| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Experimental Studies |
1 From the Laboratoire de Résonance Magnétique Nuclèaire Unité Mixte de Recherche, Villeurbanne, France (J.M.S., L.C., A.B.); Department of Radiology (J.M.S., P.C.D., J.M.C.) and Centre de Recherche et d’Applications en Traitement d l’Image et du Signal (J.M.S., P.C.D.), Hôpital Cardiovasculaire et Pneumologique L. Pradel, Bron, France; Department of Radiology, Johns Hopkins University School of Medicine, Baltimore, Md (J.M.S.); and Department of Vascular Surgery, Hôpital Edouart Herriot, Lyon, France (A.T.). From the 1999 RSNA scientific assembly. Received December 6, 1999; revision requested January 18, 2000; final revision received August 15; accepted September 12. Supported by a grant from La Fondation pour la Recherche Médicale, Paris, France. Address correspondence to J.M.S., Department of Diagnostic and Therapeutic Imaging, Hôpital Cardio-Vasculaire, B.P. Lyon Montchat, 69394 Lyon Cedex 03, France (e-mail: jserfaty@mri.jhu.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Thirty human arteries and 11 carotid endarterectomy specimens from 31 patients underwent T2-weighted MR imaging (2-T magnet; repetition time, 2,000 msec; echo time, 50 msec) at room temperature. After imaging, Bouin fixative was used to fix 26 arteries, and the other 15 arteries were fixed by means of freezing. Specimens were stained with hematoxylin-eosin and safranin or Sudan lipid stain. MR images and histologic slices were classified independently by two radiologists and a pathologist, respectively, on the basis of the American Heart Association classification.
RESULTS: Results with MR imaging were the following: type I–II plaques, sensitivity of 67% and specificity of 100%; type IV–Va plaques, sensitivity of 74% and specificity of 85%; type Vb plaques, sensitivity of 90% and specificity of 100%; type Vc plaques, sensitivity of 80% and specificity of 90%. No type III plaque was diagnosed in the study. The overall 
value was 0.68.
CONCLUSION: High-spatial-resolution MR imaging with T2 weighting alone can help accurate classification of fibrocalcic plaques (type Vb), but it is subject to limitations for the classification and analysis of other types of atherosclerotic plaques.
Index terms: Arteriosclerosis, 56.754, 9*.7212 • Magnetic resonance (MR), high-resolution, 56.121411, 9*.721 • Magnetic resonance (MR), tissue characterization, 56.121411, 9*.12916 • Magnetic resonance (MR), vascular studies, 9*.12916
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Findings in some studies have suggested that magnetic resonance (MR) imaging with T2-weighted contrast only can help distinguish between lipid cores, fibrotic caps, and calcifications (6–10). Thus with T2-weighted contrast, it should be possible to distinguish the different types of plaques in the AHA classification and thus distinguish unstable plaques from stable plaques. The purpose of this study was to verify the hypothesis that T2-weighted contrast only could help accurate classification of in vitro atherosclerotic plaques according to the AHA classification. In addition, we studied the ability of T2-weighted MR imaging to help assess plaque stability by evaluating the ratio between the diameter of the lipid core and the entire plaque.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Concurrently, to include high-grade-stenosis plaques, 11 advanced plaques were excised from 11 symptomatic patients (eight men and three women; age range, 50–76 years; mean age, 66 years) undergoing carotid endarterectomy. The patients presented with middle cerebral artery ischemia (transient, rapidly regressive) or amaurosis fugax.
Because some authors (11) describe full reversibility of temperature-dependent changes at 4°C, samples were stored at 4°C in sealed tubes for as long as 6 days before imaging. Tubes were fitted to the size of the specimen to ensure that the least amount of air was present and thereby limit dehydration. Before MR imaging, specimens were warmed at room temperature for 30 minutes and cut into 12-mm-long segments with each end perpendicular to the long axis. The tubes were filled with saline solution, because findings in a preliminary study showed no significant changes in T2 for samples placed in NaCl for 5 hours.
After imaging, 26 of the 30 excised specimens were fixed in a mixture composed of 10% formalin, picric acid, and acetic acid (Bouin fixative) to be stained with hematoxylin-eosin and safranin. Hematoxylin-eosin and safranin staining can help characterize calcifications, fibrosis, and large lipid cores. The remaining four excised specimens that showed large lipid cores on MR images were fixed with freezing, because freezing allowed equivalent hematoxylin-eosin and safranin staining and specific lipid staining (either Sudan III or Black Sudan stain). The 11 endarterectomy specimens were also fixed with freezing because specimens from symptomatic patients often contain lipid infiltration.
After fixation, all arteries were brought to the pathologist, who cut the samples into eight equal tissue rings (Fig 1). One or two tissue rings per artery (two if a plaque showed large macroscopic heterogeneity from one histologic slice to another on MR images) were cut into 5-µm-thick slices. After Bouin fixation, 37 histologic slices were stained with hematoxylin-eosin and safranin to characterize calcifications, fibrosis, and large lipid cores. After fixation with freezing, 30 histologic slices were stained at room temperature with hematoxylin-eosin and safranin or Sudan III and Black Sudan to characterize lipid infiltrates in collagenous deposits. A total of 67 tissue rings were analyzed.
|
The MR imaging protocol included a T2-weighted (repetition time msec/echo time msec = 2,000/50) spin-echo (SE) sequence with a spectral bandwidth of 20 kHz. The choice of echo time was based on lipid-to-water ratios (6,12) and T2 (6,13) of fibrous caps and lipid cores, as reported previously. By deriving the difference between [SIL = SI0L x e(-TE/T2L)] and [SIC = SI0C x e(-TE/T2C)]—where SIL is the signal intensity from water protons in the lipid core, SI0L is the signal intensity from water protons in the lipid core at echo time (TE) of 0 msec, T2L is the T2 of the lipid core, SI0C is the signal intensity from water protons in the fibrous cap at echo time of 0 msec, SIC is the signal intensity from water protons in the fibrous cap, and T2C is the T2 of the fibrous cap—it is possible to find the echo time that gives the maximum signal intensity difference between two tissues. For a lipid-to-water ratio of 0.1 in the lipid core (6,12), T2 of lipid cores between 30 and 55 msec, and T2 of fibrous caps between 80 and 200 msec, an echo time of 50 msec is optimal at 1.5 and 2.0 T for the differentiation of lipid cores and fibrous caps. These theoretic findings were confirmed in preliminary in vitro experiments. All images were acquired with a matrix of 128 x 128, section thickness of 1.5 mm, and field of view of 15 mm, which resulted in a spatial resolution of 117 µm. Two signals were acquired per image with an 8.53-minute acquisition time. Three transverse MR imaging sections were obtained on each side of the center histologic slice of the specimen.
Correlation between MR Images and Histologic Slices
Correlation between MR images and histopathologic slices was obtained by dividing the arteries and endarterectomy specimens in half with a transverse cut and then dividing the remaining tissue rings in half again to obtain eight histologic slices (Fig 1). The cut sides of the eight histologic slices all corresponded to the middle of each MR imaging section.
Interpretation Method
The primary objective of our study was to evaluate whether high-spatial-resolution MR imaging could help accurate classification of atherosclerotic plaques according to the AHA classification. Two radiologists (J.M.S., P.C.D.) independently classified 67 MR imaging sections. One pathologist (A.T.) classified, in a blinded study, the corresponding histologic slices according to a simplified AHA classification based on the original AHA classification system modified for use with MR imaging (Table 1). This modification was necessary because MR imaging cannot depict or help differentiate macrophages from muscle cells, hemorrhage from thrombus, and proteoglycan from fibrosis.
|
During the interpretation sessions, the radiologists studied the T2-weighted images for the presence of calcification (area with no signal intensity), lipid core (area with low signal intensity), and fibrosis (area with high signal intensity) and then classified the plaques. These signal intensities were based on the results of two previously published studies (6,13) that reported the signal intensities of plaque for each tissue with SE T2-weighted MR imaging (Table 2).
|
|
The degree of agreement between (a) observers in the classification of MR images, (b) observers in the quantification of lipid cores (readers were considered in agreement if the difference between the two readers was 10% or less), and (c) the pathology reader and one MR imaging reader (J.M.S.) were determined with pairwise statistics. Agreement was interpreted as follows: very good, > 0.81; good, = 0.80–0.61; moderate, = 0.60–0.41; poor, = 0.40–0.21; and bad, < 0.21.
To determine the trend to under- or overestimate the ratio between the diameters of the lipid core and the entire plaque on the basis of T2-weighted contrast, we plotted for each type IV–Va plaque the ratio between the lipid core and the plaques with high-spatial-resolution MR imaging (the interpretation of only J.M.S.) against that at histologic examination. We calculated the linear regression line by using the method of the least squares. In addition, we tested the significance of the slope coefficients by means of a Student t test on the null hypothesis of no relationship between histologic examination and MR imaging (the null hypothesis is that the coefficient of the regression is zero).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Value of T2 Weighting in Classifying Atherosclerotic Plaques
Table 3 summarizes the results of high-spatial-resolution MR imaging and pathologic examination (hematoxylin-eosin and safranin staining) in the classification of 67 histologic slices. Sensitivities and specificities are shown in Table 4. For the diagnosis of type I–II plaques, the two errors were due to similar signal intensities between thin plaques and their adjacent media, which could not be differentiated (Fig 3).
|
|
|
|
|
|
|
|
|
|
|
|
The overall score between the MR imaging and AHA classifications was 0.68, which corresponds to a good agreement. The degree of agreement between observers in the classification of MR images was very good, with a score of 0.85.
Quantification of Lipid Cores with High-Spatial-Resolution MR Imaging
The degree of agreement between observers in the quantification of lipid cores was very good, with a of 0.82. Of the 14 plaques with a true-positive lipid core at T2-weighted MR imaging, six (43%) had a ratio between the diameters of the lipid core and plaque that was overestimated by more than 30% with MR imaging compared with histologic examination (hematoxylin-eosin and safranin staining) (Fig 7). At histologic examination around the lipid core by the pathologist, these six plaques showed lipid infiltration with Sudan staining and intermediate or dense fibrotic tissue with hematoxylin-eosin and safranin staining. At statistical analysis of the data, the null hypothesis that there was no relationship between MR imaging and histologic examination could not be rejected (P = .311). It was not possible to conclude that MR imaging could provide a valuable technique for predicting the ratio between the diameters of the lipid core and plaque on the basis of histologic examination.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Although findings in a recent study suggest that T2-weighted contrast alone is not sufficient to help identify these components (15), the suggestion was based on the analysis of only five arteries, and no sensitivity and specificity rates were given. In addition, experiments were performed at 9.5 T, rather than at the 1.5 T used in many clinical MR imagers, with only endarterectomy specimens. In addition, lipid cores at histologic examination and on MR images were compared without an assessment of their size. Thus, MR imaging was assessed for only its ability to depict the presence of a lipid core, with no consideration of size, an indicator of risk in terms of plaque rupture.
Therefore, the purpose of our study was to provide sensitivities and specificities for T2-weighted MR images alone in the differentiation of fibrosis, lipid cores, and calcifications and evaluation of the size of a lipid core. A simplified AHA classification was used in the evaluation of MR images to assess the evolution of a plaque with a practical method that could be applied easily in vivo to differentiate stable from unstable plaques and young plaques from more evolved plaques. Because recent publications have shown that high-spatial-resolution T2-weighted MR images can be obtained in vivo in human arteries and animal models (16–18), it seemed important to confirm the potential of such a technique to characterize atherosclerotic plaques in a way that can be easily applied clinically.
Value of High-Spatial-Resolution MR Imaging in the Classification of Plaques according to AHA Classification
Our study showed a high accuracy for T2-weighted high-spatial-resolution MR imaging in the classification of atherosclerotic plaques. When the plaques were analyzed by type, however, our study showed important differences in sensitivity and specificity.
Normal arteries (type 0) were easily detected with high-spatial-resolution T2-weighted MR imaging, which yielded intermediate signal intensity in the media and low signal intensity in the adventitia, both of which allowed easy demarcation between the two structures and clear delineation of the arterial contour (Fig 3). The 117-µm spatial resolution in our study provided clear delineation of the different layers, even in the coronary arteries.
Differentiation of small young plaques (type I–II) from normal arteries was more problematic (sensitivity, 67%). Only plaques with a signal intensity different from that of the media were easily detected. Plaques with an intermediate homogeneous signal intensity were confused with the media. Thus, as is true with vessel wall ultrasonography (US), measurement of intimal media thickness with MR imaging was important to avoid false-positive diagnoses.
In advanced plaques (type IV–Va), our results showed that thromboses (with high or low T2-weighted contrast) could lead to additional misinterpretations. As is shown in Figure 5, arterial dissection may give false-positive results for type IV–Va plaques, because the internal media may simulate the fibrous cap and the thrombosis may simulate the core. In addition, a high-signal-intensity thrombosis was misinterpreted as a fibrous cap. MR imaging with specific sequences may help differentiate these three components (15,19).
No such problems were encountered in plaques composed of calcifications (type Vc). High-spatial-resolution T2-weighted MR imaging is an accurate tool for studying calcified plaques; it is much more effective than either US, which is unable to depict anything beyond calcified tissue (20), or radiographic angiography, which is limited by the low contrast between calcifications and fibrotic tissues (21).
Value of High-Spatial-Resolution MR Imaging in Quantifying Lipid Cores
We also evaluated the ability of MR imaging to help accurate quantification of the ratio between the diameters of the lipid core and the entire plaque and thereby determine plaque stability. Our results show that T2 weighting often results in overestimation of this ratio, especially in the case of complex tissues composed of a collagenous fiber matrix infiltrated with extracellular lipids. These fibrolipidic tissues gave low T2-weighted signal intensity and simulated that of lipid cores. The use of a second echo time (15) may be a potential solution to this problem, but the variable infiltration of lipids in the fibrous cap makes such detection difficult.
As a result of our analysis, we found physical differences between lipid cores in histologic slices. When the ratios of core to plaque diameters were concordantly quantified with MR imaging and histologic examination with hematoxylin-eosin and safranin staining (eight plaques), we found that different kinds of lipid cores had identical T2-weighted signal intensity. We differentiated (a) granular lipid cores (very soft and unstable, composed of necrosis with a toothpaste consistency, often surrounded by a circumferential dense fibrous cap) (Fig 4) and (b) organized lipid cores composed of extracellular lipids confined in a more or less dense network of collagenous fibers (less soft, more stable, and resistant to mechanical pressure) (Fig 8). High-spatial-resolution T2-weighted MR imaging helps differentiate these different entities, which may present very different mechanical properties. These findings show again the need for use of MR imaging pulse sequences in addition to the T2-weighted sequence that are more able to specifically depict collagenous fibers and lipidic components (22–24).
|
|
|
Limitations of Our Study
Procurement of plaques of all types was difficult. Fibrocalcic plaques are more often found on iliac and aortic arteries, whereas fibrolipidic plaques are more often found on carotid and coronary arteries (5). To overcome this problem, we studied plaques from all sites, and we excised carotid endarterectomy specimens to allow evaluation of as many type IV–Va plaques as Vb–Vc plaques. Although this choice was not sufficient to obtain a large population of plaques of types I–III, it did help us obtain 19 type IV–Va plaques and 31 type Vb–Vc plaques, which are more important for the assessment of the potential of MR imaging to help differentiate stable and unstable plaques. Results from the lower groups (types I–III) should be interpreted with caution, however, because of the small number of plaques included. An ideal study design would have involved collection of samples until a representative number of samples were included in each of the AHA categories. Such a study would have taken a much longer time, however, because the choice of patients and arteries in our study was random.
All our MR imaging experiments were performed at room temperature, which is lower than body temperature. Some of the plaque lipids tend to liquefy at body temperature, which may result in a higher signal intensity on T2-weighted images in the lipid cores and a lower contrast between lipid cores and fibrous caps. Thus, our results may have overestimated the capability of high-spatial-resolution T2-weighted MR imaging to help differentiate plaques with lipid cores and fibrous caps (type IV–Va). Also, plaque perfusion could not be assessed in vitro, which may have affected the results. When high-spatial-resolution T2-weighted MR images of atherosclerotic plaques become routinely available with clinical imagers, in vivo studies will be needed.
Practical application: On the basis of our data, we conclude that high-spatial-resolution T2-weighted MR imaging has limitations in the classification of atherosclerotic plaques, according to a simplified AHA classification. Although the classification of fibrocalcic plaques (type Vb) may be accurate, the assessment of other types of atherosclerotic plaques, including lipid core size quantification, has limitations. Atherosclerotic plaques comprise too many different tissues (ie, calcification, hemorrhage, thrombosis, granulomatous core, lipid core, fibrolipidic core, loose fibrosis, intermediate fibrosis, dense fibrosis) to be distinguished with only one MR contrast (eg, T2-weighted contrast). The use of multiple MR imaging sequences is necessary to improve characterization and classification of plaques and quantification of lipid cores and fibrous caps.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Abbreviations: AHA = American Heart Association, SE = spin-echo
Author contributions: Guarantors of integrity of entire study, J.M.S., L.C., P.C.D.; study concepts, J.M.S., P.C.D., L.C., A.B.; study design, J.M.S., P.C.D., L.C.; literature research, J.M.S.; clinical studies, P.C.D., J.M.S., J.M.C.; experimental studies, J.M.S., L.C., A.T., J.M.C.; data acquisition, J.M.S., L.C.; data analysis/interpretation, J.M.S., A.T., P.C.D.; statistical analysis, J.M.S., P.C.D.; manuscript preparation, J.M.S., P.C.D.; manuscript definition of intellectual content, J.M.S.; manuscript editing, J.M.S.; manuscript revision/review, J.M.S., P.C.D., L.C.; manuscript final version approval, J.M.S., P.C.D.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  T. Saam, T. S. Hatsukami, N. Takaya, B. Chu, H. Underhill, W. S. Kerwin, J. Cai, M. S. Ferguson, and C. Yuan The Vulnerable, or High-Risk, Atherosclerotic Plaque: Noninvasive MR Imaging for Characterization and Assessment Radiology, July 1, 2007; 244(1): 64 - 77. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. Touze, J.-F. Toussaint, J. Coste, E. Schmitt, F. Bonneville, P. Vandermarcq, J.-Y. Gauvrit, F. Douvrin, J.-F. Meder, J.-L. Mas, et al. Reproducibility of High-Resolution MRI for the Identification and the Quantification of Carotid Atherosclerotic Plaque Components: Consequences for Prognosis Studies and Therapeutic Trials Stroke, June 1, 2007; 38(6): 1812 - 1819. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. M. Malleier, O. Oskolkova, V. Bochkov, I. Jerabek, B. Sokolikova, T. Perkmann, J. Breuss, B. R. Binder, and M. Geiger Regulation of protein C inhibitor (PCI) activity by specific oxidized and negatively charged phospholipids Blood, June 1, 2007; 109(11): 4769 - 4776. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Reiter, R. Horvat, S. Puchner, W. Rinner, P. Polterauer, J. Lammer, E. Minar, and R.A Bucek Plaque Imaging of the Internal Carotid Artery--Correlation of B-Flow Imaging with Histopathology AJNR Am. J. Neuroradiol., January 1, 2007; 28(1): 122 - 126. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. S. Pessanha, K. Potter, F. D. Kolodgie, A. Farb, R. Kutys, E. K. Mont, A. P. Burke, T. J. O'Leary, and R. Virmani Characterization of Intimal Changes in Coronary Artery Specimens with MR Microscopy Radiology, October 1, 2006; 241(1): 107 - 115. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. Saam, J. Cai, L. Ma, Y.-Q. Cai, M. S. Ferguson, N. L. Polissar, T. S. Hatsukami, and C. Yuan Comparison of Symptomatic and Asymptomatic Atherosclerotic Carotid Plaque Features with in Vivo MR Imaging. Radiology, August 1, 2006; 240(2): 464 - 472. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Cai, T. S. Hatsukami, M. S. Ferguson, W. S. Kerwin, T. Saam, B. Chu, N. Takaya, N. L. Polissar, and C. Yuan In Vivo Quantitative Measurement of Intact Fibrous Cap and Lipid-Rich Necrotic Core Size in Atherosclerotic Carotid Plaque: Comparison of High-Resolution, Contrast-Enhanced Magnetic Resonance Imaging and Histology Circulation, November 29, 2005; 112(22): 3437 - 3444. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Ouhlous, H. Z. Flach, T. T. de Weert, J. M. Hendriks, M. R. H. M. van Sambeek, D. W. J. Dippel, P. M. T. Pattynama, and A. van der Lugt Carotid Plaque Composition and Cerebral Infarction: MR Imaging Study AJNR Am. J. Neuroradiol., May 1, 2005; 26(5): 1044 - 1049. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. V. Hofmann, R. P. Liddell, J. Eng, B. A. Wasserman, A. Arepally, D. S. Lee, and D. A. Bluemke Human Peripheral Arteries: Feasibility of Transvenous Intravascular MR Imaging of the Arterial Wall Radiology, May 1, 2005; 235(2): 617 - 622. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J.K. Lovett, J.N.E. Redgrave, and P.M. Rothwell A Critical Appraisal of the Performance, Reporting, and Interpretation of Studies Comparing Carotid Plaque Imaging With Histology Stroke, May 1, 2005; 36(5): 1085 - 1091. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Saam, M.S. Ferguson, V.L. Yarnykh, N. Takaya, D. Xu, N.L. Polissar, T.S. Hatsukami, and C. Yuan Quantitative Evaluation of Carotid Plaque Composition by In Vivo MRI Arterioscler. Thromb. Vasc. Biol., January 1, 2005; 25(1): 234 - 239. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Kampschulte, M.S. Ferguson, W.S. Kerwin, N. L. Polissar, B. Chu, T. Saam, T.S. Hatsukami, and C. Yuan Differentiation of Intraplaque Versus Juxtaluminal Hemorrhage/Thrombus in Advanced Human Carotid Atherosclerotic Lesions by In Vivo Magnetic Resonance Imaging Circulation, November 16, 2004; 110(20): 3239 - 3244. [Abstract] [Full Text] [PDF]
|
|
30%) with MR imaging for 43% of the arteries. The graph shows the linear regression line and the linear regression equation and its regression R2 coefficient value.
