This is an investigator-initiated, single-center, retrospective study performed at the Hospital Clinic Barcelona (Barcelona, Spain). All consecutive ≥18year-old patients presenting symptomatic coronary artery disease (CAD) with de novo lesions (type A, B1 and B2) [12] in vessels > 2.5 mm, who underwent PCI with implantation of a single DES and subsequent assessment by both OCT and 3DStent were included. PCI was performed according to standard clinical practice. All participants in this study provided written informed consent for percutaneous coronary intervention. The present study was approved by the local ethics committee.

The Allia IGS 7 with AutoRight™ (GE Healthcare, Chicago, IL, USA) was used for 3DStent acquisition. 3DStent reconstruction is performed using C-arm Motion Compensated Computed Tomography (CMCT) resulting in an intraprocedural 3D visualization. The patient lays in supine position with one arm above the head in order to optimize the 3DStent image quality. 3DStent relies on a 200 degrees rotational angiography. The acquisition workflow is largely automated: the operator centers the stent in 2 angulations and verifies during an X-ray–free test that the rotation is collision-free. After an automatic test spin, 3DStent rotational cine acquisition is performed with an automatic synchronization of X-ray exposure, gantry rotation and it can be launched from the control room with no impact on the operator dose. During the acquisition, similarly to digital stent enhancement, a deflated balloon must be kept inside the stent to allow stent localization by the software. 3DStent acquisition has two possible rotation speeds: 10°/s and 20°/s. The radiation dose is lower (half dose) when 20°/s rotation is used, however 10°/s rotation offers higher image quality: in our study 10°/s rotation speed was used.

3DStent software - with an axial and longitudinal resolution of 100 μm - automatically generates a 3D model and multiplanar reconstruction from the acquired frames, enabling a multislice cross-sectional evaluation of the stent architecture.

Standard OCT imaging was performed using Dragonfly™ OPTIS™ imaging catheter (Abbott Vascular, Lake County, IL, USA) after stent implantation. The OCT acquisition was performed using a commercially available system for intracoronary imaging (LightLab Imaging Inc, Westford, MA, USA). Automated pull-back at 25 mm/s was performed in concordance with blood clearance by the contrast injection [9].

Both 3DStent or OCT were performed at the same stage of the PCI procedure in order to allow for a direct comparison between the two techniques in terms of stent diameter and area.

Two interventional cardiologists independently performed offline 3DStent and OCT analysis, blinded to the alternative technique. Stent area and diameter were assessed at contiguous cross-sections every 1 mm in both modalities in order to obtain comparable cross sections in 3DStent and OCT pullbacks. (Fig. 1)

Study design. CAD = Coronary Artery Disease; OCT = Optical Coherence Tomography; PCI = Percutaneous Coronary Intervention

3DStent reconstructions were analyzed using the GE HealthCare Advantage review workstation of cathlab, which allows measurement of stent area and diameter at each frame (10 frames per mm). To the aim of this analysis, one cross section was selected at every 1 mm [13, 14]. Unlike OCT, which has clear guidelines for stent measurements, there are currently no guidelines about stent area contouring with the 3DStent: therefore, per each 3DStent cross-section analyzed, we measured stent area and diameters by delineating three different contours at the abluminal, mid, and endoluminal side of the stent. (Fig. 1)

OCT measurements of stent area and diameter were performed as usual [9]: offline OCT data analysis was carried out using specific proprietary software for off-line analysis (LightLab Imaging Inc, Westford, MA, USA) and software-enabled automatic strut detection with manual corrections allowed only in case of huge anomalies.

For the intra-observer agreement, the same analyst repeated the measurements on the same 3DStent reconstruction 3 months later, by using the same methodology described above. For the inter-observer agreement, a third analyst – with the same experience than the other two - was involved to perform the 3DStent analysis on the same cross-sections analyzed by the first analyst, by using the same methodology.

Continuous variables are presented as mean ± standard deviation (SD), while categorical variables are represented as numbers and frequencies. 3DStent and OCT measurements were analyzed at the cross-sectional level using multilevel adjusted General Estimating Equations (GEE). Agreement between 3DStent and OCT analysis of stent area and diameter was also established by the Bland-Altman test.

Intra-observer and inter-observer agreements regarding stent area and diameter by 3DStent were assessed using the Spearman test for correlation. Regarding the intra- and inter-observer agreement, Spearman r value of 0.01–0.19 indicates no or negligible agreement, 0.20–0.29 indicates weak agreement, 0.30–0.39 indicates moderate agreement, 0.40–0.69 indicates strong agreement, ≥ 0.79 indicates very strong agreement [15, 16].

A two-tailed p-value ≤ 0.05 was deemed statistically significant. Statistical analyses were conducted using IBM SPSS 20.0 Statistics software (SPSS Inc., Chicago, IL, USA).