Author + information
- Received July 7, 2004
- Revision received August 14, 2004
- Accepted September 21, 2004
- Published online June 21, 2005.
- Jacob Schneiderman, MD⁎,
- Robert L. Wilensky, MD, FACC†,⁎ (, )
- Assaf Weiss, BSc‡,
- Eitzek Samouha, BSc‡,
- Lev Muchnik, BSc‡,
- Malca Chen-Zion, PhD‡,
- Mordechay Ilovitch, PhD‡,
- Erez Golan, MSc‡,
- Aharon Blank, PhD‡,
- Moshe Flugelman, MD§,
- Yosef Rozenman, MD, FACC∥ and
- Renu Virmani, MD, FACC¶
- ↵⁎Reprint requests and correspondence:
Dr. Robert L. Wilensky, Hospital of the University of Pennsylvania, 3400 Spruce Street, 9 Gates, Philadelphia, Pennsylvania 19104
Objectives We sought to correlate findings obtained from a self-contained magnetic resonance imaging (MRI) probe with plaque morphology of ex vivo human aortas and coronary arteries.
Background Early detection of thin-cap fibroatheromas (TCFAs) may allow for early preventive treatment of acute coronary syndromes. We developed an intravascular MRI catheter capable of imaging the arterial wall without external magnets or coils by differentiating lipid-rich and fibrotic-rich areas of the atherosclerotic plaque on the basis of differential water diffusion.
Methods Aortic samples (n = 16) and coronary arteries were obtained within 12 h of death. Coronary specimens were intermediate in angiographic severity (30% to 60% luminal narrowing, n = 18). Blinded histologic and immunohistochemical analyses of the tissues were performed and correlated to MRI findings.
Results The 16 aortic lesions included four ulcerated plaques, two TCFAs, two thick-cap fibrous atheromas, two intimal xanthomas, and six adaptive intimal thickenings. The MRI scan correctly correlated with the histologic diagnosis in 15 (94%) of 16 lesions. The 18 coronary lesions included one plaque rupture, three TFCAs, seven thick-cap fibrous atheromas, four fibrocalcific plaques, two intimal xanthomas, and one adaptive intimal thickening. The MRI scan correlated with the histologic diagnosis in 16 of 18 lesions (sensitivity 100%, specificity 89%).
Conclusions The self-contained intravascular MRI catheter successfully identified TCFA and may prove to be an important diagnostic approach to determining the presence of lesions with increased risk of causing death or myocardial infarction.
Coronary plaque rupture with subsequent thrombus formation is the underlying cause of the majority of acute coronary syndromes (1,2). The conversion of an inactive plaque into an unstable, rupture-prone lesion is linked to the interplay between the fibrous cap thickness, necrotic core size, and extent of macrophage infiltration (3,4). A thin fibrous cap covering a large necrotic core is the morphologic characteristic indicative of thin-cap fibroatheroma (TCFA), considered to be a vulnerable plaque. Early diagnosis of TCFAs may lead to more aggressive treatment of symptomatic and asymptomatic coronary artery disease in order to reduce the incidence of death and myocardial infarction (5–9).
Magnetic resonance imaging (MRI) provides detailed tissue characterization of aortic and carotid atherosclerotic plaques and has identified carotid plaques that cause symptoms (10). However, it is difficult to apply MRI to coronary lesions due to cardiac motion caused by myocardial contraction and respiration (11). In addition, the need for external magnets has made MRI evaluations within the catheterization laboratory impractical.
Using MRI combined with pulsed field gradient methods, Toussaint et al. (12) calculated the apparent water diffusion coefficient (ADC) of various components of the atherosclerotic vascular wall. Their results showed decreased and isotropic water diffusion within the atherosclerotic plaque, compared with the fibrous cap and medial smooth muscle layer, a finding that can be exploited, by MRI, to develop an index of arterial wall lipid infiltration. The extent and location of increased vascular lipid infiltration can then be used to determine the presence of a lesion with an increased likelihood of subsequent clinical instability (i.e., a thin fibrous cap with increased luminal lipid deposition overlying a lipid-rich necrotic core). A self-contained intravascular MRI probe was designed to obtain high-resolution MRI, thereby revealing the depth and size of the necrotic core and assess fibrous cap thickness. The current study was performed to evaluate the sensitivity and specificity of this novel intravascular MRI catheter for the detection of human aortic and coronary TCFAs.
The intravascular MRI system (TopSpin Medical Israel Inc., Lod, Israel) consists of a miniature self-contained MRI catheter and a portable control unit (Fig. 1).Magnetic fields generated by the probe located at the tip of the catheter create a sector-shaped sensitive region looking sideways into the artery wall. The MRI probe is characterized by very strong static local gradients in the imaging region, which ranges between 150 and 200 T/m (13).
The MRI catheter provided a depth of view (DOV) imaging in the radial (or x) plane of 250 μm, lateral resolution of a 60° sector, and slice thickness of 2 mm (Fig. 1). To achieve high-resolution imaging, the MRI catheter is stabilized against the arterial wall by gentle inflation of a partially occlusive side balloon. Although evaluation of arterial wall depth >250 μm is possible, the signal-to-noise ratio (SNR) for voxels >300 μm from the probe surface requires long averaging times (>90 s), which potentially prohibit such acquisitions within coronary arteries. In order to evaluate the thickness of the fibrous cap with regard to TCFAs, magnetic resonance (MR) assessment is performed simultaneously in two separate bands: a superficial, luminal 0 to 100 μm band and a deeper 100 to 250 μm band. Physically speaking, the xplane resolution is <1 μm due to the high gradient. However, the particle resolution is limited by the SNR and the available acquisition time, so that the resolution in the present case is 100 μm.
Signal acquisition per sector requires ∼40 s of averaging to obtain an appropriate SNR. The subsequent data readout is instantaneous. The acquisition time was ∼2.5 to 4 min per sector to acquire detailed parametric data (actual Tcor ADC), as opposed to simple diffusion-weighted signal averaging. Intraplaque lipids have a very short T2that results in low SNR, and so the pulse sequence used was CPMG (Carr-Purcell-Meiboom-Gill) with an extremely short time between each echo in the train (denoted TE) of ∼12 μs (14). This allows acquisition of more than a thousand echoes per single T2with accumulation of ∼135,000 echoes over the 40-s averaging period and improves the single-shot SNR by a factor of ∼365. In addition, the strong inherent static gradient created by the probe makes self-diffusion the most dominant factor in echo decay over the echo train.
Magnetic resonance acquisition and data processing
A water diffusion time constant was derived from the acquired MRI measurement by fitting the echo train to an exponential function using a previously described algorithm that calculates nuclear magnetic resonance signals of multi-echo pulse sequence with arbitrary position dependent B0and B1fields, taking into account relaxation and spin diffusion (15). The derived time constant relates specifically to the diffusion coefficient of the tissue and is the basis for the lipid fraction (LF) that represents the percentage of lipid within the evaluated volume. The term “lipid fraction” rather than lipid diffusion is used, as it is more intuitive to clinicians. In order to determine the lipid fraction of the arterial wall, MRI catheter measurements were performed in four quadrants (Fig. 2).Within each DOV, the percentage of lipid was assessed in each quadrant in both the superficial and deep bands simultaneously, and the data were integrated to produce a circular color-coded display (see subsequent text). The resulting image, therefore, represents the sum of the four superficial and four deep field of view (FOV) bands from the four quadrants. The FOV defines the angular, longitudinal, and radial dimensions of the measured volume. The MRI determinants of TCFA were defined as the presence of an increased lipid fraction within the superficial band (0 to 100 μ) of the arterial wall, which in turn denotes the presence of a thin fibrous cap, as well as increased lipid in the deep DOV, which indicates the presence of a necrotic core or an increased concentration of lipid-rich cells. Conversely, the absence of lipid within the superficial DOV indicates a thick fibrous cap, which is associated with more stable lesions. A sizeable necrotic core within the deep band associated with a shallow band devoid of lipid renders the diagnosis of a stable lesion or thick-cap fibroatheroma.
Hearts and aortas were obtained within 8 to 12 h of death from 14 autopsy cases suspected of sudden coronary death. Unfixed tissues were initially assessed with postmortem fluoroscopy for the extent of calcium deposition. Sixteen segments from the proximal aortic arch were selected, representing a variety of noncalcified atherosclerotic lesions. All hearts were screened for coronary atherosclerosis by selective coronary angiography.
Determination of lipid fraction index
Fresh aortic samples were initially evaluated for the presence of atherosclerotic lesions. These samples were then separated into various plaque components to yield homogeneous arterial contents, including normal arterial tissue, fibrous tissue, foam cells, and necrotic cores. The following MRI parameters were used: mean gradient 175 T/m, T1for lipid samples 200 ms and for fibrous samples 450 ms, infinite T2with a TEof 22 μs to obtain the ADC using MR diffusion simulation previously published (15). The T2value was set to infinity, as it has no significant effect on the diffusion, because self-diffusion is the most dominant factor in echo decay (see the previous text). By histopathology, two extremes, with regard to lipid content, were defined: a lipid-rich necrotic core was defined as 100% lipid fraction and a completely fibrous tissue specimen, and devoid of lipid was defined as 0%. A linear scale was then designed so that LF = (TC[measured] − TC[fibrous])/(TC[necrotic core] − TC[fibrous tissue]) and designed to produce a read-out from blue to yellow such that pure blue corresponds to 0% LF and pure yellow to 100% LF.
Samples collected from the aortic arch were highly variable in appearance and included specimens thought to represent a wide range of atherosclerosis: lipid-rich necrotic core lesions (i.e., ulcerated plaques, TFCAs, and thick-cap fibroatheromas), lesions with lipid pools and/or fatty streaks, and adaptive intimal hyperplasia. Selected rectangular segments were washed in saline, trimmed to 1 to 3 cm, mounted on a small tray, and immersed in a saline bath at 37°C. The MRI catheter was approximated sideways against the tissue facing the site of interest. After the acquisition period, the segment was tagged to denote the area of interrogation on the luminal surface. The tissues were then fixed in 4% formalin and processed for histology.
All 14 hearts underwent selective coronary angiography of the left and right coronary arteries. Of the 14 hearts, 7 had a total of 18 intermediate coronary lesions, ranging from 30% to 60% in diameter stenosis. The location of each lesion was marked on the epicardial surface, and the heart was then mounted within a conical saline bath warmed to 37°C. The MRI catheter was introduced through a guiding catheter placed in the coronary artery orifice and advanced into the lesion. Pulsatile saline flow at 37°C was generated within the coronary artery through the guiding catheter. The side balloon of the catheter was gently inflated to approximate the probe against the arterial surface, and MR measurement in one quadrant was obtained. The balloon was deflated and the probe rotated by using a manual rotation device in fixed increments of 90° to provide additional measurements. The circumferential orientation was confirmed using ink markers on the catheter. The exact longitudinal position was confirmed using a transmit loop antenna, which applies external radiofrequency (RF) energy to the arterial surface, allowing for precise localization of the MRI receiving coil within the ex vivo artery. Measurements at four quadrants were performed in most instances (in total, 68 measurements in the 18 lesions). Immediately after MRI evaluation, the examined arteries were removed and arterial segments 3 to 4 mm length containing the interrogated lesion were processed for histologic evaluation.
Arterial segments were fixed in 4% formalin for 8 to 12 h, embedded in paraffin, and serially sectioned at 5 to 6 μ intervals through the area of interest. Sections 100 μ apart were stained with hematoxylin-eosin and Movat’s pentachrome stains. The presence of macrophages was determined by immunohistochemical staining for CD-68. Staining with oil red O was used to document the presence of lipid within the arterial wall within foam cells, cholesterol crystals, and necrotic core (Fig. 3).Lesions on the Movat pentachrome sections were then catalogued according to the recently published classification (4). The TFCAs (vulnerable plaques) were defined as lesions containing necrotic cores covered by thin (<75 μm) fibrous caps. The presence of any portion of the sector of the coronary artery containing an area of fibrous cap <75 μm in thickness was sufficient to classify that arterial segment as TFCA. Fatty streaks (intimal xanthomas) were lesions abundant with foam cells, regardless of fibrous cap thickness. Thick-cap fibroatheromas had a necrotic core, but fibrous caps >75 μ thick. Fibrocalcific lesions contained few foam cells and no necrotic cores but were predominantly fibrous with or without calcification. Adaptive intimal thickening was tissue rich in proteoglycans and collagen and smooth muscle cells without foam cells or necrotic core. Histologic results were compared with MRI results.
Numerical data are presented as the mean value ± SD, unless otherwise noted. An unpaired ttest was used to determine significant difference, and a p value of <0.05 was considered statistically significant.
Determination of lipid fraction index
Significant differences in ADC between fibrous tissue (or fibrous cap), fatty streak, and lipid-rich necrotic core (intact or ruptured) were demonstrated (Table 1).Further studies on aortic samples demonstrated that the MRI measurement at any site within each band demonstrated increased LF values in lesions designated by histology (Fig. 4)as containing a necrotic core (n = 13, LF index: 69 ± 14%) and foam cells (n = 7, LF index: 46 ± 12%; p < 0.001 vs. necrotic core), compared with fibrous tissue (n = 11, LF index: 13 ± 13%; p < 0.0001 vs. necrotic core).
Sixteen aortic specimens, representing a wide range of lesion types, were diagnosed as follows: ulcerated plaques (absence of fibrous cap, n = 4), TCFAs (n = 2), thick-cap fibroatheromas (n = 2), intimal xanthomas (n = 2), and adaptive intimal thickening (n = 6). Data derived from the MRI catheter correlated to histology in 15 (94%) of 16 cases. One thick-cap fibroatheroma was misdiagnosed as TCFA by MRI. Examples are shown in Figure 5.
The MRI diagnosis of lesion characteristics correlated with histologic diagnosis in 16 (89%) of 18 lesions. Lesions were classified as adaptive intimal thickening (n = 1), fatty streak (intimal xanthoma, n = 2), thick-cap fibroatheroma (n = 4), TCFA (vulnerable plaque, n = 3; Fig. 3), ruptured plaque with intraplaque hemorrhage (n = 1), plaque hemorrhage (n = 1), healed plaque rupture (n = 1), and fibrocalcific plaque (n = 5). The MRI diagnosis for every specimen was obtained by integrating lipid fraction data, which was assessed within both bands simultaneously, within each quadrant. Individual results are shown in Table 2.Examples are shown in Figure 6.Results of histology in addition to the aortic data resulted in a sensitivity of 100% and a specificity of 89%.
This study was performed to evaluate the diagnostic capabilities of a novel, self-contained intravascular MRI catheter system to document the presence of atherosclerotic lesions within human coronary arteries and aortae that possess increased intramural lipid. We elected to interrogate intermediate coronary stenoses, as these are the most likely to undergo transformation into an unstable lesion. Such vulnerable lesions often do not exhibit hemodynamic significance but may possess an increased lipid content and macrophage count, making them more susceptible to rupture (16,17). The results demonstrate the potential feasibility of the intravascular MRI catheter in interrogating lesions possessing a large lipid core, thin fibrous cap, and increased macrophage infiltration—hallmarks of a potentially unstable plaque. In addition, analysis of aortic lesion composition demonstrated the ability of the MRI catheter to distinguish between necrotic core, foam cells, and fibrous tissue based on the tissue’s lipid fraction. Hence, discrimination can be made between early atherosclerotic lesions containing primarily foam cells and advanced, potentially vulnerable lesions with relatively large necrotic cores.
Coronary lesions with the propensity of rupture or erosion resulting in acute thrombosis represent the underlying pathophysiologic substrate of acute coronary syndromes. These lesions are frequently <50% in angiographic severity before clinical presentation (2,18–21). Although the diagnosis of vulnerable lesions is made on necropsy, there are examples in which invasive methods of coronary lesion appearance correlated with subsequent clinical events. Uchida et al. (22), using coronary angioscopy in patients with stable angina, demonstrated that glistering yellow lesions, thought to reflect increased underlying lipid content, were more likely to have an acute coronary event in the subsequent year, compared with white lesions. Approximately two-thirds of the lesions that ruptured did so within two months. Yamagishi et al. (23) used intravascular ultrasound to identify lesions with superficial echolucent areas, thought to represent necrotic cores in patients with stable and unstable angina. Of the 12 patients who had an acute coronary syndrome during the two-year follow-up, 10 had echolucent areas (83%) on the initial study, versus only 19 (21%) of 90 patients without a subsequent acute coronary syndrome.
The approach presented in this report is novel and differs fundamentally from conventional MRI with or without intravascular coils. Conventional MRI techniques are based on creating the magnetic fields (static, gradient, and RF) from outside the patient. The highly homogeneous fields and the linear gradients created by such an approach allow for acquisition of large FOV images. Using intravascular reception coils improves the SNR obtained in the immediate surroundings of the coil, whereas other aspects of MRI remain the same (e.g., image coordinates, contrast methods). However, the use of external magnets and field sources results in two fundamental limitations of conventional techniques: relative motion of coronary arteries with respect to gradient fields does not allow for continuous signal integration and limits the in-slice spatial resolution. This poses a significant limitation in imaging of small necrotic cores in coronary arteries. In addition, using large and expensive MRI scanners in dedicated MRI laboratories limits their use in interventional procedures. These limitations are not present in the intravascular MRI catheter system. Furthermore, given the effective approximation of the catheter to the vessel wall, gating techniques to ensure adequate image acquisition are unnecessary.
Intravascular MRI using this probe is not a reflection of the actual morphology of the plaque but provides a simplified spatial representation of its lipid-rich component. The probe cannot differentiate between the fibrous cap and the normal medial layer using the current pulse sequence. This is not unexpected, as Toussaint et al. (12) showed that the fibrous cap and normal arterial media possess similar biophysical constraints, and hence, similar water diffusion coefficients. The core of an atheromatous intermediate lesion has no confining structures and water diffuses isotropically, allowing differentiation of the lipid core from the fibrous cap and the normal medial layer. Differentiation of the fibrous cap and the normal medial layer is not considered clinically important in determining the vulnerability of an arterial lesion (Fig. 6B, quadrant 5).
The limitations of the current MRI catheter are its size (6-F) and the need to compromise coronary blood flow during catheter stabilization with balloon inflation and to mechanically rotate the catheter to achieve circumferential sector assessment of the arterial wall. Current catheter developments have reduced the catheter size to 5.2-F. The presented catheter configuration provides a DOV of 250 μ and a lateral resolution of 60°. Hence, to obtain a circumferential study per tissue slice, it was necessary to perform separate measurements in four quadrants along the arterial circumference. Because the catheter is designed to be effective in intermediate lesions with arterial luminal diameters ranging from 2 to 3 mm, each 60° sector interrogates a circumferential arc of ∼1 mm. For 2- to 3-mm diameter arteries, one would expect interrogation of 4 mm of the 6- to 9-mm circumference, which is equal to 67% of the circumference of a 2-mm artery and 45% of a 3-mm artery. This extent of interrogation may be sufficient to identify TCFAs in vivo, as recent histopathologic data evaluating the circumferential distribution of the necrotic core of TCFAs have shown that vulnerable plaques have an angular span of at least 120° in over 75% of the cases (24).
An important limitation of all techniques evaluating potential vulnerable plaques is that current knowledge is based on pathologic studies showing that coronary instability may be caused by either rupture of TFCAs or plaque erosion. Whether any technique, including intravascular MRI, can document the presence of both pathologic subsets is unknown and awaits prospective clinical studies. Likewise, the validity of determining the exact location of a TCFA in order to prophylactically reduce coronary death or myocardial infarction must be tested in a randomized, prospective clinical trial.
Miniaturization of MRI capabilities into the intravascular catheter, free of external magnets or coils, may make coronary artery assessment of potentially vulnerable, intermediate lesions feasible in clinical practice. Reliable assessment of intermediate lesions may provide important data on the composition of the atherosclerotic lesion, thereby predicting the likelihood of eventual rupture and clinical instability and allowing guidance of local treatment.
Drs. Schneiderman, Wilensky, and Virmani are consultants for Topspin Medical. A. Weiss, E. Samouha, L. Muchnik, M. Chen-Zion, M. Ilovitch, E. Golan, and A. Blank are employees of Topspin Medical.
- Abbreviations and Acronyms
- apparent diffusion coefficient
- depth of view
- field of view
- lipid fraction
- magnetic resonance imaging
- signal to noise ratio
- thin-cap fibroatheroma
- Received July 7, 2004.
- Revision received August 14, 2004.
- Accepted September 21, 2004.
- American College of Cardiology Foundation
- Falk E.,
- Shah P.K.,
- Fuster V.
- Virmani R.,
- Kolodgie F.D.,
- Burke A.P.,
- Farb A.,
- Schwartz S.M.
- Kereiakes D.J.
- Maseri A.,
- Fuster V.
- Casscells W.,
- Naghavi M.,
- Willerson J.T.
- Yuan C.,
- Zhang S.X.,
- Polissar N.L.,
- et al.
- Toussaint J.F.,
- Southern J.F.,
- Fuster V.,
- Kantor H.L.
- Blank A.,
- Alexandrowicz G.,
- Muchnik L.,
- et al.
- Meiboom S.,
- Gill D.
- Zur Y.
- Pasterkamp G.,
- Schoneveld A.H.,
- van der Wal A.C.,
- et al.
- Varnava A.M.,
- Mills P.G.,
- Davies M.J.
- Brown B.G.,
- Gallery C.A.,
- Badger R.S.,
- et al.
- Little W.C.,
- Constantinescu M.,
- Applegate R.J.,
- et al.
- Alderman E.L.,
- Corley S.D.,
- Fisher L.D.,
- et al.
- Yamagishi M.,
- Terashima M.,
- Awano K.,
- et al.
- Virmani R.,
- Burke A.P.,
- Kolodgie F.D.,
- Farb A.