Evaluation of Left Main Coronary Artery Using Optical Frequency Domain Imaging and Its Pitfalls

Objectives We aimed to assess the quality of optical frequency domain imaging (OFDI) of the left main (LM) arterial wall and describe and analyse potential artefacts in this setting. Background OFDI is increasingly used to assess ambiguous lesions and optimize LM percutaneous coronary intervention. However, its ability to provide artefact-free high-quality images of coronary ostia and large segments such as the LM remains uncertain. Methods We included 42 consecutive patients who underwent OFDI, including LM imaging. Each OFDI frame was subdivided into four quadrants and analysed. The number of quadrants with artifacts was calculated within the proximal, mid, and distal LM and the first 5 mm of the left anterior descending artery (LAD) and/or left circumflex artery (LCX). Results The quadrants analysis showed an overall artifact rate of 8.9%, mostly out-of-field (45.1%) or residual blood (44.7%) artefacts. Most artifacts were located in the proximal LM (18.6%) with a stepwise reduction of artifact rates towards distal segments (mid LM 5.8%; distal LM 3.6%, ostial LAD 2.6%, and ostial LCX 0%; p < 0.001). While 20 (48.8%) patients had angiographically visible plaques, OFDI showed plaques in 32 patients (76.2%; p=0.007). Conclusion OFDI can accurately evaluate the LM and detect and assess angiographically unvisualized atherosclerotic plaques providing accurate assessment of >90% of the quadrants of the LM and the ostia of its bifurcation branches. However, artifacts mainly located in the proximal LM and decreasing distally in a stepwise fashion should be considered in the interpretation of OFDI in this setting.


Introduction
Accurate assessment of the left main (LM) coronary artery disease is crucial to determine treatment strategies and improve prognosis [1]. Significant LM stenosis accounts for 4.8% of all coronary angiograms and is rarely isolated [2]. e diagnosis of LM disease is based on coronary angiography in routine practice. However, due to a short vessel segment, lack of a reference segment in presence of diffuse atheroma, frequent overlapping daughter branches, and foreshortening [3][4][5], the accuracy of angiographic determination of LM narrowing may be limited [4]. Percutaneous coronary intervention (PCI) of unprotected LM has been increasingly performed over the last decade [6][7][8]. LM PCI is a high risk and challenging procedure which requires high precision given the large amount of myocardium at risk [1].
Optical coherence tomography (OCT) provides high image quality offering a unique insight into plaque characterisation and detailed structural information pre-and post-PCI [9]. Initially, OCT was not considered suitable for the assessment of the LM because of the large coronary size and poor blood washing [10,11]. e recently developed optical frequency domain imaging (OFDI) provides higher acquisition speed and larger field of view compared with prior generation time-domain OCT which may potentially overcome previous limitations. OCT and OFDI are increasingly used to assess ambiguous lesions and guide PCI [12]. Nevertheless, dedicated studies testing the possible pitfalls of OFDI imaging in LM remain scarce, especially in a population without previous LM stenting. e aim of our study was to assess the quality of ODFI imaging and to describe and analyse its potential artefacts in the assessment of LM arterial wall in coronary artery disease patients with or without detectable angiographic LM lesions.

Study Population.
We retrospectively included all consecutive patients who underwent OFDI including the whole LM and the ostia of its bifurcation branches in our center between May 2015 and August 2018. Patients with previous LM stenting were excluded.

OFDI Image Analysis.
e OFDI procedure (Lunawave ® , FastView ® , Terumo Europe, Leuven, Belgium) was performed as previously described [13]. Images were analysed offline by 2 investigators (VR and IR) using previously validated criteria for OCT plaque characterisation [14][15][16]. We analysed the whole LM from the LM ostium or catheter tip to the ostia of its bifurcation branches defined as the first 5 mm of the left anterior descending artery (LAD) and/or left circumflex artery (LCX). LM length was obtained from OFDI longitudinal reconstructions and defined as the distance between the first distal frame of the LM at crosssectional image and the last proximal LM frame before aorta or catheter visualisation. Proximal, mid, and distal LM were defined, respectively, as the first, the second, and the last third of the whole LM [17]. Reference lumen area and percent area stenosis were calculated as previously described [18]. When atherosclerotic plaque was identified, at least 3 measurements of the intima and media thickness were performed where the plaque was largest. When disappearance of the media was observed, a value of 10 µm (normal value usually observed in our practice) was given to allow calculation of the intima/media ratio. e feasibility of OFDI assessment was measured in the following regions: proximal LM, mid LM, distal LM, and the ostia of its bifurcation branches [17]. For each region, we analysed all frames. Each frame was subdivided into four quadrants: quadrant 1 � 0-90°, quadrant 2 � 90-180°, quadrant 3 � 180-270°, and quadrant 4 � 270-360°. e number of fully assessable quadrants per frame and consequently quadrants with artifacts were calculated [19]. Artifacts were defined as follows: (i) Quadrants out of the field of view: "out-of-screen" loss of image; (ii) Residual blood: suboptimal vessel flushing causing signal-rich blood swirls in the lumen. Residual blood attenuates the OCT light beam and may defocus the beam if red cell density is high; (iii) Sew-up or seam artifacts: rapid movement of the artery or the imaging catheter during the acquisition of a single cross-sectional image causing misalignment along the circumference of the image; (iv) Related to eccentric wire position: when the vessel is large in size and in vessel curvature, the catheter alignment could be noncoaxial, and the resulting image appears elliptical. is makes the measurements inaccurate.

Quantitative Coronary Angiography.
Quantitative coronary angiography (QCA) was performed offline (CAAS II, Pie Medical, Maastricht, Netherlands) using validated quantitative methods [20]. e following angiographic data were calculated: vessels reference diameter, percentage diameter stenosis, and LM length.

Statistical Analysis.
Continuous and categorical variables were expressed as mean ± standard deviation and numbers of patients and percentages and compared between groups identified by the presence or not of at least one artefact using Student's test or the chi square test, respectively. A p value of <0.05 was considered statistically significant. SAS software version 9.4 (SAS Institute, Cary, NC) was used for statistical analysis.

Results
A total of 42 patients fulfilled the inclusion criteria. LM OFDI was performed to evaluate ambiguous LM lesions in 14 and as a part of a long run for other patients. Baseline clinical and angiographic characteristics are reported in Table 1. Most patients presented with myocardial infarction (n � 38; 90.5%) and had single vessel disease (n � 26; 61.9%). OFDI analysis is reported in Table 2. Atherosclerotic disease preferentially affected distal LM (n � 13, 40.6%). e quadrants analysis showed an overall artifact rate of 8.9%, which was significantly different across the considered LM segments (p < 0.001, Figure 1). Most artifacts were located in the proximal LM (18.6%) and rates decreased distally in a stepwise fashion (mid LM 5.8%; distal LM 3.6%, ostial LAD 2.6%, and ostial LCX 0%). ere was a trend towards shorter LM length assessed with OFDI as compared with QCA (10.1 ± 4.5 vs 12 ± 4.38 mm, respectively; p � 0.0504). Most artifacts ( Figure 2) were related to quadrants out of the field of view and residual blood (45.1% and 44.7%, respectively). Other artefacts were sew-up or seam artifacts (7.2%) or related to eccentric wire position (3%). e only variable significantly associated with higher rates of artefacts was the length of LM (11.3 ± 4.8 vs 7.9 ± 2.8 mm; p � 0.016; Supplementary Table S1). e correlation between artifact's rate and LM diameter was low (r � 0.34; p � 0.026).

Discussion
Our study showed that overall more than 90% of the quadrants of LM were adequately assessable using OFDI. OFDI detected almost 60% more atherosclerotic plaques than angiography. Artefact rates were low and significantly different between analysed segments, most artifacts being located in the proximal LM. Artefacts were mainly due to quadrants out of the field of view or residual blood effect. When present, plaques were preponderantly fibrous or fibrocalcific with intimal hyperplasia on OFDI analysis. e angiographic evaluation of LM disease severity can be doubtful or discordant between angiographic views or between operators because of the inherent limitations of angiographic assessment at this site [3][4][5]. OCT and OFDI are increasingly used to assess doubtful lesions and optimize complex PCI [12]. However, their ability to provide highquality images in ostial and large calibre segments such as left main coronary artery remains challenged. Our study showed that overall more than 90% of the quadrants of the LM were adequately assessable by OFDI. Previous studies in the setting of pre-or post-PCI have reported poorer performances of   OCT in assessing the LM [19,21]. e most important pitfall reported in one study before PCI was the ability of FD-OCT to fully assess ostial LM in only 12.5% of cases [21]. Parodi et al., who studied LM stenting in 15 patients, reported that 69 ± 20% of the stent inner area (or 2.7 ± 0.8 quadrants/slice) were analyzable, mostly because of quadrants out of screen [19]. e small cohorts and the different OCT systems may participate to the different reported performances. We used OFDI technology which provides a better signal-to-noise ratio, higher acquisition speed during automatic motorized  pullback, and larger field of view as compared to latter OCT systems [22]. Additionally, our patients did not have tight stenoses as the OFDI catheter may not cross tight calcified lesions in rare cases. Most artifacts were located at the proximal part of the LM, as previously reported [21]. is is mostly due to incomplete blood displacement by iodine contrast which is more difficult to obtain for large and proximal segments and quadrants out of the imaging field of view. Other observed image artifacts were sew-up artifacts as the result of rapid artery or imaging system movements during the acquisition of a single cross-sectional image and eccentricity artefact due to the wire position causing elliptical images that attenuate the signal [22]. Nonuniform rotational distortion, fold-over, air bubbles, and saturation artifacts were not observed in our study but may still occur with OFDI.
OFDI provided a more precise evaluation of LM atherosclerotic plaques undetected by angiography in one-third of our patients. OFDI showed mostly fibrous or fibrocalcific plaques with intimal hyperplasia in concordance with previous intravascular ultrasound studies [23,24]. e lipid core content of LM lesions is less important than in other coronary segments [23,24] and may explain the greater ability of OFDI to assess the LM wall in absence of attenuation of optical signal beyond the lipid-rich plaques. We found that atherosclerotic disease preferentially affected distal LM encroaching LAD or LCX ostia. Indeed, LM disease is rather diffuse than focal [25,26] and this point should be considered when treating LM stenosis, as distal disease is associated with worse outcome [27,28].
Our study showed that OFDI can adequately evaluate LM bifurcation which is the most commonly diseased segment of the LM [26]. On the contrary, operators should consider the limitations of OFDI to assess the ostial LM wall. Tortuous, very large, and short LM may not be ideal for OFDI evaluation. For ostial LM assessment, the choice of the guiding catheter may differ [3]. We preferentially used an extra backup (EBU) guiding catheter which allowed deeper engagement of the LM and consequently adequate contrast flushing. But the true anatomical ostium may have been missed as supported by the shorter LM length observed with OFDI compared to angiography.
is is important for optimal stent sizing as malapposition is more common in the proximal than distal LM [29]. e Judkins left (3.5 or 4) guiding catheter can be more easily positioned at the ostium but at the cost of lower flushing and image quality as well as less support in case of further PCI.

Limits.
Although larger than most studies on the subject, our study is based on a relatively small sized cohort. We did not assess patients with ostial or very severe LM stenosis which are usually not suitable for OFDI assessment.

Conclusion
Our study showed that OFDI can accurately evaluate the LM and detect and assess angiographically unvisualized atherosclerotic plaques, providing accurate assessment of >90% of the quadrants of the LM and the ostia of its bifurcation branches. Most artifacts were located in the proximal LM and their rate decreased distally. OFDI can accurately evaluate LM bifurcation, which is the most commonly diseased segment for LM stenosis, and provides a precise evaluation of LM atherosclerotic plaques. However, out-of-field and residual blood-related artifacts should be considered when using OFDI in the ostial or proximal LM.

Data Availability
e datasets used and/or analysed during the current study are available from the corresponding author upon reasonable request.

Conflicts of Interest
e authors have no conflicts of interest to declare.

Supplementary Materials
Supplementary  Journal of Interventional Cardiology 5