JACC: Cardiovascular Interventions

Volume 8, Issue 12, October 2015 DOI: 10.1016/j.jcin.2015.04.031
PDF Article

Impact of Glycoprotein IIb/IIIa Inhibition in Contemporary Percutaneous Coronary Intervention for Acute Coronary Syndromes
Insights From the National Cardiovascular Data Registry

David M. Safley, Lakshmi Venkitachalam, Kevin F. Kennedy and David J. Cohen

Author + information

vol. 8 no. 12 1574-1582
DOI: 
https://doi.org/10.1016/j.jcin.2015.04.031
Published By: 
JACC: Cardiovascular Interventions
Print ISSN: 
1936-8798
Online ISSN: 
1876-7605
History: 
  • Received January 20, 2015
  • Revision received April 13, 2015
  • Accepted April 23, 2015
  • Published online October 1, 2015.
Copyright & Usage: 
2015 American College of Cardiology Foundation

Author Information

  1. David M. Safley, MD∗,†∗ (dsafley{at}saint-lukes.org),
  2. Lakshmi Venkitachalam, PhD‡,
  3. Kevin F. Kennedy, MS∗ and
  4. David J. Cohen, MD, MSc∗,†
  1. ∗Saint Luke’s Mid America Heart Institute, Kansas City, Missouri
  2. †University of Missouri–Kansas City, Kansas City, Missouri
  3. ‡Department of Biomedical & Health Informatics, University of Missouri-Kansas City School of Medicine, Kansas City, Missouri
  1. ↵∗Reprint requests and correspondence:
    Dr. David M. Safley, Saint Luke’s Mid America Heart Institute, University of Missouri–Kansas City, 4401 Wornall Road, Kansas City, Missouri 64111.

Abstract

Objectives This study investigates the effects of glycoprotein IIb/IIIa inhibitors (GPIs) on outcomes after percutaneous coronary intervention (PCI).

Background Ischemic complications are reduced after PCI when a GPI is added to heparin. However, there are limited data on the safety and efficacy in contemporary PCI.

Methods We used the National Cardiovascular Data Registry CathPCI Registry data to assess the association between GPI use and PCI outcomes for acute coronary syndrome between July 2009 and September 2011. The primary outcome was all-cause in-hospital mortality. The secondary outcome was major bleeding. To adjust for potential bias, we used multivariable logistic regression, propensity-matched (PM) analysis, and instrumental variable analysis (IVA).

Results There were 970,865 patients included; 326,283 (33.6%) received a GPI. Unadjusted mortality and major bleeding were more common with a GPI (2.4% vs. 1.4% and 3.7% vs. 1.5%, respectively; p < 0.001 for both). In contrast, GPI use was associated with lower mortality on adjusted analyses; relative risks range from 0.72 (95% confidence interval [CI]: 0.50 to 0.97) with IVA to 0.90 (95% CI: 0.86 to 0.95) with PM. The association of GPI use with bleeding remained in adjusted analyses (multivariable relative risk: 1.93, 95% CI: 1.83 to 2.04; PM relative risk: 1.83, 95% CI: 1.74 to 1.92; and IVA relative risk: 1.53, 95% CI: 1.27 to 2.13). Subgroup analysis revealed enhanced risk reduction with ST-segment elevation myocardial infarction, high predicted mortality, and heparin-based anticoagulation.

Conclusions In unselected acute coronary syndrome patients undergoing PCI, GPI use was associated with reduced in-hospital mortality and increased bleeding. In the modern era of PCI, there may still be a role for the judicious use of GPIs.

Key Words

For patients with an acute coronary syndrome, clinical guidelines recommend timely percutaneous coronary intervention (PCI) to reduce complications including death and recurrent myocardial infarction (MI) (1–3). Ongoing national initiatives target reductions in ischemic and bleeding complications (4,5). An early strategy to reduce ischemic complications of PCI was the use of parenteral glycoprotein IIb/IIIa inhibitors (GPIs) (6–9). The benefit of GPIs in early studies was driven primarily by reductions in periprocedural MI (10–12). Advances in both stent design (e.g., thin-strut designs, biocompatible polymers) and adjunct pharmacology (e.g., preloading with potent oral antiplatelet agents) have led to improved outcomes and a diminished role for GPIs as reflected in current PCI guidelines (13).

Although randomized clinical trials are the gold standard for evaluation of alternative treatments by virtue of their ability to limit confounding, they have important limitations. In particular, randomized trials often use composite endpoints and may have only modest power to examine individual endpoints (14). Moreover, most clinical trials enroll highly select individuals at lower risk than patients treated in routine clinical practice (15). If a therapy provides preferential benefits among high-risk patients, these benefits might be missed when studied in a low-risk population.

We hypothesized that these issues might be particularly relevant for GPIs, which have been studied in predominantly low- to moderate-risk patients (to limit bleeding complications) and have been powered to demonstrate reductions in myonecrosis rather than mortality (16,17). To address these limitations, we designed a retrospective observational study to examine the association between the use of GPIs and outcomes of PCI in acute coronary syndrome patients using data from the National Cardiovascular Data Registry (NCDR). In addition to conventional risk-adjustment techniques such as multivariable (MV) and propensity-matched (PM) analyses, we used instrumental variable analysis (IVA), an alternative risk-adjustment technique that addresses both measured and unmeasured confounding (18–20).

Methods

Inclusion & exclusion criteria

The study population consisted of all patients included in the NCDR CathPCI Registry between July 2009 and September 2011 with an acute coronary syndrome (unstable angina, non–ST-segment elevation MI, or ST-segment elevation MI) undergoing an initial PCI. The beginning of the study period was defined by the adoption of version 4.0 of the registry data collection, and the end of the study period was defined by the most recent quarter of data available when the study protocol approval was granted by the NCDR and the study sponsor. Patients were excluded if they underwent elective PCI, were missing data on clinical presentation, or were enrolled at hospitals that treated fewer than 50 patients in a given calendar quarter.

Exposure

Patients were defined as having received a GPI if they had received a GPI at the time of the PCI or within 24 h before.

Outcomes and definitions

The primary outcome was all-cause in-hospital mortality. The registry includes data collected at participating hospitals by trained staff using standard elements. Data must pass quality standards for inclusion. The key secondary endpoint was major bleeding, defined as any access or nonaccess site bleeding requiring a transfusion, prolonged hospital stay, or leading to a decrease in hemoglobin >3.0 g/dl. Access site bleeding was identified as external bleeding or a hematoma >2 cm for radial access or >10 cm for femoral access. Data definitions for bleeding and other registry-reported outcomes are available from the CathPCI Registry Web site (21).

Statistical analysis

Baseline patient characteristics, procedural data, medication use, and in-hospital outcomes were compared between patients who did and did not receive a GPI. Univariate comparisons of continuous variables were performed by unpaired t tests for normally distributed variables or Wilcoxon rank sum tests for nonnormally distributed variables. Categorical variables were compared with the chi-square or Fisher exact test, as appropriate. Risk-adjusted analyses for each outcome were performed using risk-adjustment techniques: MV adjustment, PM analysis, and the IVA method.

MV risk adjustment

Hierarchical logistic regression models including hospital as a random effect were used to estimate the adjusted odds ratio (OR) and 95% confidence interval (CI) associated with GPI use for each outcome. Because of the large sample size, statistical differences between the groups on univariate analysis was considered of limited clinical utility. Selection of the variables for inclusion in the risk-adjustment models was therefore made by consensus of the authors that variables were likely to have a meaningful impact on the outcomes of interest. In the event of disagreement, variables were included in the model. Variables in the MV model included age, sex, race, insurance status, current smoking, hypertension, dyslipidemia, family history of early coronary artery disease, previous MI, heart failure, valve surgery, PCI, coronary artery bypass grafting, stroke, peripheral arterial disease, hemodialysis, chronic lung disease, diabetes, ST-segment elevation MI at presentation, complex (type C) lesion, multivessel coronary artery disease, bifurcation lesion, non–ST-segment elevation MI, shock, cardiac arrest, recent heart failure (within 2 weeks), and use of bivalirudin, heparin (unfractionated or low molecular weight), arterial closure devices, and radial access.

PM analysis

For the PM analysis, we first used logistic regression to develop a propensity score reflecting the probability of receiving a GPI conditional on the same covariates used in the MV risk adjustment. We then matched each patient in the GPI group with a patient in the no GPI group using 1:1 nearest-neighbor matching without replacement, with a caliper width of 0.2 times the SD (22) and matched by hospital. The success of matching was examined using standardized differences between the 2 groups; a difference of <10% was considered acceptable. Finally, conditional logistic regression was used to compare the risk of outcomes with GPI use in the matched cohort.

The IVA method

In observational studies, treatment allocation is nonrandom and likely influenced by participant, physician, and/or hospital characteristics that may also affect treatment outcomes. In these cases, traditional regression-based models are used to account for these factors but are limited to available data and generate biased effect estimates as a result of residual confounding (18,23). In the IVA method, comparison groups are defined on the basis of the likelihood of receiving treatment, determined by a randomly distributed “instrument,” rather than the observed biased treatment received (which may be biased) (19,20,24–26). A commonly used instrument is one on the basis of a treatment-prescribing preference that meets 3 common assumptions: 1) it is strongly associated with the treatment of interest; 2) is not independently associated with the outcome of interest, other than as related to the treatment of interest; and 3) is unrelated to measured and unmeasured characteristics that may influence treatment selection and outcomes (24,27).

In keeping with a previous study (24), our choice of hospital-based preference is on the basis of the notion that variations in GPI use introduced by unmeasured hospital factors such as costs and formulary decisions are independent of perceived treatment impact on outcomes. To account for temporal change in hospital practices, we examined the use of GPIs within the calendar quarter during which the patient underwent PCI and identified hospitals with consistently low versus high rates of GPI use. Patients treated at a hospital with <10% GPI use were considered less likely to receive a GPI, whereas those patients treated at hospitals with >70% GPI use were considered more likely. Patients treated at hospitals with GPI use between these 2 thresholds were excluded from IVA because GPI use in this intermediate range may be driven by patient or disease characteristics that also influence outcomes (24).

We then performed IVA using 2-stage linear regression to estimate absolute risk differences in outcomes by GPI use. These models were also adjusted for hospital characteristics (region, government vs. university vs. private, rural vs. urban, annual PCI volume). Finally, to assess the quality of the instrument, we performed a “falsification analysis” to demonstrate a lack of association between the instrument (high vs. low likelihood of GPI exposure) and acute kidney injury (which would not be expected to vary systematically on the basis of GPI exposure) (28). The results of each analysis (unadjusted, MV, PM, and instrumental variable) are presented as both absolute risk differences and relative risks (RRs) to allow direct comparison of results (26).

To evaluate the extent to which unmeasured confounding could explain the difference in outcomes, we also used the method of Lin et al. (29) to assess whether the observed differences in death and bleeding would be explained by the existence of an unmeasured variable with differential distribution between patients on the basis of GPI exposure.

Subgroup analyses

Stratified analyses were performed to examine the association between GPI use and clinical outcomes within pre-specified subgroups and by adding interaction terms to the MV logistic regression analyses to test whether the treatment effect differed by subgroup. Subgroups were selected to focus on patient types that would be expected to derive particular benefit (or harm) from GPI use and included the type of presentation (ST-segment elevation MI vs. unstable angina/non–ST-segment elevation MI), type of anticoagulant agent used (bivalirudin vs. heparin), and access site (radial vs. femoral). In addition, patients were categorized according to their predicted risk of bleeding and the tertile of predicted risk of in-hospital mortality on the basis of validated risk models (30–32). These analyses were confined to the primary and key secondary endpoints (mortality and major bleeding). This study was approved by the Saint Luke’s Hospital Institutional Review Board.

Results

Patient population

Between July 2009 and September 2011, there were 1,465,498 PCIs included in the CathPCI Registry. After excluding 407,942 elective procedures and 86,691 performed at low-volume sites, 970,865 procedures constituted the analytic cohort for our study. Within this cohort, GPIs were administered to 326,283 (GPI group), leaving 644,582 patients unexposed (no GPI group). After 1:1 propensity matching, the analytic cohort consisted of 135,620 patients in each treatment group.

Baseline characteristics of the overall and matched patient populations are summarized in Table 1. Patients treated with GPIs were older and less likely female. They were more likely to be uninsured and current smokers. Hypertension and dyslipidemia were less common among GPI-treated patients. There was also a lower prevalence of previous MI, congestive heart failure or previous coronary revascularization, hemodialysis, cerebral or peripheral arterial disease, diabetes, and use of antianginal medications in the GPI group. Clopidogrel was the P2Y12 antagonist used in the majority of patients, although a minority of patients were treated with prasugrel (11.1% GPI vs. 10.3% no GPI). Antecedent shock and cardiac arrest were more common in GPI patients, as was MI (with or without ST-segment elevation).

View this table:
Table 1

Baseline Clinical, Anatomic, and Procedural Characteristics Before and After Propensity Matching

There were several differences in anatomic and procedural characteristics between groups. Specifically, the GPI group had longer stents with larger diameters, more complex lesions, less bivalirudin use, and fewer vascular closure devices. There were no differences in the prevalence of multivessel coronary artery disease, bifurcation lesions, saphenous vein grafts as target lesions, or use of radial access. After propensity matching, differences between the GPI and no GPI groups were attenuated.

In-hospital outcomes

Unadjusted outcomes are summarized in Table 2. Virtually all endpoints including mortality and major bleeding were increased in patients who received GPIs. After adjustment for baseline factors using multiple logistic regression, in-hospital mortality was lower for the GPI group (adjusted OR: 0.83, 95% CI: 0.79 to 0.88; p < 0.001) (Table 2). For major bleeding (adjusted OR: 1.56, 95% CI: 1.51 to 1.61; p < 0.001), the adjusted ORs remained higher for the GPI group, although the RRs were attenuated substantially. The incidence of nonaccess site bleeding was higher in the GPI group (42.5% of bleeds vs. 33.9% of bleeds in non-GPI patients, p < 0.001). To further investigate the effect of GPIs on bleeding on the basis of radial or femoral access for PCI, separate subgroup analyses were performed using the outcomes of access site and nonaccess site bleeding on the basis of access site. GPI use increased the risk for both access site bleeds (radial access, OR: 1.85, 95% CI: 1.35 to 2.55; femoral access, OR: 1.71, 95% CI: 1.63 to 1.78; interaction p value = 0.372) and nonaccess site bleeds (radial access, OR: 2.93, 95% CI: 2.18 to 3.94; femoral access, OR: 2.19, 95% CI: 2.06 to 2.33; interaction p value = 0.331).

View this table:
Table 2

Unadjusted and Risk-Adjusted In-Hospital Outcomes

The results of the PM analysis are summarized in Table 3. After matching, there was excellent balance between groups with standardized risk differences <10% for all covariates. The PM analysis demonstrated a statistically significant mortality reduction associated with GPI use (2.1 vs. 2.3%, OR: 0.90, p < 0.001). The rate of major bleeding (3.1 vs. 1.7%, OR: 1.83, p < 0.001) remained higher among the GPI group.

View this table:
Table 3

In-Hospital Outcomes: Propensity-Matched Analysis

The IVA method demonstrated that GPI use was associated with a modest but statistically significant reduction in mortality (absolute risk difference: −0.39% [95% CI: −0.04% to −0.70%], RR: 0.72 [95% CI: 0.40 to 0.97]; p = 0.03) (Table 4). The IVA also demonstrated that GPI use was associated with major bleeding (absolute risk difference: 0.80% [95% CI: 0.4 to 1.7%], RR: 1.53 [95% CI: 1.27 to 2.13]; p < 0.001).

View this table:
Table 4

Instrumental Variable Analysis Results and Comparison of ARD and RR Differences

Subgroup analyses

The subgroup analyses are summarized in Figures 1 and 2. There were significant interactions between patient characteristics and the benefit of GPI (Figure 1). Specifically, GPI use was associated with reduced mortality in patients with ST-segment elevation MI (vs. unstable angina/non–ST-segment elevation MI), in those who did not receive bivalirudin, and in patients at high predicted mortality risk. However, GPI use was associated with higher mortality with non–ST-segment elevation MI and in bivalirudin-treated patients. There was no significant interaction between access site and mortality. With respect to major bleeding, there were quantitative interactions according to the type of presentation and the use of bivalirudin (Figure 2). Nonetheless, the adjusted OR for the association between GPI use and major bleeding remained >1.5 for all subgroups.

Figure 1
Figure 1

Subgroup Analysis: In-Hospital Mortality

Dot-whisker plot demonstrating the association of GPI use with in-hospital mortality. CI = confidence interval; GPI = glycoprotein IIb/IIIa inhibitor; NSTEMI = non–ST-segment elevation myocardial infarction; OR = odds ratio; STEMI = ST-segment elevation myocardial infarction.

Figure 2
Figure 2

Subgroup Analysis: Major Bleeding

Dot-whisker plot demonstrating the association of GPI use with major bleeding. Abbreviations as in Figure 1.

Falsification analysis

In unadjusted analyses, there was a correlation between GPI exposure and the pre-specified falsification endpoint, acute kidney injury (unadjusted OR: 1.31, 95% CI: 1.27 to 1.36; p < 0.0001). However, this association was abolished in MV analysis (adjusted OR: 1.02, 95% CI: 0.99 to 1.06) and nearly completely attenuated in PM analysis (OR: 1.04, 95% CI: 1.01 to 1.06) and IVA (absolute risk difference: −0.007%, 95% CI: −0.002% to −0.013%).

Unmeasured cofounding

Sensitivity analysis to assess the potential effect of a hypothetical unmeasured cofounder reveals that if a variable had an OR of 2 to cause bleeding and was present in 10% of patients in the no GPI group, it would have to be present in 67% of GPI patients to explain the results of this study. Similarly, if it were present in 20% or 30% of no GPI patients, the prevalence in GPI patients would need to be 82% and 97%, respectively, to account for the observed differences (Online Figure 1). Analysis of a hypothetical cofounder with an OR of 2 for mortality revealed similar results. A prevalence of 10%, 20%, or 30% in GPI patients would require a 25%, 37%, and 48% prevalence in no GPI patients to account for these results (Online Figure 2).

Discussion

In this study of nearly 1 million patients undergoing PCI for acute coronary syndrome, we found that GPI use was associated with lower mortality but higher rates of major bleeding. These results were similar regardless of the analytic approach used to adjust for potential differences between those who did and did not receive a GPI. Subgroup analysis demonstrated that the mortality benefit of GPIs in this setting was restricted to patients with ST-segment elevation MI and those treated with heparin anticoagulation. There was no evidence of differential benefit or harm from GPI use according to access site.

When the study population is examined, there are multiple differences between patients treated with GPIs and those not treated with GPIs. Patients who received GPIs tended to be sicker and more complex. As a result, in unadjusted analyses, the use of GPIs was associated with higher mortality. However, in adjusted analyses using MV risk adjustment, propensity matching, and IVA, the use of GPI was associated with reduced mortality. RR reductions ranged from 10% to 28%. In contrast, GPI use was associated with an increased risk of major bleeding in unadjusted and risk-adjusted analyses, although the risk was attenuated in adjusted analyses. The increased mortality among non−ST-segment elevation acute coronary syndromes was also an unexpected finding that warranted further evaluation. To more thoroughly investigate the potential effect of unmeasured confounding in this analysis, we applied the IVA method to this subgroup. The mortality risk was tempered in the IVA: absolute risk difference 0.28 (95% CI: 0.03 to 0.53), p = 0.082.

Whether these findings truly represent a causal relationship cannot be addressed through observational studies. Both MV risk adjustment and propensity matching are accepted methods to adjust for measured confounders, but cannot adjust for unmeasured factors. However, IVA is a technique that should adjust for both measured and unmeasured confounding (33), provided an appropriate instrument is available. In our study, we took advantage of the large number of centers that participate in the NCDR CathPCI Registry to use site-level GPI use as the instrument. Thus, as long as patients do not differ systematically across PCI centers (an unlikely scenario), patients are effectively “randomized” across centers with high versus low GPI use, and any outcome differences identified by the IVA should be attributable to the differential use of GPI. It is important to keep in mind that observed differences in outcome only apply to the “marginal” patients who would have received a GPI at a high-use site but not at a low-use site. Moreover, it is possible that a high rate of GPI use is a site-level marker for other aspects of care that were not assessed directly but are associated with mortality. Finally, the fact that the results of the same analyses applied to the falsification endpoint (acute kidney injury) demonstrated that a neutral effect further supports the validity of our findings (34).

Comparison with previous studies

Over the past 2 decades since their introduction, numerous studies have examined the benefits and risks of GPIs in the acute coronary syndrome PCI setting. In a meta-analysis of more than 30,000 patients with acute coronary syndromes (24% of whom underwent PCI), Boersma et al. (35) found that treatment with GPIs led to a 9% reduction in the RR of death or MI with a concurrent 1% absolute increase in major bleeding. More recently, Sethi et al. (36) performed a meta-analysis of randomized trials of GPI use in patients undergoing primary PCI. In this study of more than 7,000 patients, GPI use was associated with a 25% reduction in mortality. Meta-regression suggested that the benefits of GPI use were confined to patients at highest risk of mortality. Finally, Winchester et al. (37) performed a meta-analysis of GPI use in acute coronary syndromes and PCI on the basis of trials performed in the contemporary era of stents and dual antiplatelet therapy. Among acute coronary syndrome patients, GPI use was associated with a significant reduction in nonfatal MI and an increase in minor bleeding but no differences in mortality or major bleeding. Thus, although there are numerous differences in patient populations, timing of drug administration, and concomitant medical therapy, most studies have tended to demonstrate that GPI use in acute coronary syndromes and PCI (particularly in the highest risk patients) leads to modest benefits in terms of ischemic complications with a concomitant increase in bleeding—generally consistent with the results of our study. The novel findings of the current study are that these results are reproduced in a large observational database that chronicles the clinical care of most patients undergoing PCI in the United States in a contemporary setting and that established bleeding and mortality risk models define those most likely to benefit from aggressive pharmacotherapy such as GPI use.

Several large-scale clinical trials have evaluated the role of GPIs in invasive management of acute coronary syndromes. There was no difference in outcomes among 9,492 heparin-treated acute coronary syndrome patients randomized to receive GPIs before versus after coronary angiography (16). A trial of 13,819 patients with non–ST-segment elevation acute coronary syndrome randomized to heparin + a GPI, bivalirudin monotherapy, or bivalirudin + a GPI revealed that ischemic outcomes were similar across all 3 regimens, but bivalirudin monotherapy was associated with less major bleeding (38). Finally, bivalirudin monotherapy was compared with heparin + GPI in 3,602 patients undergoing primary PCI (39). Although overall ischemic outcomes were similar, bivalirudin was associated with a 40% reduction in major bleeding and a 34% reduction in 30-day mortality.

Clinical implications

On the basis of our findings, GPI use in patients undergoing PCI for acute coronary syndromes appears to carry both benefits and risks. Thus, it is not possible to derive a “one-size-fits-all” recommendation on the basis of our findings. Rather, it would appear prudent to reserve GPI use for patients who are at high risk of early ischemic complications and for heparin-treated patients. On the other hand, in patients for whom the predominant short-term risk is major bleeding, avoidance of GPIs unless faced with extreme thrombotic risk would appear to be reasonable. Finally, transradial versus transfemoral access should not affect the decision to use GPIs because outcomes were consistent regardless of access site.

Study limitations

There are several limitations to be considered when interpreting this study. First, given the study’s observational nature, our results may be influenced by unmeasured confounding and should be considered hypothesis generating only. Nonetheless, our findings were consistent across 3 alternative approaches to risk adjustment. It is also reassuring that our results are consistent with those of previous meta-analyses. Second, we limited our analysis to in-hospital outcomes, and it is possible that downstream effects of in-hospital treatments (and complications) may affect long-term outcomes. Third, the NCDR CathPCI Registry does not capture the specific GPI used or the timing of its use; we are therefore unable to comment on the potential of differential outcomes on the basis of the use of a particular GPI or differences with upstream versus intraprocedural GPI use. Fourth, patients treated with bivalirudin may have been treated with a bailout GPI with a different threshold than patients treated with heparin-based anticoagulation, potentially biasing subgroup analysis. Finally, although these data span the period July 2009 to September 2011, it is possible that novel PCI devices, pharmacological adjuncts, and techniques alter these results. For instance, ticagrelor was not approved for use during the study period, and prasugrel was only used in a minority of patients (11.1% of the GPI group and 10.3% of the no GPI group).

Conclusions

In PCI for unstable angina/non–ST-segment elevation MI and ST-segment elevation MI, GPI use is associated with lower mortality but higher bleeding rates. These results were confirmed with multiple analytic methods. Survival was better with GPI use in important patient subgroups including ST-segment elevation MI and heparin-based anticoagulation, whereas the benefits of GPI use were consistent regardless of access site.

Perspectives

WHAT IS KNOWN? This study examines the use of GPIs in contemporary PCI for acute coronary syndromes, including unstable angina, non–ST-segment elevation MI and ST-segment elevation MI. Previous studies established the utility of GPI in reducing ischemic complications in this population, but few studies evaluated current treatment patterns and outcomes in clinical practice. Approximately one-third of these patients are treated with GPIs in this cohort, and ischemic outcomes (including mortality) were reduced, whereas bleeding was increased.

WHAT IS NEW? Importantly, subgroup analyses demonstrated that the mortality benefits of GPI use were restricted to several key subgroups including patients treated with heparin anticoagulation (but not bivalirudin), patients undergoing PCI for ST-segment elevation MI (primary PCI), and patients at high risk of mortality.

WHAT IS NEXT? Future research is needed to define the utility of GPIs in modern PCIs performed with patients on newer antiplatelet and anticoagulant regimens.

Appendix

Online Figures 1 and 2 [S1936879815011358_mmc1.docx]

Footnotes

  • This project was funded by an investigator-initiated research grant from Merck Pharmaceuticals. Dr. Cohen is a consultant for and has received research support from Eli Lilly, AstraZeneca, Merck, Abbott Vascular, Medtronic, and Boston Scientific; and has received speaker fees from Eli Lilly and AstraZeneca. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.

Abbreviations and Acronyms
CI
confidence interval
GPI
glycoprotein IIb/IIIa inhibitor
IVA
instrumental variable analysis
MI
myocardial infarction
MV
multivariable
NCDR
National Cardiovascular Data Registry
PCI
percutaneous coronary intervention
PM
propensity matched
OR
odds ratio
RR
relative risk
  • Received January 20, 2015.
  • Revision received April 13, 2015.
  • Accepted April 23, 2015.

References

  1. ↵
    1. Canadian Cardiovascular Society, American Academy of Family Physicians, American College of Cardiology/American Heart Association,
    2. Antman E.M.,
    3. Hand M.,
    4. Armstrong P.W.,
    5. et al.
    (2008) 2007 Focused Update of the ACC/AHA 2004 Guidelines for the Management of Patients With ST-Elevation Myocardial Infarction: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol 51:210–247.
    OpenUrlFREE Full Text
    1. Jneid H.,
    2. Anderson J.L.,
    3. Wright R.S.,
    4. et al.
    (2012) 2012 ACCF/AHA focused update of the guideline for the management of patients with unstable angina/non-ST-elevation myocardial infarction (updating the 2007 guideline and replacing the 2011 focused update): a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol 60:645–681.
    OpenUrlFREE Full Text
    1. O'Gara P.T.,
    2. Kushner F.G.,
    3. Ascheim D.D.,
    4. et al.
    (2013) 2013 ACCF/AHA guideline for the management of ST-elevation myocardial infarction: a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol 61:e78–e140.
    OpenUrlFREE Full Text
  4. ↵
    1. Kadakia M.B.,
    2. Desai N.R.,
    3. Alexander K.P.,
    4. et al.
    (2010) Use of anticoagulant agents and risk of bleeding among patients admitted with myocardial infarction: a report from the NCDR ACTION Registry–GWTG (National Cardiovascular Data Registry Acute Coronary Treatment and Intervention Outcomes Network Registry–Get With the Guidelines). J Am Coll Cardiol Intv 3:1166–1177.
    OpenUrl
    1. Mudrick D.W.,
    2. Chen A.Y.,
    3. Roe M.T.,
    4. et al.
    (2010) Changes in glycoprotein IIb/IIIa inhibitor excess dosing with site-specific safety feedback in the Can Rapid risk stratification of Unstable angina patients Suppress ADverse outcomes with Early implementation of the ACC/AHA guidelines (CRUSADE) initiative. Am Heart J 160:1072–1078.
    OpenUrlCrossRefPubMed
  6. ↵
    (1994) Use of a monoclonal antibody directed against the platelet glycoprotein IIb/IIIa receptor in high-risk coronary angioplasty. The EPIC Investigation. N Engl J Med 330:956–961.
    OpenUrlCrossRefPubMed
  7. (1997) Platelet glycoprotein IIb/IIIa receptor blockade and low-dose heparin during percutaneous coronary revascularization. N Engl J Med 336:1689–1696.
    OpenUrlCrossRefPubMed
  8. (1998) Randomised placebo-controlled and balloon-angioplasty-controlled trial to assess safety of coronary stenting with use of platelet glycoprotein-IIb/IIIa blockade. Lancet 352:87–92.
    OpenUrlCrossRefPubMed
    1. O'Shea J.C.,
    2. Hafley G.E.,
    3. Greenberg S.,
    4. et al.
    (2001) Platelet glycoprotein IIb/IIIa integrin blockade with eptifibatide in coronary stent intervention: the ESPRIT trial: a randomized controlled trial. JAMA 285:2468–2473.
    OpenUrlCrossRefPubMed
  10. ↵
    1. Lindsey J.B.,
    2. Kennedy K.F.,
    3. Stolker J.M.,
    4. et al.
    (2011) Prognostic implications of creatine kinase-MB elevation after percutaneous coronary intervention: results from the Evaluation of Drug-Eluting Stents and Ischemic Events (EVENT) registry. Circ Cardiovasc Interv 4:474–480.
    OpenUrlAbstract/FREE Full Text
    1. Lindsey J.B.,
    2. Marso S.P.,
    3. Pencina M.,
    4. et al.
    (2009) Prognostic impact of periprocedural bleeding and myocardial infarction after percutaneous coronary intervention in unselected patients: results from the EVENT (evaluation of drug-eluting stents and ischemic events) registry. J Am Coll Cardiol Intv 2:1074–1082.
    OpenUrl
    1. Moussa I.D.,
    2. Klein L.W.,
    3. Shah B.,
    4. et al.
    (2013) Consideration of a new definition of clinically relevant myocardial infarction after coronary revascularization: an expert consensus document from the Society for Cardiovascular Angiography and Interventions (SCAI). J Am Coll Cardiol 62:1563–1570.
    OpenUrlFREE Full Text
  13. ↵
    1. King S.B. 3rd.,
    2. Smith S.C. Jr..,
    3. Hirshfeld J.W. Jr..,
    4. et al.
    (2008) 2007 Focused update of the ACC/AHA/SCAI 2005 Guideline Update for Percutaneous Coronary Intervention: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol 51:172–209.
    OpenUrlFREE Full Text
  14. ↵
    1. Tomlinson G.,
    2. Detsky A.S.
    (2010) Composite end points in randomized trials: there is no free lunch. JAMA 303:267–268.
    OpenUrlCrossRefPubMed
  15. ↵
    1. Kandzari D.E.,
    2. Roe M.T.,
    3. Chen A.Y.,
    4. et al.
    (2005) Influence of clinical trial enrollment on the quality of care and outcomes for patients with non-ST-segment elevation acute coronary syndromes. Am Heart J 149:474–481.
    OpenUrlCrossRefPubMed
  16. ↵
    1. Giugliano R.P.,
    2. White J.A.,
    3. Bode C.,
    4. et al.
    (2009) Early versus delayed, provisional eptifibatide in acute coronary syndromes. N Engl J Med 360:2176–2190.
    OpenUrlCrossRefPubMed
    1. Kastrati A.,
    2. Neumann F.J.,
    3. Schulz S.,
    4. et al.
    (2011) Abciximab and heparin versus bivalirudin for non-ST-elevation myocardial infarction. N Engl J Med 365:1980–1989.
    OpenUrlCrossRefPubMed
  18. ↵
    1. Bowden R.J.,
    2. Turkington D.A.
    (1984) Instrumental Variables (Cambridge University Press, Cambridge, MA).
  19. ↵
    1. McClellan M.,
    2. McNeil B.J.,
    3. Newhouse J.P.
    (1994) Does more intensive treatment of acute myocardial infarction in the elderly reduce mortality? Analysis using instrumental variables. JAMA 272:859–866.
    OpenUrlCrossRefPubMed
    1. Venkitachalam L.,
    2. Lei Y.,
    3. Magnuson E.A.,
    4. et al.
    (2011) Survival benefit with drug-eluting stents in observational studies: fact or artifact? Circ Cardiovasc Qual Outcomes 4:587–594.
    OpenUrlAbstract/FREE Full Text
  21. ↵
    CathPCI Registry. Available at: https://www.ncdr.com/webncdr/cathpci/home/datacollection. Accessed September 10, 2015.
  22. ↵
    1. Rosenbaum P.R.,
    2. Rubin D.B.
    (1985) Constructing a control group using multivariate matched sampling methods that incorporate the propensity score. Am Stat 39:6.
    OpenUrl
    1. Newhouse J.P.,
    2. McClellan M.
    (1998) Econometrics in outcomes research: the use of instrumental variables. Annu Rev Public Health 19:17–34.
    OpenUrlCrossRefPubMed
  24. ↵
    1. Rassen J.A.,
    2. Mittleman M.A.,
    3. Glynn R.J.,
    4. Alan Brookhart M.,
    5. Schneeweiss S.
    (2009) Safety and effectiveness of bivalirudin in routine care of patients undergoing percutaneous coronary intervention. Eur Heart J 31:561–572.
    OpenUrlPubMed
    1. Ryan J.W.,
    2. Peterson E.D.,
    3. Chen A.Y.,
    4. et al.
    (2005) Optimal timing of intervention in non-ST-segment elevation acute coronary syndromes: insights from the CRUSADE (Can Rapid risk stratification of Unstable angina patients Suppress ADverse outcomes with Early implementation of the ACC/AHA guidelines) Registry. Circulation 112:3049–3057.
    OpenUrlAbstract/FREE Full Text
  26. ↵
    1. Stukel T.A.,
    2. Fisher E.S.,
    3. Wennberg D.E.,
    4. Alter D.A.,
    5. Gottlieb D.J.,
    6. Vermeulen M.J.
    (2007) Analysis of observational studies in the presence of treatment selection bias: effects of invasive cardiac management on AMI survival using propensity score and instrumental variable methods. JAMA 297:278–285.
    OpenUrlCrossRefPubMed
    1. Brookhart M.A.,
    2. Schneeweiss S.
    (2007) Preference-based instrumental variable methods for the estimation of treatment effects: assessing validity and interpreting results. Int J Biostat 3:14.
    OpenUrl
  28. ↵
    1. Morgan S.L.,
    2. Harding D.J.
    (2006) Matching estimators of causal effects: prospects and pitfalls in theory and practice. Sociol Methods Res 35:3–60.
    OpenUrlCrossRef
  29. ↵
    1. Lin D.Y.,
    2. Psaty B.M.,
    3. Kronmal R.A.
    (1998) Assessing the sensitivity of regression results to unmeasured confounders in observational studies. Biometrics 54:948–963.
    OpenUrlCrossRefPubMed
  30. ↵
    1. Mehta S.K.,
    2. Frutkin A.D.,
    3. Lindsey J.B.,
    4. et al.
    (2009) Bleeding in patients undergoing percutaneous coronary intervention: the development of a clinical risk algorithm from the National Cardiovascular Data Registry. Circ Cardiovasc Interv 2:222–229.
    OpenUrlAbstract/FREE Full Text
    1. Rao S.V.,
    2. McCoy L.A.,
    3. Spertus J.A.,
    4. et al.
    (2013) An updated bleeding model to predict the risk of post-procedure bleeding among patients undergoing percutaneous coronary intervention: a report using an expanded bleeding definition from the National Cardiovascular Data Registry CathPCI Registry. J Am Coll Cardiol Intv 6:897–904.
    OpenUrl
    1. Peterson E.D.,
    2. Dai D.,
    3. DeLong E.R.,
    4. et al.
    (2010) Contemporary mortality risk prediction for percutaneous coronary intervention: results from 588,398 procedures in the National Cardiovascular Data Registry. J Am Coll Cardiol 55:1923–1932.
    OpenUrlFREE Full Text
  33. ↵
    1. Tu J.V.,
    2. Ko D.T.
    (2008) Ecological studies and cardiovascular outcomes research. Circulation 118:2588–2593.
    OpenUrlFREE Full Text
  34. ↵
    1. Prasad V.,
    2. Jena A.B.
    (2013) Prespecified falsification end points: can they validate true observational associations? JAMA 309:241–242.
    OpenUrlCrossRefPubMed
  35. ↵
    1. Boersma E.,
    2. Harrington R.A.,
    3. Moliterno D.J.,
    4. White H.,
    5. Simoons M.L.
    (2002) Platelet glycoprotein IIb/IIIa inhibitors in acute coronary syndromes. Lancet 360:342–343.
    OpenUrlCrossRefPubMed
  36. ↵
    1. Sethi A.,
    2. Bajaj A.,
    3. Bahekar A.,
    4. et al.
    (2013) Glycoprotein IIb/IIIa inhibitors with or without thienopyridine pretreatment improve outcomes after primary percutaneous coronary intervention in high risk patients with ST elevation myocardial infarction–a meta-regression of randomized controlled trials. Catheter Cardiovasc Interv 82:171–181.
    OpenUrlPubMed
  37. ↵
    1. Winchester D.E.,
    2. Brearley W.D.,
    3. Wen X.,
    4. Park K.E.,
    5. Bavry A.A.
    (2012) Efficacy and safety of unfractionated heparin plus glycoprotein IIb/IIIa inhibitors during revascularization for an acute coronary syndrome: a meta-analysis of randomized trials performed with stents and thienopyridines. Clin Cardiol 35:93–100.
    OpenUrlCrossRefPubMed
  38. ↵
    1. Stone G.W.,
    2. McLaurin B.T.,
    3. Cox D.A.,
    4. et al.
    (2006) Bivalirudin for patients with acute coronary syndromes. N Engl J Med 355:2203–2216.
    OpenUrlCrossRefPubMed
  39. ↵
    1. Stone G.W.,
    2. Witzenbichler B.,
    3. Guagliumi G.,
    4. et al.
    (2008) Bivalirudin during primary PCI in acute myocardial infarction. N Engl J Med 358:2218–2230.
    OpenUrlCrossRefPubMed
View Abstract
Back to top

Toolbox

Print PDF
Email
Citation

Metrics

Advertisement

Article Outline

Figures in Article

Figures

  • Figure 1
  • Figure 2

Supplementary Materials

Similar Articles

Advertisement