The relationship of pre-procedural Dmax based sizing to lesion level outcomes in Absorb BVS and Xience EES treated patients in the AIDA trial

Due to expansion limits of the Absorb bioresorbable scaffold a meticulous implantation with correct sizing is required. We sought to investigate the clinical outcomes based on the sizing of the device related to the maximal lumen diameter measured by quantitative coronary angiography in Absorb BVS and Xience EES treated lesions in the AIDA trial. Sizing of Absorb bioresorbable vascular scaffold (BVS) and Xience everolimus eluting stent (EES) was graded according to the definitions of device non-oversize and device oversize on pre-procedural angiography. Lesion-oriented outcomes (LOCE) (device thrombosis, TLR and TVMI) that occurred during 2 years follow-up were related to device non-oversized or oversized status. In the Absorb BVS group, LOCE occurred in 48 (7.4%) lesions in the oversized group and in 32 (8.2%) lesions in the non-oversized group (HR 0.91; 95% CI 0.58–1.42; p = 0.681), whereas TLR occurred in 34 (5.3%) lesions and in 23 lesions (5.9%), respectively (HR 0.89; 95% CI 0.52–1.51; p = 0.666). Definite scaffold thrombosis occurred in 11 (1.7%) device oversized treated lesions against 16 (4.1%) device non-oversized treated lesions (HR 0.41; 95% CI 0.19–0.89; p = 0.020). There were no differences in event rates between oversized and non-oversized groups in lesions treated with Xience EES. There was no significant difference in LOCE between oversized and non-oversized treated Absorb BVS and Xience EES treated lesions. Non-oversized Absorb BVS implantation was associated with a higher risk of scaffold thrombosis at complete 2 years follow-up. The majority of very late scaffold thrombosis occurred in properly sized devices. Electronic supplementary material The online version of this article (10.1007/s10554-019-01576-y) contains supplementary material, which is available to authorized users.

The limitations of the device, as well as implantation techniques or strategies, have been proposed as possible causes of adverse events [8]. Proper sizing of the scaffold may be of great importance acutely after scaffold implantation. The limited ability to over-expand makes the Absorb BVS prone to scaffold disruption when an undersized device is implanted and aggressively post-dilated. Furthermore, implanting an oversized device has been shown to cause more edge dissections and peri-procedural myocardial infarctions [9][10][11].
Ishibashi et al. demonstrated that implantation of an oversized Absorb BVS was associated with a higher risk of major adverse cardiac events (MACE) at 1 year [12]. Whereas, Katagiri et al. demonstrated that implantation of an undersized Absorb BVS was associated with a higher risk of MACE between 1 and 3 years [13]. This association was investigated in trials, which included only simple lesions and where quantitative assessment of the target vessel diameter by online quantitative coronary angiography (QCA) was required for enrollment in the trial [1][2][3]14].
The Amsterdam Investigator-Initiated Absorb Strategy All-comers (AIDA) trial was an all-comers randomized controlled trial that analyzed Absorb BVS versus Xience EES in routine percutaneous coronary intervention (PCI). Although it demonstrated non-inferiority in the composite endpoint of target vessel failure (TVF) of Absorb BVS versus Xience EES at 2 years, Absorb BVS was associated with a higher risk of definite/probable device thrombosis [15]. In the current report we investigate whether proper sizing is a key factor leading to adverse events after device implantation and therefore we evaluated the relationship of maximum lumen diameter (Dmax) based device sizing on QCA, to lesion oriented outcomes in both the Absorb BVS and Xience EES arms of the AIDA trial.

Study design AIDA trial
The AIDA compared Absorb BVS (Abbott Vascular, Santa Clara, CA, USA) with Xience EES (Abbott Vascular, Santa Clara, CA, USA) in routine PCI. The study design [16], the preliminary safety report [6], and the 2-year results [15] have been published previously.
Briefly, between August 2013 and December 2015, 1845 consecutive patients with coronary artery disease undergoing PCI with one or more target lesions suitable for drugeluting stent implantation were included in AIDA. Key exclusion criteria were target lesions longer than 70 mm, a visually estimated reference vessel diameter of < 2.5 mm of > 4.0 mm, treatment of a true bifurcation lesion with a priori planned two device strategy, and treatment of in-stent restenosis. Patients were randomized after successful predilatation of the first lesion. Device sizing was based on visual assessment by the operator. Online QCA, or periprocedural testing of cardiac biomarkers, were not mandatory. Device implantation strategy was planned according to the instructions for use of the implanted device.

QCA analysis
The current analysis includes all study lesions treated with at least one study device of which baseline angiogram suited for offline QCA analysis was available. QCA assessment of device sizing relied on the angiographic diameter function curve of the pre-treatment vessel segment consisting of three non-ambiguous luminal dimensions; namely, the minimal lumen diameter (MLD) and the Dmax in the proximal segment (proximal Dmax) and in the distal segment of interest (distal Dmax) with respect to the MLD.
Pre-and post-procedural measurements were performed in either (1) multiple matched views or (2) a single matched view. If no matched views were available, measurements were done within the view with visually the highest stenosis grade. Proximal and distal Dmax were measured in (a) pre-procedural projection(s). In case of a total occlusion, the MLD was considered to be 0 mm; the proximal Dmax was measured in a pre-procedural projection; and the distal Dmax was considered similar to the proximal Dmax. Offline QCA analyses were initially performed on the post-procedural angiograms within the Absorb BVS arm, and were performed by seven experienced readers [6]. We later added 5 experienced readers to our academic Corelab, in order to complete pre-procedural measures on the Absorb BVS arm, and complete the pre-procedural and post-procedural measures within the Xience EES arm. All QCA analyses were performed using Cardiovascular Angiography Analyses System [CAAS], version 5.11 (Pie Medical Imaging, Maastricht, the Netherlands). All QCA readers were blinded for clinical events, and were supervised by one cardiologist [YO], who is an expert in this field.

Definitions
The study population in both arms was stratified by the difference between both the proximal and distal angiographic Dmax and the nominal diameter of the implanted device. The selection of device size was considered "oversized" when the proximal and distal Dmax of the treated lesion were smaller than the nominal size of the device. Lesions with either a proximal or a distal Dmax or both Dmax larger than the nominal size of the device constituted the "nonoversized group". When a patient received 2 or 3 overlapping devices in a long lesion, the nominal size of the proximally implanted device was compared with the proximal Dmax, whereas the nominal size of the distally implanted device was compared with the distal Dmax. Devices were considered properly sized when the ratio of the proximal or distal Dmax minus the device diameter fell within the range of -0.5 to + 0.5 mm.

Lesion oriented endpoints
The lesion-oriented composite outcome (LOCE) of this analysis included definite device thrombosis, target lesion revascularization (TLR) and target vessel myocardial infarction (TVMI). All myocardial infarctions were defined by the Third Universal Myocardial Infarction definitions and other events were defined according to the Academic Research Consortium definitions. All reported events were adjudicated by an independent Clinical Events Committee (Cardialysis BV., Rotterdam, the Netherlands).

Statistical analysis
This report provides information on the lesion-oriented outcomes subdivided by their device non-oversize or oversize status. All analyses were conducted using the per-protocol population. Descriptive statistics were used to present the data. Groups were compared using Chi square test and student t-test. Kaplan-Meier method, log-rank test and Cox proportional hazards models were used for time-to-event analysis. Landmark analyses were performed at 31 days and 365 days after index procedure. A 2-sided p value < 0.05 was considered significant. The prognostic value for LOCE of device oversize was assessed in a univariate Cox proportional-hazard analysis. In a multivariate Cox regression analysis we identified other statistically significant predictors of LOCE. Next we performed a multivariate regression analysis for device oversize with adjustment for other statistically significant predictors of LOCE.
The entry criterium for the multivariable analysis was set at p < 0.1. Device oversize was forced into the multivariate model. All statistical tests were performed using SPSS software, version 23.0 (IBM SPSS Statistics, IBM Chicago, IL, USA).

Results
In the AIDA trial, 924 patients were randomized to treatment with Absorb BVS and 921 patients to treatment with Xience EES. A total of 2446 lesions were treated; 1237 lesions within the Absorb BVS arm and 1209 within the Xience EES arm. We excluded 89 lesions that did not receive any study device. Another 179 lesion were not analyzable, and were therefore excluded. Pre-procedural Dmax assessment was available in 2152 lesions (87.9%). The full study flowchart is shown in Fig. 1.
Baseline and angiographic characteristics of the oversized and non-oversized groups for both study devices are displayed in Tables 1 and 2. In 653 (62.0%) of Absorb BVS treated lesions, and 648 (59.0%) of Xience EES treated lesions an oversized device was implanted. Oversized devices were more frequently implanted in the left coronary artery, and in lesions with a TIMI flow of 0/1. In both study devices pre-procedural reference vessel diameter (RVD), both proximal and distal Dmax and minimal lumen diameter (MLD), were significantly smaller in the oversized groups compared with the non-oversized groups. Moreover, pre-and post-dilatation were less frequently performed in Fig. 1 The study flowchart. AIDA Amsterdam Investigator-Initiated Absorb Strategy All-comers, QCA quantitative coronary angiography, Dmax maximum diameter the Absorb BVS oversized group compared with the nonoversized groups.

Discussion
The main findings of this AIDA sub study are: 1. We found no difference in Lesion-oriented Composite Endpoint between oversized and non-oversized Absorb BVS treated lesions.

Definite Scaffold Thrombosis and Target Vessel Myo-
cardial Infarction occurred more frequently in non-oversized Absorb BVS treated lesions, especially between 1 and 2 year follow-up. 3. Most cases of definite scaffold thrombosis occurred in properly sized Absorb BVS according to the Instructions for Use of Absorb BVS and Dmax assessment. 4. We found no difference in lesion-oriented outcomes between oversized and non-oversized Xience EES treated lesions.
Scaffold thrombosis is an unpredictable severe complication, which usually results in myocardial infarction, or even cardiac death. Several studies have investigated whether Absorb BVS optimized implantation techniques are able to resolve the scaffold thrombosis issue. The results of these studies are, however, equivocal. Correct scaffold sizing remains a challenge in routine PCI because appropriate device selection relies on the device to coronary diameter ratio and its potential mismatch [12,17]. Correct scaffold sizing is particularly difficult in long tapered lesions. It is known that the avoidance of a scaffold to coronary diameter mismatch is associated with less frequent ischemia driven TLR, whereas scaffold oversizing may be associated with higher rates of MACE at 1 year follow-up [12]. Whether this theory also applies to metallic stents remains unknown.
In the current analysis we investigated the relationship between Dmax based device sizing and lesion-oriented outcomes in both the Absorb BVS as the Xience EES treated arm of the AIDA trial. Dmax based device sizing is a lesion specific characteristic, and therefore we present specific lesion-oriented outcomes. In contrast with the findings of Ishibashi et al, our analysis showed that oversized Absorb BVS were not associated with higher rates of lesion oriented events. This may be due to the difference in MI definition in the two studies, namely in AIDA trial there was no routine post-procedure marker measurement and only clinically significant post-procedural myocardial infarctions were captured. However, from 1 to 2 year follow-up, similarly to Katagiri et al. we found significant increase in the rates of device thrombosis in non-oversized Absorb BVS Fig. 4 Landmark analysis of definite scaffold thrombosis at 31 days (a), 31 to 365 days (b) and 1-2 years (c). BVS bioresorbable vascular scaffold, CI confidence interval, no. number treated lesions. Furthermore, most cases of very late scaffold thrombosis occurred in properly sized devices [13]. This later finding may implicate the complex interaction between vessel wall and the device resorption and dismantling process rather that any initial procedural technique as a major factor in very late stent thrombosis cases.
Our study is the first to investigate the effect of Dmax based device sizing in Xience EES treated lesions. We found no difference in event rates between oversized and non-oversized Xience EES treated lesions. Definite stent thrombosis is unpredictable and does not depend on the Xience EES / coronary artery diameter mismatch. Moreover, all three acute stent thrombosis occurred in properly sized devices, in patients taking dual antiplatelet therapy, suggesting that other factors than implantation technique play a role in the occurrence. These results can potentially be attributed to the better device characteristics of the Xience EES such as thinner struts, better expansion limits, higher tensile and radial strength as compared to Absorb BVS.
The insights of this AIDA sub-study can be useful for the development of the next generation coronary bioresorbable scaffolds, with wider expansion limits and better tensile and radial strength, potentially resolving the scaffold sizing issues and difficulties. In addition, QCA or intravascular imaging should be strongly considered in forthcoming trials of new generation bioresorbable coronary scaffolds.

Limitations
The current analysis has several limitations. First, in AIDA, device sizing based on online QCA was not mandatory, and therefore incorrect device sizing is likely to have occurred more frequently. Second, routine intravascular imaging has not been performed and therefore the study does not provide mechanistic insights in the occurrence of events due to a device / artery diameter mismatch. Third, the role of lesion morphology and typology stratification as a prognostic facture for device failure has not been analyzed in the current explorative analysis. The gold standard for lesion morphology and typology stratification is intracoronary imaging. The AIDA population reflects routine PCI, and since intracoronary imaging is not a part of routine PCI (used in less than 10% of the cases), it was not possible to explore the exact role of these factors in this study. Fourth, as with all post-hoc analysis, this AIDA sub-study is subject to under powering.

Conclusions
In this AIDA trial sub-study, we found no significant difference in LOCE between oversized and non-oversized treated Absorb BVS and Xience EES treated lesions. However, nonoversized Absorb BVS implantation was associated with a higher risk of scaffold thrombosis at complete 2 years follow-up, in particular between 1 and 2 years follow-up.