Outcomes of Vascular Closure Device Use After Transfemoral Coronary Intervention: Insights From the EXCEL Trial

Objectives. To assess the safety and efficacy of using vascular closure devices (VCDs) in percutaneous coronary intervention (PCI) for left main coronary artery disease (LM-CAD). Background. VCDs provide rapid hemostasis for patients undergoing PCI with transfemoral access (TFA); however, the safety and efficacy of VCDs continues to be debated. Methods. We analyzed data from the EXCEL trial in patients with LM-CAD in whom PCI was performed via TFA with vs without VCD. The primary endpoint was a composite of death, myocardial infarction (MI), or stroke. Bleeding Academic Research Consortium (BARC) type 2-5 bleeding at 30 days was also assessed. Propensity-score matching analysis was used. Results. Among 694 patients with LM-CAD undergoing TFA-PCI, 423 (61.0%) received VCDs (collagen plug, 320 [75.7%]; suture mediated, 55 [13.0%]; others, 48 [11.3%]). Patients with and without VCD use had similar 30-day rates of BARC type 2-5 bleeding (5.0% vs 6.7%, respectively; P=.30) and BARC type 3-5 bleeding (2.1% vs 3.7%, respectively; P=.20). There were no significant differences in the rates of death, MI, or stroke in patients with and without VCD use at 30 days (4.7% vs 4.1%, respectively; P=.74) or at 5 years (20.3% vs 24.2%, respectively; P=.16). These results were similar after adjustment. Conclusion. In the EXCEL trial, LM-CAD PCI via TFA using VCD was associated with similar 30-day rates of bleeding and comparable early and late major adverse cardiovascular events compared with manual compression.

Key words: drug-eluting stent, femoral, left main, vascular access closure device

Periprocedural bleeding is the most common complication of percutaneous coronary intervention (PCI) and is associated with increased mortality.^1,2 Vascular closure devices (VCDs) provide rapid hemostasis and enable early mobilization for patients undergoing PCI; however, studies of bleeding risk with VCDs have reported mixed results.^3-5 The most recent American College of Cardiology/American Heart Association guidelines (from 2011) provide a class IIa recommendation for faster hemostasis and a shorter duration of bed rest with VCD use, and a class III recommendation against the routine use of VCDs to reduce vascular complications, including bleeding.⁶ To our knowledge, recommendations for VCD use are not directly addressed in the European Union guidelines on myocardial revascularization.

PCI of left main (LM) coronary artery disease (CAD) may be technically challenging, especially when the distal LM bifurcation is involved. Although radial access may be preferred due to fewer bleeding-related complications, access-site preference varies by country and region. Femoral access is still the most common access route in the United States and facilitates use of a large sheath size, which may be beneficial when performing a complex 2-stent strategy.^7-9 Few data are available regarding VCD use for LM-CAD PCI. To address this issue, we performed a post hoc analysis of outcomes in patients with LM-CAD randomized to PCI with everolimus-eluting stents who underwent a transfemoral approach (TFA) with vs without VCD from the EXCEL (Evaluation of XIENCE Versus Coronary Artery Bypass Surgery for Effectiveness of Left Main Revascularization) trial.¹⁰

The EXCEL trial design and principal results have been previously reported.^10-12 In brief, EXCEL was an international, open-label, multicenter, randomized trial that compared PCI using cobalt-chromium fluoropolymer-based everolimus-eluting stents (Xience; Abbott Vascular) vs coronary artery bypass grafting in 1905 patients with LM-CAD. Inclusion criteria were LM diameter stenosis ≥70% as estimated visually or stenosis of 50% to <70% if hemodynamically significant by non-invasive or invasive testing, plus a low or intermediate (≤32) site-determined SYNTAX (Synergy Between PCI with Taxus and Cardiac Surgery) score. Consensus among the members of the heart team for revascularization with either PCI or CABG was required. Clinical follow-up was performed at 1 month, 6 months, 1 year, and then annually through 5 years. The investigation was approved by the ethics committee or institutional review board at each center, and all patients signed informed consent. The current study cohort comprised 694 patients who underwent TFA-PCI (Figure 1). Patients who underwent staged PCI were included (n = 68).

The primary effectiveness endpoint for the EXCEL trial was the composite of death from any cause, stroke, or myocardial infarction (MI) at 5 years, as previously defined.¹³ For the present analysis, the principal safety endpoint was periprocedural (30-day) bleeding complications including Thrombolysis in Myocardial Infarction (TIMI)-scale bleeding¹⁴ and Bleeding Academic Research Consortium (BARC)-scale bleeding.¹⁵ Access-site related bleeding complications occurring within 72 hours after PCI were also assessed as the principal safety endpoints. Bleeding events were monitored but were not centrally adjudicated. We also assessed the rates of vascular complications including pseudoaneurysm requiring treatment, arteriovenous fistulas, local infection requiring oral or intravenous antibiotics, limb ischemia, or any complication requiring vascular surgery, if reported.

Statistical analysis. Categorical variables were compared using the Chi-square test or Fisher’s exact test. Continuous variables were compared using the Student’s t-test or the Wilcoxon rank-sum test for non-normally distributed data. Event rates were calculated using Kaplan-Meier estimates in time-to-first-event analyses and were compared with the log-rank test between patients with and without VCD use. Odds ratios and 95% confidence intervals were estimated from logistic regression with follow-up time included as a log-transformed offset variable. We determined a propensity score for VCD use by logistic regression using the following covariates: age, sex, body mass index, prior MI, diabetes mellitus, prior transient ischemic attack or cerebrovascular disease, congestive heart failure, peripheral artery disease, chronic obstructive pulmonary disease, history of anemia, history of carotid artery disease, smoking, hypertension, hyperlipidemia, prior PCI, atrial fibrillation, chronic kidney disease (creatinine clearance ≤60 mL/min), type of P2Y₁₂ inhibitor (clopidogrel vs prasugrel and ticagrelor), presentation with recent ST-segment elevation MI vs non-ST segment elevation MI vs unstable angina vs stable angina, hemodynamic support device insertion, guiding catheter size (6 Fr vs 7 Fr vs 8 Fr), distal LM bifurcation involvement, planned 2-stent strategy, intraprocedural anticoagulant (heparin or bivalirudin), and use of glycoprotein IIb/IIIa receptor blockers. A 1:1 match was performed with the nearest neighbor within a caliper width of 0.1. Patients were also subdivided into 3 sheath sizes group (6 Fr, 7 Fr, and 8 Fr) and analyzed with crude and propensity-matched analysis for 30-day BARC type 2-5 and BARC type 3-5 bleeding. A 2-sided P<.05 was considered statistically significant. All statistical analyses were conducted with SAS, version 9.4 (SAS Institute).

Patients. Among the 694 patients undergoing LM-CAD PCI by TFA, 423 (61.0%) received VCDs (collagen plug [AngioSeal; Terumo], 320 [75.7%]; suture mediated [Perclose; Abbott Cardiovascular], 55 [13.0%]; others, 48 [11.3%]). Baseline characteristics are shown in Table 1. Patients who received VCDs were less likely to have prior cerebrovascular disease, diabetes mellitus, peripheral vascular disease, and carotid artery disease compared with those treated with manual compression (MC). Angiographic and procedural characteristics are shown in Table 2. Patients with VCD use had angiographically less complex coronary anatomy and received more 8 Fr sheaths compared with those treated with MC.

Clinical outcomes. Table 3 shows 30-day and 5-year clinical events. Patients with and without VCD use had similar 30-day rates of the primary effectiveness endpoint, TIMI major or minor bleeding, TIMI major bleeding, BARC type 2-5 bleeding, and BARC type 3-5 bleeding. Of note, of the 39 BARC type 2-5 bleeding events within 30 days, 28 occurred before hospital discharge, including 17 (4.0%) in the VCD group and 11 (4.1%) in the MC group (P=.98). Patients with and without VCD use also had similar rates of access-site related bleeding complications within 72 hours after PCI (Table 3). Two cases of vascular complications (both pseudoaneurysms requiring vascular repair, 1 surgical, 1 percutaneous) occurred, both in patients with VCD use (vascular complications in VCD group vs MC group, 0.5% vs 0.0%; P=.52). There were no arteriovenous fistulas, infections requiring oral or intravenous antibiotics, limb ischemia, or other complications requiring vascular surgery. At 5 years, there were no differences in the primary endpoint or bleeding in patients with VCD use vs MC use (Table 3; Figures 2A and 2B). 

Among patients with VCD use, those treated with the AngioSeal device (the most commonly used VCD) had lower rates of the primary effectiveness endpoint (composite of death, stroke, or MI) at 30 days compared with other devices (3.4% vs 8.7%, respectively; P=.03); however, BARC type 2-5 and BARC type 3-5 bleeding rates at 30 days were not significantly different between the devices (4.4% vs 6.8% [P=.30] and 2.2% vs 1.9% [P=.91], respectively). At 5 years, there were no differences in the primary endpoint, BARC type 2-5, or BARC type 3-5 bleeding rates in patients treated with AngioSeal device vs others (19.8% vs 22.0% [P=.58], 9.5% vs 9.8% [P=.86], and 4.1% vs 3.0% [P=.62], respectively).

Propensity-matched analysis. Propensity-score matching resulted in 2 groups of 233 patients each that were well matched for baseline features (Supplemental Tables S1 and S2). Propensity-matched patients treated with and without VCDs had non-significantly different 30-day rates of the primary endpoint and all measures of bleeding (Table 4). At 5 years, there were no differences in the primary endpoint or bleeding complications between the groups (Figures 2C and 2D). Other ischemic event rates, including cardiac death, were also similar at 30 days and 5 years in the 2 groups.

No significant differences in 30-day bleeding rates between the VCD and MC groups were noted in either the crude or propensity-matched analyses according to femoral artery sheath size (Table 5).

The major findings from the present analysis investigating the use of VCDs vs MC in patients with LM-CAD undergoing PCI via TFA are the following: (1) VCDs were used in approximately 60% of patients, and patients treated with a VCD had fewer comorbidities compared with those treated with MC. In a propensity-matched analysis, the 30-day rate of BARC 2-5 bleeding was not significantly different between VCD and MC. Nor were there significant differences in the 30-day or 5-year risks of ischemic events in patients with vs without VCD use. The present study supports the safety of VCD use in patients undergoing LM-CAD PCI by TFA, with MC as an acceptable alternative.

Despite the recent upsurge in the use of radial access, TFA was the predominant vascular access approach in the international EXCEL trial. Furthermore, adoption of the transradial approach has been slower in the United States compared with other countries and regions.^7,9,16,17 A prior study showed the rate of transradial access during 2014-2015 in the United States was only 17.8%, substantially lower than many other countries.^18,19 Access-site selection also varies widely by country and region.²⁰ Thus, examination of optimal TFA closure strategies remain relevant. Although the capability of VCD use to accelerate time to hemostasis and ambulation is unquestioned, previous studies have found mixed results in terms of safety and effectiveness of VCD use. An analysis from the  National Cardiovascular Data Registry CathPCI registry reported that VCDs were associated with fewer bleeding complications compared with MC,³ although patients at high risk for bleeding were less likely to receive a bleeding avoidance strategy. A separate study reported that bleeding complications were increased after emergent PCI with the use of VCDs.⁴ In contrast, a meta-analysis showed no increase in vascular complications, but a significantly higher risk of infection with VCD use.⁵ Although VCDs are considered a potential bleeding avoidance strategy,²¹ a scientific statement from the American Heart Association nearly a decade ago (nonetheless, the most recent guidance on this topic) provides a class III recommendation for the routine use of VCDs to reduce vascular complications.²² Moreover, few data are available regarding VCD use after LM-CAD PCI. The present study is thus useful to inform the safety of VCD use following LM-CAD intervention by TFA.

In our study, the use of VCDs was not associated with greater contrast volume, fluoroscopy time, or catheterization duration. Thirty-day bleeding rates were not significantly different between VCD and MC. Overall 5-year clinical outcomes did not differ significantly between patients using VCDs vs MC in both unadjusted and propensity-matched analyses to account for differences in baseline clinical and angiographic characteristics associated with VCD use. These findings support the safety of VCD use in PCI for LM-CAD, with MC as an acceptable alternative.

Study limitations. The present study has several limitations. The choice to use a VCD was not randomized. Despite mitigating the impact of measured confounders via propensity-matched analysis, we cannot rule out an effect from unmeasured confounders, such as femoral vascular disease. The variety of VCDs used in the present study, while reflecting real-world usage patterns, confounds analysis of any single VCD vs MC. The present post hoc results should thus be considered hypothesis generating. Time to hemostasis and ambulation, which are well established to be reduced with VCDs, were not assessed in the present study. Nor did we not collect data on operator volume and experience, which could affect complication rates of VCD use.²³ Moreover, although the largest study to date on the impact of VCD use after LM-CAD PCI, our sample size is modest, which precludes detecting small differences between groups. For example, in the propensity-matched analysis, major bleeding with VCD vs MC was numerically but not significantly less frequent at 30 days (1.3% vs 3.9%; hazard ratio, 0.31; 95% confidence interval, 0.08-1.16; P=.08). This difference may (or may not) have become significant had more patients been enrolled.

These limitations notwithstanding, we can conclude that in the EXCEL trial, PCI in patients with LM-CAD using TFA with VCD use was associated with similar early rates of major bleeding, and comparable early and late ischemic outcomes compared with MC. These findings support the safety of VCD use in patients undergoing LM-CAD PCI by TFA, with MC as an acceptable alternative.