Catheterization and Cardiovascular Interventions 85:371–379 (2015)
Femoral Vascular Closure Device Use, Bivalirudin Anticoagulation, and Bleeding after Primary Angioplasty for STEMI: Results from the HORIZONS-AMI Trial Timothy A. Sanborn,1* MS, MD, Matthew I. Tomey,2 MD, Roxana Mehran,2,3 MD, ne reux,3,4,5 MD, Bernhard Witzenbichler,6 MD, Sorin J. Brener,3,7 MD, Philippe Ge Ajay J. Kirtane,3,4 MD, SM, Thomas C. McAndrew,3 MS, Ran Kornowski,8 MD, Dariusz Dudek,9 MD, Eugenia Nikolsky,10 MD, PhD, and Gregg W. Stone,3,4 MD Objective: To assess the relationship of femoral vascular closure device (VCD) use to bleeding and ischemic events in patients undergoing primary percutaneous coronary intervention (PCI) for ST-segment elevation myocardial infarction (STEMI) via different anticoagulation strategies. Background: It is unknown whether femoral VCD reduce major bleeding after primary PCI for STEMI using bivalirudin anticoagulation. Methods: We compared VCD-treated patients with propensity-matched controls in the HORIZONS-AMI trial with respect to net adverse clinical events (NACE), defined as the composite of major bleeding unrelated to coronary artery bypass graft surgery (CABG) and major adverse cardiac events (comprised of death, reinfarction, ischemia-driven target vessel revascularization, and stroke), at 30 days and 1 year. Results: Among 3,602 patients enrolled in HORIZONS-AMI, 2,948 underwent primary PCI via femoral arterial access and 896 (30%) received VCDs, of whom 642 were included in our model along with 642 propensity-matched controls. At 30 days, VCD-treated patients had significantly less NACE (6.7% vs. 10.8%, HR: 0.61, 95% CI: 0.42–0.89, P 5 0.009), driven by a lower rate of non-CABG related major bleeding (5.0% vs. 8.1%, HR: 0.61, 95% CI: 0.39–0.94, P 5 0.02). Bleeding reduction was maintained at one year and consistent in magnitude regardless of randomization to bivalirudin or unfractionated heparin plus a glycoprotein IIb/IIIa inhibitor (P for interaction 5 0.84). Conclusion: In patients undergoing transfemoral primary PCI for STEMI, VCD use was associated with significantly
1
NorthShore University HealthSystem, Evanston, Illinois Icahn School of Medicine at Mount Sinai, New York, New York 3 Cardiovascular Research Foundation, New York, New York 4 Columbia University Medical Center, New York, New York 5 ^ pital du Sacre -Coeur de Montre al, Montre al, Que bec, Ho Canada 6 Helios Amper-Klinikum, Dachau, Germany 7 New York Methodist Hospital, Brooklyn, New York 8 Rabin Medical Center, Petach-Tikva, Israel 9 Department of Interventional Cardiology, Jagiellonian University Medical College, Krakow, Poland 10 Rambam Health Care Campus and the Technion – Israel Institute of Technology, Haifa, Israel.
Medtronic, Abbott Vascular, Abiomed, St. Jude Medical, Vascular Dynamics, and Eli Lilly. Dariusz Dudek: Consulting and lecture fees: Abbott, Adamed, Adyton Medical Polska, Abiomed Europe, AstraZeneca, Biotronik, Balton, Bayer, BBraun, BioMatrix, Boston Scientific, Boehringer Ingelheim, Bracco, Bristol-Myers Squibb, Comesa Polska, Cordis, Cook, Covidien Polska Sp. z o. o., DRG MedTek, Eli Lilly, EuroCor, Hammermed, GE Healthcare, Glaxo, InspireMD, Iroko Cardio International, Medianet Sp. z o.o., Medtronic, The Medicines Company, Meril Life Sciences, Merck Sharp & Dohme, Orbus-Neich, Pfizer, Possis, ProCardia Medical, Promed, REVA Medical, Sanofi-Aventis, Siemens, Solvay, Stentys, St. Jude Medical, Terumo, Tyco, Volcano. Other authors: None. The HORIZONS-AMI Trial was funded by Boston Scientific and The Medicines Company.
Conflict of interest: Roxana Mehran: Institutional Research Grant Support: The Medicines Company, Bristol-Myers Squibb/Sanofi, Eli Lilly and Company/Daiichi-Sankyo, Regado Biosciences, and STENTYS; consulting: Abbott Vascular, AstraZeneca, Boston Scientific, Covidien, CSL Behring, Janssen Pharmaceuticals, Maya Medical, and Merck & Co.; advisory board: Covidien, Janssen Pharmaceuticals, Merck, and Sanofi. Philippe Genereux: Speaker honoraria—Abbott Vascular, CSI; consultant—CSI. Bernhard Witzenbichler: Consultant—Volcano. Ajay J. Kirtane: Institutional research grants to Columbia University from Boston Scientific,
*Correspondence to: Timothy A. Sanborn, MS, MD, Cardiology Division, NorthShore University HealthSystem, 2650 Ridge Ave., Walgreen Building, Third Floor, Evanston, IL 60201. E-mail: [email protected]
2
C 2014 Wiley Periodicals, Inc. V
Received 27 August 2014; Revision accepted 29 August 2014 DOI: 10.1002/ccd.25663 Published online 1 September 2014 in Wiley Online Library (wileyonlinelibrary.com)
372
Sanborn et al. lower non-CABG major bleeding irrespective of anticoagulation strategy.
C 2014 Wiley V
Periodicals, Inc.
Key words: vascular closure device; primary percutaneous coronary intervention; ST-segment elevation myocardial infarction; bivalirudin; heparin
INTRODUCTION
Femoral vascular closure devices (VCD) improve patient comfort, expedite ambulation, and reduce access site-related bleeding complications in patients with acute coronary syndromes managed with percutaneous coronary intervention (PCI) [1–4]. Their role as a “bleeding avoidance strategy” has remained controversial, however, in comparison with other strategies including radial access and bivalirudin anticoagulation [5]. Factors contributing to this uncertainty include perceived risk of vascular complications, drawn from early VCD experience [6], and increased attention to the importance of non-access site related bleeding [7]. In the setting of acute ST-segment elevation myocardial infarction (STEMI), growing evidence shows superiority of radial over femoral access as a strategy to reduce bleeding and vascular complications of primary PCI [8–11]. Uptake of radial access has been heterogeneous, however, and indeed, femoral access still predominates in the United States. Accordingly, the challenge of avoiding bleeding after transfemoral primary PCI remains highly relevant. The Harmonizing Outcomes with Revascularization and Stents in Acute Myocardial Infarction (HORIZONS-AMI) trial, which primarily involved femoral access, established bivalirudin anticoagulation as an effective bleeding avoidance strategy in primary PCI [12]. It has yet to be demonstrated whether femoral VCD confer additional bleeding reduction in this population on a backdrop of bivalirudin anticoagulation. To address this question, we performed a propensitymatched comparison of patients treated with or without VCD in the HORIZONS-AMI trial. METHODS Study Design and Participants
We identified patients for the present analysis from the HORIZONS-AMI trial. Briefly, HORIZONS-AMI was a prospective, open-label, randomized, multicenter trial enrolling 3,602 patients with STEMI who presented within 12 hr of symptom onset for primary PCI. Patients were randomized in a 22 factorial design to receive bivalirudin or unfractionated heparin plus a glycoprotein IIb/IIIa inhibitor (UFHþGPI) (1:1) and then, where eligible for PCI, to paclitaxel-eluting or bare metal stents (3:1). Clinical follow-up was con-
ducted at the prespecified time points of 30 days, 6 months, 1 year, 2 years, and 3 years. The study was approved by the institutional review board or ethics committee at every participating center and all patients provided written informed consent. Full details of the HORIZONS-AMI study protocol, inclusion and exclusion criteria have been reported previously (www.clinicaltrials.gov:NCT00433966) [13]. Of note, pertinent to the present analysis, VCD allocation was not randomized and rather left to the discretion of treating physicians. Patients with extensive peripheral vascular disease were excluded by design. For the present study, we restricted analysis to patients undergoing primary PCI via femoral arterial access with information available about VCD use. We did not exclude patients for whom external devices were used as adjuncts to manual hemostasis and analyzed these participants as part of the no VCD group. We excluded patients treated with intra-aortic balloon counterpulsation due to potential confounding of both VCD assignment and clinical outcomes. End Points and Definitions The primary outcome measure for this analysis was net adverse clinical events (NACE) at 30 days. NACE was defined as the composite of major adverse cardiac events (MACE, comprised of death, reinfarction, ischemia-driven target vessel revascularization, and stroke) and major bleeding unrelated to coronary artery bypass graft surgery (CABG). Key secondary endpoints included major bleeding and MACE at 30 days. All outcomes were followed to one year. Major bleeding was defined as intracranial or intraocular hemorrhage; bleeding at the access site, with a hematoma 5 cm or larger or requiring intervention; decrease in hemoglobin concentration by 4 g/dL without an overt bleeding source or 3 g/dL with an overt bleeding source; reoperation for bleeding; or blood transfusion. Bleeding was also assessed and adjudicated on the basis of the Thrombolysis In Myocardial Infarction (TIMI) and Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary Arteries (GUSTO) scales [14,15]. Detailed definitions of all study endpoints have been reported previously [13]). An independent clinical events committee blinded to treatment assignments adjudicated all end point events.
Catheterization and Cardiovascular Interventions DOI 10.1002/ccd. Published on behalf of The Society for Cardiovascular Angiography and Interventions (SCAI).
Vascular Closure Devices in STEMI
Statistical Analysis As a preliminary analysis, we divided participants into four groups based on randomization to bivalirudin or UFHþGPI and use or nonuse of VCD. We compared these four groups with respect to baseline characteristics, medication use, treatment strategies, and clinical outcomes. Categorical variables are presented as percentages and compared using the chi-squared test or Fisher’s exact test, as appropriate. Continuous variables are presented as median with interquartile range except where noted and compared using the Wilcoxon rank sum test. Event rates at follow-up are Kaplan– Meier estimates and compared using the log-rank test. All statistical tests were 2-tailed. P values less than 0.05 were deemed statistically significant. For the main analysis, in order to control for confounders influencing patient selection for VCD use, we developed a propensity model for VCD use. Analysis was conducted using greedy matching technique and involved several rounds of matching with caliper widths of 0.0001, 0.001, 0.01, and 0.1. No time window was considered in matching. Variables included in the propensity model included age, sex, body mass index, diabetes mellitus, current smoking, hypertension, peripheral vascular disease, prior myocardial infarction, prior PCI, United States location, bivalirudin use, GPI use, pre-randomization heparin use, final TIMI flow, baseline creatinine clearance, anemia, platelet count, warfarin use, clopidogrel loading dose, multivessel PCI, bare metal stent implantation, and left ventricular ejection fraction. Operator data was not available for inclusion in the model. We then compared VCDtreated patients with propensity-matched controls with respect to NACE, major bleeding, and MACE at 30 days and one year.
RESULTS
Of the 3,602 patients enrolled in HORIZONS-AMI, 3,134 underwent primary PCI via femoral arterial access. A single patient crossing over from radial to femoral access was not included in this analysis. After exclusion of 6 patients without information about VCD use and 80 additional patients treated with intra-aortic balloon counterpulsation, 2,948 patients remained for inclusion in the final analysis. Of these patients, 896 received VCD (30%), including 450 patients randomized to bivalirudin and 446 patients randomized to UFHþGPI. A summary of baseline features for participants treated with bivalirudin or UFHþGPI and with VCD or no VCD is presented in Table I. VCD-treated patients randomized to bivalirudin were less likely to have diabetes mellitus, peripheral vascular disease, re-
373
nal impairment, prior myocardial infarction, or prior PCI. VCD-treated patients, in general, had a lower prevalence of anemia and a higher likelihood of receiving pre-procedure heparin or the 600 mg clopidogrel loading dose. Groups were otherwise similar. The most commonly used VCD was Angio-Seal (St. Jude Medical, St. Paul, MN), followed by StarClose (Abbott Vascular, Santa Clara, CA) and Perclose (Abbott Vascular). Only six patients in total were treated with VasoSeal (DataScope, Montvale, NJ). Among patients not treated with VCD, 17% had an external FemoStop device, C-clamp or subcutaneous patch as an adjunct to manual hemostasis. Unadjusted ischemic and bleeding events at 30 days are presented in Table II. Rates of major bleeding and NACE differed significantly across groups. The highest rates of major bleeding and NACE were observed in participants treated with UFHþGPI and no VCD (8.5% and 11.7%, respectively). The lowest rates of major bleeding and NACE were observed in participants treated with bivalirudin and VCD (3.8% and 5.1%, respectively). Cross-group comparisons in bleeding were similar using TIMI and GUSTO scales. VCD-treated patients were next compared with propensity-matched controls. Upon matching, our model included 642 patients treated with VCD (72% of all VCD-treated patients) and 642 propensity-matched patients not treated with VCD. Groups were wellmatched (Table III). Comparisons of VCD-treated patients versus propensity-matched controls with respect to NACE, major bleeding, and MACE at 30 days and one year are presented in Fig. 1 panels A–F. At 30 days, VCD use was associated with a significantly lower incidence of NACE (6.7% vs. 10.8%, HR: 0.61, 95% CI: 0.42–0.89, P ¼ 0.009) and major bleeding (5.0% vs. 8.1%, HR: 0.61, 95% CI: 0.39–0.94, P ¼ 0.02); these differences persisted at one year. The main forms of major bleeding were drops in hemoglobin 4 g/dL without overt source, large hematomas, blood product transfusion, and drops in hemoglobin 3 g/dL with overt source (Fig. 2). Differences between VCD-treated patients and propensity-matched controls with respect to specific types of bleeding were not statistically significant. In order to assess whether the observed relationship between VCD use and major bleeding avoidance depended on anticoagulation strategy, we conducted formal interaction testing (Fig. 3). The magnitude of reduction in non-CABG related major bleeding was conserved regardless of randomization to bivalirudin or UFHþGPI, with no significant interaction. A trend toward lower incidence of MACE among VCD-treated patients at 30 days did not remain at one year. Review of specific ischemic events contributing
Catheterization and Cardiovascular Interventions DOI 10.1002/ccd. Published on behalf of The Society for Cardiovascular Angiography and Interventions (SCAI).
374
Sanborn et al.
TABLE I. Baseline, Clinical, and Procedural Characteristics Bivalirudin
Age, years Female sex BMI Diabetes mellitus PVD Prior MI Prior PCI Anemiaa CrCl, mL/min CrCl < 60 mL/min Closure device Angio-Seal VasoSeal Perclose StarClose Other No closure device Manual pressure only FemoStop C-clamp Subcutaneous patch Prolonged sheath dwell Other Pre-procedure heparin Bivalirudin in cath lab Clopidogrel loading dose 300 mg 600 mg Discharge medications ACEI or ARB Thienopyridine Aspirin
UFHþGPI
VCD (n ¼ 450)
No VCD (n ¼ 1,028)
VCD (n ¼ 446)
No VCD (n ¼ 1,024)
P value
59.3 [50.8, 68.6] 21.1 27.1 [24.2, 30.4] 11.3 2.4 7.3 6.9 8.4 95.4 [72.6, 119.1] 12.2
59.8 [52.3, 69.8] 22.5 27.1 [24.6, 30.4] 16.9 5.0 10.3 11.2 11.8 87.4 [66.5, 112.2] 16.6
61.6 [52.3, 70.5] 24.7 27.7 [24.5, 30.0] 15.5 3.1 10.1 10.1 8.6 88.9 [68.8, 116.2] 17.0
60.0 [53.2, 69.6] 22.9 27.1 [24.7, 29.8] 16.9 5.0 12.0 11.4 11.9 87.9 [68.7, 109.4] 17.6
0.33 0.64 0.98 0.03 0.06 0.06 0.05 0.09 0.005 0.08
54.9 0.9 10.2 34.0 0.2
61.7 0.4 7.4 30.5 0.0
0.04 0.42 0.14 0.26 0.32
72.4 98.9
82.0 12.3 4.4 0.2 0.8 1.1 64.1 98.7
83.0 0.7
81.2 12.7 4.3 0.1 1.3 1.7 71.7 0.4
21.4 78.1
36.1 63.0
24.9 73.7
36.0 62.5
79.9 99.3 98.2
80.8 97.8 99.5
81.9 98.6 97.5
84.5 97.9 98.8
0.63 0.78 0.92 0.56 0.27 0.25
Femoral Vascular Closure Device Use, Bivalirudin Anticoagulation, and Bleeding after Primary Angioplasty for STEMI: Results from the HORIZONS-AMI Trial Timothy A. Sanborn,1* MS, MD, Matthew I. Tomey,2 MD, Roxana Mehran,2,3 MD, ne reux,3,4,5 MD, Bernhard Witzenbichler,6 MD, Sorin J. Brener,3,7 MD, Philippe Ge Ajay J. Kirtane,3,4 MD, SM, Thomas C. McAndrew,3 MS, Ran Kornowski,8 MD, Dariusz Dudek,9 MD, Eugenia Nikolsky,10 MD, PhD, and Gregg W. Stone,3,4 MD Objective: To assess the relationship of femoral vascular closure device (VCD) use to bleeding and ischemic events in patients undergoing primary percutaneous coronary intervention (PCI) for ST-segment elevation myocardial infarction (STEMI) via different anticoagulation strategies. Background: It is unknown whether femoral VCD reduce major bleeding after primary PCI for STEMI using bivalirudin anticoagulation. Methods: We compared VCD-treated patients with propensity-matched controls in the HORIZONS-AMI trial with respect to net adverse clinical events (NACE), defined as the composite of major bleeding unrelated to coronary artery bypass graft surgery (CABG) and major adverse cardiac events (comprised of death, reinfarction, ischemia-driven target vessel revascularization, and stroke), at 30 days and 1 year. Results: Among 3,602 patients enrolled in HORIZONS-AMI, 2,948 underwent primary PCI via femoral arterial access and 896 (30%) received VCDs, of whom 642 were included in our model along with 642 propensity-matched controls. At 30 days, VCD-treated patients had significantly less NACE (6.7% vs. 10.8%, HR: 0.61, 95% CI: 0.42–0.89, P 5 0.009), driven by a lower rate of non-CABG related major bleeding (5.0% vs. 8.1%, HR: 0.61, 95% CI: 0.39–0.94, P 5 0.02). Bleeding reduction was maintained at one year and consistent in magnitude regardless of randomization to bivalirudin or unfractionated heparin plus a glycoprotein IIb/IIIa inhibitor (P for interaction 5 0.84). Conclusion: In patients undergoing transfemoral primary PCI for STEMI, VCD use was associated with significantly
1
NorthShore University HealthSystem, Evanston, Illinois Icahn School of Medicine at Mount Sinai, New York, New York 3 Cardiovascular Research Foundation, New York, New York 4 Columbia University Medical Center, New York, New York 5 ^ pital du Sacre -Coeur de Montre al, Montre al, Que bec, Ho Canada 6 Helios Amper-Klinikum, Dachau, Germany 7 New York Methodist Hospital, Brooklyn, New York 8 Rabin Medical Center, Petach-Tikva, Israel 9 Department of Interventional Cardiology, Jagiellonian University Medical College, Krakow, Poland 10 Rambam Health Care Campus and the Technion – Israel Institute of Technology, Haifa, Israel.
Medtronic, Abbott Vascular, Abiomed, St. Jude Medical, Vascular Dynamics, and Eli Lilly. Dariusz Dudek: Consulting and lecture fees: Abbott, Adamed, Adyton Medical Polska, Abiomed Europe, AstraZeneca, Biotronik, Balton, Bayer, BBraun, BioMatrix, Boston Scientific, Boehringer Ingelheim, Bracco, Bristol-Myers Squibb, Comesa Polska, Cordis, Cook, Covidien Polska Sp. z o. o., DRG MedTek, Eli Lilly, EuroCor, Hammermed, GE Healthcare, Glaxo, InspireMD, Iroko Cardio International, Medianet Sp. z o.o., Medtronic, The Medicines Company, Meril Life Sciences, Merck Sharp & Dohme, Orbus-Neich, Pfizer, Possis, ProCardia Medical, Promed, REVA Medical, Sanofi-Aventis, Siemens, Solvay, Stentys, St. Jude Medical, Terumo, Tyco, Volcano. Other authors: None. The HORIZONS-AMI Trial was funded by Boston Scientific and The Medicines Company.
Conflict of interest: Roxana Mehran: Institutional Research Grant Support: The Medicines Company, Bristol-Myers Squibb/Sanofi, Eli Lilly and Company/Daiichi-Sankyo, Regado Biosciences, and STENTYS; consulting: Abbott Vascular, AstraZeneca, Boston Scientific, Covidien, CSL Behring, Janssen Pharmaceuticals, Maya Medical, and Merck & Co.; advisory board: Covidien, Janssen Pharmaceuticals, Merck, and Sanofi. Philippe Genereux: Speaker honoraria—Abbott Vascular, CSI; consultant—CSI. Bernhard Witzenbichler: Consultant—Volcano. Ajay J. Kirtane: Institutional research grants to Columbia University from Boston Scientific,
*Correspondence to: Timothy A. Sanborn, MS, MD, Cardiology Division, NorthShore University HealthSystem, 2650 Ridge Ave., Walgreen Building, Third Floor, Evanston, IL 60201. E-mail: [email protected]
2
C 2014 Wiley Periodicals, Inc. V
Received 27 August 2014; Revision accepted 29 August 2014 DOI: 10.1002/ccd.25663 Published online 1 September 2014 in Wiley Online Library (wileyonlinelibrary.com)
372
Sanborn et al. lower non-CABG major bleeding irrespective of anticoagulation strategy.
C 2014 Wiley V
Periodicals, Inc.
Key words: vascular closure device; primary percutaneous coronary intervention; ST-segment elevation myocardial infarction; bivalirudin; heparin
INTRODUCTION
Femoral vascular closure devices (VCD) improve patient comfort, expedite ambulation, and reduce access site-related bleeding complications in patients with acute coronary syndromes managed with percutaneous coronary intervention (PCI) [1–4]. Their role as a “bleeding avoidance strategy” has remained controversial, however, in comparison with other strategies including radial access and bivalirudin anticoagulation [5]. Factors contributing to this uncertainty include perceived risk of vascular complications, drawn from early VCD experience [6], and increased attention to the importance of non-access site related bleeding [7]. In the setting of acute ST-segment elevation myocardial infarction (STEMI), growing evidence shows superiority of radial over femoral access as a strategy to reduce bleeding and vascular complications of primary PCI [8–11]. Uptake of radial access has been heterogeneous, however, and indeed, femoral access still predominates in the United States. Accordingly, the challenge of avoiding bleeding after transfemoral primary PCI remains highly relevant. The Harmonizing Outcomes with Revascularization and Stents in Acute Myocardial Infarction (HORIZONS-AMI) trial, which primarily involved femoral access, established bivalirudin anticoagulation as an effective bleeding avoidance strategy in primary PCI [12]. It has yet to be demonstrated whether femoral VCD confer additional bleeding reduction in this population on a backdrop of bivalirudin anticoagulation. To address this question, we performed a propensitymatched comparison of patients treated with or without VCD in the HORIZONS-AMI trial. METHODS Study Design and Participants
We identified patients for the present analysis from the HORIZONS-AMI trial. Briefly, HORIZONS-AMI was a prospective, open-label, randomized, multicenter trial enrolling 3,602 patients with STEMI who presented within 12 hr of symptom onset for primary PCI. Patients were randomized in a 22 factorial design to receive bivalirudin or unfractionated heparin plus a glycoprotein IIb/IIIa inhibitor (UFHþGPI) (1:1) and then, where eligible for PCI, to paclitaxel-eluting or bare metal stents (3:1). Clinical follow-up was con-
ducted at the prespecified time points of 30 days, 6 months, 1 year, 2 years, and 3 years. The study was approved by the institutional review board or ethics committee at every participating center and all patients provided written informed consent. Full details of the HORIZONS-AMI study protocol, inclusion and exclusion criteria have been reported previously (www.clinicaltrials.gov:NCT00433966) [13]. Of note, pertinent to the present analysis, VCD allocation was not randomized and rather left to the discretion of treating physicians. Patients with extensive peripheral vascular disease were excluded by design. For the present study, we restricted analysis to patients undergoing primary PCI via femoral arterial access with information available about VCD use. We did not exclude patients for whom external devices were used as adjuncts to manual hemostasis and analyzed these participants as part of the no VCD group. We excluded patients treated with intra-aortic balloon counterpulsation due to potential confounding of both VCD assignment and clinical outcomes. End Points and Definitions The primary outcome measure for this analysis was net adverse clinical events (NACE) at 30 days. NACE was defined as the composite of major adverse cardiac events (MACE, comprised of death, reinfarction, ischemia-driven target vessel revascularization, and stroke) and major bleeding unrelated to coronary artery bypass graft surgery (CABG). Key secondary endpoints included major bleeding and MACE at 30 days. All outcomes were followed to one year. Major bleeding was defined as intracranial or intraocular hemorrhage; bleeding at the access site, with a hematoma 5 cm or larger or requiring intervention; decrease in hemoglobin concentration by 4 g/dL without an overt bleeding source or 3 g/dL with an overt bleeding source; reoperation for bleeding; or blood transfusion. Bleeding was also assessed and adjudicated on the basis of the Thrombolysis In Myocardial Infarction (TIMI) and Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary Arteries (GUSTO) scales [14,15]. Detailed definitions of all study endpoints have been reported previously [13]). An independent clinical events committee blinded to treatment assignments adjudicated all end point events.
Catheterization and Cardiovascular Interventions DOI 10.1002/ccd. Published on behalf of The Society for Cardiovascular Angiography and Interventions (SCAI).
Vascular Closure Devices in STEMI
Statistical Analysis As a preliminary analysis, we divided participants into four groups based on randomization to bivalirudin or UFHþGPI and use or nonuse of VCD. We compared these four groups with respect to baseline characteristics, medication use, treatment strategies, and clinical outcomes. Categorical variables are presented as percentages and compared using the chi-squared test or Fisher’s exact test, as appropriate. Continuous variables are presented as median with interquartile range except where noted and compared using the Wilcoxon rank sum test. Event rates at follow-up are Kaplan– Meier estimates and compared using the log-rank test. All statistical tests were 2-tailed. P values less than 0.05 were deemed statistically significant. For the main analysis, in order to control for confounders influencing patient selection for VCD use, we developed a propensity model for VCD use. Analysis was conducted using greedy matching technique and involved several rounds of matching with caliper widths of 0.0001, 0.001, 0.01, and 0.1. No time window was considered in matching. Variables included in the propensity model included age, sex, body mass index, diabetes mellitus, current smoking, hypertension, peripheral vascular disease, prior myocardial infarction, prior PCI, United States location, bivalirudin use, GPI use, pre-randomization heparin use, final TIMI flow, baseline creatinine clearance, anemia, platelet count, warfarin use, clopidogrel loading dose, multivessel PCI, bare metal stent implantation, and left ventricular ejection fraction. Operator data was not available for inclusion in the model. We then compared VCDtreated patients with propensity-matched controls with respect to NACE, major bleeding, and MACE at 30 days and one year.
RESULTS
Of the 3,602 patients enrolled in HORIZONS-AMI, 3,134 underwent primary PCI via femoral arterial access. A single patient crossing over from radial to femoral access was not included in this analysis. After exclusion of 6 patients without information about VCD use and 80 additional patients treated with intra-aortic balloon counterpulsation, 2,948 patients remained for inclusion in the final analysis. Of these patients, 896 received VCD (30%), including 450 patients randomized to bivalirudin and 446 patients randomized to UFHþGPI. A summary of baseline features for participants treated with bivalirudin or UFHþGPI and with VCD or no VCD is presented in Table I. VCD-treated patients randomized to bivalirudin were less likely to have diabetes mellitus, peripheral vascular disease, re-
373
nal impairment, prior myocardial infarction, or prior PCI. VCD-treated patients, in general, had a lower prevalence of anemia and a higher likelihood of receiving pre-procedure heparin or the 600 mg clopidogrel loading dose. Groups were otherwise similar. The most commonly used VCD was Angio-Seal (St. Jude Medical, St. Paul, MN), followed by StarClose (Abbott Vascular, Santa Clara, CA) and Perclose (Abbott Vascular). Only six patients in total were treated with VasoSeal (DataScope, Montvale, NJ). Among patients not treated with VCD, 17% had an external FemoStop device, C-clamp or subcutaneous patch as an adjunct to manual hemostasis. Unadjusted ischemic and bleeding events at 30 days are presented in Table II. Rates of major bleeding and NACE differed significantly across groups. The highest rates of major bleeding and NACE were observed in participants treated with UFHþGPI and no VCD (8.5% and 11.7%, respectively). The lowest rates of major bleeding and NACE were observed in participants treated with bivalirudin and VCD (3.8% and 5.1%, respectively). Cross-group comparisons in bleeding were similar using TIMI and GUSTO scales. VCD-treated patients were next compared with propensity-matched controls. Upon matching, our model included 642 patients treated with VCD (72% of all VCD-treated patients) and 642 propensity-matched patients not treated with VCD. Groups were wellmatched (Table III). Comparisons of VCD-treated patients versus propensity-matched controls with respect to NACE, major bleeding, and MACE at 30 days and one year are presented in Fig. 1 panels A–F. At 30 days, VCD use was associated with a significantly lower incidence of NACE (6.7% vs. 10.8%, HR: 0.61, 95% CI: 0.42–0.89, P ¼ 0.009) and major bleeding (5.0% vs. 8.1%, HR: 0.61, 95% CI: 0.39–0.94, P ¼ 0.02); these differences persisted at one year. The main forms of major bleeding were drops in hemoglobin 4 g/dL without overt source, large hematomas, blood product transfusion, and drops in hemoglobin 3 g/dL with overt source (Fig. 2). Differences between VCD-treated patients and propensity-matched controls with respect to specific types of bleeding were not statistically significant. In order to assess whether the observed relationship between VCD use and major bleeding avoidance depended on anticoagulation strategy, we conducted formal interaction testing (Fig. 3). The magnitude of reduction in non-CABG related major bleeding was conserved regardless of randomization to bivalirudin or UFHþGPI, with no significant interaction. A trend toward lower incidence of MACE among VCD-treated patients at 30 days did not remain at one year. Review of specific ischemic events contributing
Catheterization and Cardiovascular Interventions DOI 10.1002/ccd. Published on behalf of The Society for Cardiovascular Angiography and Interventions (SCAI).
374
Sanborn et al.
TABLE I. Baseline, Clinical, and Procedural Characteristics Bivalirudin
Age, years Female sex BMI Diabetes mellitus PVD Prior MI Prior PCI Anemiaa CrCl, mL/min CrCl < 60 mL/min Closure device Angio-Seal VasoSeal Perclose StarClose Other No closure device Manual pressure only FemoStop C-clamp Subcutaneous patch Prolonged sheath dwell Other Pre-procedure heparin Bivalirudin in cath lab Clopidogrel loading dose 300 mg 600 mg Discharge medications ACEI or ARB Thienopyridine Aspirin
UFHþGPI
VCD (n ¼ 450)
No VCD (n ¼ 1,028)
VCD (n ¼ 446)
No VCD (n ¼ 1,024)
P value
59.3 [50.8, 68.6] 21.1 27.1 [24.2, 30.4] 11.3 2.4 7.3 6.9 8.4 95.4 [72.6, 119.1] 12.2
59.8 [52.3, 69.8] 22.5 27.1 [24.6, 30.4] 16.9 5.0 10.3 11.2 11.8 87.4 [66.5, 112.2] 16.6
61.6 [52.3, 70.5] 24.7 27.7 [24.5, 30.0] 15.5 3.1 10.1 10.1 8.6 88.9 [68.8, 116.2] 17.0
60.0 [53.2, 69.6] 22.9 27.1 [24.7, 29.8] 16.9 5.0 12.0 11.4 11.9 87.9 [68.7, 109.4] 17.6
0.33 0.64 0.98 0.03 0.06 0.06 0.05 0.09 0.005 0.08
54.9 0.9 10.2 34.0 0.2
61.7 0.4 7.4 30.5 0.0
0.04 0.42 0.14 0.26 0.32
72.4 98.9
82.0 12.3 4.4 0.2 0.8 1.1 64.1 98.7
83.0 0.7
81.2 12.7 4.3 0.1 1.3 1.7 71.7 0.4
21.4 78.1
36.1 63.0
24.9 73.7
36.0 62.5
79.9 99.3 98.2
80.8 97.8 99.5
81.9 98.6 97.5
84.5 97.9 98.8
0.63 0.78 0.92 0.56 0.27 0.25
