Duration of eptifibatide infusion after percutaneous coronary intervention and outcomes among high-risk patients with non-ST-segment elevation acute coronary syndrome: insights from EARLY ACS

Abstract

Background and Objectives:

Eptifibatide is indicated during percutaneous coronary intervention (PCI) with continuation for 18–24 hours post procedure but is associated with bleeding. We examined the efficacy and safety of shorter post-PCI eptifibatide infusions in high-risk non-ST-segment elevation acute coronary syndrome (NSTE ACS) patients.

Methods:

EARLY ACS patients treated with PCI and eptifibatide were grouped by post-procedure infusion duration: <10, 10–13, 13–17, and 17–25 (per protocol) hours. Adjusted estimated event rates for 96-hour death/myocardial infarction (MI)/recurrent ischaemia requiring urgent revascularization (RIUR), 30-day death/MI, post-PCI packed red blood cell (PRBC) transfusion, and GUSTO (Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary Arteries) moderate/severe bleeding were obtained using inverse-propensity weighting to account for informative censoring of infusions.

Results:

Among 3271 eptifibatide-treated PCI patients, there were 66 96-hour death/MI/RIUR events, 94 30-day death/MI events, 127 PRBC transfusions, and 115 GUSTO moderate/severe bleeds. Compared with per protocol, patients receiving post-PCI infusions <10 hours had similar adjusted estimated rates of 96-hour death/MI/RIUR (absolute difference 0.021 higher; 0.040 vs. 0.019, 95% CI −0.023 to 0.064; p=0.35) and 30-day death/MI (0.020 higher; 0.046 vs. 0.026, 95% CI −0.021 to 0.062; p=0.34). There were also no differences in ischaemic outcomes between infusions of 10–17 hours and per-protocol infusions. Adjusted estimated rates of PRBC transfusion were higher for the <10-hour infusion group compared with per protocol (0.048 higher; 0.079 vs. 0.031, 95% CI 0.005 to 0.091, p=0.03) but were similar for other groups. Adjusted GUSTO moderate/severe bleeding rates were similar to per-protocol rates for all groups.

Conclusions:

In high-risk NSTE ACS patients, post-PCI eptifibatide infusions <18 hours were not associated with worse ischaemic outcomes. Shorter eptifibatide infusions in this population may be feasible.

Introduction

Thrombosis is a key contributor to ischaemic complications of percutaneous coronary intervention (PCI) and can be mitigated by platelet glycoprotein (GP) IIb/IIIa receptor antagonists. Previous randomized clinical trials showed that inhibiting GP IIb/IIIa improves outcomes in patients with acute coronary syndrome (ACS) and after PCI.¹ Eptifibatide is an intravenous, cyclic heptapeptide that reversibly inhibits GP IIb/IIIa. The efficacy of eptifibatide for PCI and current drug-dosing regimen were established in ESPRIT (Enhanced Suppression of the Platelet IIb/IIIa Receptor with Integrilin Therapy), a placebo-controlled, randomized trial in which high-dose eptifibatide (two boluses of 180 µg/kg given 10 minutes apart followed by infusion of 2.0 µg/kg/min for 18–24 hours post PCI) was associated with a 35% reduction in 30-day death or myocardial infarction (MI).² Professional society guidelines recommend treatment with eptifibatide in high-risk patients presenting with non-ST-segment elevation (NSTE) ACS undergoing PCI either with [American College of Cardiology/American Heart Association (ACC/AHA) class IIb, level of evidence B; European Society of Cardiology (ESC) class IIb, level of evidence C] or without (ACC/AHA class I, level of evidence A; ESC class IIa, level of evidence C) clopidogrel preloading.^3,4

Despite the well-established anti-ischaemic benefits of GP IIb/IIIa inhibitors for patients with ACS, extended use of GP IIb/IIIa inhibitors has disadvantages. Greater risk of bleeding was observed in clinical trials,¹ and post-PCI bleeding and MI were similarly associated with 1-year mortality.^5,6 Routine use of prolonged infusions also has practical implications, including longer hospital stay and higher drug cost. Clinical trials of bivalirudin bolus plus brief infusion versus heparin plus ≥18 hours of GP IIb/IIIa inhibition have demonstrated less bleeding but perhaps more early ischaemia.^7–9 Previous analyses examined the feasibility of abbreviated post-procedure GP IIb/IIIa inhibitor infusions for elective and uncomplicated PCIs,^10–12 but optimal post-PCI eptifibatide infusion duration in high-risk NSTE ACS patients is unknown.

EARLY ACS (Early Glycoprotein IIb/IIIa Inhibition in Non-ST-Segment Elevation Acute Coronary Syndrome) was a large randomized clinical trial evaluating early versus delayed provisional eptifibatide use at PCI in high-risk NSTE ACS patients expected to undergo an early invasive strategy. The protocol-specified duration of eptifibatide infusion after PCI was 18–24 hours but was variable among patients due to multiple factors, including adverse events and differences in local practice patterns. We exploited this variability to examine the association of post-PCI eptifibatide infusion duration with outcomes. In this study, we tested the primary hypothesis that shorter infusions of eptifibatide after PCI in a high-risk NSTE ACS population would not be associated with more ischaemic events. We also secondarily evaluated whether shorter eptifibatide therapy might be associated with reduced bleeding.

Methods

Study population

The EARLY ACS trial methods have been described previously.¹³ Briefly, the study was conducted from 2004–08 in 29 countries. Patients were enrolled if they were ≥18 years old and presented with rest ischaemia lasting ≥10 minutes within 24 hours before randomization plus ≥2 of the following: positive cardiac biomarkers; ischaemic changes on electrocardiography (ST-segment depression/transient ST-segment elevation); aged ≥60 years; or aged 50–59 years with documented coronary, cerebrovascular, or peripheral artery disease. Enrolment was contingent upon randomization within 12 hours of hospital presentation and planned invasive strategy ≥12 hours post-randomization.

EARLY ACS randomized 9406 patients to early administration of double bolus plus infusion of eptifibatide or placebo versus delayed, provisional use of eptifibatide after angiography but before PCI at the discretion of the treating cardiologist, with the option of intra-procedure eptifibatide ‘thrombotic bailout’ for predefined procedural complications [major dissection, decrease in Thrombolysis In Myocardial Infarction (TIMI) flow grade, abrupt vessel closure, no reflow, thrombus formation, side branch closure, and distal embolization]. Of 9406 patients, 5554 underwent PCI and had PCI data available. Patients with complete data who received eptifibatide through early randomization, transitioned to active drug just before PCI, or received thrombotic bailout during PCI were included, resulting in a final analysis population of 3271 patients (Figure 1).

Figure 1.

Flow diagram of patient selection.

PCI, percutaneous coronary intervention.

The EARLY ACS protocol complied with the Declaration of Helsinki and was approved by the ethics committee or institutional review board of each participating site. All randomized patients provided written informed consent. The use of EARLY ACS data for our study was approved by the Duke University Medical Center Institutional Review Board. The complete trial database resides at the Duke Clinical Research Institute, and all analyses were conducted independently of the trial sponsor.

Outcomes

Our primary outcomes were the composite ischaemic endpoints of 96-hour death, spontaneous MI, or recurrent ischaemia requiring urgent revascularization (RIUR) and 30-day death or spontaneous MI. Secondary endpoints included packed red blood cell (PRBC) transfusion and GUSTO (Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary Arteries) moderate bleeding (transfusion required but no haemodynamic compromise) or severe bleeding (intracranial bleeding or haemodynamic compromise requiring intervention) from the time of PCI through hospital discharge. An independent clinical events committee blinded to knowledge of the randomized treatment assignment adjudicated all suspected ischaemic events.

Statistical methods

Based on the distribution of observed post-PCI eptifibatide infusion times, the following study infusion duration intervals were chosen to allow the largest sample sizes possible for statistical comparisons among shorter infusion duration groups representing clinically relevant intervals: <10, 10–13, 13–17, 17–25, and >25 hours. The 17–25-hour interval represents the recommended 18–24-hour infusion ± 1 hour (per protocol). Demographics, qualifying event details, and clinical characteristics until the time of PCI were summarized using medians with interquartile ranges for continuous variables and frequencies and percentages for categorical variables.

We performed a landmark analysis from the time of PCI using the inverse propensity-weighted estimation method of Johnson and Tsiatis¹⁴ to account for confounding and informative censoring. Patients for whom infusions were terminated due to adverse events or protocol-specified reasons were considered informatively censored; the decision to stop treatment could be related to the outcomes of interest. Otherwise, infusions were assumed to have been completed at the discretion of the treating physician. A logistic model for discrete hazards was fit for the chosen infusion intervals and adjusted for covariates up to the time of PCI. In this context, a discrete hazard describes the probability of completing eptifibatide infusion during a fixed interval, assuming treatment has continued up to the start of the interval. Functions of the estimated discrete hazards can be viewed as the propensity to infuse eptifibatide for a given duration following PCI. Inverse propensity scores were used to weight patients and yield estimated outcome event rates for each interval while accounting for informative censoring by calculating a weighted mean frequency for each outcome. This method has been shown to yield unbiased estimated event rates when confounding and informative censoring are present, if the discrete hazards model is correctly specified.¹⁴

Adjustment baseline covariates for our discrete hazards models were chosen based on previous EARLY ACS outcomes models and also included important PCI-related variables that were post randomization but before landmark analyses were performed. These variables included age, female sex, weight, race, region, heart rate, creatinine clearance, platelet count, white blood cell count, haemoglobin level, haematocrit, positive cardiac markers, and electrocardiogram changes at the qualifying event, history of coronary artery disease (previous MI, coronary artery bypass graft surgery, PCI, angina), cerebrovascular disease or peripheral vascular disease, cardiogenic shock, congestive heart failure, Killip class, use of intraaortic balloon pump, PCI circumstance (elective or urgent/emergent), access site, culprit vessel size, lesion length, PCI complication, clinical instability during PCI as determined by the treating physician, prolonged ischaemia (ischaemic symptoms or ST-segment changes) during PCI, use of bailout eptifibatide kit, PCI outcome (successful or unsuccessful), and time from hospitalization to randomization. To avoid omission of incomplete records during multivariable modelling, multiple imputation was performed using SAS version 9 (SAS Institute, Cary, NC, USA). Forty-one baseline and catheterization variables were imputed. All tables presented and subsequent multivariable modeling used imputed data.

Event rate estimates and associated standard errors and 95% confidence intervals (CIs) were computed for each infusion duration interval. Crude estimates included an overall event rate and an event rate among uncensored patients (e.g. those without a treatment-terminating event). Unadjusted and adjusted estimated event rates were calculated, with the former assuming no confounding and the latter adjusting for covariates in the discrete hazards models. Differences in adjusted estimated event rates for patients receiving shorter versus per-protocol infusion durations were calculated, and corresponding 95% CIs and p-values were determined using Wald tests. All tests and 95% CIs were two-sided, and p-values <0.05 were considered statistically significant. No adjustments were made for multiple comparisons.

Sensitivity analyses

Prespecified sensitivity analyses were performed to assess the stability of the primary results. Ischaemic event rates were recalculated for each of the following patient categories: (1) no GUSTO moderate/severe or TIMI major bleeding; (2) randomization to early eptifibatide with continued infusion post PCI; (3) randomization to placebo with crossover to active drug at PCI; and (4) randomization to placebo with receipt of thrombotic bailout eptifibatide. Bleeding endpoint rates were recalculated for patients randomized to early eptifibatide who continued blinded infusion post PCI.

Results

Patient population

The distribution and baseline characteristics of the 3271 patients in our final analysis population across infusion duration intervals are shown in Table 1. Most patients (70.4%) received eptifibatide for 17–25 hours post PCI, as recommended by the EARLY ACS protocol. Physicians in Western Europe most frequently treated patients with infusions lasting <18–24 hours. Overall, patients receiving shorter post-PCI infusions were older, were more frequently female, had lower creatinine clearance, and were more likely to have prior/current congestive heart failure and/or cardiogenic shock.

Table 1.

Baseline patient characteristics, treatments, and PCI characteristics according to eptifibatide infusion duration.

Characteristic	Infusion duration (h)
	<10 (n=326)	10–13 (n=163)	13–17 (n=213)	17–25a (n=2303)
Demographics
Age (years)	67.8 (60.2–73.9)	68.1 (59.2–76.9)	67.6 (59.2–74.4)	65.4 (58.3–73.2)
Female sex	31.3	31.3	27.2	27.1
Region of enrolment
North America	39.4	39.9	42.3	33.9
Western Europe	43.9	45.4	38.0	32.4
Eastern Europe	6.1	3.1	3.3	12.3
Middle East, Africa, Asia-Pacific	19.6	11.7	16.4	21.3
Clinical characteristics
Medical history
Current smoker	26.7	34.4	22.5	31.7
Diabetes mellitus	30.4	28.8	31.0	27.5
Dyslipidaemia	54.9	55.2	59.2	59.3
Hypertension	70.9	74.2	66.2	66.5
Previous CABG	13.8	14.7	16.0	13.8
Previous MI	27.6	22.1	31.5	26.1
Previous PCI	23.3	27.0	30.0	25.1
Previous CVA	4.6	4.3	5.2	3.9
Previous PVD	6.4	9.2	9.4	8.7
Previous CHF	12.6	14.1	7.5	8.5
Estimated creatinine clearanceb (ml/min)	73.6 (57.3–100.3)	75.9 (53.9–103.4)	77.6 (58.3–98.6)	79.5 (60.1–100.1)
Killip class II, III, IV	11.7	12.3	10.3	6.9
Elevated troponin	83.0	78.4	81.3	84.6
Qualifying high-risk features
Age ≥60 years + CM, ST changes	16.6	16.6	20.2	17.8
Age ≥60 years + CM	49.5	37.4	37.6	42.6
CM, ST changes	13.5	17.8	15.0	18.5
Age ≥60 years, ST changes	8.0	16.6	14.1	8.6
Age 50–60 years + CM, previous vascular disease	6.2	5.5	7.0	6.0
Cardiogenic shock	2.5	0.6	0.0	0.4
Treatment during index hospitalization
Unfractionated heparin	46.3	32.9	42.3	43.7
Low-molecular-weight heparin	60.7	70.6	66.7	65.3
Aspirin	97.8	99.4	97.6	98.2
Clopidogrel
Early use intended	81.3	71.8	81.2	79.9
At any time during hospitalization	95.7	98.2	98.1	98.1
Beta-blocker	85.6	85.9	86.4	89.1
Statin	86.8	79.8	89.2	88.9
ACEI	66.6	55.2	62.0	69.5
ARB	9.7	12.5	13.0	8.8
Insertion of intraaortic balloon pump	3.1	1.8	0.5	0.6
CABG surgery	1.8	1.2	0.0	0.2
PCI characteristics
Circumstance
Elective	95.1	93.3	95.8	96.0
Urgent/emergent	4.9	6.7	4.2	4.0
Access site
Femoral	87.1	91.4	90.1	86.6
Radial	12.6	8.6	9.9	13.1
Brachial	0.3	0.0	0.0	0.3
Culprit vessel
LAD	38.7	34.4	36.6	34.9
RCA	23.8	26.4	28.2	28.4
LCx	39.7	29.7	28.6	28.9
Left main	1.8	1.8	1.9	0.9
Bypass graft	4.0	6.7	4.7	7.0
Culprit vessel size (mm)
<2.5	9.8	6.1	7.0	7.1
2.5–3.5	80.7	81.0	84.5	81.8
>3.5	9.5	12.9	8.5	11.2
Lesion length in culprit vessel (mm)
<10	14.4	11.7	9.9	9.2
10–20	55.8	60.1	62.4	61.7
>20	29.8	28.2	27.7	29.1
PCI complicationsc	13.8	9.2	9.4	10.6
Clinical instability or prolonged ischaemiad	2.5	1.8	0.5	2.1
Successful PCI	94.5	98.2	98.1	98.0

Values are % or median (interquartile range).

^aPer protocol.

^bCreatinine clearance was calculated using the Cockcroft–Gault formula.

^cPredefined as major dissection, decrease in Thrombolysis In Myocardial Infarction flow grade, abrupt vessel closure, no reflow, thrombus formation, side branch closure, or distal embolization. 

^dClinical instability during percutaneous coronary intervention as determined by the treating physician; prolonged ischaemia defined as ischaemic symptoms or ST-segment changes during percutaneous coronary intervention.

ACEI, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; CABG, coronary artery bypass graft; CHF, congestive heart failure; CM, cardiac markers; CVA, cerebrovascular accident; LAD, left anterior descending artery; LCx, left circumflex artery; MI, myocardial infarction; PCI, percutaneous coronary intervention; PVD, peripheral vascular disease; RCA, right coronary artery.

Data from patients receiving infusions >25 hours (n=266) are not included, as our analyses were not relevant to this group.

Ischaemic outcomes

There were 66 death/MI/RIUR events at 96 hours, including 15 deaths, 37 MIs, and 25 incidents of RIUR. At 30 days, there were 94 death/MI events, comprising 41 deaths and 61 MIs. Table 2 lists the crude, estimated, and adjusted event rates for these outcomes. Adjusted estimated rates of 96-hour death/MI/RIUR and 30-day death/MI rates for patients receiving <10 hours of post-PCI eptifibatide compared with those receiving infusions per protocol were not significantly different (96-hour death/MI/RIUR 0.021 higher, 95% CI −0.023 to 0.064; p=0.35; 30-day death/MI 0.020 higher, 95% CI −0.021 to 0.062; p=0.34; Figure 2). Adjusted estimated 96-hour death/MI/RIUR and 30-day death/MI rates for patients receiving eptifibatide infusions of 10–13 hours or 13–17 hours were also similar to those of the per-protocol group.

Table 2.

Event rates according to eptifibatide infusion duration.

Infusion duration (h)	Completed infusion (n)	Discontinued infusion (n)	Overall event rate	Uncensored event rate	Estimated event rate (SE)	Adjusted estimated event rate (SE)
96-h death, spontaneous MI, or RIUR
<10	220	118	0.064	0.016	0.021 (0.008)	0.040 (0.022)
10–13	150	12	0.018	0.013	0.018 (0.009)	0.023 (0.011)
13–17	194	19	0.026	0.014	0.019 (0.008)	0.020 (0.008)
17–25	2274	22	0.014	0.014	0.020 (0.003)	0.019 (0.003)
30-day death or spontaneous MI
<10	220	109	0.060	0.029	0.032 (0.010)	0.046 (0.021)
10–13	150	12	0.030	0.019	0.023 (0.011)	0.028 (0.013)
13–17	194	19	0.034	0.023	0.027 (0.010)	0.028 (0.010)
17–25	2281	22	0.023	0.023	0.028 (0.003)	0.026 (0.003)
PRBC transfusion
<10	219	117	0.145	0.070	0.077 (0.016)	0.079 (0.022)
10–13	149	15	0.082	0.058	0.067 (0.018)	0.054 (0.015)
13–17	192	20	0.034	0.028	0.040 (0.011)	0.035 (0.009)
17–25	2273	25	0.021	0.019	0.032 (0.003)	0.031 (0.003)
GUSTO moderate/severe bleeding
<10	211	134	0.179	0.039	0.053 (0.012)	0.056 (0.018)
10–13	149	15	0.064	0.026	0.042 (0.013)	0.038 (0.010)
13–17	189	23	0.030	0.014	0.033 (0.008)	0.033 (0.008)
17–25	2268	20	0.014	0.013	0.033 (0.003)	0.032 (0.003)

GUSTO, Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary Arteries; MI, myocardial infarction; PRBC, packed red blood cell; RIUR, recurrent ischaemia requiring urgent revascularization; SE, standard error.

Data from patients receiving infusions >25 hours (n=266) are not included, as our analyses were not relevant to this group.

Figure 2.

Differences in adjusted ischaemic event rates between shorter and per-protocol eptifibatide infusions.

Each square represents the difference of event rates (shorter minus per-protocol infusion). Horizontal lines indicate 95% confidence intervals.

CI, confidence interval; MI, myocardial infarction; RIUR, recurrent ischaemia requiring urgent revascularization.

Bleeding outcomes

There were 127 PRBC transfusions and 115 GUSTO moderate/severe bleeding events. As shown in Table 2, patients receiving <10-hour and 10–13-hour post-PCI infusions had the highest crude rates of PRBC transfusion and GUSTO moderate/severe bleeding. Compared with the per-protocol group, the adjusted estimated PRBC transfusion rate was significantly higher for patients receiving infusions <10 hours (0.048 higher, 95% CI 0.005 to 0.091; p=0.03) but not for those receiving infusions of 10–13 hours or 13–17 hours (Figure 3). Adjusted estimated rates of GUSTO moderate/severe bleeding among patients in shorter infusion duration intervals were similar to those in the per-protocol group.

Figure 3.

Differences in adjusted bleeding and transfusion rates between shorter and per-protocol eptifibatide infusions.

Each square represents the difference of event rates (shorter minus per-protocol infusion). Horizontal lines indicate 95% confidence intervals.

CI, confidence interval; GUSTO, Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary Arteries; mod, moderate; PRBC, packed red blood cell.

Sensitivity analyses

Sensitivity analyses were performed to test the stability of the estimates under the primary analysis methods (supplementary data). Including only patients without GUSTO moderate/severe or TIMI major bleeding (n=3129) did not affect our primary ischaemic event results. Analysis of patients randomized to early eptifibatide who continued blinded eptifibatide post PCI (n=1421) showed lower rates of 96-hour death/MI/RIUR and 30-day death/MI for 13–17-hour eptifibatide infusions compared with per-protocol infusions. Outcomes were similar among patients receiving infusions for <10 hours or 10–13 hours versus the per-protocol duration. Among patients who received eptifibatide after angiography but before PCI (n=1263) or who were given eptifibatide for thrombotic bailout (n=582), efficacy outcomes for all shorter durations were similar to those in the per-protocol group.

Among patients randomized to early eptifibatide who continued blinded therapy post PCI, compared with per-protocol infusions, there was no difference in rates of PRBC transfusion with infusions shorter than per protocol. However, we found lower rates of GUSTO moderate/severe bleeding associated with 13–17-hour (0.010 lower, 0.021 vs. 0.031, 95% CI −0.018 to −0.002; p= 0.010) and 10–13-hour (0.011 lower, 0.020 vs. 0.031, 95% CI −0.020 to −0.003; p=0.005) infusions compared with per-protocol treatment.

Discussion

Patients presenting with high-risk NSTE ACS treated with an early invasive strategy benefit from GP IIb/IIIa inhibitors,³ but associated bleeding contributes to morbidity and mortality.¹⁵ Eptifibatide is indicated for treatment in this population with continuation of drug for 18–24 hours post PCI; shortening this duration in an effort to minimize bleeding has not been well studied. In our analysis, we found that post-PCI infusions of eptifibatide <18 hours in a high-risk NSTE ACS population were not associated with worse in-hospital or 30-day ischaemic outcomes. Furthermore, although eptifibatide infusions <10 hours were associated with higher adjusted rates of PRBC transfusion, they were not associated with more clinical bleeding events. We hypothesize that the absence of lower bleeding and transfusion rates with infusions shorter than per protocol was likely due to residual confounding from early termination of eptifibatide driven by clinical reluctance to continue therapy based on high risk for bleeding and/or the occurrence of minor and unreported bleeding events that could not be accounted for by our statistical methods.

Infusion duration and outcomes

Efforts to minimize bleeding risk associated with GP IIb/IIIa inhibitors have included studies of abbreviated courses of therapy. Current guidelines and package labelling recommend sustained infusions of abciximab (12 hours), tirofiban (12–18 hours), or eptifibatide (18–24 hours). Although bolus-plus-infusion dosing of abciximab has been associated with incremental ischaemic benefit compared with bolus-only treatment at a cost of increased bleeding, a statistically insignificant reduction in ischaemic outcomes was also observed with bolus-only drug compared with placebo.¹⁶ However, these observations occurred in the pre-stent era, when early return to the catheterization laboratory for suspected ischaemia and revascularization was much more common than it is in contemporary PCI practice. In the more recent ESPRIT substudy, continuing eptifibatide infusions for >16 hours provided no further clinical benefit in elective PCI,¹² suggesting that limited infusions might provide ischaemic benefit and reduce bleeding risk. This is especially relevant given current PCI practice patterns that incorporate frequent use of stenting.

The only randomized clinical trial of shorter eptifibatide infusion was BRIEF-PCI (Brief Infusion of Eptifibatide Following Percutaneous Coronary Intervention), which randomized patients with stable angina or NSTE ACS undergoing successful, uncomplicated PCI to a double bolus of eptifibatide plus an 18-hour vs. a 2-hour infusion.¹⁰ Compared with standard treatment, shorter infusions did not change peri-procedural MI or 30-day death/MI/target vessel revascularization, but they did significantly reduce major bleeding.

Two studies compared bolus-only versus bolus-plus standard infusion in mixed stable angina and NSTE ACS cohorts. In the first study, bolus-only abciximab was not inferior to standard bolus-plus infusion for ischaemic events in patients undergoing transradial PCI and same-day discharge.¹⁷ The second study found similar ischaemic complication rates but fewer vascular/bleeding complications in patients receiving bolus-only versus standard dosing of post-PCI abciximab or eptifibatide.¹¹

Our study expands current knowledge regarding abbreviated post-PCI eptifibatide treatment by examining shorter infusions in high-risk NSTE-ACS patients and including complicated, unsuccessful interventions. Although previous studies evaluated bolus-only GP IIb/IIIa inhibitors, generalization of results may be limited. Such studies tested abciximab in transradial PCI patients stable for same-day discharge or included abciximab and eptifibatide, 2 very different drugs. Abciximab has a short plasma half-life but a high affinity for the GP IIb/IIIa receptor and a functional half-life of up to 7 days. Eptifibatide has a longer plasma half-life but dissociates quickly from the GP IIb/IIIa receptor. Previous study results may have been driven by pharmacokinetic differences and/or may not apply to our study population.

Abbreviated eptifibatide infusions and residual confounding: the need for a randomized clinical trial

We found no association between abbreviated eptifibatide infusions and increased ischaemic events. Although we could not completely eliminate confounding from our analyses, the lack of signal for increased ischaemia with shorter eptifibatide infusions in multiple sensitivity analyses supports our primary results. This was especially reassuring among the subgroups that were randomized to placebo but received active treatment pre-PCI or bailout eptifibatide, as these patients would be at highest risk for ischaemic outcomes. Furthermore, it is important to note that these patients were also at highest risk for bleeding, yet no increase in ischaemic outcomes was observed in these subgroups.

Although use of abbreviated eptifibatide infusions clinically might be expected to reduce bleeding, we did not observe this association in our study. This result is partly due to confounding related to the structure of the study case report form. Despite EARLY ACS protocol recommendations for 18–24 hours of eptifibatide post PCI, 65.2% of patients in the <10-hour group were either identified by sites as having had their infusions ‘completed per protocol’ or did not provide a reason for drug discontinuation. Since the protocol allowed for drug discontinuation for bleeding, infusions stopped before 18 hours due to bleeding may have been marked as ‘completed per protocol’ but without a concurrent report of bleeding. In addition, although we observed higher transfusion rates in the <10-hour subgroup, this was not accompanied by an increase in clinical bleeding, raising the possibility that these patients may have received ‘prophylactic’ transfusions for low haemoglobin and haematocrit without actual bleeding. Taken together, these potential causes for residual confounding could not be addressed with our statistical methods and could explain both the higher transfusion rate with <10-hour infusions and the absence of lower bleeding rates with shorter infusions compared with the per-protocol group. Despite potential confounding, in a sensitivity analysis among patients randomized to early eptifibatide who continued blinded treatment post PCI as randomized, we did observe a significant reduction in GUSTO moderate/severe bleeding with shorter eptifibatide infusions compared with per-protocol infusions, suggesting that a randomized study might prove a true safety advantage of abbreviated therapy.

Another important consideration is that multiple characteristics that define patients at higher risk for ischaemic events are also associated with increased bleeding in NSTE ACS and after PCI.^18,19 Compared with previous studies of abbreviated GP IIb/IIIa inhibitor therapy, our overall study population likely had higher ischaemic and bleeding risk, and the overlap of risk factors in these patients may have resulted in a treatment paradox. High-risk patients derive ischaemic benefit from longer infusions at a cost of increased bleeding; although bleeding is a reason to stop treatment, prolonged ischaemia is an indication for continuing it. Such complex interactions may preclude accurate definition of relationships among infusion duration, bleeding events, and ischaemic outcomes without a randomized clinical trial explicitly addressing the question.

Shorter GP IIb/IIIa infusions and other PCI treatment strategies

Despite the availability of newer anticoagulants, GP IIb/IIIa inhibitors remain an important treatment for NSTE ACS, especially among high-risk patients undergoing PCI. Use of bivalirudin in this population has increased due to an improved safety profile compared with heparin plus a GP IIb/IIIa inhibitor.⁷ However, GP IIb/IIIa inhibitors are still used in 41.2% of US patients presenting with NSTEMI,²⁰ and bivalirudin is not routinely used worldwide. Furthermore, despite the safety benefit of bivalirudin, data suggest that there may be an increase in early ischaemia.^8,9 A shorter eptifibatide infusion may be an alternative antithrombotic strategy for PCI that offers the bleeding benefits of bivalirudin while also providing sufficient platelet blockade to avoid any early ischaemic risk associated with bivalirudin; however, confirming this hypothesis requires a randomized trial.

Limitations and strengths

Our study has limitations and strengths. Patients were not prospectively randomized post PCI to different eptifibatide infusion durations and, despite adjusting for informative censoring, residual confounding likely remains. The relatively small number of patients in shorter infusion duration groups and low event rates affected the power of our analyses, suggesting cautious interpretation of results. In addition, our database was not suited to studying the influence of varying use of upstream clopidogrel or other antithrombotic regimens on ischaemic events or bleeding rates, particularly in the shorter infusion duration groups. Our study was, however, strengthened by additional analyses examining the sensitivity of our primary estimates to model perturbations. Ultimately, a well-designed randomized clinical trial of shorter post-PCI eptifibatide infusions and clinical outcomes in the NSTE ACS population is needed to definitively answer the question. A forthcoming phase II trial of abbreviated tirofiban versus standard-dose eptifibatide for PCI may provide further insight on the issue.²¹

Conclusions

Despite reductions in post-PCI ischaemic complications with use of GP IIb/IIIa inhibitors, bleeding remains a serious risk. Current guidelines and package labelling recommend eptifibatide infusions for 18–24 hours post PCI. Although our analysis was likely confounded with respect to bleeding outcomes, we observed no increase in ischaemic events with shorter eptifibatide infusions. Amidst current pressures to minimize bleeding risk and reduce hospital length of stay, our results support further investigation into abbreviated post-PCI eptifibatide infusions in patients with NSTE ACS.