Most cases of myocardial infarction and stroke occur when a thrombus is formed on a previously stable plaque that ruptures.1 Persons at risk typically have no premonitory symptoms, and angiographic studies of coronary arteries in patients with non-fatal acute coronary syndromes (ACS) showed that most such events are due to rapid progression of mild, hemodynamically insignificant lesions.2–4 Pathological and more recent intravascular ultrasound (IVUS) studies of ruptured plaques showed that underlying lesions actually present a large plaque burden, often without much lumen compromise, a positive remodeling, and a thin, macrophage-rich cap that covers a large necrotic core.5–12 IVUS provides high-resolution, realtime, cross-sectional images of blood vessels and accurate images of the extent of atherosclerosis, whereas angiography depicts only a two-dimensional silhouette of the arterial lumen. However, the only consistent IVUS finding in ACS lesions is the presence of positive remodeling: grayscale IVUS is unable to assess the other histomorphological features associated with rupture-prone plaques. It has been suggested that large eccentric plaques containing a superficial echolucent area on grayscale IVUS present an increased risk of future rupture.13 However, there is low inter-observer concordance for echolucent area detection,14 and it is well recognized that IVUS is limited in discriminating fibrous from fatty plaque.15,16
New Intravascular Ultrasound Methods for Plaque Characterization
Post-processing of the backscattered (reflected) radio frequency (RF) IVUS signal has been developed in order to better characterize plaque composition. Tissue characterization refers to computer-assisted methods of analyzing a medical image and identifying the component tissue types. It has been shown that the ultrasound RF signals provide quantitative information on tissue microstructures.17,18 For IVUS, Nair et al. have proposed spectral analysis of IVUS signals combined with classification trees.19 High accuracies (>85%) were reported for differentiating fibrous, fibrofatty, calcified, and necrotic regions using 30MHz, unfocused, mechanically rotating catheters. This approach— known as ‘virtual histology’ (IVUS-VH)—is based on seven spectral parameters (intercept, slope, mid-band fit, and minimum and maximum powers and their corresponding frequencies) extracted from calibrated tissue spectra. Tissue maps are reconstructed where dark green represents fibrotic tissue, light green represents fibrofatty tissue, red represents necrotic core, and white codes for dense calcium (see Figure 1).
This methodology has now been implemented in the Volcano (Rancho Cordova, CA) IVUS clinical scanners that offer near-realtime tissue characterization in vivo using a 20MHz phased-array transducer. However, a peer-reviewed independent assessment of the specificity and sensitivity of IVUS-VH with this new transducer technology is still lacking, while a report in a porcine model recently showed an accuracy for detecting fibrous, fibrofatty, and necrotic core tissue of only 58%, 38%, and 38%, respectively.20 A limitation might be that these porcine lesions, created by liposome injection, do not mimic exactly human atherosclerosis. Other limitations of IVUS-VH are a limited penetration of ultrasound behind large calcified plaques21 and the absence of thrombus in the classification tree.
Kawasaki et al. have described an alternative tissue classification scheme based solely on another RF-derived parameter: the integrated backscatter (IB). This system is presently distributed only in Japan (YD Co. Ltd, Tokyo) and uses the IVUS catheter from Boston Scientific (Fremont, CA) based on a 40MHz single rotating crystal.22,23 IVUS-IB values are divided into five categories to construct color-coded maps: thrombus, intimal hyperplasia or lipid core, fibrous tissue, mixed lesions, and calcification. Comparisons of IVUS-IB with histopathology demonstrated a high sensitivity for characterizing calcification, fibrosis, and lipid pool (100%, 94%, and 84%, respectively) and a high specificity (99%, 84%, and 97%, respectively).24 Other methodologies of IVUS plaque characterization are in various stages of development and are not yet available for clinical use, e.g. angledependent echo-intensity variation to map the fibrous cap of atherosclerotic plaques,25 or wavelet analysis methods.26 Palpography is an alternative method that derives the mechanical properties of rupture-prone plaques by strain evaluation.27 A significant reduction in abnormal strain patterns has recently been reported in a follow-up of 67 patients, mainly in those who presented an ST-elevation myocardial infarction.28 A comparison of palpography and IVUS-VH has shown that the mean strain value is higher in IVUS cross-sections in which a necrotic core is in contact with the lumen.29
Radio Frequency-derived Plaque Characterization in the Clinical Setting
Reports on IVUS-VH and IVUS-IB so far mostly concentrated on the description of the plaque content in different patient populations and/or different grayscale IVUS image subsets.30 Fibrofatty plaque area was significantly greater in lesions with positive remodeling than in lesions with intermediate/negative remodeling: 1.2±0.7mm2 versus 0.8±0.4mm2, p=0.001 (or 26.3±6.6% versus 19.8±5.7%, p=0.001) of total plaque volume. This has been confirmed by another group,31 whereas Rodriguez et al. reported that positively remodeled lesions show a larger necrotic core.32 This calls for caution in interpreting the conclusions of early reports that were derived from observations in a limited number of patients. IVUS-VH is still a methodology under development, and we have seen several iterations of the classification tree, as well as an early implementation using rotating 30MHz IVUS catheters and now 20MHz phased-array ones. The validity of pooling the data from both systems as recently reported remains uncertain.33 Sometimes, conflicting data are even published by the same group of investigators, e.g. reporting once that the area of the necrotic core is not related to plaque size,34 while publishing elsewhere the correlation of necrotic core and plaque sizes.35
IVUS-VH findings in non-culprit vessels in 51 consecutive patients showed that there were no significant differences in relative plaque composition along the vessel with respect to fibrous, fibrofatty, and calcified tissue; however, the percent necrotic core was increased in the first 10mm segment.36 Multivariable regression analysis demonstrated that the distance from the ostium was an independent predictor of relative necrotic core size. This has been confirmed more recently by the same group in 72 patients, where it could also be demonstrated that the proportion of necrotic core is larger in ACS (median 11.4%, interquartile range 5.5; 19.8) than in stable angina (7.3%, 3.2; 12.9, p=0.0013).34 The authors hypothesized that the greater number of ACS events in proximal coronary arteries37 might be related to these axial differences in plaque composition. In 15 patients with versus 15 patients without ACS undergoing directional coronary atherectomy, histopathology of the retrieved atherectomy specimens was compared with that of IVUS-VH. Predictive accuracy of IVUS-VH was 87% for fibrous tissue, 87% for fibrofatty tissue, 88% for necrotic core, and 97% for dense calcium regions. The relative size of the necrotic core was also significantly larger in the ACS group than in the stable angina group (24.5% versus 10.4%, p=0.002).38
Conversely, Sumerly reported that lesions of ACS patients have more fibrotic tissue than stable angina (66±11% versus 61±9%, p=0.034) and less necrotic core (7±6% versus 11±8%, p=0.02).39 Prospective and longitudinal data of IVUS-VH are still missing in order to assess the predictive value of the additional diagnostic information provided by this method. On the other hand, Sano et al. studied prospectively 160 coronary lesions without significant baseline stenosis in 140 patients with angina pectoris.40 During the follow-up period of 30±7 months, 12 plaques caused ACS, 10 of which had IVUS parameters recorded at baseline. These 10 plaques were classified as vulnerable plaques and were then compared with 143 plaques that were classified as stable. Plaque burden (60±9% versus 52±9%, p=0.014), eccentricity (0.70±0.10 versus 0.55±0.17, p=0.013), remodeling index (1.30±0.08 versus 1.16±0.16, p=0.006) and percent lipid area (72±10% versus 50±16%, p<0.0001) were greater in vulnerable than in stable plaques, while percent fibrous area (23±6% versus 47±14%, p<0.0001) was lower in vulnerable than in stable plaques. In another longitudinal study, Kawasaki et al. reported the effect of statin therapy in 52 patients with hyperlipidemia who were randomized to six-month treatment with pravastatin 20mg/day (n=17), atorvastatin 20mg/day (n=18), or diet (n=17).41 In statin-treated patients, there was a significant increase in fibrosis volume (pravastatin: 25.4±6.5% to 28.1±6.1%; atorvastatin: 26.2±5.7% to 30.1±5.5%) and mixed lesion volume (atorvastatin: 25.5±6.6% to 28.7±5.1%) and a significant reduction in lipid volume (pravastatin: 25.5±5.7% to 21.9±5.3%; atorvastatin: 26.5±5.2% to 19.9±5.5%) without an improvement in stenosis severity.
Towards a Classification of Lesion Types
A qualitative assessment scheme that parallels the modified American Heart Association (AHA) classification introduced by Virmani et al.8 has been introduced to better characterize a coronary lesion besides its quantitative content of fibrotic tissue, necrotic core, and other constituents.42,43 A lesion should have a minimum plaque burden >40%. A fibrotic lesion will be comprised of nearly all fibrous tissue and will have <10% necrotic core, <10% dense calcium, and <10% fibrofatty plaque. A fibrocalcific lesion will be mainly fibrous with dense calcium; the necrotic core will be <10% and dense calcium will be >10% of plaque volume.
Pathological intimal thickening (PIT) will be mainly a mixture of fibrous and fibrofatty tissue; necrotic core and dense calcium must be <10%. A fibroatheroma (FA) has a confluent necrotic core occupying >10% of the plaque area in three successive frames. FAs can be subtyped as having single or multiple necrotic cores, with or without an evident fibrotic cap, and with or without dense calcium (>10% plaque volume). Since the axial resolution of IVUS is in the order of 120–200μm and a thin-cap FA on histology has a cap thickness <65μm, we define a thin-cap FA (TCFA) on IVUS-VH as a necrotic core that touches the lumen surface in three consecutive cross-sections for at least one-third of the arc encompassed by the necrotic core.
The PROSPECT trial, with 700 enrolled patients, is the largest prospective natural history study to date utilizing multimodality imaging procedures to attempt to characterize vulnerable plaque and patients at risk for future cardiovascular events. Quantitative coronary angiography, IVUS, IVUS-VH, and palpography, as well as multiple serum markers, were employed for comprehensive patient and plaque characterization. Patients are to be followed for up to five years with repeat imaging in case of events. Proprietary softwares were developed in our core lab at the Cardiovascular Research Foundation that allow co-registration of angiographic, IVUS, IVUS-VH, and palpographic data for each millimeter of the coronary tree imaged. Baseline images of culprit sites responsible for a new ACS event during the follow-up will be compared with the images of stable sites in order to identify which combinations of IVUS, IVUS-VH, and palpography are predictive of future events.
There has never been a better time for IVUS catheters. They recently started to see in color, and this is more than just an esthetic upgrade such as that for television 50 years ago: the additional information shown is extracted from the spectral content of the RF IVUS signal and appears to be related to the atherosclerotic plaque composition.
These new methods are still in their infancy and we will need further clarifications from large-scale cross-sectional studies in order to understand differences in IVUS-RF parameters in several patient groups (diabetes, ACS, etc.) or grayscale IVUS subset lesions. More importantly, we must await the results of prospective natural history studies such as PROSPECT to establish the predictive value of these RF-derived lesion types such as an IVUS-VH TCFA.