Methods
The DEB-AMI (Drug Eluting Balloon in Acute Myocardial Infarction) trial was a randomized, international, 2-center, single-blinded, 3-arm study, aimed at comparing BMS implantation (group A), versus sequential DEB dilatation and BMS implantation (group B) and paclitaxel DES implantation (group C) in patients presenting with STEMI. In order to minimize confounding, in both group A and B, an identical stent platform was used, and the same drug (paclitaxel) was compared in groups B and C.
The study, conducted according to the Declaration of Helsinki, was approved by the ethics committees of both participating centers, and signed informed consent was obtained from all included patients.
Patient Selection
Patients between 18 and 80 years of age, presenting in the first 12 h after the onset of STEMI (diagnosed by the presence of anginal symptoms associated with electrocardiographic ST-segment elevation of >1 mm in >2 contiguous leads or new left bundle branch block), and undergoing primary PCI, with angiographic evidence of a single culprit lesion in the target vessel, after successful thrombus aspiration (defined by no angiographically evident flow-limiting residual thrombus at the site of [sub]occlusion, and Thrombolysis In Myocardial Infarction [TIMI] flow grade >1) were deemed eligible for inclusion.
Major clinical and procedural exclusion criteria were contraindications to study medications (acetylsalicylic acid, clopidogrel, paclitaxel), life expectancy <12 months, lesion length >25 mm, reference vessel diameter <2.5 mm and >4.0 mm, severe triple vessel disease, left main stenosis >50%, and a combination of type C coronary lesion and diabetes mellitus (in which DES was favored).
Devices
The DEB used in this study was the second-generation DIOR coronary angioplasty balloon. This DEB has a coating consisting of a 1:1 mixture of paclitaxel with shellac applied to the balloon by a micro-pipetting procedure. This device is coated with 3 μg of paclitaxel/mm of balloon surface. The DIOR DEB is available in the following lengths: 15, 20, 25, and 30 mm; and diameters: 2.0, 2.25, 2.5, 2.75, 3.0, 3.5, and 4.0 mm. The shellac coating protects the drug from a wash-off effect during tracking in the guiding catheter and in the coronary vasculature. The minimal inflation time is 30 s (recommended 45 to 60 s) to allow sufficient drug release into the vessel wall in order to achieve the required effective tissue dosages of paclitaxel to inhibit smooth muscle cell proliferation.
The BMS (Genius Magic stent, Eurocor) is a new cobalt chromium stent platform with a strut thickness of 60 μm. The DES (Taxus Liberté, Boston Scientific, Natick, Massachusetts) has a stainless steel stent platform coated with a permanent polymer that allows the release of paclitaxel (1 μg/mm). The total strut thickness including stent and polymer is 132 μm.
Randomization and Interventional Procedure
All patients received routinely in the ambulance or at the first medical contact a loading dose of acetylsalicylic acid (325 to 500 mg) and of clopidogrel (600 mg). Heparin was administered before and during the procedure in order to maintain an activated clotting time ≥250 s. Additional administration of glycoprotein IIb/IIIa inhibitors was recommended, but was left to the physician's discretion.
After fulfilling angiographic inclusion criteria, patients underwent mandated thrombus aspiration of the culprit lesion with a manual thrombus aspiration device. Sequentially, if thrombus aspiration was successful according to definition, patients were randomly assigned to 1 of the 3 treatment strategies: group A: pre-dilation with a standard balloon (balloon-to-artery ratio 0.8:1) followed by BMS implantation; group B: pre-dilation with a standard balloon (balloon-to-artery ratio 0.8:1) followed by sequential dilation with DEB (balloon-to-artery ratio 1:1 and at least 5 mm longer than the normal balloon in order to avoid geographic miss) for at least 30 s, and BMS implantation; or group C: pre-dilation with a standard balloon (balloon-to-artery ratio 0.8:1) followed by implantation of a paclitaxel-eluting DES. Randomization was obtained by means of sequentially numbered, opaque sealed envelopes containing the assigned group letter (A, B, or C). The coupling between number of the envelope and group letter was automatically generated by a computer, and the envelopes were sealed by an independent employee. The allocation ratio to the 3 groups was 1:1:1.
Concerning the DEB use, a 1:1 balloon-to-artery ratio, nominal inflation pressures, and an inflation time of >30 s were mandated. Special care was taken to center the DEB in the lesion and to use a DEB at least 5 mm longer than the normal balloon used and the intended stent length, in order to avoid potential geographic miss. In any group, additional bailout stenting (with stents of the same randomization group) was performed in case of residual edge dissections or incomplete lesion coverage. In this setting in group B, additional DEB were also mandated to cover the complete stented segment. In case of multiple DEB use, care was taken to avoid excessive DEB overlap (to avoid double-dose release). Additional post-dilation was left to the physician's discretion. Discharge medications included acetylsalicylic acid 80 to 100 mg per day lifelong and clopidogrel 75 mg for 12 months.
Follow-up and Clinical Endpoints
All patients were contacted by phone call 1 month after the procedure and underwent clinical and angiographic follow-up at 6 months. In case an event occurred, detailed review of the related hospital files was performed.
Major adverse cardiac events (MACE) were defined as a hierarchical composition of death, any myocardial infarction (MI), and target vessel revascularization (TVR). All definitions followed the Academic Research Consortium criteria. Target lesion revascularization (TLR) was defined as any repeat percutaneous or surgical intervention due to a restenosis in the treated segment (including the stent and 5 mm proximal and distal). A TLR was considered clinically indicated in case of restenosis >50% by quantitative coronary angiography (QCA), associated with recurrent angina and/or objective signs of silent ischemia (stress tests or fractional flow reserve), or in case of restenosis >70% by QCA without the aforementioned signs or symptoms. Stent thrombosis was defined as a definite, probable, or possible and early or late, also according to the Academic Research Consortium criteria. Angiographic success was defined as achievement of a TIMI flow grade 3 and final residual stenosis <30%, using any percutaneous method. Device success was defined as angiographic success using the randomized device. Procedural success was defined as angiographic success without the occurrence of in-hospital MACE. All patients' files were independently monitored and all outcomes were adjudicated by an independent clinical events committee.
Quantitative Coronary Angiography
QCA was performed according to standard procedures, using dedicated software (CAAS 5.9.1 research edition, Pie Medical Imaging, Maastricht, the Netherlands). Images were analyzed by an independent core laboratory with operators not involved with the procedure and blinded to randomization assignment. As the vessel was totally occluded in most of the baseline images, the pre-procedural images analyzed were those after thrombus aspiration. The number of patients with a total occlusion was assessed, and the values for diameter stenosis and minimal luminal diameter (MLD) were changed in these patients in 100% and 0 mm, respectively. In the post-procedural and follow-up images, the stent(s) and additional 5-mm segments proximal and distal to the stent(s) edges were analyzed. MLD and lesion length were directly measured by the QCA software, whereas reference vessel diameter was estimated by an interpolation method, and percent diameter stenosis was subsequently computed. Binary restenosis was defined as a diameter stenosis ≥50% at angiographic follow-up. Late-luminal loss was defined as the difference between post-procedural MLD and MLD at follow-up in the same segment (proximal to the stent, in-stent, distal to the stent, in-segment). In-stent late-luminal loss was the primary endpoint of the study.
Optical Coherence Tomography
At the same time as the main randomization, each patient was also randomized to undergo, at 6 months, only angiographic follow-up or optical coherence tomography (OCT) investigation and endothelial function testing together with the angiographic control (4:1 ratio).
Time-domain or frequency-domain OCT systems (M3 LightLab Imaging, Westford, Massachusetts, or C7XR, St. Jude Medical, St. Paul, Minnesota) were used. All images were analyzed by an independent core laboratory.
OCT imaging of the target lesion was obtained after 200-μg intracoronary nitroglycerin infusion, and OCT pullback images were acquired during continuous infusion of contrast from the guiding catheter, by means of a controlled injection (2 to 4 ml/s contrast with 200 to 300 psi, depending on the coronary assessed), as previously described. Images were acquired with automated pullback at a rate of 2, 3, or 20 mm/s (according to the type of OCT system). All cross-sectional images (frames) were initially screened for quality assessment and excluded if any portion of the stent was out of the screen, images were not analyzable due to side branches, or images had poor quality caused by residual blood, artifacts, or reverberation.
A dedicated semiautomated contour-detection system (Curad BV, Wijk bij Duurstede, the Netherlands) was used. Two contours were delineated: the lumen contour (for each cross-sectional image) and the stent contour (every 0.2 to 0.3 mm, depending on the pullback speed, thus, for example, every 5 frames for 2 to 3 mm/s). For stent analysis, an automated stent contour interpolation was performed between frames. Manual corrections were applied if needed. Lumen, stent, and neointimal hyperplasia diameters, areas, and volumes were automatically calculated. Stent struts were semiautomatically classified as covered embedded, covered protruding, uncovered apposed, and malapposed. Covered embedded struts were defined as covered by tissue and not otherwise interrupting the smooth lumen contour; covered protruding struts were defined as covered with tissue but extending into the lumen (however, not greater than the stent strut thickness: 132 μm for DES and 60 μm for BMS); uncovered apposed struts were defined by the same distance from the lumen border as covered protruding struts, without the presence of tissue coverage, however; malapposed struts were defined as those not abutting the lumen border: ≥132 μm for DES and ≥60 μm for BMS. The distance was measured between the center of the stent strut and the lumen border, and was preset in the software and manually corrected if necessary. Hence, all struts were semiautomatically defined depending on their distance.
Endothelial Function Testing
The assessment of the endothelium-dependent vasomotor function was performed by the selective infusion of the endothelium-dependent vasodilator acetylcholine into the target coronary artery. After baseline angiography, acetylcholine was infused via an infusion microcatheter located at the level of the stent into the target coronary artery at incremental concentrations of 10, 10, and 10 mol/l/ml. Acetylcholine was infused for 3 min at each concentration, with a 3-min interval between each infusion. Angiography was obtained after each infusion. The infusion was terminated when the largest dose of acetylcholine was reached or in case of coronary vasoconstriction >50% by visual estimation. Nitroglycerin was then injected as an intracoronary bolus (100 μg) through the guiding catheter to evaluate the endothelium-independent coronary vasoactive response of the coronary artery. Angiography was also performed after nitroglycerin infusion.
Coronary responses to acetylcholine and nitroglycerin were analyzed offline, using dedicated QCA software allowing for segmental analysis by the same core laboratory. All images were recorded in identical gantry position, allowing for accurate consecutive analysis, and were analyzed during diastole. The images obtained at baseline and after each infusion were analyzed for MLD in consecutive 5-mm segments, distal to the stent. Shoulder patterns on QCA analysis were used to identify the distal stent edge. The first 15 mm were considered in the analysis, and the worst MLD per each 5-mm segment was chosen. The MLD obtained after each infusion was compared with baseline values. Baseline MLD was set as 0. Endothelial function was quantified as percentage change of MLD from baseline, with negative values expressing paradoxical vasoconstriction and positive values representing physiological vasodilatation.
Statistical Analysis
The primary endpoint of the study was in-stent late-luminal loss. The sample size calculation was based on the direct comparison between BMS (group A) and DEB plus BMS (group B). The study tested the hypothesis that late-luminal loss in group B would be significantly better than in group A. A sample size of 43 subjects per group was estimated to show a significant reduction in late-luminal loss of 50% (from 0.70 to 0.35 mm) with a 2-tailed p value of 0.05 and a power of 90%, assuming a standard deviation of 0.50 mm. To accommodate a 15% loss in angiographic follow-up, 50 patients per group were enrolled. A third DES arm with the same number of patients was added as an exploratory arm, to test at the same time this device in the same setting. Concerning OCT and endothelial function, the number of patients for the substudy was 10 per group, and the substudy was specifically powered (Online Appendix).
Continuous variables are presented as mean ± SD if normally distributed, or median [interquartile range] if not normally distributed. Categorical variables are presented as counts and percentages. Continuous variables were compared between 2 groups using the Student t test or its nonparametric equivalent Mann-Whitney U test. In case of a between-groups comparison (A vs. B vs. C), analysis of variance or its nonparametric equivalent Kruskal-Wallis test was applied. Categorical variables were compared using the chi-square or Fischer exact test. All analyses were performed according to the intention-to-treat> principle. A 2-tailed p value of 0.05 was considered statistically significant.