Note: This copy is for your personal non-commercial use only. To order presentation-ready copies for distribution to your colleagues or clients, contact us at www.rsna.org/rsnarights.

REVIEW The Vulnerable, or High-Risk, Ⅲ Atherosclerotic Plaque: Noninvasive MR Imaging for Characterization and Assessment1 REVIEWS AND COMMENTARY Tobias Saam, MD “Vulnerable” plaques are atherosclerotic plaques that have Thomas S. Hatsukami, MD a high likelihood to cause thrombotic complications, such Norihide Takaya, MD, PhD as myocardial or . Plaques that tend to Baocheng Chu, MD, PhD progress rapidly are also considered to be vulnerable. Hunter Underhill, MD Besides luminal , plaque composition and mor- William S. Kerwin, PhD phology are key determinants of the likelihood that a Jianming Cai, MD, PhD plaque will cause cardiovascular events. Noninvasive mag- Marina S. Ferguson, MT netic resonance (MR) imaging has great potential to en- Chun Yuan, PhD able characterization of atherosclerotic plaque composi- tion and morphology and thus to help assess plaque vulner- ability. A classification for clinical, as well as pathologic, evaluation of vulnerable plaques was recently put forward in which five major and five minor criteria to define vulner- able plaques were proposed. The purpose of this review is to summarize the status of MR imaging with regard to depiction of the criteria that define vulnerable plaques by using existing MR techniques. The use of MR imaging in animal models and in human disease in various vascular beds, particularly the carotid , is presented.

௠ RSNA, 2007

1 From the Departments of Radiology (T.S., N.T., B.C., H.U., W.S.K., J.C., M.S.F., C.Y.) and Surgery (T.S.H.), Uni- versity of Washington, Seattle, Wash; and Surgical Ser- vice, VA Puget Sound Health Care System, Seattle, Wash (T.S.H.). Received October 31, 2005; revision requested December 14; revision received January 17, 2005; ac- cepted March 2; final version accepted July 17; final re- view and update by T.S. January 9, 2007. Address cor- respondence to T.S., Department of Clinical Radiology, Grosshadern Campus, University of Munich, Marchioni- nistr 15, 81377 Munich, Germany (e-mail: Tobias_ [email protected]).

஽ RSNA, 2007

64 Radiology: Volume 244: Number 1—July 2007 REVIEW: MR Imaging of the Vulnerable Atherosclerotic Plaque Saam et al

omplications of cardiovascular must be able to help determine the pa- will focus on the current status of MR as disease, including stroke, myocar- tient-specific risk of experiencing a car- a diagnostic imaging method to help Cdial infarction, and sudden car- diovascular event. identify the features of vulnerable diac death, are the most common Imaging methods have the potential plaques. The structure of this article is causes of death in the world (1). Ath- to not only be used as a screening tool based on two recently published con- erosclerotic disease accounts for ap- for the presence of but sensus documents (3,4) by a group of proximately 25% of ischemic also to help distinguish stable from vul- experienced researchers in atheroscle- (2) and for the majority of myocardial nerable plaques and ultimately to distin- rosis, including pathologists, clinicians, and sudden cardiac deaths guish patients with low versus those molecular biologists, and imaging scien- (3,4). Despite major advances in the with high risk of cardiovascular compli- tists, in which the key features of the treatment of atherosclerosis, a larger cations. Methods commonly used in vulnerable plaque were defined. The au- percentage of individuals with the dis- atherosclerosis imaging include B-mode thors of those documents argue that ease who are apparently healthy die ultrasonography (US), intravascular knowledge of luminal diameter is not without prior symptoms (3). The chal- US, conventional , com- sufficient to determine the vulnerability lenge for screening and diagnostic puted tomography (CT), and magnetic of an atherosclerotic lesion, and they methods is to identify patients at high resonance (MR) imaging. Each of these propose five major and five minor crite- risk who have lesions that are vulnera- imaging modalities has advantages and ria for the detection of vulnerable ble to thrombosis, so-called vulnerable disadvantages, and all have been re- plaques. These plaque features were plaques, before the event occurs. To tai- viewed elsewhere (5,6). This article will based on studies of coronary arteries lor and improve treatment strategies, focus on the particular potential of MR and included a thin cap with a large these screening and diagnostic methods imaging to depict the key features of the lipid-necrotic core, active inflammation, vulnerable plaque. MR imaging is well a fissured plaque, stenosis greater than suited for this role because it is noninva- 90%, endothelial denudation with or Essentials sive, does not involve ionizing radiation, without superficial aggregation Ⅲ The challenge for imaging meth- enables visualization of the vessel lumen and fibrin deposition, endothelial dys- ods is to enable identification of and wall (7,8), and can be repeated se- function, calcified nodules, intraplaque patients with high-risk lesions rially to track progression or regres- hemorrhage, glistening yellow plaques that are vulnerable to thrombosis, sion. Furthermore, the excellent soft- (seen at angioscopy), and outward re- so-called vulnerable plaques, be- tissue contrast provided by MR imaging modeling (see Table 1). fore the occurrence of cardiovas- allows evaluation of compositional and In this review, we will discuss the cular complications. morphologic features of atherosclerotic role of MR imaging in the identification Ⅲ Noninvasive MR imaging has plaques (9–15). This information is cru- of each of the features that define the great potential to help character- cial because, besides luminal stenosis, vulnerable plaque, as proposed in the ize atherosclerotic plaque compo- plaque composition and morphology are consensus documents (3,4). sition and morphology and thus to key determinants of a plaque’s vulnera- enable assessment of plaque vul- bility with regard to causing cardiovas- Major Criteria nerability. cular events (3,4). Ⅲ Serial MR imaging of atheroscle- Authors of several review articles rotic plaques can provide useful (7,8,16) have described in detail the Thin Cap with Large Lipid-Necrotic Core insights on the natural history of technical aspects of the MR imaging Virmani et al (17) defined the fibrous vulnerable plaques and might be characterization of human atheroscle- cap as a distinct layer of connective tis- useful in identifying plaques that rotic plaque, including hardware con- sue completely covering the lipid-ne- are progressing toward a vulnera- siderations, imaging sequences, and im- crotic core. The lipid-necrotic core, ble state. aging protocols. Results of the accuracy which is frequently surrounded by mac- Ⅲ A recent prospective study dem- of MR imaging, compared with histo- rophages, consists of large amounts of onstrated that certain vulnerable logic findings, for measurements of extracellular lipid, crystals, plaque features identified on MR plaque burden, tissue characterization, and necrotic debris (17,18). Lesions images are associated with the and fibrous cap status have also been occurrence of subsequent cere- reported (7,8,16). Choudhury et al (7) brovascular events. focused on present and future MR appli- Published online Ⅲ Most of the plaque imaging data cations for the characterization of ath- 10.1148/radiol.2441051769 are based on larger vessels, such erosclerotic plaques, including real-time Radiology 2007; 244:64–77 as the carotid arteries, and fur- vascular intervention, new contrast ther advances in temporal and agents, and molecular imaging. Abbreviation: TOF ϭ time of flight spatial resolution are needed for The present review is a continuation coronary plaque imaging. of previous review articles (8,16) and Authors stated no financial relationship to disclose.

Radiology: Volume 244: Number 1—July 2007 65 REVIEW: MR Imaging of the Vulnerable Atherosclerotic Plaque Saam et al

with a large lipid-necrotic core and a identified the major components of ca- wall, was similar on MR images and thin fibrous cap are considered to be rotid atherosclerotic plaque on MR im- histologic specimens, with values of most likely to rupture (19). ages and serially sectioned carotid end- 30.1% Ϯ 12.5 (standard deviation) and MR imaging is able to help charac- arterectomy specimens, respectively 32.7% Ϯ 12.3, respectively. Further- terize all major plaque components, in- (Fig 1). MR measurements of the lipid- more, area measurements were strongly cluding the lipid-necrotic core, by de- necrotic core did not differ significantly correlated (r ϭ 0.85; P Ͻ .001). picting particular combinations of signal from findings on histologic specimens The ability of MR imaging to help intensities of each component on im- (23.7% vs 20.3%; P ϭ .1), and a strong determine the status and thickness of ages obtained with different contrast correlation between MR and histologic the fibrous cap will be discussed in the weightings (see Table 2 for details). Ini- area measurements was found (r ϭ following section. tial experiments involving ex vivo imag- 0.75; P Ͻ .001) (9). ing of endarterectomy specimens by us- Comparison of T1-weighted images Fissured Plaque ing T1-, T2-, and intermediate-weighted obtained before and after contrast ma- Virmani et al (17) described plaque rup- imaging demonstrated that tissue com- terial enhancement may improve the ac- ture as an area of fibrous cap disruption ponents, including the lipid-necrotic curacy of MR for differentiating the whereby the overlying is in core, hemorrhage, and calcification, lipid-necrotic core from surrounding fi- continuity with the underlying lipid- could be detected with sensitivities and brous tissue. The lipid-necrotic core necrotic core. Typically, these lesions specificities ranging from 84% to 100% shows little if any enhancement on post- have a large lipid-necrotic core and a (12). Translation of these findings to in contrast T1-weighted images, and dif- disrupted fibrous cap infiltrated by mac- vivo imaging showed that the lipid-ne- ferentiation from the surrounding en- rophages and lymphocytes; smooth crotic core and intraplaque hemorrhage hancing fibrous tissue is thereby facili- muscle cell content within the fibrous could be identified with a sensitivity of tated (21). Using pre- and postcontrast cap at the rupture site may be sparse 85% and a specificity of 92% (20). The T1-weighted images, Cai et al (22) (17). ability of MR imaging to demonstrate showed that the lipid-necrotic core, Hatsukami et al (15) reported the plaque composition in vivo enables clas- measured as proportion of the vessel use of a three-dimensional TOF bright- sification of human carotid atheroscle- rotic plaque, according to criteria devel- Table 1 oped by the American Heart Associa- tion. Cai et al (14) showed that in vivo Criteria for Defining Vulnerable Plaques on the Basis of the Study of “Culprit” Plaques MR imaging can help identify sections Type of Criteria Criterion that have either a lesion with a confluent extracellular lipid-necrotic core (Ameri- Major Active inflammation: and and sometimes T-cell infiltration can Heart Association type IV lesion) or Thin cap with large lipid-necrotic core a lipid-necrotic core covered by a fi- Endothelial denudation with superficial platelet aggregation brous cap (type V lesion), with a sensi- Fissured plaque tivity of 84% and a specificity of 90%. Stenosis Ͼ90% While initial studies showed that Minor Superficial calcified nodule MR imaging is able to depict the lipid- Glistening yellow plaque seen at angioscopy necrotic core, recent reports have fo- Intraplaque hemorrhage Endothelial dysfunction cused on quantifying it. In an ex vivo MR Outward (positive) remodeling imaging study of carotid endarterec- tomy specimens by Clarke et al (10), who used diffusion-weighted imaging in Figure 1 addition to T1-, intermediate-, and T2- weighted imaging, pixels were classified into different plaque components and were then compared with histologic specimens. On the basis of this classifi- cation, the sensitivity and specificity for necrotic tissue were 83.9% and 75%, respectively. Quantification of the lipid- necrotic core was recently translated to in vivo plaque imaging by using TOF and Figure 1: Pie charts show plaque composition as percentage of vessel wall area, calculated per and T1-, T2-, and intermediate-weighted se- then averaged across all arteries, for MR imaging and histologic specimens. Percentages were compared by quences (9). In a study with 40 subjects, using paired t tests. (Reprinted, with permission, from reference 9.) a radiologist and a pathologist manually

66 Radiology: Volume 244: Number 1—July 2007 REVIEW: MR Imaging of the Vulnerable Atherosclerotic Plaque Saam et al

blood imaging technique (multiple over- imaging has high test sensitivity (81%) weighted black-blood spin-echo se- lapping thin slab angiography, or and specificity (90%) for the identifica- quences and gadopentetate dimeglu- MOTSA [23]) to help identify a rup- tion of a thin or ruptured cap. Trivedi et mine–enhanced T1-weighted images. tured fibrous cap in atherosclerotic hu- al (25) used a short-tau inversion-recov- The ␬ value for agreement between the man carotid arteries in vivo. By using ery sequence to quantify the fibrous cap number of tissue components demon- preoperative images of 22 consecutive and lipid-necrotic core of 25 recently strated by MR imaging and identified at endarterectomy patients, the in vivo symptomatic patients and correlated pathologic examination was 0.79. These state of the fibrous cap was character- the results with the excised carotid end- components included the fibrous cap, ized on the basis of its appearance on arterectomy specimens. The authors thrombus, and/or the lipid-necrotic MR images as being intact and thick, showed good agreement between MR core. The authors found that contrast intact and thin, or ruptured (Fig 2). The and histologic quantifications of both enhancement improved the delineation authors found a high level of agreement the fibrous cap and the lipid-necrotic of the fibrous cap, which was found to between the MR findings and the histo- core content, with good interobserver be hyperintense relative to the lipid-ne- logic state of the fibrous cap, with Co- agreement (intraclass correlation coef- crotic core on T2-weighted images. hen ␬ϭ0.83 (95% confidence inter- ficients of 0.88–0.94) (25). Kramer et Wassermann et al (27), in a study of val ϭ 0.67, 1.00) and weighted ␬ϭ al (26) studied aortic atherosclerotic nine subjects with carotid atherosclero- 0.87. An MR investigation by Mitsumori plaques in 23 patients with abdominal sis, found that gadolinium-enhanced T1- et al (24) demonstrated that in vivo MR aortic aneurysm by using T1- and T2- weighted images helped discriminate

Table 2

MR Parameters in Selected Studies Setting* Magnet (T) Contrast Weighting† Contrast Agent‡ Voxel Size (␮m)

Human carotid (13) 1.5 Dual-echo T2 (R-R interval/20, 55), T1 (500/18) None 390 ϫ 390 ϫ 5000 Ex vivo human carotid (12) 9.4 Intermediate to T2 (1000–2000/13–50), T1 (300–700/13) None 256 ϫ 256 ϫ 500 Human carotid (15) 1.5 3D multiple overlapping thin-slab MR angiography None 510 ϫ 510 ϫ 1000 (34 or 22/2.9–4.4) Dogs (42) 1.5 3D T1 (24/8.1), 3D phase-contrast MR angiography (15/5.3) Fibrin-targeted Not applicable nanoparticle Rabbit aorta (54) 1.5 3D MR angiography (6.7/1.6) USPIOs 1100 ϫ 1000 ϫ 1400 Human coronary arteries (62) 1.5 3D MR angiography (2.4/8.8) None 700 ϫ 1000 ϫ 1500 In vitro human carotid (11) 2.0 T2 (2000/50) None 117 ϫ 117 ϫ 1500 Human coronary arteries (103) 1.5 3D black-blood spiral acquisition (30/2) None 1000 ϫ 1000 ϫ 1000 Swine carotid (37) 1.5 T1 (700/11), T2 (2000/42) None 470 ϫ 470 ϫ 3000 Human carotid and aorta (101) 1.5 Intermediate (2 R-R intervals/12), T2 (2 R-R intervals/45) None 4692–7802 ϫ 3000 Ex vivo human carotid (10) 1.5 T1, T2, intermediate, diffusion-weighted, fast imaging employing None 156 ϫ 156 ϫ 200 steady state (see reference 10 for details) ␣ ␤ ϫ ϫ Rabbit aorta (56) 1.5 T1 (380/11) v 3-integrin–targeted 250 250 5000 nanoparticles Human carotid (55) 1.5 T1 (41–44/8.0–9.2), T2* (R-R interval/20), intermediate USPIOs 208 ϫ 256 ϫ 5000 (2–3 R-R intervals/20) Human carotid (86) 1.5 T1 turbo spin echo (900/10.3), T1 turbo field echo (570/14) None 390 ϫ 490 ϫ 2500 Human aorta (27) 1.5 T1 breath hold (R-R interval/32), T2 breath hold None 1210 ϫ 620 ϫ 7000 (2 R-R intervals/80) Swine coronary arteries (43) 1.5 3D turbo field echo (3.8/1.9), 3D inversion recovery MR Fibrin-targeted agent 1250 ϫ 1250 ϫ 3000 angiography (4.7/1, inversion time, 4 msec) Human coronary arteries (64) 1.5 Whole-heart 3D steady-state free precession MR angiography None 700 ϫ 1000 ϫ 3000 (4.7–31/1.7–2.3) Human carotid (9) 1.5 T1 (800/9.3–11), intermediate to T2 (3–4 R-R intervals/10–70), None 600 ϫ 600 ϫ 2000 3D TOF (23/3.5) Human aorta (104) 1.5 Intermediate (2 R-R intervals/10), T2 (2 R-R intervals/60) None 780 ϫ 780 ϫ 5000 Human carotid (49) 1.5 Two-dimensional T1 spoiled gradient-recalled echo (100/3.5) Gadolinium chelate 600 ϫ 800 ϫ 3000 Human carotid (22) 1.5 Enhanced and unenhanced T1 (800/9.3–11) Gadolinium chelate 600 ϫ 600 ϫ 2000

* Except where otherwise noted, studies were in vivo. Number in parentheses is the reference number. † Data in parentheses are repetition time msec/echo time msec. TOF ϭ time of flight, 3D ϭ three-dimensional. ‡ USPIO ϭ ultrasmall paramagnetic iron oxide.

Radiology: Volume 244: Number 1—July 2007 67 REVIEW: MR Imaging of the Vulnerable Atherosclerotic Plaque Saam et al

Figure 2

Figure 2: Transverse MR images depict fibrous cap rupture (arrows) in right common carotid artery. The hyperintense region (chevron) on TOF (repetition time msec/ echo time msec, 23.0/3.8) and T1-weighted (T1W) (800/9.3) images and hypointense region on intermediate-weighted (IMW) (2500/10.0) and T2-weighted (T2W) (2500/39.9) images is a lipid-necrotic core with type I hemorrhage. Parts of remaining fibrous cap (arrowheads) are visible on TOF image. the fibrous cap from the lipid-necrotic occlusion and acute cardiovascular zation were clearly depicted on T1- and core, with a contrast-to-noise ratio as events, or may layer and orga- T2-weighted MR images and were ex- good as or better than that of T2- nize, further contributing to plaque pro- pressed as alterations in signal intensity weighted MR images but with approxi- gression (17,29). In this section, we will relative to the signal intensity of adja- mately twice the signal-to-noise ratio discuss the ability of MR imaging to help cent muscle. Authors of another study (postcontrast images, 36.6 Ϯ 3.6; T2- identify platelet aggregation or fibrin (37) showed the potential of gadolin- weighted images, 17.5 Ϯ 2.1; P Ͻ .001). deposition. (The ability of MR to help ium-enhanced MR imaging to help dis- Cai et al (22) used unenhanced T1- evaluate endothelial denudation or dys- criminate different types of intracardiac weighted and contrast-enhanced T1- function will be discussed in the section clots (subacute and organized) in the weighted images to measure the intact Endothelial Dysfunction.) cardiac chambers in 15 patients sched- fibrous cap. The authors showed mod- The detection of thrombus and the uled for cardiotomy. erate to good correlation between find- determination of its age on MR images Kampschulte et al (38) used an in ings from carotid MR imaging and the are mainly related to the physical char- vivo multisequence protocol (TOF and excised histologic specimen for maximal acteristics and visual appearance of intermediate-, T1-, and T2-weighted thickness (r ϭ 0.78, P Ͻ .001), length thrombus (30), which originate from imaging) in 26 patients scheduled for (r ϭ 0.73, P Ͻ .001), and area (r ϭ the cross linking of the fibrin strands, carotid endarterectomy, to differentiate 0.90, P Ͻ .001) of intact fibrous cap. thrombus organization (31), the struc- between intraplaque hemorrhage (Fig Yuan et al (28) used three-dimen- ture of hemoglobin, and the oxidation 3a) and juxtaluminal hemorrhage/ sional TOF and intermediate-, T1-, and state (32). Toussaint et al (31) used the thrombus (Fig 3b). For locations in T2-weighted images in 28 symptomatic water diffusion properties of athero- which MR and histologic findings were and 25 asymptomatic subjects to deter- thrombotic tissue to distinguish fresh in agreement regarding the presence of mine whether fibrous cap thinning or thrombi from 1-week-old thrombi and any type of hemorrhage, MR imaging rupture, identified on MR images, is as- occluding old thrombi on MR images. enabled differentiation of juxtaluminal sociated with a history of recent tran- Recently, a similar technique (diffusion- hemorrhage and thrombus from in- sient ischemic attack or stroke. Com- weighted MR) was successfully applied traplaque hemorrhage with an accuracy pared with patients with a thick fibrous for the detection of cerebral venous of 96%, with histologic findings as the cap, patients with a ruptured cap were thrombosis (33) and dural sinus throm- reference standard. This technique was 23 times more likely to have had a re- bosis (34). Gradient-recalled-echo MR used to evaluate whether in vivo MR cent transient ischemic attack or stroke imaging is frequently applied in the imaging can depict differences between (95% confidence interval: 3, 210) (28). study of cerebral hematoma and venous symptomatic and asymptomatic carotid thrombosis (35). atherosclerotic plaques in the same pa- Endothelial Denudation with Superficial Corti et al (36) demonstrated the tient at the same point in time (39). Platelet Aggregation and Fibrin potential of black-blood MR imaging (T1 Compared with asymptomatic plaques, Deposition and T2 weighted) for help in the detec- plaques associated with neurologic Atherosclerotic plaques with endothe- tion of thrombus and thrombus age in symptoms were associated with a lial denudation are characterized either pigs. Thrombi induced in the carotid higher occurrence of juxtaluminal hem- by superficial erosion and platelet ag- arteries were monitored at 6 hours; 1 orrhage and thrombus (61% vs 26%; gregation or by fibrin deposition (17). day; and 1, 2, 3, 6, and 9 weeks. Tem- P ϭ .039), whereas intraplaque hemor- Platelet aggregation may lead to luminal poral changes due to thrombus organi- rhage was common in both symptom-

68 Radiology: Volume 244: Number 1—July 2007 REVIEW: MR Imaging of the Vulnerable Atherosclerotic Plaque Saam et al

atic and asymptomatic plaques (91% vs under development (30). Fibrin can be (41). Botnar et al (42) used a fibrin- 83%; P ϭ .5). identified by using lipid-encapsulated binding gadolinium-labeled peptide in New targeted contrast agents to en- perfluorocarbon paramagnetic nano- an experimental rabbit model of plaque able detection of local small thrombi are particles in vitro (40,41) and in vivo rupture and thrombosis to detect acute and subacute thrombosis. Similar Figure 3 agents have been successfully used in swine to detect coronary thrombus and in- thrombosis (43), as well as pul- monary emboli (44).

Active Inflammation Inflammation plays a critical role in plaque initiation, progression, and dis- ruption and represents an emerging tar- get in the treatment of atherosclerosis (45). Systemic indicators of inflamma- tion such as C-reactive protein and in- terleukin-6 titers can be easily obtained by testing blood samples, and it has been shown that a variety of circulating inflammatory markers are associated with the risk of (45). However, it appears unlikely that C-reactive protein or any of the other markers actually causes the disease (45). Instead, they reflect the local in- flammatory process in the artery and perhaps other tissues (eg, adipose tis- sue) (45). It is this local inflammatory process that can be assessed by nonin- vasive imaging methods. Plaques with active inflammation may be identified by the presence of extensive macrophage accumulation (46), and recent investigations with MR imaging have shown an association be- tween contrast enhancement and the degree of macrophage infiltration of plaque (47,48). Lin et al (49) were, to our knowledge, the first to use gadolin- ium-based contrast agents in athero- sclerotic plaque imaging in an animal model, and they demonstrated strong contrast enhancement of the vessel wall. Lin et al concluded that the con- trast enhancement of the vessel wall that they observed was caused by an increased vascular supply associated with thrombosis and neointimal thick- Figure 3: (a)Left: Transverse TOF (23/3.8), T1-weighted (T1W) (800/9.3), intermediate-weighted (IMW) (3488/20), ening. Aoki et al (50) used dynamic MR and T2-weighted (T2W) (3488/40) MR images depict intraplaque hemorrhage (arrows) deep within highly stenosed ca- imaging to correlate carotid wall en- ϭ rotid plaque with smooth luminal surface. Arrowheads lumen. Right: Endarterectomy specimen. (b) Left: Transverse hancement changes with pathologic TOF (23/3.8), T1-weighted (T1W) (550/9), intermediate-weighted (IMW) (3093/20), and T2-weighted (T2W) (3093/40) conditions. The dominant feature they with luminal surface irregularity ,(ء) MR images depict juxtaluminal hemorrhage/thrombus (arrows) adjacent to the lumen noted was a bright outer rim of the ves- (chevrons). A calcified area (arrowheads) is hypointense on MR images obtained with all four contrast weightings. Right: sel wall, which was attributed to growth Endarterectomy specimen. (Reprinted, with permission, from reference 38.) of the in the adventitia.

Radiology: Volume 244: Number 1—July 2007 69 REVIEW: MR Imaging of the Vulnerable Atherosclerotic Plaque Saam et al

Figure 4 Yuan et al (21) used contrast-enhanced ␣v␤3-Integrin is a molecule that is T1-weighted images and found the believed to be associated with angiogen- greatest enhancement to be associated esis and inflammation. An ␣v␤3-inte- with areas of dense neovasculature, grin–targeted agent was shown to yield which are thought to contribute to greater enhancement in cholesterol-fed plaque destabilization (51). rabbits than in controls, and cholester- An advantage of using contrast ol-fed rabbits also exhibited increased agents that is only beginning to be ex- development of the adventitial vasa va- plored in plaque imaging is the ability sorum (55). to extract quantitative measurements of enhancement. To date, such studies Severe Stenosis have focused on the use of kinetic Arterial stenosis is an abnormal nar- modeling and dynamic contrast-en- rowing of a blood vessel that reduces Figure 4: Scatterplot demonstrates correlation hanced MR imaging to measure the blood flow in the lumen, which may between dynamic contrast-enhanced MR imaging rate of enhancement. Kerwin et al cause ischemic cardiovascular compli- and corresponding histologic findings. There is a have identified a fast enhancement cations. On the surface of atheroscle- highly significant association between Ktrans as component, vp, associated with plaque rotic plaques associated with severe ste- measured on MR images and degree of macro- neovasculature (47) and a slower en- nosis, shear stress imposes a risk of phage infiltration of the plaque on histologic spec- hancement component, Ktrans (Fig 4), thrombosis and sudden occlusion (3). imens (fractional macrophage area) (Pearson R ϭ associated with permeability (48). It Therefore, a stenotic plaque may be a 0.75, P Ͻ .001). (Reprinted, with permission, has recently been demonstrated (48) vulnerable plaque regardless of pres- from reference 48.) that these dynamic enhancement pa- ence or absence of ischemia (3). More- rameters are strongly associated with over, a stenotic plaque may indicate the macrophage density in advanced ca- presence of many nonstenotic or less- raphy, with conventional coronary an- rotid plaque (Fig 4). Macrophage den- stenotic plaques that could be vulnera- giography as the reference standard. In sity was estimated from histologic ble to rupture and thrombosis (56,57). a separate analysis, the authors calcu- specimens by using software (Image- MR angiography of the carotid ar- lated a sensitivity of 100% for the iden- Pro Plus; Media Cybernetics, Be- teries has gone through a long evolu- tification of either left main coronary thesda, Md) to measure the total area tionary period to become a routine im- artery disease or three-vessel disease, that stained positive for aging modality for evaluation of stenosis conditions in which revascularization is and normalizing according to the at many centers (58,59). In carotid ar- of particular therapeutic value. How- plaque cross-sectional area (48). teries, the ability of MR angiography to ever, current spatial (approximately An alternative method for imaging demonstrate less than 70% versus 1 ϫ 1 ϫ 2-mm) and temporal (approxi- macrophage content is to use ultra- 70%–99% stenosis is much better than mately 60–125-msec) resolutions per- small superparamagnetic iron oxide that of duplex US (60). Also, MR angiog- mit imaging only of the proximal two- (USPIO) particles that are suspended raphy is a sensitive and specific test thirds of the coronary arteries, exclu- in solution and injected into patients. when digital subtraction angiography is sive of the major side branches (62). USPIOs alter MR reaction times, and used as the reference standard (61). Though coronary MR angiography is prom- the propensity of macrophages to However, noninvasive imaging of ising, further development is needed before phagocytose the USPIOs results in the coronary arteries is a formidable it can enter routine clinical use (62). USPIOs being a macrophage-specific task: The small diameter of the coro- Recently, alternative acquisition tech- agent (52). If sufficient time for mac- nary arteries (approximately 2–4 mm), niques, such as steady-state free preces- rophage uptake is allowed—at least 24 their tortuous course, their close ana- sion (64), have been proposed (63) that hours—T2*-weighted images show tomic relationship to the coronary veins establish different contrast behavior and substantial signal intensity reduction and cardiac chambers, and, finally, use different k-space acquisition schemes, due to the susceptibility effects of their continuous rapid motion due to such as spiral (65) or radial (also known USPIOs (53). In experiments with hu- cardiac contractions and respiratory ex- as projection-reconstruction) (66) read- man carotid endarterectomy subjects, cursions create obstacles that are diffi- out patterns. Furthermore, a whole-heart Kooi et al (54) showed that macro- cult to overcome. Kim et al (60) used a steady-state free-precession coronary phages within plaques were positive free-breathing, navigator-gated, three- MR angiography technique has been pro- for iron at histologic and electron mi- dimensional MR technique to depict the posed (63) that improves visible vessel croscopy evaluations of endarterec- coronary arteries and help detect coro- length and facilitates high-quality coro- tomy specimens. On MR images of nary stenoses in a series of 109 patients nary MR angiography of the complete corresponding locations, substantial (Fig 5). In segments that could be evalu- coronary tree in a single measurement. signal intensity reductions were ob- ated, 78 of 94 stenoses of more than These newer sequences are reported to served. 50% were detected by using MR angiog- provide several advantages over the free-

70 Radiology: Volume 244: Number 1—July 2007 REVIEW: MR Imaging of the Vulnerable Atherosclerotic Plaque Saam et al

breathing T2-prepared segmented gradi- rotid Surgery Trial (ECST) criteria, the For evaluation of the degree of ste- ent-recalled-echo sequence (67,68). The common carotid method, or the so- nosis on projection angiograms, the advantages include increases in signal-to- called eyeballing method (69). While it minimum luminal diameter at a target noise ratio, contrast-to-noise ratio, vessel has been demonstrated that the degree site is determined. The projection im- sharpness, and coverage of vessel length, of ipsilateral carotid stenosis, as mea- age should be generated at an angle that as well as reduction in acquisition time. sured with the techniques used in the allows measurement of the minimum lu- Traditionally, the degree of stenosis ECST and NASCET, clearly predict minal diameter; this dimension may not is assessed on a projection angiogram stroke risk (70,71), all of these tech- be measurable in cases of eccentric ste- by using the North American Symptom- niques have sizable disagreements nosis with images generated at subopti- atic Carotid Endarterectomy Trial among repeated measurements for mal angles (72). In contrast, on cross- (NASCET) criteria, the European Ca- some vessels (68). sectional images the minimum luminal diameter can be measured accurately Figure 5 without difficulty (72,73), and some in- vestigators state that a reduction in cross-sectional area correlates better with the hemodynamic effect of stenosis than does a reduction in diameter (74). In the future, new computerized carotid stenosis measurement systems that measure the stenosis by using trans- verse MR angiographic images (75,76) have the potential to further reduce variability.

Minor Criteria

Superficial Calcified Nodule The consensus documents (3,4) de- scribe a type of plaque that has a calci- fied nodule within or very close to the fibrous cap, and this structure pro- trudes through and can rupture the cap (3). It has been shown in coronary ar- teries that the presence of calcified nod- ules is not necessarily associated with severe coronary calcification and a high calcium score (17). In vivo MR imaging is able to depict overall plaque calcification with good sensitivity and specificity (9). Further- more, a strong correlation between MR and histologic area measurements has been achieved (r ϭ 0.74, P Ͻ .001) (9). However, juxtaluminal calcification can be more difficult to identify than calcifi- Figure 5: Coronary angiography in a 53-year-old man with exertional chest pain. AA ϭ ascending aorta, cation that is deeper within the plaque, LA ϭ left atrium, LV ϭ left ventricle, PA ϭ pulmonary artery, RV ϭ right ventricle, RVOT ϭ right ventricular because juxtaluminal calcification ap- outflow tract. (a) Coronary MR angiogram (left) along left main and left anterior descending coronary arteries pears as a hypointense signal on TOF and corresponding conventional coronary angiogram in right anterior oblique projection (right) indicate se- and T1-, intermediate-, and T2- vere stenosis at bifurcation of left anterior descending and left circumflex coronary arteries involving the left weighted images and is not distinguish- main coronary artery (solid arrows) and more distal focal stenosis of left circumflex coronary artery (dotted able from the lumen on black-blood T1-, arrows). (b) Coronary MR angiogram (left) in coronal orientation along course of right coronary artery and intermediate-, and T2-weighted images. corresponding conventional angiogram in left anterior oblique projection (right) indicate two stenoses of However, the addition of bright-blood proximal (solid arrows) and middle (dotted arrows) right coronary artery. (Reprinted, with permission, from TOF imaging facilitates identification of reference 61.) juxtaluminal calcification, in that it is

Radiology: Volume 244: Number 1—July 2007 71 REVIEW: MR Imaging of the Vulnerable Atherosclerotic Plaque Saam et al

Figure 6 clearly distinguishable from the bright lumen on TOF images. Recent reports (20,38) have provided preliminary evi- dence that the combined use of black- blood and bright-blood images makes feasible the detection of calcified nod- ules and/or juxtaluminal calcification in the carotid arteries (Fig 6).

Glistening Yellow Plaques at Angioscopy Yellow plaques, particularly glistening ones, may indicate a large lipid-necrotic core and a thin fibrous cap, suggesting a high risk of rupture. However, because plaques in different stages can be yellow and not all lipid-laden plaques are des- tined to rupture or undergo thrombosis, this criterion may lack sufficient speci- ficity (77,78). Furthermore, this crite- rion is determined solely on the basis of Figure 6: Left: Transverse MR images demonstrate usefulness of a multisequence protocol. TOF image angioscopic observations. Angioscopy is (21.0/3.2) reveals two distinct areas (arrows A and B) with irregular lumen boundaries that are not well defined an intravascular imaging method that on T1-weighted (T1W, 550/9.0), intermediate-weighted (IMW, 1528/28), and T2-weighted (T2W, 1528/56) achieves high spatial resolution. How- images, consistent with juxtaluminal calcification. Right: Histologic specimens demonstrate that material ever, the information is limited to the intruding into the lumen contains calcified nodules. Boxes A and B correspond to areas labeled on TOF image. luminal surface, the technique is inva- sive and highly operator dependent, and it is currently not approved by the Food 3a) with good sensitivity and moderate lipid-necrotic core volume (28.4% vs and Drug Administration for use in the to good specificity (13,20,83–85). Fur- Ϫ5.2%, P ϭ .001) was significantly United States. Obviously, MR images thermore it has been shown that MR higher in the hemorrhage group than in cannot depict the color of the plaque’s imaging can help differentiate between the control group (Fig 7). Further, pa- surface. However, as mentioned earlier hemorrhages of different ages (83) with tients with intraplaque hemorrhage at in the section Thin Cap with Large the use of modified criteria for cerebral baseline showed a far greater suscepti- Lipid-Necrotic Core, invivo MR imaging hemorrhage. Moody et al (84) showed bility to repeat plaque hemorrhages is able to depict lipid-necrotic cores and that T1-weighted images of the carotid (87). Therefore, intraplaque hemor- the status of the fibrous cap, both fea- arteries can be used to accurately iden- rhage may represent a critical transi- tures of angioscopically glistening yel- tify histologically confirmed complicated tion, promoting the conversion of a sta- low plaques. plaque with hemorrhage or thrombus. ble to an unstable plaque phenotype Though intraplaque hemorrhage is (86). Intraplaque Hemorrhage listed as a minor criterion, recent arti- Extravasation of red blood cells or iron cles (79,86,87) suggest a greater role of Endothelial Dysfunction accumulation in plaque may represent hemorrhage in plaque destabilization Studies have shown that endothelial plaque instability and promote plaque than was previously thought. Results of dysfunction is associated with coronary progression (3,79). The mechanism a study of 24 patients (79) who had died heart disease and stroke (88,89). Fur- that results in intraplaque hemorrhages suddenly of coronary causes suggested thermore, vulnerable plaques with ac- is not completely understood: Constan- that hemorrhage contributed to the tive inflammation and oxidative stress tinides (80) and others (81) have sug- deposition of free cholesterol, infiltra- are likely to be associated with impaired gested that hemorrhage into a plaque tion by macrophages, and enlargement endothelial function (3). occurs from cracks or fissures that orig- of the necrotic core, thereby represent- Although in vivo MR imaging is not inate at the luminal surface, while ing a potent atherogenic stimulus. Con- able to depict the directly, Paterson (82) has proposed that in- sistent with this notion, a recent pro- MR and a number of other imaging mo- traplaque hemorrhage is secondary to spective longitudinal MR investigation dalities can be used to measure arterial rupture of the vasa vasorum, a common (87) of 31 patients showed that hemor- stiffness noninvasively (90). Arterial feature of advanced lesions with plaque rhage into a carotid plaque accelerated stiffness and endothelial dysfunction rupture and luminal thrombi. plaque progression in a period of 18 commonly coexist in patients at in- In vivo MR imaging is able to depict months. The percent change in wall vol- creased risk of cardiovascular disease carotid intraplaque hemorrhage (Fig ume (6.8% vs Ϫ0.15%, P ϭ .009) and (90), for example in patients with diabe-

72 Radiology: Volume 244: Number 1—July 2007 REVIEW: MR Imaging of the Vulnerable Atherosclerotic Plaque Saam et al

tes (91) and in smokers (92). Indeed, namic contrast-enhanced MR imaging in carotid lesions and by 16% and 16%, some studies in children have directly (48). This parameter quantifies the in- respectively, in thoracic lesions at 24 related increased stiffness with im- creased vascularity, permeability, and months (Fig 8). Lumen area increased paired endothelial function (93,94). extracellular water content typically as- by only 5% after 24 months of statin Although MR imaging can help as- sociated with the inflammatory re- therapy. Recently, Yonemura et al sess various stiffness parameters such sponse. (101) reported the results of a prospec- as vascular distensibility, flow measure- tive, randomized, open-label trial to elu- ments, and pulse wave velocity by using Outward (Positive) Remodeling cidate the effects on atherosclerotic true fast imaging with steady-state free Many nonstenotic lesions undergo “ex- plaques in the thoracic aorta of atorva- precession (95) and gradient-recalled- pansive,” “positive,” or “outward” re- statin dosages of 20 mg/d versus 5 echo pulse sequences with a velocity- modeling: namely, compensatory en- mg/d. Their results indicated that the encoding gradient for phase-contrast largement that permits growth in over- 20 mg/d dose reduced maximal vessel MR (96,97), only a few such investiga- all plaque size without reduction in wall thickness and vessel wall area of tho- tions have been performed in this field luminal area. Several authors (98,99) racic aortic plaques (Ϫ12% and Ϫ18%, (90). This is due to the large number of have suggested that such remodeling is respectively; P Ͻ .001), whereas 5 mg/d other imaging modalities, such as Dopp- a potential surrogate marker of plaque did not (ϩ1% and ϩ4%, respectively). ler US or applanation tonometry, which vulnerability. MR imaging is ideally Lumen area increased 5% in the high- also allow evaluation of arterial stiffness suited for serial and noninvasive assess- dose group and 0% in the low-dose (90). However, a recent study showed ment of arterial remodeling, because group. that MR imaging offers insights not oth- MR provides information about the lu- In a study that included six healthy erwise possible (90) with regard to de- men and arterial wall simultaneously. subjects and six subjects with nonsignifi- scribing an age-related increase in pulse Clinical trials of MR imaging to ex- cant coronary artery disease (10%–50% wave velocity in the proximal aorta rel- amine the results of statin therapy on diameter reduction on conventional an- ative to that in the distal aorta (97). In plaque progression or regression have giogram), Kim et al (102) introduced a another MR investigation (95) in which demonstrated a greater effect on vessel noninvasive MR imaging method to dem- central vascular distensibility was as- wall area than on changes in lumen di- onstrate expansive remodeling in coro- sessed, pulse wave velocity and flow- mensions, which suggests a remodeling nary arteries. Free-breathing three-di- mediated dilatation showed global im- process on the vessel wall. Corti et al mensional black-blood coronary MR im- pairment of brachial, carotid, and aortic (100) performed serial carotid and aor- aging with isotropic resolution depicted vascular function in young smokers. tic MR imaging in patients treated with an increased coronary vessel wall thick- Another indirect MR measure of the simvastatin and found that vessel wall ness with preservation of lumen size in endothelial function is to measure the area and vessel wall thickness de- patients with nonsignificant coronary ar- transfer constant Ktrans by using dy- creased by 18% and 19%, respectively, tery disease, consistent with expansive arterial remodeling. A recent study (103) Figure 7 in 15 volunteers demonstrated the feasi- bility of coronary wall MR imaging with free-breathing and breath-hold two-di- mensional black-blood turbo spin-echo sequences at 3 T. Although measure- ments of coronary wall thickness and area and lumen diameter and area were consistent with previous MR measure- ments at 1.5 T, the authors concluded that further improvement in resolution and image quality is required to enable detection and characterization of coro- nary plaques (103).

Summary, Clinical Perspectives, and Conclusion Figure 7: Representative transverse T1-weighted MR images (800/9.3) of the progression of atherosclero- Noninvasive MR imaging has the poten- sis with intraplaque hemorrhage. Right carotid artery was imaged (a) at baseline and (b) 18 months later. Lu- tial to help identify, directly or indi- decreased and wall area (arrowheads) increased in each section on b. Bif ϭ bifurcation, CCA ϭ (ء) men area rectly, most of the consensus criteria common carotid artery, ECA ϭ external carotid artery, ICA ϭ internal carotid artery. (Reprinted, with permis- for the identification of vulnerable sion, from reference 87.) plaque in larger vessels by using a vari-

Radiology: Volume 244: Number 1—July 2007 73 REVIEW: MR Imaging of the Vulnerable Atherosclerotic Plaque Saam et al

Figure 8 ety of proposed image-acquisition and contrast enhancement techniques (see Table 2). Thus far, there is not a single MR technique that can be used to assess all the vulnerable plaque criteria, and it remains to be seen which imaging ap- proach, if any, will be commonly used in clinical practice. While some of the pro- posed image acquisition techniques rely on one (11) or two (26,36,85,101) MR sequences, others are based on the ac- quisition of multiple contrast-weighted images (9,10,104). The number and types of sequences used depend on which plaque charac- teristics are to be studied. For instance, rapid assessment of overall plaque bur- den is feasible by using one or two MR sequences (100,105). To evaluate plaque composition and morphology in one im- aging session, however, most studies rely on multiple MR sequences (9,10,104). Black-blood T1-, intermedi- ate-, and T2-weighted images with fat Figure 8: Bar graphs show changes in atherosclerotic vessel wall dimensions (in square millimeters) after (8) and flow suppression (106) enable statin treatment: mean vessel wall area (top) and mean lumen area (bottom) at baseline (BL) and after 6, 12, 18, visualization of the full vessel wall and and 24 months of simvastatin treatment for aorta (left) and carotid arteries (right). ANOVA ϭ analysis of vari- are needed to characterize the major ance, error bars ϭ standard error of the mean. (Reprinted, with permission, from reference 100.) plaque components, such as a lipid- necrotic core, calcification, and a loose fibrous matrix (9). Bright-blood TOF that the consensus document (3,4) cri- 4.7; 95% confidence interval: 1.6, 14.0; images are needed for evaluation of the teria that define the vulnerable plaque P ϭ .004), larger mean intraplaque status of the fibrous cap (15,28) and are based on histologic studies of culprit hemorrhage area (hazard ratio for 10- identification of calcified nodules (see plaques. Histologic examination can mm2 increase: 2.4; 95% confidence in- Fig 6) (20). Other sequences, such as only capture information regarding the terval: 1.4, 4.1; P ϭ .008), thin or rup- contrast-enhanced T1-weighted images plaque from a single time point. Thus, tured fibrous cap (hazard ratio: 9.4; (22,27) and dynamic contrast-enhanced little is known about the evolution of 95% confidence interval: 2.1, 42.1; P Ͻ two-dimensional spoiled gradient-re- vulnerable plaques, and noninvasive se- .001), larger maximum percentage lip- called-echo images (48) are useful for rial MR imaging of such plaques can, id-rich–to-necrotic core ratio (hazard quantifying the size of the lipid-necrotic therefore, provide useful insights into ratio for 10% increase: 1.4; 95% confi- core (22) and evaluating plaque inflam- their natural history and might help to dence interval: 1.1, 1.9; P ϭ .01), and mation (47,48). Furthermore, the use identify plaques that are on a trajectory maximum wall thickness (hazard ratio of sophisticated contrast agents, such as of evolution toward a vulnerable state. for 1-mm increase: 1.6; 95% confidence ultrasmall superparamagnetic iron ox- In a recent prospective in vivo MR interval, 1.1, 2.1; P ϭ .007) were asso- ides (53,54) and fibrin-targeted (41,42) imaging study (107) that used T1-, T2-, ciated with the cerebrovascular events. and ␣v␤3-integrin–targeted (55) nano- and intermediate-weighted images and Although initial findings of this pro- particles offers the promise of in vivo TOF images, the authors investigated spective study (107) were promising, targeted imaging of the plaque (7). the association between carotid plaque several challenges remain. Most of the The ability to distinguish stable from characteristics and subsequent ische- in vivo MR studies in humans cited in vulnerable plaques may ultimately per- mic cerebrovascular events in 154 sub- this article were based on analysis of mit identification of patients with low jects who initially had an asymptomatic data from relatively large human ves- versus those with high risk of cardiovas- 50%–79% carotid stenosis seen at du- sels, such as the carotid arteries. To cular complications. Furthermore, in plex US. During a mean follow-up of achieve similar results in the much vivo MR imaging has potential as a tool 38.2 months, 12 carotid cerebrovascu- smaller coronary arteries, substantial for screening for atherosclerosis and for lar events occurred. Cox regression advances in temporal and spatial resolu- serial monitoring of disease progression analysis indicated that plaques with in- tion are necessary. This may be accom- or regression. It is important to note traplaque hemorrhage (hazard ratio: plished with improvements in pulse se-

74 Radiology: Volume 244: Number 1—July 2007 REVIEW: MR Imaging of the Vulnerable Atherosclerotic Plaque Saam et al

quence design and MR hardware (eg, rotic plaques: classification and character- in atherosclerotic carotid plaque: compari- higher field strength, new coil designs). ization with T2-weighted high-spatial-reso- son of high-resolution, contrast-enhanced Therefore, prospective studies in large lution MR imaging—an in vitro study. magnetic resonance imaging and histology. Radiology 2001;219:403–410. Circulation 2005;112:3437–3444. populations are needed to determine the predictive value of MR-based as- 12. Shinnar M, Fallon JT, Wehrli S, et al. The 23. Parker DL, Yuan C, Blatter DD. MR angiog- sessment of vulnerable plaque features diagnostic accuracy of ex vivo MRI for hu- raphy by multiple thin slab 3D acquisition. man atherosclerotic plaque characteriza- Magn Reson Med 1991;17:434–451. for subsequent ischemic events. tion. Arterioscler Thromb Vasc Biol 1999; 24. Mitsumori LM, Hatsukami TS, Ferguson Acknowledgment: We thank Andrew An Ho, 19:2756–2761. MS, Kerwin WS, Cai J, Yuan C. In vivo MA, for editing this manuscript. 13. Toussaint JF, LaMuraglia GM, Southern JF, accuracy of multisequence MR imaging for Fuster V, Kantor HL. Magnetic resonance identifying unstable fibrous caps in ad- References images lipid, fibrous, calcified, hemor- vanced human carotid plaques. J Magn Re- son Imaging 2003;17:410–420. 1. World Health Organization. World Health Sta- rhagic, and thrombotic components of hu- man atherosclerosis in vivo. Circulation tistics 2005. http://www.who.int/healthinfo/ 25. Trivedi RA, King-Im J, Graves MJ, et al. 1996;94:932–938. statistics/whsatsdownloads/en/index.html. Multi-sequence in vivo MRI can quantify Published 2005. Accessed April 18, 2007. 14. Cai JM, Hatsukami TS, Ferguson MS, Small fibrous cap and lipid core components in human carotid atherosclerotic plaques. Eur 2. Weinberger J. Diagnosis and prevention of R, Polissar NL, Yuan C. Classification of J Vasc Endovasc Surg 2004;28:207–213. atherosclerotic cerebral infarction. CNS human carotid atherosclerotic lesions with in vivo multicontrast magnetic resonance Spectr 2005;10:553–564. 26. Kramer CM, Cerilli LA, Hagspiel K, DiMa- imaging. Circulation 2002;106:1368–1373. ria JM, Epstein FH, Kern JA. Magnetic res- 3. Naghavi M, Libby P, Falk E, et al. From onance imaging identifies the fibrous cap in vulnerable plaque to vulnerable patient: a 15. Hatsukami TS, Ross R, Polissar NL, Yuan atherosclerotic abdominal aortic aneu- call for new definitions and risk assessment C. Visualization of fibrous cap thickness rysm. Circulation 2004;109:1016–1021. strategies—part I. Circulation 2003;108: and rupture in human atherosclerotic ca- rotid plaque in vivo with high-resolution 1664–1672. 27. Wasserman BA, Smith WI, Trout HH III, magnetic resonance imaging. Circulation Cannon RO 3rd, Balaban RS, Arai AE. Ca- 4. Naghavi M, Libby P, Falk E, et al. From 2000;102:959–964. rotid artery atherosclerosis: in vivo mor- vulnerable plaque to vulnerable patient: a phologic characterization with gadolinium- call for new definitions and risk assessment 16. Yuan C, Kerwin WS. MRI of atherosclerosis. enhanced double-oblique MR imaging—ini- strategies—part II. Circulation 2003;108: J Magn Reson Imaging 2004;19:710–719. tial results. Radiology 2002;223:566–573. 1772–1778. 17. Virmani R, Kolodgie FD, Burke AP, Farb A, 28. Yuan C, Zhang SX, Polissar NL, et al. Iden- 5. Mistretta CA. Relative characteristics of Schwartz SM. Lessons from sudden coro- tification of fibrous cap rupture with mag- MR angiography and competing vascular nary death: a comprehensive morphologi- netic resonance imaging is highly associ- imaging modalities. J Magn Reson Imaging cal classification scheme for atherosclerotic ated with recent transient ischemic attack 1993;3:685–698. lesions. Arterioscler Thromb Vasc Biol 2000;20:1262–1275. or stroke. Circulation 2002;105:181–185. 6. Saam T, Kerwin WS, Yuan C. Vascular im- 18. Stary HC, Chandler AB, Dinsmore RE, et al. 29. Falk E, Shah PK, Fuster V. Coronary plaque aging. In: Wnek GE, Bowlin GL, eds. Ency- A definition of advanced types of atheroscle- disruption. Circulation 1995;92:657–671. clopedia of biomaterials and biomedical en- rotic lesions and a histological classification of gineering. New York, NY: Dekker, 2004. 30. Wentzel JJ, Aguiar SH, Fayad ZA. Vascular atherosclerosis: a report from the Committee MRI in the diagnosis and therapy of the high 7. Choudhury RP, Fuster V, Badimon JJ, Fisher on Vascular Lesions of the Council on Arte- risk atherosclerotic plaque. J Interv Cardiol EA, Fayad ZA. MRI and characterization of riosclerosis, American Heart Association. 2003;16:129–142. atherosclerotic plaque: emerging applications Circulation 1995;92:1355–1374. and molecular imaging. Arterioscler Thromb 31. Toussaint JF, Southern JF, Fuster V, Kan- 19. Burke AP, Weber DK, Kolodgie FD, Farb Vasc Biol 2002;22:1065–1074. tor HL. Water diffusion properties of hu- A, Taylor AJ, Virmani R. Pathophysiology man atherosclerosis and thrombosis mea- 8. Yuan C, Mitsumori LM, Beach KW, Mara- of calcium deposition in coronary arteries. sured by pulse field gradient nuclear mag- villa KR. Carotid atherosclerotic plaque: Herz 2001;26:239–244. netic resonance. Arterioscler Thromb Vasc noninvasive MR characterization and iden- 20. Yuan C, Mitsumori LM, Ferguson MS, et al. Biol 1997;17:542–546. tification of vulnerable lesions. Radiology In vivo accuracy of multispectral magnetic 2001;221:285–299. 32. Bradley WG Jr. MR appearance of hemor- resonance imaging for identifying lipid-rich rhage in the brain. Radiology 1993;189:15– 9. Saam T, Ferguson MS, Yarnykh VL, et al. necrotic cores and intraplaque hemorrhage 26. Quantitative evaluation of carotid plaque in advanced human carotid plaques. Circu- composition by in vivo MRI. Arterioscler lation 2001;104:2051–2056. 33. Chu K, Kang DW, Yoon BW, Roh JK. Dif- Thromb Vasc Biol 2005;25:234–239. fusion-weighted magnetic resonance in ce- 21. Yuan C, Kerwin WS, Ferguson MS, et al. rebral venous thrombosis. Arch Neurol 10. Clarke SE, Hammond RR, Mitchell JR, Rutt Contrast-enhanced high resolution MRI for 2001;58:1569–1576. BK. Quantitative assessment of carotid atherosclerotic carotid artery tissue char- plaque composition using multicontrast acterization. J Magn Reson Imaging 2002; 34. Yoshikawa T, Abe O, Tsuchiya K, et al. MRI and registered histology. Magn Reson 15:62–67. Diffusion-weighted magnetic resonance im- Med 2003;50:1199–1208. aging of dural sinus thrombosis. Neuroradi- 22. Cai J, Hatsukami TS, Ferguson MS, et al. In ology 2002;44:481–488. 11. Serfaty JM, Chaabane L, Tabib A, Cheval- vivo quantitative measurement of intact fi- lier JM, Briguet A, Douek PC. Atheroscle- brous cap and lipid-rich necrotic core size 35. Taber KH, Hayman LA, Herrick RC, Kirk-

Radiology: Volume 244: Number 1—July 2007 75 REVIEW: MR Imaging of the Vulnerable Atherosclerotic Plaque Saam et al

patrick JB. Importance of clot structure in rotid atherosclerotic plaque. Circulation conventional angiography. Radiology 2001; gradient-echo magnetic resonance imaging 2003;107:851–856. 218:138–143. of hematoma. J Magn Reson Imaging 1996; 48. Kerwin WS, O’Brien KD, Ferguson MS, 60. Kim WY, Danias PG, Stuber M, et al. Cor- 6:878–883. Polissar N, Hatsukami TS, Yuan C. Inflam- onary magnetic resonance angiography for 36. Corti R, Osende JI, Fayad ZA, et al. In vivo mation in carotid atherosclerotic plaque: a the detection of coronary stenoses. N Engl noninvasive detection and age definition of dynamic contrast-enhanced MR imaging J Med 2001;345:1863–1869. arterial thrombus by MRI. J Am Coll Car- study. Radiology 2006;241(2):459–468. 61. Nederkoorn PJ, van der Graff Y, Hunink diol 2002;39:1366–1373. 49. Lin W, Abendschein DR, Haacke EM. Con- MG. Duplex ultrasound and magnetic reso- 37. Paydarfar D, Krieger D, Dib N, et al. In vivo trast-enhanced magnetic resonance angiog- nance angiography compared with digital magnetic resonance imaging and surgical raphy of carotid arterial wall in pigs. J subtraction angiography in carotid artery histopathology of intracardiac masses: dis- Magn Reson Imaging 1997;7:183–190. stenosis: a systematic review. Stroke 2003; 34:1324–1332. tinct features of subacute thrombi. Cardiol- 50. Aoki S, Aoki K, Ohsawa S, Nakajima H, ogy 2001;95:40–47. Kumagai H, Araki T. Dynamic MR imaging 62. Ropers D, Regenfus M, Wasmeier G, 38. Kampschulte A, Ferguson MS, Kerwin WS, of the carotid wall. J Magn Reson Imaging Achenbach S. Non-interventional cardiac et al. Differentiation of intraplaque versus 1999;9:420–427. diagnostics: computed tomography, mag- netic resonance and real-time three-dimen- juxtaluminal hemorrhage/thrombus in ad- 51. McCarthy MJ, Loftus IM, Thompson MM, sional echocardiography—techniques and vanced human carotid atherosclerotic le- et al. Angiogenesis and the atherosclerotic clinical applications. Minerva Cardioangiol sions by in vivo magnetic resonance imag- carotid plaque: an association between 2004;52:407–417. ing. Circulation 2004;110:3239–3244. symptomatology and plaque morphology. J 39. Saam T, Cai J, Ma L, et al. Comparison of Vasc Surg 1999;30:261–268. 63. Weber OM, Pujadas S, Martin AJ, Higgins symptomatic and asymptomatic atheroscle- 52. Schmitz SA, Taupitz M, Wagner S, Wolf KJ, CB. Free-breathing, three-dimensional cor- rotic carotid plaque features with in vivo Beyersdorff D, Hamm B. Magnetic resonance onary artery magnetic resonance angiog- MR imaging. Radiology 2006;240(2):464– imaging of atherosclerotic plaques using su- raphy: comparison of sequences. J Magn 472. perparamagnetic iron oxide particles. J Magn Reson Imaging 2004;20:395–402. Reson Imaging 2001;14:355–361. 40. Yu X, Song SK, Chen J, et al. High-resolu- 64. Deshpande VS, Shea SM, Laub G, Simon- tion MRI characterization of human throm- 53. Ruehm SG, Corot C, Vogt P, Kolb S, Deba- etti OP, Finn JP, Li D. 3D magnetization- bus using a novel fibrin-targeted paramag- tin JF. Magnetic resonance imaging of ath- prepared true-FISP: a new technique for netic nanoparticle contrast agent. Magn Re- erosclerotic plaque with ultrasmall super- imaging coronary arteries. Magn Reson son Med 2000;44:867–872. paramagnetic particles of iron oxide in hy- Med 2001;46:494–502. perlipidemic rabbits. Circulation 2001;103: 41. Flacke S, Fischer S, Scott MJ, et al. Novel 65. Bornert P, Stuber M, Botnar RM, et al. 415–422. MRI contrast agent for molecular imaging of Direct comparison of 3D spiral vs. Carte- fibrin: implications for detecting vulnerable 54. Kooi ME, Cappendijk VC, Cleutjens KB, et sian gradient-echo coronary magnetic reso- plaques. Circulation 2001;104:1280–1285. al. Accumulation of ultrasmall superpara- nance angiography. Magn Reson Med 2001; magnetic particles of iron oxide in human 46:789–794. 42. Botnar RM, Perez AS, Witte S, et al. In vivo atherosclerotic plaques can be detected by molecular imaging of acute and subacute 66. Katoh M, Stuber M, Buecker A, Gunther in vivo magnetic resonance imaging. Circu- thrombosis using a fibrin-binding magnetic RW, Spuentrup E. Spin-labeling coronary lation 2003;107:2453–2458. resonance imaging contrast agent. Circula- MR angiography with steady-state free pre- tion 2004;109:2023–2029. 55. Winter PM, Morawski AM, Caruthers SD, cession and radial k-space sampling: initial et al. Molecular imaging of angiogenesis in results in healthy volunteers. Radiology 43. Botnar RM, Buecker A, Wiethoff AJ, et al. early-stage atherosclerosis with alpha(v) 2005;236(3):1047–1052. In vivo magnetic resonance imaging of cor- beta3-integrin-targeted nanoparticles. Cir- 67. Botnar RM, Stuber M, Danias PG, Kiss- onary thrombosis using a fibrin-binding mo- culation 2003;108:2270-2274. lecular magnetic resonance contrast agent. inger KV, Manning WJ. Improved coronary Circulation 2004;110:1463–1466. 56. Goldstein JA, Demetriou D, Grines CL, artery definition with T2-weighted, free- Pica M, Shoukfeh M, O’Neill WW. Multiple breathing, three-dimensional coronary 44. Spuentrup E, Buecker A, Katoh M, et al. complex coronary plaques in patients with MRA. Circulation 1999;99:3139–3148. Molecular magnetic resonance imaging of acute . N Engl J Med 68. Stuber M, Botnar RM, Danias PG, et al. coronary thrombosis and pulmonary em- 2000;343:915–922. boli with a novel fibrin-targeted contrast Double-oblique free-breathing high resolu- agent. Circulation 2005;111:1377–1382. 57. Asakura M, Ueda Y, Yamaguchi O, et al. tion three-dimensional coronary magnetic Extensive development of vulnerable resonance angiography. J Am Coll Cardiol 45. Hansson GK. Inflammation, atherosclero- plaques as a pan-coronary process in pa- 1999;34:524–531. sis, and coronary artery disease. N Engl tients with myocardial infarction: an angio- 69. Young GR, Humphrey PR, Nixon TE, Smith J Med 2005;352:1685–1695. scopic study. J Am Coll Cardiol 2001;37: ET. Variability in measurement of extracra- 1284–1288. 46. Bassiouny HS, Sakaguchi Y, Mikucki SA, et nial internal carotid artery stenosis as dis- al. Juxtalumenal location of plaque necrosis 58. Rapp JH, Saloner D. Current status of ca- played by both digital subtraction and mag- and neoformation in symptomatic carotid rotid imaging by MRA. Cardiovasc Surg netic resonance angiography: an assessment stenosis. J Vasc Surg 1997;26:585–594. 2003;11:445–447. of three caliper techniques and visual impres- sion of stenosis. Stroke 1996;27:467–473. 47. Kerwin W, Hooker A, Spilker M, et al. 59. Huston J III, Fain SB, Wald JT, et al. Ca- Quantitative magnetic resonance imaging rotid artery: elliptic centric contrast-en- 70. North American Symptomatic Carotid End- analysis of neovasculature volume in ca- hanced MR angiography compared with arterectomy Trial. Methods, patient char-

76 Radiology: Volume 244: Number 1—July 2007 REVIEW: MR Imaging of the Vulnerable Atherosclerotic Plaque Saam et al

acteristics, and progress. Stroke 1991;22: plaque: a high-resolution MRI study. Stroke lution magnetic resonance imaging. J Am 711–720. 2004;35:1079–1084. Coll Cardiol 2004;44:2056–2064.

71. European Carotid Surgery Trialists’ Collab- 84. Moody AR, Murphy RE, Morgan PS, et al. 96. Mohiaddin RH, Firmin DN, Longmore DB. orative Group. MRC European Carotid Sur- Characterization of complicated carotid Age-related changes of human aortic flow gery Trial: interim results for symptomatic plaque with magnetic resonance direct wave velocity measured noninvasively by patients with severe (70–99%) or with mild thrombus imaging in patients with cerebral magnetic resonance imaging. J Appl Physiol (0–29%) carotid stenosis. Lancet 1991; ischemia. Circulation 2003;107:3047–3052. 1993;74:492–497. 337:1235–1243. 85. Cappendijk VC, Cleutjens KB, Heeneman 97. Rogers WJ, Hu YL, Coast D, et al. Age- 72. Ota H, Takase K, Rikimaru H, et al. Quan- S, et al. In vivo detection of hemorrhage in associated changes in regional aortic pulse titative vascular measurements in arterial human atherosclerotic plaques with mag- wave velocity. J Am Coll Cardiol 2001;38: occlusive disease. RadioGraphics 2005;25: netic resonance imaging. J Magn Reson Im- 1123–1129. 1141–1158. aging 2004;20:105–110. 98. Smits PC, Pasterkamp G, Quarles van Uf- 73. Morasch MD, Gurjala AN, Washington E, 86. Virmani R, Kolodgie FD, Burke AP, et al. ford MA, et al. Coronary artery disease: et al. Cross-sectional magnetic resonance Atherosclerotic plaque progression and vul- arterial remodelling and clinical presenta- angiography is accurate in predicting de- nerability to rupture: angiogenesis as a tion. Heart 1999;82:461–464. gree of carotid stenosis. Ann Vasc Surg source of intraplaque hemorrhage. Arterio- 99. Varnava AM, Mills PG, Davies MJ. Rela- 2002;16:266–272. scler Thromb Vasc Biol 2005;25:2054–2061. tionship between coronary artery remodel- 74. Dodds SR. The haemodynamics of asym- 87. Takaya N, Yuan C, Chu B, et al. Presence of ing and plaque vulnerability. Circulation metric stenoses. Eur J Vasc Endovasc Surg intraplaque hemorrhage stimulates progres- 2002;105:939–943. 2002;24:332–337. sion of carotid atherosclerotic plaques: a 100. Corti R, Fuster V, Fayad ZA, et al. Lipid 75. Kim DY, Park JW. Computerized quantifi- high-resolution magnetic resonance imaging lowering by simvastatin induces regression cation of carotid artery stenosis using MRA study. Circulation 2005;111:2768–2775. of human atherosclerotic lesions: 2 years’ follow-up by high-resolution noninvasive axial images. Magn Reson Imaging 2004;22: 88. Halcox JP, Schenke WH, Zalos G, et al. magnetic resonance imaging. Circulation 353–359. Prognostic value of coronary vascular endo- 2002;106:2884–2887. 76. van Bemmel CM, Viergever MA, Niessen thelial dysfunction. Circulation 2002;106: WJ. Semiautomatic segmentation and ste- 653–658. 101. Yonemura A, Momiyama Y, Fayad ZA, et al. Effect of lipid-lowering therapy with nosis quantification of 3D contrast-en- 89. Targonski PV, Bonetti PO, Pumper GM, atorvastatin on atherosclerotic aortic hanced MR angiograms of the internal ca- Higano ST, Holmes DR Jr, Lerman A. Cor- plaques detected by noninvasive magnetic rotid artery. Magn Reson Med 2004;51: onary endothelial dysfunction is associated resonance imaging. J Am Coll Cardiol 2005; 753–760. with an increased risk of cerebrovascular 45:733–742. 77. Kodama K, Asakura M, Ueda Y, Yamaguchi events. Circulation 2003;107:2805–2809. 102. Kim WY, Stuber M, Bornert P, Kissinger O, Hirayama A. The role of plaque rupture 90. Oliver JJ, Webb DJ. Noninvasive assess- KV, Manning WJ, Botnar RM. Three-di- in the development of acute coronary syn- ment of arterial stiffness and risk of athero- mensional black-blood cardiac magnetic drome evaluated by the coronary angio- sclerotic events. Arterioscler Thromb Vasc resonance coronary vessel wall imaging de- scope. Intern Med 2000;39:333–335. Biol 2003;23:554–566. tects positive arterial remodeling in pa- 78. Takano M, Mizuno K, Okamatsu K, 91. De Vriese AS, Verbeuren TJ, Van de tients with nonsignificant coronary artery Yokoyama S, Ohba T, Sakai S. Mechanical Voorde J, Lameire NH, Vanhoutte PM. En- disease. Circulation 2002;106:296–299. and structural characteristics of vulnerable dothelial dysfunction in diabetes. Br J Phar- 103. Koktzoglou I, Simonetti O, Li D. Coronary ar- plaques: analysis by coronary angioscopy macol 2000;130:963–974. and . J Am Coll tery wall imaging: initial experience at 3 tesla. J Cardiol 2001;38:99–104. 92. Zeiher AM, Schachinger V, Minners J. Magn Reson Imaging 2005;21:128–132. Long-term cigarette smoking impairs endo- 79. Kolodgie FD, Gold HK, Burke AP, et al. 104. Cappendijk VC, Cleutjens KB, Kessels AG, thelium-dependent coronary arterial vaso- Intraplaque hemorrhage and progression of et al. Assessment of human atherosclerotic dilator function. Circulation 1995;92:1094– coronary . N Engl J Med 2003; carotid plaque components with multi- 1100. 349:2316–2325. sequence MR imaging: initial experience. 93. Martin H, Hu J, Gennser G, Norman M. Radiology 2005;234(2):487–492. 80. Constantinides P. Plaque fissuring in hu- Impaired endothelial function and in- man coronary thrombosis. J Atheroscler 105. Lima JA, Desai MY, Steen H, Warren WP, creased carotid stiffness in 9-year-old chil- Res 1966;6:1–17. Gautam S, Lai S. Statin-induced cholesterol dren with low birthweight. Circulation lowering and plaque regression after 6 months 81. Davies MJ, Thomas AC. Plaque fissuring: 2000;102:2739–2744. of magnetic resonance imaging-monitored the cause of acute myocardial infarction, therapy. Circulation 2004;110:2336–2341. 94. Tounian P, Aggoun Y, Dubern B, et al. sudden ischaemic death, and crescendo an- Presence of increased stiffness of the com- 106. Yarnykh VL, Yuan C. T1-insensitive flow sup- gina. Br Heart J 1985;53:363–373. mon carotid artery and endothelial dysfunc- pression using quadruple inversion-recovery. 82. Paterson JC. rupture with intimal tion in severely obese children: a prospec- Magn Reson Med 2002;48:899–905. hemorrhage as a causative factor in coro- tive study. Lancet 2001;358:1400–1404. 107. Takaya N, Yuan C, Chu B, et al. Association nary thrombosis. Arch Pathol 1938;25: 95. Wiesmann F, Petersen SE, Leeson PM, et between carotid plaque characteristics and 474–487. al. Global impairment of brachial, carotid, subsequent ischemic cerebrovascular events: 83. Chu B, Kampschulte A, Ferguson MS, et al. and aortic vascular function in young a prospective assessment with MRI—initial Hemorrhage in the atherosclerotic carotid smokers: direct quantification by high-reso- results. Stroke 2006;37(3):818–823.

Radiology: Volume 244: Number 1—July 2007 77