide regimen and forego the infusion after successful stent implantation.^9,10 This practice has been growing in recent years although there are limited data on the clinical safety and efficacy of such an approach. Accordingly, we used advanced matching techniques to assess the efficacy and safety of this off-label regimen and compared it with the conventional bolus plus infusion regimen in a large cohort of patients undergoing contemporary PCI.

Methods

Study Population

We included data from patients undergoing PCI at 47 hospitals participating in the Blue Cross Blue Shield of Michigan Cardiovascular Consortium (BMC2). The details of the Blue Cross Blue Shield of Michigan Cardiovascular Consortium registry and its data collection and auditing process have been described previously.^11–14 Briefly, procedural data on all patients undergoing PCI at participating hospitals are collected using standardized data collection forms. Baseline data include clinical, demographic, procedural, and angiographic characteristics, as well as medications used before, during, and after the procedure, and in-hospital outcomes. All data elements have been prospectively defined, and the protocol is approved by local institutional review boards at each hospital. In addition to a random audit of 2% of all cases, medical records of all patients undergoing multiple procedures or coronary artery bypass grafting and of patients who died in the hospital are reviewed routinely to ensure data accuracy.

The study population for this analysis included all consecutive patients who underwent PCI between January 1, 2010, and December 31, 2011. We excluded patients who were treated with bivalirudin (n=34635) or abciximab (n=1468), patients undergoing rescue (n=523) or salvage PCI (n=108), patients who presented with cardiac arrest (n=1518) or cardiogenic shock (n=1320), patients in whom an intra-aortic balloon pump (n=2035) or left ventricular assist device (n=480) was used, and patients who died in the catheterization laboratory (n=63) or those who underwent emergent coronary artery bypass grafting (n=165) during the same hospitalization. Furthermore, we excluded patients who developed coronary perforation (n=335) or pericardial tamponade (n=53) in the catheterization laboratory (because it would be customary to stop the eptifibatide infusion), those with significant untreated residual dissection (because it would be common to administer a prolonged infusion; n=1494), patients with a known contraindication to aspirin (n=731), and those with recent surgery (n=712) or gastrointestinal bleeding (n=914).

Study End Points

Our study had 3 end points for this analysis: in-hospital death, bleeding, and need for transfusion. The secondary end points include myocardial infarction, repeat PCI on the same lesion, and stent thrombosis. In-hospital death was defined as death from any cause before discharge from the hospital after PCI. Bleeding and myocardial infarction were defined as per the Cath PCI National Cardiovascular Data Registry registry definitions. Transfusion was defined as administration of whole blood or packed red cells irrespective of total number of units infused. Stent thrombosis was defined by angiographic evidence of stent thrombosis on repeat angiography during the same hospitalization.

Statistical Analysis

We categorized patients into 2: those treated with eptifibatide bolus plus infusion (standard therapy) and those treated with a catheterization laboratory–only eptifibatide regimen. Patients were considered to have been treated with an infusion if the drug was being infused at the time patient left the catheterization laboratory irrespective of the total time of drug administration after the end of the procedure. Patients who only received eptifibatide during the PCI procedure and did not have an infusion after leaving the catheterization laboratory were considered to be treated with a catheterization laboratory–only regimen.

We used a combination of exact matching and optimal full matching on the propensity score to account for nonrandom assignment of a standard versus catheterization laboratory–only regimen in patients undergoing PCI. A propensity score, or the probability of receiving the catheterization laboratory–only regimen, was estimated using a nonparsimonious logistic regression model created from a variety of preprocedure patient characteristics and therapies. These included weight, height, race, cardiac status, presenting symptoms or diagnosis, comorbidities, angiographic characteristics, and therapies administered before PCI. In addition, to allow the estimated propensity scores to vary by hospital, the propensity score model included a fixed effect for hospital.

Optimal full matching is preferred over greedy pair matching because it allows units to be arranged into sets of varying sizes and does not depend on the ordering of units to be matched.¹⁵ For computational efficiency and to facilitate subsequent subgroup analyses, we performed propensity score matching within exact matched subsets defined by age, sex, bifurcation lesions, presence or absence of thrombus, PCI for ST-segment–elevation myocardial infarction, and PCI for stable angina. Optimal full matching considers all possible arrangements of patients into matched sets of varying sizes in which either 1 patient who received standard therapy is matched to ≥1 patients with similar propensity scores and received catheterization laboratory–only eptifibatide or 1 patient with catheterization laboratory–only eptifibatide regimen is matched to ≥1 patients with similar scores who received the standard therapy. The optimal full matching routine chooses the arrangement of patients into sets that minimize the total propensity score distance between units in matched sets.¹⁶

Optimal full matching retained ≈99.8% of the data overall, excluding 34 (0.2%) standard regimen patients who are unlike catheterization laboratory–only patients. Standard regimen patients are ≈3× more prevalent than catheterization laboratory–only patients, so all catheterization laboratory–only regimen patients were assigned a matched set.

Adequacy of covariate balance (as a proxy approximation of a block-randomized experiment) was assessed using 2 strategies. First, we assessed covariate balance incorporating the matched sets to ensure that few statistically significant imbalances exist for the covariates included in the propensity score.^16,17 For each variable, tests assess whether the distribution of the variable across groups of patients treated with the catheterization laboratory–only regimen compared with the standard regimen within matched sets is as though treatment were randomly assigned. These tests compute test statistics by standardizing and comparing weighted averages or proportions of a particular variable for patients treated with standard therapy versus catheterization laboratory–only therapy.¹⁸ Second, we assessed for differences in unrelated outcomes (contrast-induced nephropathy and nephropathy requiring dialysis) that would not be expected to be influenced by eptifibatide infusion.¹⁹ The presence of a differing outcome between the 2 groups in the absence of biological plausibility would be suggestive of residual confounding.

After the matching procedure and balance assessment, we determined the effect of using a catheterization laboratory–only regimen with standard therapy on bleeding and need for transfusion in addition to other outcomes. We used a conditional logistic regression model with additional confounder control for covariates highly predictive of the outcome of interest and a fixed effect for the matched set, estimating effects as though the model could be separately fit within each set and aggregated across sets.²⁰ The association of the catheterization laboratory–only regimen with these complications was thus assessed in the entire population and within the prespecified subgroups. The statistical software package R-2.15.0 was used for all analysis.

Results

Of the 84 678 PCIs performed during the study period (calendar year 2010–2011), and meeting our inclusion criteria, eptifibatide was administered to 21 296 patients. Of these, a bolus-only strategy was used in 4511 (21.2%) patients, whereas 16 785 patients were treated with a bolus and an infusion. The rate of the use of the standard treatment (bolus and infusion) varied across hospitals from 35% to 100%.

In unadjusted analysis, significant differences were noted between patients who were treated with catheterization laboratory–only therapy versus standard therapy with patients who received the standard treatment being more likely to be older, women, and have a history of diabetes mellitus and hypertension (Table 1). The matched population was well balanced (Table 1)

In unadjusted and unmatched analysis, patients treated with the standard dosing regimen were more likely to have a bleed (2.7% versus 2.3%) or need a transfusion (3.8% versus 2.7%), and roughly as likely to die (0.42% versus 0.41%) or have vascular complications (0.5% versus 0.7%).

Of the 21 262 patients included in the matched analysis, 558 patients (2.6%) required a blood transfusion and 764 patients (3.6%) had a bleeding event (Table 2). In optimally matched analysis, the catheterization laboratory–only eptifibatide was associated with a significantly decreased risk of bleeding (odds ratio [OR], 0.74; 95% confidence interval [CI], 0.58–0.93; P=0.014) and a decreased need for transfusion (OR, 0.70; 95% CI, 0.52–0.92; P=0.012). The catheterization laboratory–only approach was associated with a significantly decreased likelihood of gastrointestinal bleeding (OR, 0.40; 95% CI, 0.22–0.74; P=0.003), but there was no difference in retroperitoneal bleeding.

There was no significant difference between the standard regimen and catheterization laboratory–only therapy in terms of subsequent vascular complications (OR, 1.40; 95% CI, 0.79–2.46; P=0.249), death (OR, 1.25; 95% CI, 0.65–2.42; P=0.508), stent thrombosis (OR, 0.49; 95% CI, 0.12–1.96; P=0.312), repeat PCI on the same lesion (OR, 0.58; 95% CI, 0.28–1.18; P=0.132), or myocardial infarction post procedure (OR, 0.76; 95% CI, 0.55–1.05; P=0.097).

Within the subgroups that were studied, significant interaction was seen with respect to the presenting diagnosis (ST-segment–elevation myocardial infarction versus other) with the lower risk of bleeding with the catheterization laboratory–only approach being evident in those with diagnosis other than ST-segment–elevation myocardial infarction (Figure 1). A weaker interaction was also observed with respect to the presence of angiographic thrombus with the risk of bleeding with the catheterization laboratory–only regimen being less among those patients who lacked angiographic evidence of thrombus. Similarly, the need for transfusion in association with the standard therapy was greater among those patients who were undergoing PCI for a diagnosis other than ST-segment–elevation myocardial infarction and in those who did not have angiographic evidence of thrombus (Figure 2). No subgroups were identified where the standard approach provided better outcomes than a catheterization laboratory–only approach.

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

There was no difference in the unrelated outcome of contrast-induced nephropathy or need for dialysis.

Discussion

The key finding of our study is that the use of a catheterization laboratory–only dosing strategy of eptifibatide is associated with a clinically significant reduction in bleeding and need for transfusion without an increase in risk of thrombotic events. These findings have important clinical implications because the demonstrated reduction in periprocedural ischemic complications in association with the routine use of glycoprotein IIb/IIIa inhibitors is counterbalanced by an increased risk of bleeding and significant drug cost.^21–24

The largest body of data to support the need for a prolonged infusion in association with glycoprotein IIb/IIIa inhibitors comes from the Evaluation of 7E3 for the Prevention of Ischaemic Complications (EPIC) trial.²² In this trial of patients undergoing angioplasty or atherectomy for unstable coronary lesions, when compared with placebo, abciximab bolus and infusion resulted in a greater reduction in the primary end point when compared with bolus only. When the individual end points were evaluated, the key difference between the bolus-only group versus bolus and drip was in emergency percutaneous transluminal coronary angioplasty (3.6% versus 0.8%) and bailout stent placement (1.7% versus 0.6%). The incidence of major bleeding was lower in the group treated with the bolus only versus those treated with bolus and drip (11% versus 14%).

Subsequently the strategy of using glycoprotein IIb/IIIa inhibitors as a bolus and drip has become the cornerstone of modern interventional therapy. The same strategy has been used for patients treated with small-molecule glycoprotein IIb/IIIa inhibitor eptifibatide or tirofiban although no adequately powered comparison of bolus-only versus bolus and drip strategy has been published.

The relevance of routine glycoprotein IIb/IIIa infusion after the loading bolus dose in contemporary interventional PCI is, however, questionable. The dose and need for the infusion were derived from the degree of platelet inhibition required to maintain patency of a coronary artery injury model in vivo and after angioplasty in humans. The applicability of these models to contemporary stent-based PCI is uncertain.

Because the use of a drip after PCI adds significant cost (direct drug cost and nursing costs) and increases the hazard of bleeding while providing unclear benefits, many investigators and operators have evaluated the safety and efficacy of this strategy in small studies. The only study that tested this in a randomized fashion in patients undergoing contemporary PCI for stable and unstable coronary artery disease demonstrated no difference in ischemic events with use of an abbreviated infusion compared with the standard technique, while a reduction in bleeding events was noted.⁸

In 3 observational studies, 3 high-volume centers reported their experience with using bolus-only glycoprotein IIb/IIIa inhibitors. In a study from a single Michigan center, Fischell et al¹⁰ reported favorable outcomes in 401 patients treated with a single bolus of eptifibatide. Similar results have been described by investigators from the University of New York, SUNY Downstate Medical Center.⁹ Finally in the largest series to date, Kini et al²⁵ described a reduction in bleeding and cost with the use of a bolus-only approach. These groups of investigators have drawn on pharmacological principles and have argued that routine glycoprotein IIb/IIIa infusion is unnecessary because the antiplatelet effects of the bolus last for ≈2 hours and overlaps with the oral antiplatelet agents that are routinely administered before or immediately after the completion of PCI. Recent work by Valgimigli et al²⁶ demonstrating rapid and sustained inhibition of platelet aggregation by a combination of high-dose tirofiban bolus and oral prasugrel provides support for this argument. Second, the bolus-only group may avoid the risk of delayed bleeding seen with the standard therapy. In a study that specifically focused on time to bleeding in association with eptifibatide bolus and drip, the median time to bleed was 7.0 hours (interquartile range, 3.1–14.0 hours) after the first eptifibatide bolus and occurred in 3.2% of the patients.²⁷ Given the short half-life of eptifibatide, the catheterization laboratory–only approach is expected to result in less bleeding by decreasing the time of exposure.

Our findings corroborate and significantly extend previous work. Our study population is derived from an ongoing collaborative quality initiative with robust data quality. Our study population is larger than reported in any of the previous studies and reflects contemporary interventional practice. Furthermore, we used advanced matching methods and the results of the study are likely extant. Although observational data in general cannot supplant clinical trials, large studies such as ours can be helpful when randomized data are sparse. Furthermore, given the lack of economic incentives, it is unlikely that any large randomized controlled trial will be performed in the near-future to evaluate the clinical efficacy of a bolus-only eptifibatide strategy. The bolus contributes a small fraction of the total drug cost, and a switch to this dosing regimen would be financially disadvantageous for the drug manufacturer. However, this reduction in drug cost, reduced nursing and pharmacy time, as well as the lower risk of bleeding, and reduced transfusion observed in association with the abbreviated therapy likely translates into major cost savings for the healthcare system.

Our results should be interpreted with certain caveats. The Blue Cross Blue Shield of Michigan Cardiovascular Consortium-PCI registry is a regional collaborative with an active focus on multicentric quality improvement and may or may not be representative of the wider population of patients undergoing PCI in the United States.¹² A proportion of patients had their infusion stopped before leaving the catheterization laboratory because of occurrence of bleeding complications in the catheterization laboratory, and we were not able to identify this from the registry data. Inclusion of these patients in the catheterization laboratory–only group would tend to negate the benefit of using such an approach and likely reduce the observed reduction in bleeding complications. The duration of total infusion time was not recorded, and it is unclear as to how many patients actually received the label recommend dose of ≥12 hours. It is likely that study was also subject to unmeasured residual confounding, despite optimally matched analyses that might influence any of the noted associations.

Conclusions

A catheterization laboratory–only dosing of eptifibatide is associated with a reduction in bleeding complications and a reduced need for transfusion without any observed excess of ischemic complications. The improved clinical outcomes observed with the use of this approach, combined with lower cost and similar results observed in a previous randomized trial, suggest that such a strategy should be preferentially considered when the use of eptifibatide is contemplated.

Footnotes