Assessed by Intravascular Ultrasound Virtual Histology and Myocardial Ischemia Assessed by Quantitative Flow Ratio

Coronary plaque composition may play an important role in the induction of myocardial ischemia. Our objective was to further clarify the relation between coronary plaque composition and myocardial ischemia in patients with chest pain symptoms. The study population consisted of 103 patients who presented to the outpatient clinic or emergency department with chest pain symptoms and were referred for diagnostic invasive coronary angiography. Intravascular ultrasound virtual histology was used for the assessment of coronary plaque composition. A noncalciﬁed plaque was deﬁned as a combination of necrotic core and ﬁbrofatty tissue. Quantitative ﬂow ratio (QFR), which is a coronary angiography-based technique used to calculate fractional ﬂow reserve without the need for hyperemia induction or for a pressure wire, was used as the reference standard for the evaluation of myocardial ischemia. Coronary artery plaques with QFR of ≤ 0.80 were considered abnormal — that is, ischemia-generating. In total, 149 coronary plaques were analyzed, 21 of which (14%) were considered abnormal according to QFR. The percentage of noncalciﬁed tissue was signiﬁcantly higher in plaques with abnormal QFR (38.2 § 6.5% vs 33.1 § 9.0%, p = 0.014). After univariable analysis, both plaque load (odds ratio [OR] per 1% increase 1.081, p < 0.001)

During the past 2 decades, multiple randomized controlled trials have shown that, in terms of clinical outcome, physiology-guided revascularization is superior to revascularization driven by angiographic stenosis severity. 1,2 In addition, angiographic stenosis severity is only modestly associated with the functional consequences of a coronary stenosis. 3 Therefore, interest in the hemodynamic impact, rather than anatomic characteristics, of coronary artery plaques has increased. The "mismatch" between anatomic stenosis severity-assessed by either invasive coronary angiography (ICA) or coronary computed tomography angiography (CTA)-and the hemodynamic impact of a coronary stenosis cannot be fully explained by technical imaging limitations (such as the finite resolution of radiographic imaging). Coronary plaque composition may play an important role in the induction of myocardial ischemia. 4 Although several studies have suggested a pathophysiologic mechanism that underlies the relation between (low-density) noncalcified plaque and inducible myocardial ischemia, conflicting results were found. 5−10 In the present study, our aim was to further clarify the relation between coronary plaque composition and myocardial ischemia in patients with chest pain symptoms. For this purpose, we used intravascular ultrasound virtual histology (IVUS-VH) for the assessment of coronary plaque composition. Quantitative flow ratio (QFR), which is an ICA-based technique used to calculate invasive fractional flow reserve (FFR) without the need for hyperemia induction or for a pressure wire, was used as the reference standard for the evaluation of myocardial ischemia.

Methods
The study population consisted of 103 patients who presented to the outpatient clinic or emergency department of the Leiden University Medical Center with chest pain symptoms. Based on the clinical presentation and/or imaging results, ICA and IVUS-VH were performed for clinical indications to further evaluate the extent and severity of coronary artery disease. Patients with suspected acute or chronic coronary syndrome were evaluated. Clinical data were prospectively entered in the electronic patient file and retrospectively analyzed. The Medical Ethical Committee of the Leiden University Medical Center, The Netherlands, approved this retrospective evaluation of clinically collected data and waived the need for written informed consent.
IVUS-VH was performed during conventional ICA according to standard clinical protocol using a dedicated IVUS console (s5 Imaging System, Volcano Corporation, Rancho Cordova, California). Contraindications for IVUS-VH were (1) severe luminal narrowing, (2) total or subtotal vessel occlusion, and (3) severe vessel tortuosity. After local intracoronary injection of nitroglycerin 200 mg, IVUS-VH was performed with a 20-MHz, 2.9-F phasedarray IVUS catheter (Eagle Eye, Volcano Corporation). The IVUS catheter was positioned distally in the coronary artery, and an automatic motorized pullback was performed at a constant speed of 0.5 mm/s until the IVUS catheter reached the guiding catheter. Images were subsequently stored digitally for further offline analysis.
IVUS-VH analysis was performed using a dedicated software package (QCU-CMS 4.59, Medis Medical Imaging, Leiden, The Netherlands and Volcano software, Volcano Corporation, Rancho Cordova, CA). See Figure 1 for the complete workflow of QFR and IVUS-VH analyses. All IVUS-VH examinations were evaluated by 3 experienced observers who were blinded to the clinical data and QFR analysis. Contour detection of the vessel wall, which was defined by the media-adventitia border, and lumen was automatically performed and manually adjusted if needed. Images were analyzed on a per-plaque basis at the site with the smallest cross-sectional luminal area (i.e., the minimal luminal area site). For the analysis, coronary plaques were defined as structures >1 mm 2 that were within and/or adjacent to the coronary artery lumen and could be clearly distinguished from the vessel lumen. 11 Plaque area was calculated as the absolute difference between vessel area and luminal area, whereas localized plaque load was calculated as plaque area / vessel area £ 100; both plaque and vessel areas were determined at the minimal luminal area site. Radiofrequency backscatter analysis was used to determine coronary plaque composition. A total of 4 types of plaque composition were differentiated: (1) fibrotic tissue (displayed in dark green on the IVUS-VH images), (2) fibrofatty tissue (displayed in light green), (3) necrotic core tissue (displayed in red), and (4) dense calcium tissue (displayed in white). A noncalcified plaque was defined as a combination of necrotic core and fibrofatty tissue. 12 Plaque composition was reported in both absolute values and percentages of plaque area.
QFR analysis was performed offline using a dedicated software package (QAngio XA 3D/QFR, Medis Medical Imaging, Leiden, The Netherlands). The complete QFR workflow, which is based on 3-dimensional (3D) quantitative coronary angiography and computational fluid dynamics, has been described in detail previously. 13 In short, 2 angiographic views of the coronary artery of interest were selected. Angiographic views were required to be ≥25å part and to contain minimum vessel overlap and/or foreshortening. Lumen contours were automatically traced on both angiographic views and were manually adjusted if necessary. Subsequently, a 3D model of the coronary artery was reconstructed based on the selected angiographic views. The thrombolysis in myocardial infarction frame count method was used to calculate contrast flow velocity, which was based on the length of the coronary artery segment and the transport time of the injected contrast medium. For this purpose, the angiographic view with the best image quality for contrast flow was used. Finally, hyperemic flow velocity was calculated based on contrast flow velocity, and QFR was computed. In addition, standard quantitative coronary angiography parameters (i.e., diameter of stenosis, lesion length, area of stenosis, and minimal luminal diameter [MLD]) were available for each coronary plaque. Exclusion criteria for QFR analysis were (1) excessive vessel overlap and/or foreshortening, (2) insufficient angiographic image quality, (3) absence of angiographic views ≥25˚apart, (4) presence of ostial stenosis, and (5) presence of a coronary stent or bypass graft. Coronary artery plaques with QFR of ≤0.80 were considered abnormal.
QFR was determined distal to all coronary plaques detected with IVUS-VH. When multiple plaques were present per vessel, the most proximal plaque was used for the analysis, and the QFR value was obtained distal to this plaque. To ensure that identical plaques were analyzed with IVUS-VH and QFR, coronary artery segments with plaque were assessed according to the modified 17-segment American Heart Association classification. 14 For this purpose, slice thickness for IVUS-VH and 3D reconstruction for QFR were used to define anatomic markers.
The distribution of continuous variables was determined using histograms and normal quantile-quantile plots. Continuous variables are presented as mean § SD or as median and interquartile range, as appropriate, and were compared using the independent-sample Student's t test and Mann-Whitney U test, respectively. Categorical variables are presented as number and percentages and were compared using the chi-square test. A univariable logistic regression analysis was performed to assess the association between IVUS-VH plaque characteristics and abnormal QFR. All variables The American Journal of Cardiology (www.ajconline.org) with p value <0.10 were included in a multivariable analysis. Receiver operating characteristic curves were constructed for the discrimination between normal and abnormal QFR using IVUS-VH plaque characteristics. The receiver operating characteristic area under the curve (AUC) was compared between IVUS-VH plaque characteristics, using the method of DeLong et al. 15

Results
A total of 103 patients (mean age 58 § 10 years, 67% male) who underwent ICA and subsequent IVUS-VH were included. Overall, 68 patients (66%) and 35 patients (34%) presented with suspected acute and chronic coronary syndrome, respectively. A complete overview of the patient characteristics is displayed in Table 1.

Discussion
In this study, we investigated the relation between coronary plaque composition and myocardial ischemia in patients with chest pain symptoms. We showed that the noncalcified plaque area was significantly higher in hemodynamically significant coronary lesions (defined by a QFR ≤0.80) than in nonsignificant lesions (defined by a QFR >0.80). Moreover, an increase in localized plaque load and noncalcified plaque area, but not in fibrotic or dense calcium plaque area, was significantly associated with abnormal QFR. However, by multivariable analysis, only plaque load remained independently associated with abnormal QFR.
In recent years, multiple studies have analyzed the relation between plaque composition and the hemodynamic significance of coronary lesions. In these studies, coronary plaque composition was assessed by either quantitative coronary CTA or IVUS-VH, whereas the presence of myocardial ischemia was determined from positron emission tomography perfusion imaging or invasive FFR. Driessen et al 6 evaluated the effects of plaque load and composition on myocardial blood flow and invasive FFR in a post hoc substudy from the PACIFIC trial. The substudy included 208 patients with suspected coronary artery disease who prospectively underwent coronary CTA, positron emission tomography perfusion imaging, and invasive FFR measurements. It was demonstrated that noncalcified plaque and positive remodeling were significantly associated with reduced myocardial blood flow and invasive FFR, independent of luminal stenosis. Also, Imai et al 7 investigated the relation between coronary plaque characteristics and invasive FFR. The study analyzed nonobstructive lesions, which were defined as <50% stenosis by quantitative coronary angiography, in 108 patients who underwent coronary CTA and ICA. Subsequently, coronary CTA characteristics of nonobstructive lesions with FFR ≤0.80 (i.e., anatomy-physiology mismatch) were compared with those of nonobstructive lesions with FFR >0.80 (i.e., anatomy-physiology match). In the absence of anatomically significant stenosis, the presence of positive remodeling, larger plaque load, and low-attenuation plaque (an indicator of necrotic core on CTA) was associated with abnormal FFR. In addition, Gaur et al 8 performed a post hoc substudy in 484 vessels in 254 patients from the NXT trial to assess the association between coronary stenosis, plaque composition, coronary CTA-derived FFR, and invasive FFR. Interestingly, lowdensity noncalcified plaque was significantly associated with myocardial ischemia assessed by invasive FFR, independent of total plaque volume. Waksman et al 9 evaluated IVUS-VH parameters in relation to invasive FFR in 350 patients with 367 intermediate coronary lesions (defined as 40% to 80% stenosis by ICA). 9 In contrast to results from most coronary CTA studies, plaque composition by IVUS-VH showed no correlation with invasive FFR for the detection of myocardial ischemia. The relation between IVUS-VH parameters and invasive FFR was also assessed by Brown et al 10 in 92 lesions in 89 patients with stable angina. The authors found no relation between IVUS-VH-defined plaque composition and invasive FFR.
FFR is the ratio of the hyperemic myocardial flow in a stenotic territory to the normal hyperemic myocardial flow (i.e., in the absence of coronary stenoses). 16 For FFR measurement, it is essential to induce maximal vasodilation of the epicardial coronary artery (e.g., by nitroglycerin administration) and the microvasculature (e.g., by adenosine administration). Only under these circumstances can FFR be correctly calculated from the ratio of distal coronary pressure to aortic pressure.
Although several studies have shown a relation between coronary plaque composition and invasive FFR, the underlying pathophysiologic link remains unclear. It has been thought that the presence of a necrotic core in noncalcified Values are mean § SD or median (interquartile range). * Noncalcified was defined as a combination of fibrofatty and necrotic core plaque.IVUS-VH = intravascular ultrasound virtual histology; QFR = quantitative flow ratio.
Coronary Artery Disease/Coronary Plaque Composition and Myocardial Ischemia 5 plaque leads to local endothelial dysfunction, oxidative stress, and inflammation. 4,17,18 This may suggest that noncalcified plaques with necrotic core do not allow adequate vasodilation of the epicardial coronary artery during nitroglycerin administration. As a consequence, vasodilation of the stenotic noncalcified coronary segment may be reduced relative to vasodilation of the remainder of the vessel, resulting in an increased pressure decrease and lower FFR value. In our study, QFR was used as the reference standard for myocardial ischemia and was compared with coronary plaque composition defined by IVUS-VH. QFR is a relatively novel technique to calculate FFR without hyperemia induction or the need for a pressure wire. 19 QFR calculation is based on coronary tree reconstruction by 3D quantitative coronary angiography and computational fluid dynamics. As opposed to invasive FFR, QFR computation is derived from angiographic views performed under baseline conditions (i.e., without hyperemia induction). Therefore, coronary flow modification induced by hyperemic agents, which leads to vasodilation of the epicardial coronary artery and microvasculature, has been omitted from QFR calculation. Instead, the hyperemic flow velocity is modeled under Figure 2. IVUS-VH plaque composition, according to MLD quartiles and QFR. In the subgroup of plaques in the lowest MLD quartile, plaques with abnormal QFR had a significantly larger noncalcified plaque load compared with plaques with preserved QFR (38.5 § 7.3% vs 30.8 § 5.7%, p = 0.001). However, compositional plaque load did not significantly differ between plaques with abnormal and normal QFR in the remaining MLD quartile subgroups.  The American Journal of Cardiology (www.ajconline.org) baseline conditions using the contrast flow velocity and is therefore based on the assumption that the hyperemic response to nitroglycerin and adenosine is predictable. 20 In this study, we found a significant relation between the noncalcified plaque area and QFR. However, this relation was not significant after adjustment for localized plaque load in multivariable analysis. These findings are in accordance with a recent substudy of the PACIFIC trial, which showed an independent association of CT-derived adverse plaque characteristics with hyperemic FFR measurements but not with nonhyperemic instantaneous wave-free ratio. 21 This may be explained by the omission of coronary plaque composition-and its potential effect on local coronary vasodilation-from QFR and instantaneous wave-free ratio computation. Our study has several limitations. First, coronary angiography images were retrospectively evaluated for QFR analysis. Therefore, 13% of coronary plaques were excluded from our analysis because of suboptimal image quality. Second, IVUS-VH analysis was not performed for the entire coronary plaque or vessel, but only for the plaque site with the smallest cross-sectional luminal area. Third, when multiple plaques per vessel were present, only the most proximal plaque was used for the analysis, and the QFR value was obtained distal to this plaque. The potential effects of distal coronary plaques on coronary hemodynamics and the QFR value were thereby ignored. Fourth, QFR analysis was performed only under baseline conditions (i.e., without hyperemia induction). Therefore, we were unable to confirm our hypothesis that noncalcified plaques with necrotic core do not allow adequate vasodilation under hyperemic conditions; the noncalcified plaque area showed no independent association with QFR. Importantly, the proposed mechanism that noncalcified plaques induce a higher risk of myocardial ischemia should be regarded as hypothesis-generating.
In conclusion, we showed that the noncalcified plaque area was significantly higher in hemodynamically significant coronary lesions (defined by a QFR ≤0.80) than in nonsignificant lesions (defined by a QFR >0.80). Although an increase in the noncalcified plaque area was significantly associated with an abnormal QFR, this association lost significance after adjustment for localized plaque load. Future research should focus on the underlying pathophysiologic mechanisms of coronary plaque composition that affect myocardial perfusion and its potential applicability in clinical practice.

Disclosures
Dr. Bax received speaker fees from Abbott Vascular. Dr. Wijns is cofounder of Argonauts, is medical advisor to Rede Optimus Research, and received institutional research grants and honoraria from MicroPort. The remaining authors have no conflicts of interest to declare. Figure 3. Receiver operating characteristic curve analysis for discrimination between abnormal (≤0.80) and normal (>0.80) QFR using IVUS-VH plaque characteristics. The AUC was significantly higher for plaque load than for the percentage of dense calcium (difference between AUCs 0.25, p = 0.012) or fibrotic (difference between AUCs 0.17, p = 0.033) tissue. However, no significant difference was found between the AUCs for plaque load and the percentage of noncalcified tissue (difference between AUCs 0.08, p = 0.31).