ABSTRACT: Background. Assessing the clinical importance of angiographically intermediate coronary artery stenoses at the time of cardiac catheterization remains a challenge. Spectral analysis of radiofrequency ultrasound backscatter signals, or virtual histology (VH), allows in vivo assessment of plaque composition. This study characterizes the VH composition of fractional flow reserve (FFR)-negative intermediate stenoses and adjacent vessel segments. Methods. Intermediate coronary artery stenoses (> 40% and < 70% diameter stenosis) were assessed by pressure wire. If the FFR was >/= 0.75, percutaneous coronary intervention was deferred and VH was performed on the lesion and adjacent segments using a commercially available system. The primary clinical endpoint was any adverse cardiac event. Results. Thirty-seven intermediate stenoses in 30 patients were studied. The reference vessel size was 3.02 ± 0.71 mm, the QCA diameter stenosis was 52 ± 6% and the FFR was 0.89 ± 0.07. The target stenoses were characterized by VH as: thin-cap fibroatheroma (VH-TCFA; n = 22); fibrous cap atheroma (n = 5), fibrocalcific lesion (n = 7) and pathological intimal thickening (n = 3). The relative contribution of each stenosis plaque component was conserved across adjacent segments (“signature” plaque). Three patients, all with VH-TCFAs at index, had events in the clinical follow-up period of 12 ± 2 months, but only 1 of these patients had an event related to the index stenosis. Conclusions. FFR-negative intermediate stenoses have heterogeneous plaque by VH, but are enriched in VH-TCFAs. Relative plaque composition is conserved along adjacent vessel segments. Although the specificity of VH-TCFA for index stenosis-related events appears low, larger trials are needed to assess the prognostic value of VH in this lesion subset.

       Assessing the clinical importance of angiographically intermediate coronary artery stenoses (> 40% and < 70%) at the time of cardiac catheterization remains a challenge. Since not all intermediate stenoses are hemodynamically significant, determining fractional flow reserve (FFR) by pressure wire measurement is a validated method for determining the physiologic import of an intermediate lesion. FFR represents maximum achievable blood flow to the myocardium supplied by a stenotic artery as a fraction of normal maximum flow. Its normal value is 1.0, and a value of < 0.75 has been reported to identify stenoses associated with inducible ischemia. Both retrospective and prospective studies have shown that deferral of angioplasty in patients with FFR > 0.75 results in low target stenosis event rates.^1–3
       Intravascular ultrasound (IVUS) has demonstrated that the extent of coronary atherosclerosis is underestimated by coronary angiography, and IVUS permits direct measurements of the lumen, coronary artery wall and components of atherosclerotic plaques.^4–6 IVUS-derived minimum lumen area (MLA) < 4.0 mm² and/or minimum lumen diameter (MLD) < 2.0 mm have been shown to predict the probability of an adverse clinical event in follow up.^7,8 Qualitative assessment of atheroma morphology has been shown to have clinical relevance in predicting vulnerable coronary plaques that may result in an acute coronary syndrome, although these studies have been limited by inadequate characterization of plaque components by traditional ultrasound technology.
       Tissue characterization of atherosclerotic plaques by IVUS can be enhanced by analysis of the back-scattered radiofrequency (RF) ultrasound signal, which contains amplitude and frequency.^9–12 In standard IVUS grayscale images, calcified regions of plaque and dense fibrous components generally reflect ultrasound energy well, and thus appear bright and homogenous, whereas regions of low echo-reflectance in IVUS images are usually labeled “soft” or “mixed” plaque. Unfortunately, utilizing such grayscale images to differentiate and quantify plaque components is technically difficult. Spectral analysis of the RF IVUS backscatter data offers an in vivo opportunity to easily assess plaque composition.¹³ The analyzed RF information is used to reconstruct tissue maps to provide “virtual histology” intravascular ultrasound (VH-IVUS) information about the vessel wall and atherosclerotic plaque. Spectral analysis of the RF ultrasound backscatter signals has been shown to have 80–92% accuracy in identifying the four possible basic tissue types in an atherosclerotic plaque: fibrous, fibrolipidic, calcified and lipid necrotic (core).^13,14
       The purpose of this pilot study was to examine the VH-IVUS morphologic characteristics of intermediate coronary artery stenoses and adjacent vessel segments with an FFR > 0.75. Since VH-IVUS is now widely available, the findings reported are relevant to the practicing interventional cardiologist.

Methods
       Study population and trial design. Subjects (n = 30) were enrolled after

Figure 1

Virtual histology morphologic plaque types. Plaques were classified as thin-cap fibroatheromas (TCFA), fibrous cap atheromas (FCA), fibrocalcific lesions (FC) and pathological intimal thickening (PIT).

obtaining informed consent through a protocol approved by our institutional review board. Baseline clinical information was collected. Inclusion criteria for participation were: (1) age >/= 18 years; (2) subjects referred for non-emergency cardiac catheterization; (3) presence of angiographically intermediate stenosis (> 40% and < 70% diameter stenosis by quantitative coronary angiography) in a native coronary artery with a reference diameter > 2.25 mm; and (4) stenosis FFR >/= 0.75. Patients were excluded if they had a stenosis FFR < 0.75 (percutaneous coronary intervention was performed), or if they had any contraindication to IVUS examination. Eligible subjects then went on to have VH-IVUS studies as described below. Clinical follow up was obtained and patients were re-enrolled in the study if they presented for

Figure 2i

Signature” plaque. (A) Relative plaque composition is preserved across a vessel in a given patient, in a pattern unique to each patient. Shown are longitudinal plots of 2 representative vessels (red, lipid core; dark green, fibrous; light green, fibrolipidic; white, calcium). Arrows denote site of maximal luminal stenosis based on IVUS. (i) Predominant pathological intimal thickening; (ii) Increased lipid core with preserved relative contribution of components across the vessel. (B) Composition of plaque at the target stenosis frame compared with the average composition among adjacent frames shows excellent correlation. Shown are percent lipid core correlation plots for index stenosis frame versus the average of the adjacent 2, 20, and 40 frames and the entire vessel for all subjects. (C) Table showing correlation values of percent plaque component at index stenosis frame versus adjacent frames for all plaque components. P < 0.01 for all r-values (next page).

clinically-indicated catheterization during the follow-up period, and repeat FFR and IVUS were performed if possible.
       Angiographic analysis. Intracoronary nitroglycerin (100–200 mg) was administered prior to coronary angiography. Quantitative coronary angiography (QCA) was performed by a trained technician blinded to the VH-IVUS results using a computer-assisted automatic edge detection algorithm (Inturis Allura, version 9.0, Phillips Medical Systems, The Netherlands). With the outer diameter of the contrast-filled catheter as the calibration standard, the minimum lumen diameter in diastole, as measured from orthogonal projections, was recorded. The reference segment diameter was obtained from an

Figure 2ii

angiographically normal segment proximal or distal to the lesion. Lesions were classified by CASS site and the modified American College of Cardiology/American Heart Association classification.¹⁵
       Determination of fractional flow reserve. Pressure wire measurements were performed using a 0.014 inch SmartWire XT/WaveMap pressure guidewire (Volcano Therapeutics, Inc., Rancho Cordova, California). Intravenous heparin sufficient to achieve an activated clotting time of > 200 seconds was administered before the wire was advanced through a standard guiding catheter, normalized and advanced distal to the stenosis. Adenosine was administered to induce maximal hyperemia, either intravenously (140–180 mg/kg/minute) or high-dose intracoronary (30–60 µg in the right, and 60–100 mg in the left coronary artery). FFR was calculated as the minimum ratio of mean hyperemic distal coronary pressure

Figure 2iii

measured by the pressure wire to the mean aortic pressure measured by the guiding catheter.
       Qualitative and quantitative virtual histology-intravascular ultrasound analysis. IVUS was performed using a solid-state, linear array, 20 MHz, 2.9 Fr Eagle Eye IVUS imaging catheter (Volcano Therapeutics). The IVUS catheter was advanced over the pressure wire to a position at least 10 mm distal to the target lesion. The entire artery was then imaged retrograde to a position at least 10 mm proximal to the stenosis, and usually to the aorto-ostial junction at a motorized pullback speed of 0.5 mm/second. IVUS studies (grayscale and VH-IVUS) were acquired and recorded using an IVG3 IVUS digital data recorder (Version 1.3, Volcano Therapeutics, Inc.). Radiofrequency backscatter IVUS data acquisition was gated to the R-wave of the electrocardiogram for determination of “virtual histology” IVUS. Each gated acquisition was defined as a VH-IVUS “frame”, which is the unit of measurement used on the VH-IVUS console. Automatic border detection with manual confirmation was performed on grayscale images for determination of lumen and media-adventitial borders. A color-coded tomographic view of each IVUS image was generated,

C

Table showing correlation values of percent plaque component at index stenosis frame versus adjacent frames for all plaque components. P < 0.01 for all r-values.

quantifying relative contributions of fibro-lipidic (light green), fibrous (green), calcific (white), and lipid necrotic core (red) components of the atherosclerotic plaque. Target stenosis” plaque composition was calculated as an average of the frame with the MLA, as well as of the immediate distal and proximal frames (3-frame average), unless otherwise specified. As previously described, atherosclerotic lesions were classified (Figure 1) as pathological intimal thickening (PIT, mainly fibrotic-fibrolipidic tissue with 0–3% lipid core CSA), fibrocalcific lesions (FC, mainly fibrotic plaques with some calcium and > 3%, but less than or equal to 10% CSA) plaques with overlying fibrous tissue], and VH-derived thin-cap fibroatheromas [VH-TCFA, lipid-rich (> 10% CSA)] plaques with no overlying fibrous tissue).¹⁶ All lesions interrogated in this study fell into one of these four categories. Note that the term “VH-TCFA” is used to distinguish this category from histologic TCFA. Since the axial resolution of VH-IVUS is ~150 microns, the prevalence of TCFA will be overestimated.

Table 1

Baseline clinical characteristics.

       Statistical analysis. Continuous data are reported as mean ± standard deviation. To investigate the possibility that there could be consistent plaque composition over a vessel, we examined the relationship between plaque composition variables at the index stenosis and the averages of these variables over portions of the vessel adjacent to the index stenosis. The averages of adjacent segments were computed over the index stenosis along with the adjacent 2 frames, over the 20 adjacent frames, over the 40 adjacent frames and over the entire imaged vessel. Pearson correlations were computed between the plaque composition variables (calcified, fibrous, fibrolipidic and lipid core) at the index stenosis and these averages for each variable. Pearson correlations were also computed between: (1) HDL, LDL and total cholesterol versus plaque composition at the target stenosis; and (2) FFR and the minimum luminal diameter at the target stenosis. A p-value < 0.05 was considered statistically significant.

Results
       Baseline demographic data are presented in Table 1. Twelve patients (40%) underwent PCI on a lesion in another vessel at the time of study. In addition to the clinical characteristics described, all patients had a left ventricular ejection fraction (LVEF) > 40%, no patients had prior coronary artery bypass graft surgery (CABG),

Table 2

Baseline angiographic, physiologic and VH-IVUS characteristics.

and diabetics were insulin-requiring (n = 2) or on oral hypoglycemics (n = 6). Angiographic, physiologic (FFR) and VH-IVUS data are summarized in Table 2. FFR did not significantly correlate with the contribution of any of the four plaque components to absolute or relative plaque area at the stenosis or in adjacent segments. FFR correlated weakly with the IVUS-derived minimum luminal diameter at the target stenosis (r = 0.35, p = 0.04), but not with the minimum luminal area at the target stenosis (r = 0.27, p = 0.11). Despite fairly homogeneous angiographic stenosis severity (> 40% and < 70%) and FFR values (all subjects >/= 0.75), there was marked heterogeneity in the composition of plaque at the stenosis sites. In order of decreasing frequency, the majority of intermediate stenoses were VH-derived thin-cap fibroatheromas (n = 22), followed by fibrocalcific atheromas (n = 7), fibrous-cap atheromas (n = 5), and pathological intimal thickening (n = 3). The relative contribution of each component in plaque remained relatively constant in the frames adjacent to the site of maximum stenosis, and correlated well at sites more distant from the index stenosis. The observation that plaque components are conserved in their relative contributions to plaque across adjacent segments is presented in Figure 2. The tightest correlation for a plaque component across a vessel was for lipid core. No significant

Table 3

Clinical event rates for intermediate coronary artery stenoses.

relationships between lipid values and plaque composition were found, although this may be due to the fact that the majority of patients were on a statin at the time of study enrollment. Presumably for the same reason, high-sensitivity C-reactive protein did not correlate with the amount of calcium in the vessel, as determined by VH-IVUS.
       Clinical follow up of 12 ± 2 months was available for 26 patients. Three patients (all with VH-TCFAs at index evaluation) experienced events in the follow-up period, but only 1 of these patients had an event at the index lesion (plaque progression resulting in angina requiring PCI). The second patient had plaque rupture at a site distant to the index stenosis, and the third patient had an unwitnessed death at home without the performance of an autopsy. The small study size precludes comment on VH-IVUS as a predictive tool, but the low specificity of TCFA for predicting events may be noted.

Discussion
       There are several important observations to be made from this pilot study of VH-IVUS as applied to intermediate coronary stenoses. First, although the relative plaque composition of intermediate stenoses is variable, the prevalence of VH-derived thin-cap fibroatheromas is high (59%) in this select lesion subset. As expected, patients’ absolute and relative component contributions of plaque were not significantly correlated with FFR, suggesting that plaque composition and physiologic import are not directly related. Since prior work has shown that positive coronary artery remodeling is related to TCFA or fibroatheromatous lesions by VH, the lumen may be relatively spared in these lesions.¹⁶ In addition, it is intuitive that the rheologic effect of a stenosis depends more on lumen morphology than on plaque composition.
       Although IVUS-derived minimum luminal diameter and FFR were weakly correlated in this study, minimum luminal area (presumably a more accurate measure of lumen dimension) did not correlate with FFR. This is likely due to the fact that this study included only flow-insignificant lesions (FFR >/= 0.75) which would weaken the known relationship between IVUS lumen measurements and flow-significant FFR.8 Studies to date have already shown that diabetic patients have significantly more lipid necrotic tissue and calcium than non-diabetic patients.¹⁷ In our study, lipid values did not correlate with plaque composition (likely because of prevalent statin use at baseline), nor did hs-CRP, in spite of prior published findings examining the relationship of hs-CRP to CT coronary calcification.¹⁸ Previous work has also shown that calcium is significantly higher in stable versus unstable angina patients, and that the amount of HDL has a significant, negative correlation with the level of lipid necrotic tissue.¹⁹
       A new observation from this study is that the pattern of the four plaque components in each patient’s index stenosis is unique and remains fairly consistent along adjacent vessel segments, thus solidifying the notion that each patient has a conserved plaque “signature” (Figure 2). Although plaque morphology is variable along a vessel, these data support the concept that in a vessel where plaque does exist, its relative composition is similar to other plaque in adjacent vessel segments. In the few patients in whom we interrogated more than 1 vessel, plaque composition appeared similar between vessels.
       The likelihood of an intermediate stenosis progressing to a clinical event is a complex interaction of environmental, patient and lesion-specific characteristics. Conventional pressure wire measurements and IVUS criteria have enabled clinicians to identify lower-risk subgroups. Several studies have shown that intermediate stenoses can progress to clinical events, but at low rates of 0–6% (Table 3). Thus, deferral of PCI in such patients has been shown to be safe, and preemptive intervention may result in worse clinical outcomes as shown by one study which employed considerable use of stand-alone angioplasty and bare-metal stents. In this study, the overall clinical event rate was 11.5%, with a confirmed target lesion event rate of 3.8%. It is not known if preemptive deployment of a drug-eluting stent will decrease event rates in this population. One can speculate that given real-world target lesion revascularization rates of 4–5% for paclitaxel- or sirolimus-eluting stents, a small absolute risk reduction may be obtained.^20,21
       Study limitations. This study’s limitations include its relatively small size and heterogeneous clinical population. In addition, with the limited axial resolution (~150 microns) of VH-IVUS, our study may overestimate the prevalence of TCFAs, which histologically have a thin fibrous cap with a thickness < 65 microns.²² In our study and others, the concept of VH-derived TCFA has been adopted as a surrogate for histologic-TCFA, and its clinical import remains to be fully determined.²³ Given the limited axial resolution of VH-IVUS to detect a thin fibrous cap, in future analyses involving larger patient cohorts, it may be reasonable to combine the VH-TCFA and FCA categories, although this approach remains to be validated. There is currently no widely accepted classification scheme for VH-derived plaque characterization, and we have adopted a user-friendly, simple categorization in this study that can be readily adopted by clinicians in the cardiac catheterization laboratory. Further refinements of VH-plaque classification will need to be performed in larger series and be based on prior comprehensive histopathological classification schemes.^24,25

Conclusion
       In summary, this study characterizes the VH-IVUS composition of FFR-negative intermediate coronary artery stenoses, and establishes the concept that plaque composition is conserved in adjacent vessel segments. The potential utility of VH-IVUS is great and future studies may allow one to follow the progression or regression of atherosclerotic plaque in different patient populations, determine the efficacy of different systemic therapies with respect to the amount and composition of plaque burden, and gain a better understanding of the underlying effect of plaque composition on the amount and type of neointimal growth after interventional procedures. It is tempting to hypothesize at this point that VH-IVUS will provide additional prognostic capability for identifying intermediate stenoses prone to rapid progression or rupture. The appropriate therapy remains unclear, however. Furthermore, it appears that not all VH-TCFAs will progress to an event, and not all will require prophylactic treatment. Larger observational trials are currently under way to answer these questions.