The Vulnerable, or High-Risk,Atherosclerotic Plaque:Noninvasive MR Imaging forCharacterization and Assessment1

Tobias Saam, MDThomas S. Hatsukami, MDNorihide Takaya, MD, PhDBaocheng Chu, MD, PhDHunter Underhill, MDWilliam S. Kerwin, PhDJianming Cai, MD, PhDMarina S. Ferguson, MTChun Yuan, PhD

“Vulnerable” plaques are atherosclerotic plaques that havea high likelihood to cause thrombotic complications, suchas myocardial infarction or stroke. Plaques that tend toprogress rapidly are also considered to be vulnerable.Besides luminal stenosis, plaque composition and mor-phology are key determinants of the likelihood that aplaque will cause cardiovascular events. Noninvasive mag-netic resonance (MR) imaging has great potential to en-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 forwardin which five major and five minor criteria to define vulner-able plaques were proposed. The purpose of this review isto summarize the status of MR imaging with regard todepiction of the criteria that define vulnerable plaques byusing existing MR techniques. The use of MR imaging inanimal models and in human disease in various vascularbeds, particularly the carotid arteries, 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 requestedDecember 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_Saam@med.uni-muenchen.de).

� RSNA, 2007

REVI

EWS

AND

COM

MEN

TARY

�RE

VIEW

64 Radiology: Volume 244: Number 1—July 2007

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.

Complications of cardiovasculardisease, including stroke, myocar-dial infarction, and sudden car-

diac death, are the most commoncauses of death in the world (1). Ath-erosclerotic disease accounts for ap-proximately 25% of ischemic strokes(2) and for the majority of myocardialinfarctions and sudden cardiac deaths(3,4). Despite major advances in thetreatment of atherosclerosis, a largerpercentage of individuals with the dis-ease who are apparently healthy diewithout prior symptoms (3). The chal-lenge for screening and diagnosticmethods is to identify patients at highrisk who have lesions that are vulnera-ble to thrombosis, so-called vulnerableplaques, before the event occurs. To tai-lor and improve treatment strategies,these screening and diagnostic methods

must be able to help determine the pa-tient-specific risk of experiencing a car-diovascular event.

Imaging methods have the potentialto not only be used as a screening toolfor the presence of atherosclerosis butalso to help distinguish stable from vul-nerable plaques and ultimately to distin-guish patients with low versus thosewith high risk of cardiovascular compli-cations. Methods commonly used inatherosclerosis imaging include B-modeultrasonography (US), intravascularUS, conventional angiography, com-puted tomography (CT), and magneticresonance (MR) imaging. Each of theseimaging modalities has advantages anddisadvantages, and all have been re-viewed elsewhere (5,6). This article willfocus on the particular potential of MRimaging to depict the key features of thevulnerable plaque. MR imaging is wellsuited for this role because it is noninva-sive, does not involve ionizing radiation,enables visualization of the vessel lumenand wall (7,8), and can be repeated se-rially to track progression or regres-sion. Furthermore, the excellent soft-tissue contrast provided by MR imagingallows evaluation of compositional andmorphologic features of atheroscleroticplaques (9–15). This information is cru-cial because, besides luminal stenosis,plaque composition and morphology arekey determinants of a plaque’s vulnera-bility with regard to causing cardiovas-cular events (3,4).

Authors of several review articles(7,8,16) have described in detail thetechnical aspects of the MR imagingcharacterization of human atheroscle-rotic plaque, including hardware con-siderations, imaging sequences, and im-aging protocols. Results of the accuracyof MR imaging, compared with histo-logic findings, for measurements ofplaque burden, tissue characterization,and fibrous cap status have also beenreported (7,8,16). Choudhury et al (7)focused on present and future MR appli-cations for the characterization of ath-erosclerotic plaques, including real-timevascular intervention, new contrastagents, and molecular imaging.

The present review is a continuationof previous review articles (8,16) and

will focus on the current status of MR asa diagnostic imaging method to helpidentify the features of vulnerableplaques. The structure of this article isbased on two recently published con-sensus documents (3,4) by a group ofexperienced researchers in atheroscle-rosis, including pathologists, clinicians,molecular biologists, and imaging scien-tists, in which the key features of thevulnerable plaque were defined. The au-thors of those documents argue thatknowledge of luminal diameter is notsufficient to determine the vulnerabilityof an atherosclerotic lesion, and theypropose five major and five minor crite-ria for the detection of vulnerableplaques. These plaque features werebased on studies of coronary arteriesand included a thin cap with a largelipid-necrotic core, active inflammation,a fissured plaque, stenosis greater than90%, endothelial denudation with orwithout superficial platelet aggregationand fibrin deposition, endothelial dys-function, calcified nodules, intraplaquehemorrhage, glistening yellow plaques(seen at angioscopy), and outward re-modeling (see Table 1).

In this review, we will discuss therole of MR imaging in the identificationof each of the features that define thevulnerable plaque, as proposed in theconsensus documents (3,4).

Major Criteria

Thin Cap with Large Lipid-Necrotic CoreVirmani et al (17) defined the fibrouscap as a distinct layer of connective tis-sue completely covering the lipid-ne-crotic core. The lipid-necrotic core,which is frequently surrounded by mac-rophages, consists of large amounts ofextracellular lipid, cholesterol crystals,and necrotic debris (17,18). Lesions

Published online10.1148/radiol.2441051769

Radiology 2007; 244:64–77

Abbreviation:TOF � time of flight

Authors stated no financial relationship to disclose.

Essentials

� The challenge for imaging meth-ods is to enable identification ofpatients with high-risk lesionsthat are vulnerable to thrombosis,so-called vulnerable plaques, be-fore the occurrence of cardiovas-cular complications.

� Noninvasive MR imaging hasgreat potential to help character-ize atherosclerotic plaque compo-sition and morphology and thus toenable assessment of plaque vul-nerability.

� Serial MR imaging of atheroscle-rotic plaques can provide usefulinsights on the natural history ofvulnerable plaques and might beuseful in identifying plaques thatare progressing toward a vulnera-ble state.

� A recent prospective study dem-onstrated that certain vulnerableplaque features identified on MRimages are associated with theoccurrence of subsequent cere-brovascular events.

� Most of the plaque imaging dataare based on larger vessels, suchas the carotid arteries, and fur-ther advances in temporal andspatial resolution are needed forcoronary plaque imaging.

REVIEW: MR Imaging of the Vulnerable Atherosclerotic Plaque Saam et al

Radiology: Volume 244: Number 1—July 2007 65

with a large lipid-necrotic core and athin fibrous cap are considered to bemost likely to rupture (19).

MR imaging is able to help charac-terize all major plaque components, in-cluding the lipid-necrotic core, by de-picting particular combinations of signalintensities of each component on im-ages obtained with different contrastweightings (see Table 2 for details). Ini-tial experiments involving ex vivo imag-ing of endarterectomy specimens by us-ing T1-, T2-, and intermediate-weightedimaging demonstrated that tissue com-ponents, including the lipid-necroticcore, hemorrhage, and calcification,could be detected with sensitivities andspecificities ranging from 84% to 100%(12). Translation of these findings to invivo imaging showed that the lipid-ne-crotic core and intraplaque hemorrhagecould be identified with a sensitivity of85% and a specificity of 92% (20). Theability of MR imaging to demonstrateplaque composition in vivo enables clas-sification of human carotid atheroscle-rotic plaque, according to criteria devel-oped by the American Heart Associa-tion. Cai et al (14) showed that in vivoMR imaging can help identify sectionsthat have either a lesion with a confluentextracellular lipid-necrotic core (Ameri-can Heart Association type IV lesion) ora lipid-necrotic core covered by a fi-brous cap (type V lesion), with a sensi-tivity of 84% and a specificity of 90%.

While initial studies showed thatMR imaging is able to depict the lipid-necrotic core, recent reports have fo-cused on quantifying it. In an ex vivo MRimaging study of carotid endarterec-tomy specimens by Clarke et al (10),who used diffusion-weighted imaging inaddition to T1-, intermediate-, and T2-weighted imaging, pixels were classifiedinto different plaque components andwere then compared with histologicspecimens. On the basis of this classifi-cation, the sensitivity and specificity fornecrotic tissue were 83.9% and 75%,respectively. Quantification of the lipid-necrotic core was recently translated toin vivo plaque imaging by using TOF andT1-, T2-, and intermediate-weighted se-quences (9). In a study with 40 subjects,a radiologist and a pathologist manually

identified the major components of ca-rotid atherosclerotic plaque on MR im-ages and serially sectioned carotid end-arterectomy specimens, respectively(Fig 1). MR measurements of the lipid-necrotic core did not differ significantlyfrom findings on histologic specimens(23.7% vs 20.3%; P � .1), and a strongcorrelation between MR and histologicarea measurements was found (r �0.75; P � .001) (9).

Comparison of T1-weighted imagesobtained before and after contrast ma-terial enhancement may improve the ac-curacy of MR for differentiating thelipid-necrotic core from surrounding fi-brous tissue. The lipid-necrotic coreshows little if any enhancement on post-contrast T1-weighted images, and dif-ferentiation from the surrounding en-hancing fibrous tissue is thereby facili-tated (21). Using pre- and postcontrastT1-weighted images, Cai et al (22)showed that the lipid-necrotic core,measured as proportion of the vessel

wall, was similar on MR images andhistologic specimens, with values of30.1% � 12.5 (standard deviation) and32.7% � 12.3, respectively. Further-more, area measurements were stronglycorrelated (r � 0.85; P � .001).

The ability of MR imaging to helpdetermine the status and thickness ofthe fibrous cap will be discussed in thefollowing section.

Fissured PlaqueVirmani et al (17) described plaque rup-ture as an area of fibrous cap disruptionwhereby the overlying thrombus is incontinuity with the underlying lipid-necrotic core. Typically, these lesionshave a large lipid-necrotic core and adisrupted fibrous cap infiltrated by mac-rophages and lymphocytes; smoothmuscle cell content within the fibrouscap at the rupture site may be sparse(17).

Hatsukami et al (15) reported theuse of a three-dimensional TOF bright-

Figure 1

Figure 1: Pie charts show plaque composition as percentage of vessel wall area, calculated per artery andthen averaged across all arteries, for MR imaging and histologic specimens. Percentages were compared byusing paired t tests. (Reprinted, with permission, from reference 9.)

Table 1

Criteria for Defining Vulnerable Plaques on the Basis of the Study of “Culprit” Plaques

Type of Criteria Criterion

Major Active inflammation: monocyte and macrophage and sometimes T-cell infiltrationThin cap with large lipid-necrotic coreEndothelial denudation with superficial platelet aggregationFissured plaqueStenosis �90%

Minor Superficial calcified noduleGlistening yellow plaque seen at angioscopyIntraplaque hemorrhageEndothelial dysfunctionOutward (positive) remodeling

REVIEW: MR Imaging of the Vulnerable Atherosclerotic Plaque Saam et al

66 Radiology: Volume 244: Number 1—July 2007

blood imaging technique (multiple over-lapping thin slab angiography, orMOTSA [23]) to help identify a rup-tured fibrous cap in atherosclerotic hu-man carotid arteries in vivo. By usingpreoperative images of 22 consecutiveendarterectomy patients, the in vivostate of the fibrous cap was character-ized on the basis of its appearance onMR images as being intact and thick,intact and thin, or ruptured (Fig 2). Theauthors found a high level of agreementbetween the MR findings and the histo-logic state of the fibrous cap, with Co-hen � � 0.83 (95% confidence inter-val � 0.67, 1.00) and weighted � �0.87. An MR investigation by Mitsumoriet al (24) demonstrated that in vivo MR

imaging has high test sensitivity (81%)and specificity (90%) for the identifica-tion of a thin or ruptured cap. Trivedi etal (25) used a short-tau inversion-recov-ery sequence to quantify the fibrous capand lipid-necrotic core of 25 recentlysymptomatic patients and correlatedthe results with the excised carotid end-arterectomy specimens. The authorsshowed good agreement between MRand histologic quantifications of boththe fibrous cap and the lipid-necroticcore content, with good interobserveragreement (intraclass correlation coef-ficients of 0.88–0.94) (25). Kramer etal (26) studied aortic atheroscleroticplaques in 23 patients with abdominalaortic aneurysm by using T1- and T2-

weighted black-blood spin-echo se-quences and gadopentetate dimeglu-mine–enhanced T1-weighted images.The � value for agreement between thenumber of tissue components demon-strated by MR imaging and identified atpathologic examination was 0.79. Thesecomponents included the fibrous cap,thrombus, and/or the lipid-necroticcore. The authors found that contrastenhancement improved the delineationof the fibrous cap, which was found tobe hyperintense relative to the lipid-ne-crotic core on T2-weighted images.Wassermann et al (27), in a study ofnine subjects with carotid atherosclero-sis, found that gadolinium-enhanced T1-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 � 390 � 5000Ex vivo human carotid (12) 9.4 Intermediate to T2 (1000–2000/13–50), T1 (300–700/13) None 256 � 256 � 500Human carotid (15) 1.5 3D multiple overlapping thin-slab MR angiography

(34 or 22/2.9–4.4)None 510 � 510 � 1000

Dogs (42) 1.5 3D T1 (24/8.1), 3D phase-contrast MR angiography (15/5.3) Fibrin-targetednanoparticle

Not applicable

Rabbit aorta (54) 1.5 3D MR angiography (6.7/1.6) USPIOs 1100 � 1000 � 1400Human coronary arteries (62) 1.5 3D MR angiography (2.4/8.8) None 700 � 1000 � 1500In vitro human carotid (11) 2.0 T2 (2000/50) None 117 � 117 � 1500Human coronary arteries (103) 1.5 3D black-blood spiral acquisition (30/2) None 1000 � 1000 � 1000Swine carotid (37) 1.5 T1 (700/11), T2 (2000/42) None 470 � 470 � 3000Human carotid and aorta (101) 1.5 Intermediate (2 R-R intervals/12), T2 (2 R-R intervals/45) None 4692–7802 � 3000Ex vivo human carotid (10) 1.5 T1, T2, intermediate, diffusion-weighted, fast imaging employing

steady state (see reference 10 for details)None 156 � 156 � 200

Rabbit aorta (56) 1.5 T1 (380/11) �v3-integrin–targetednanoparticles

250 � 250 � 5000

Human carotid (55) 1.5 T1 (41–44/8.0–9.2), T2* (R-R interval/20), intermediate(2–3 R-R intervals/20)

USPIOs 208 � 256 � 5000

Human carotid (86) 1.5 T1 turbo spin echo (900/10.3), T1 turbo field echo (570/14) None 390 � 490 � 2500Human aorta (27) 1.5 T1 breath hold (R-R interval/32), T2 breath hold

(2 R-R intervals/80)None 1210 � 620 � 7000

Swine coronary arteries (43) 1.5 3D turbo field echo (3.8/1.9), 3D inversion recovery MRangiography (4.7/1, inversion time, 4 msec)

Fibrin-targeted agent 1250 � 1250 � 3000

Human coronary arteries (64) 1.5 Whole-heart 3D steady-state free precession MR angiography(4.7–31/1.7–2.3)

None 700 � 1000 � 3000

Human carotid (9) 1.5 T1 (800/9.3–11), intermediate to T2 (3–4 R-R intervals/10–70),3D TOF (23/3.5)

None 600 � 600 � 2000

Human aorta (104) 1.5 Intermediate (2 R-R intervals/10), T2 (2 R-R intervals/60) None 780 � 780 � 5000Human carotid (49) 1.5 Two-dimensional T1 spoiled gradient-recalled echo (100/3.5) Gadolinium chelate 600 � 800 � 3000Human carotid (22) 1.5 Enhanced and unenhanced T1 (800/9.3–11) Gadolinium chelate 600 � 600 � 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 � time of flight, 3D � three-dimensional.‡ USPIO � ultrasmall paramagnetic iron oxide.

REVIEW: MR Imaging of the Vulnerable Atherosclerotic Plaque Saam et al

Radiology: Volume 244: Number 1—July 2007 67

the fibrous cap from the lipid-necroticcore, with a contrast-to-noise ratio asgood as or better than that of T2-weighted MR images but with approxi-mately twice the signal-to-noise ratio(postcontrast images, 36.6 � 3.6; T2-weighted images, 17.5 � 2.1; P � .001).Cai et al (22) used unenhanced T1-weighted and contrast-enhanced T1-weighted images to measure the intactfibrous cap. The authors showed mod-erate to good correlation between find-ings from carotid MR imaging and theexcised histologic specimen for maximalthickness (r � 0.78, P � .001), length(r � 0.73, P � .001), and area (r �0.90, P � .001) of intact fibrous cap.

Yuan et al (28) used three-dimen-sional TOF and intermediate-, T1-, andT2-weighted images in 28 symptomaticand 25 asymptomatic subjects to deter-mine whether fibrous cap thinning orrupture, identified on MR images, is as-sociated with a history of recent tran-sient ischemic attack or stroke. Com-pared with patients with a thick fibrouscap, patients with a ruptured cap were23 times more likely to have had a re-cent transient ischemic attack or stroke(95% confidence interval: 3, 210) (28).

Endothelial Denudation with SuperficialPlatelet Aggregation and FibrinDepositionAtherosclerotic plaques with endothe-lial denudation are characterized eitherby superficial erosion and platelet ag-gregation or by fibrin deposition (17).Platelet aggregation may lead to luminal

occlusion and acute cardiovascularevents, or platelets may layer and orga-nize, further contributing to plaque pro-gression (17,29). In this section, we willdiscuss the ability of MR imaging to helpidentify platelet aggregation or fibrindeposition. (The ability of MR to helpevaluate endothelial denudation or dys-function will be discussed in the sectionEndothelial Dysfunction.)

The detection of thrombus and thedetermination of its age on MR imagesare mainly related to the physical char-acteristics and visual appearance ofthrombus (30), which originate fromthe cross linking of the fibrin strands,thrombus organization (31), the struc-ture of hemoglobin, and the oxidationstate (32). Toussaint et al (31) used thewater diffusion properties of athero-thrombotic tissue to distinguish freshthrombi from 1-week-old thrombi andoccluding old thrombi on MR images.Recently, a similar technique (diffusion-weighted MR) was successfully appliedfor the detection of cerebral venousthrombosis (33) and dural sinus throm-bosis (34). Gradient-recalled-echo MRimaging is frequently applied in thestudy of cerebral hematoma and venousthrombosis (35).

Corti et al (36) demonstrated thepotential of black-blood MR imaging (T1and T2 weighted) for help in the detec-tion of thrombus and thrombus age inpigs. Thrombi induced in the carotidarteries were monitored at 6 hours; 1day; and 1, 2, 3, 6, and 9 weeks. Tem-poral changes due to thrombus organi-

zation were clearly depicted on T1- andT2-weighted MR images and were ex-pressed as alterations in signal intensityrelative to the signal intensity of adja-cent muscle. Authors of another study(37) showed the potential of gadolin-ium-enhanced MR imaging to help dis-criminate different types of intracardiacclots (subacute and organized) in thecardiac chambers in 15 patients sched-uled for cardiotomy.

Kampschulte et al (38) used an invivo multisequence protocol (TOF andintermediate-, T1-, and T2-weightedimaging) in 26 patients scheduled forcarotid endarterectomy, to differentiatebetween intraplaque hemorrhage (Fig3a) and juxtaluminal hemorrhage/thrombus (Fig 3b). For locations inwhich MR and histologic findings werein agreement regarding the presence ofany type of hemorrhage, MR imagingenabled differentiation of juxtaluminalhemorrhage and thrombus from in-traplaque hemorrhage with an accuracyof 96%, with histologic findings as thereference standard. This technique wasused to evaluate whether in vivo MRimaging can depict differences betweensymptomatic and asymptomatic carotidatherosclerotic plaques in the same pa-tient at the same point in time (39).Compared with asymptomatic plaques,plaques associated with neurologicsymptoms were associated with ahigher occurrence of juxtaluminal hem-orrhage and thrombus (61% vs 26%;P � .039), whereas intraplaque hemor-rhage was common in both symptom-

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.

REVIEW: MR Imaging of the Vulnerable Atherosclerotic Plaque Saam et al

68 Radiology: Volume 244: Number 1—July 2007

atic and asymptomatic plaques (91% vs83%; P � .5).

New targeted contrast agents to en-able detection of local small thrombi are

under development (30). Fibrin can beidentified by using lipid-encapsulatedperfluorocarbon paramagnetic nano-particles in vitro (40,41) and in vivo

(41). Botnar et al (42) used a fibrin-binding gadolinium-labeled peptide inan experimental rabbit model of plaquerupture and thrombosis to detect acuteand subacute thrombosis. Similaragents have been successfully used inswine to detect coronary thrombus andin-stent thrombosis (43), as well as pul-monary emboli (44).

Active InflammationInflammation plays a critical role inplaque 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 obtainedby testing blood samples, and it hasbeen shown that a variety of circulatinginflammatory markers are associatedwith the risk of coronary artery disease(45). However, it appears unlikely thatC-reactive protein or any of the othermarkers actually causes the disease(45). Instead, they reflect the local in-flammatory process in the artery andperhaps other tissues (eg, adipose tis-sue) (45). It is this local inflammatoryprocess that can be assessed by nonin-vasive imaging methods.

Plaques with active inflammationmay be identified by the presence ofextensive macrophage accumulation(46), and recent investigations with MRimaging have shown an association be-tween contrast enhancement and thedegree of macrophage infiltration ofplaque (47,48). Lin et al (49) were, toour knowledge, the first to use gadolin-ium-based contrast agents in athero-sclerotic plaque imaging in an animalmodel, and they demonstrated strongcontrast enhancement of the vesselwall. Lin et al concluded that the con-trast enhancement of the vessel wallthat they observed was caused by anincreased vascular supply associatedwith thrombosis and neointimal thick-ening. Aoki et al (50) used dynamic MRimaging to correlate carotid wall en-hancement changes with pathologicconditions. The dominant feature theynoted was a bright outer rim of the ves-sel wall, which was attributed to growthof the vasa vasorum in the adventitia.

Figure 3

Figure 3: (a)Left: TransverseTOF(23/3.8), T1-weighted (T1W) (800/9.3), intermediate-weighted (IMW) (3488/20),andT2-weighted (T2W) (3488/40)MRimagesdepict intraplaquehemorrhage(arrows)deepwithinhighlystenosedca-rotidplaquewithsmoothluminalsurface.Arrowheads� lumen.Right:Endarterectomyspecimen. (b)Left:TransverseTOF(23/3.8),T1-weighted (T1W) (550/9), intermediate-weighted (IMW) (3093/20),andT2-weighted (T2W) (3093/40)MRimagesdepict juxtaluminalhemorrhage/thrombus(arrows)adjacent to the lumen(�),with luminalsurface irregularity(chevrons).Acalcifiedarea(arrowheads) ishypointenseonMRimagesobtainedwithall fourcontrastweightings.Right:Endarterectomyspecimen. (Reprinted,withpermission, fromreference38.)

REVIEW: MR Imaging of the Vulnerable Atherosclerotic Plaque Saam et al

Radiology: Volume 244: Number 1—July 2007 69

Yuan et al (21) used contrast-enhancedT1-weighted images and found thegreatest enhancement to be associatedwith areas of dense neovasculature,which are thought to contribute toplaque destabilization (51).

An advantage of using contrastagents that is only beginning to be ex-plored in plaque imaging is the abilityto extract quantitative measurementsof enhancement. To date, such studieshave focused on the use of kineticmodeling and dynamic contrast-en-hanced MR imaging to measure therate of enhancement. Kerwin et alhave identified a fast enhancementcomponent, vp, associated with plaqueneovasculature (47) and a slower en-hancement component, Ktrans (Fig 4),associated with permeability (48). Ithas recently been demonstrated (48)that these dynamic enhancement pa-rameters are strongly associated withmacrophage density in advanced ca-rotid plaque (Fig 4). Macrophage den-sity was estimated from histologicspecimens by using software (Image-Pro Plus; Media Cybernetics, Be-thesda, Md) to measure the total areathat stained positive for macrophagesand normalizing according to theplaque cross-sectional area (48).

An alternative method for imagingmacrophage content is to use ultra-small superparamagnetic iron oxide(USPIO) particles that are suspendedin solution and injected into patients.USPIOs alter MR reaction times, andthe propensity of macrophages tophagocytose the USPIOs results inUSPIOs being a macrophage-specificagent (52). If sufficient time for mac-rophage uptake is allowed—at least 24hours—T2*-weighted images showsubstantial signal intensity reductiondue to the susceptibility effects ofUSPIOs (53). In experiments with hu-man carotid endarterectomy subjects,Kooi et al (54) showed that macro-phages within plaques were positivefor iron at histologic and electron mi-croscopy evaluations of endarterec-tomy specimens. On MR images ofcorresponding locations, substantialsignal intensity reductions were ob-served.

�v3-Integrin is a molecule that isbelieved to be associated with angiogen-esis and inflammation. An �v3-inte-grin–targeted agent was shown to yieldgreater enhancement in cholesterol-fedrabbits than in controls, and cholester-ol-fed rabbits also exhibited increaseddevelopment of the adventitial vasa va-sorum (55).

Severe StenosisArterial stenosis is an abnormal nar-rowing of a blood vessel that reducesblood flow in the lumen, which maycause ischemic cardiovascular compli-cations. On the surface of atheroscle-rotic plaques associated with severe ste-nosis, shear stress imposes a risk ofthrombosis and sudden occlusion (3).Therefore, a stenotic plaque may be avulnerable plaque regardless of pres-ence or absence of ischemia (3). More-over, a stenotic plaque may indicate thepresence of many nonstenotic or less-stenotic plaques that could be vulnera-ble to rupture and thrombosis (56,57).

MR angiography of the carotid ar-teries has gone through a long evolu-tionary period to become a routine im-aging modality for evaluation of stenosisat many centers (58,59). In carotid ar-teries, the ability of MR angiography todemonstrate less than 70% versus70%–99% stenosis is much better thanthat of duplex US (60). Also, MR angiog-raphy is a sensitive and specific testwhen digital subtraction angiography isused as the reference standard (61).

However, noninvasive imaging ofthe coronary arteries is a formidabletask: The small diameter of the coro-nary arteries (approximately 2–4 mm),their tortuous course, their close ana-tomic relationship to the coronary veinsand cardiac chambers, and, finally,their continuous rapid motion due tocardiac contractions and respiratory ex-cursions create obstacles that are diffi-cult to overcome. Kim et al (60) used afree-breathing, navigator-gated, three-dimensional MR technique to depict thecoronary arteries and help detect coro-nary stenoses in a series of 109 patients(Fig 5). In segments that could be evalu-ated, 78 of 94 stenoses of more than50% were detected by using MR angiog-

raphy, with conventional coronary an-giography as the reference standard. Ina separate analysis, the authors calcu-lated a sensitivity of 100% for the iden-tification of either left main coronaryartery disease or three-vessel disease,conditions in which revascularization isof particular therapeutic value. How-ever, current spatial (approximately1 � 1 � 2-mm) and temporal (approxi-mately 60–125-msec) resolutions per-mit imaging only of the proximal two-thirds of the coronary arteries, exclu-sive of the major side branches (62).Though coronary MR angiography is prom-ising, further development is needed beforeit can enter routine clinical use (62).

Recently, alternative acquisition tech-niques, such as steady-state free preces-sion (64), have been proposed (63) thatestablish different contrast behavior anduse different k-space acquisition schemes,such as spiral (65) or radial (also knownas projection-reconstruction) (66) read-out patterns. Furthermore, a whole-heartsteady-state free-precession coronaryMR angiography technique has been pro-posed (63) that improves visible vessellength and facilitates high-quality coro-nary MR angiography of the completecoronary tree in a single measurement.These newer sequences are reported toprovide several advantages over the free-

Figure 4

Figure 4: Scatterplot demonstrates correlationbetween dynamic contrast-enhanced MR imagingand corresponding histologic findings. There is ahighly significant association between Ktrans asmeasured on MR images and degree of macro-phage infiltration of the plaque on histologic spec-imens (fractional macrophage area) (Pearson R �0.75, P � .001). (Reprinted, with permission,from reference 48.)

REVIEW: MR Imaging of the Vulnerable Atherosclerotic Plaque Saam et al

70 Radiology: Volume 244: Number 1—July 2007

breathing T2-prepared segmented gradi-ent-recalled-echo sequence (67,68). Theadvantages include increases in signal-to-noise ratio, contrast-to-noise ratio, vesselsharpness, and coverage of vessel length,as well as reduction in acquisition time.

Traditionally, the degree of stenosisis assessed on a projection angiogramby using the North American Symptom-atic Carotid Endarterectomy Trial(NASCET) criteria, the European Ca-

rotid Surgery Trial (ECST) criteria, thecommon carotid method, or the so-called eyeballing method (69). While ithas been demonstrated that the degreeof ipsilateral carotid stenosis, as mea-sured with the techniques used in theECST and NASCET, clearly predictstroke risk (70,71), all of these tech-niques have sizable disagreementsamong repeated measurements forsome vessels (68).

For evaluation of the degree of ste-nosis on projection angiograms, theminimum luminal diameter at a targetsite is determined. The projection im-age should be generated at an angle thatallows measurement of the minimum lu-minal diameter; this dimension may notbe measurable in cases of eccentric ste-nosis with images generated at subopti-mal angles (72). In contrast, on cross-sectional images the minimum luminaldiameter can be measured accuratelywithout difficulty (72,73), and some in-vestigators state that a reduction incross-sectional area correlates betterwith the hemodynamic effect of stenosisthan does a reduction in diameter (74).In the future, new computerized carotidstenosis measurement systems thatmeasure the stenosis by using trans-verse MR angiographic images (75,76)have the potential to further reducevariability.

Minor Criteria

Superficial Calcified NoduleThe consensus documents (3,4) de-scribe a type of plaque that has a calci-fied nodule within or very close to thefibrous 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 withsevere coronary calcification and a highcalcium score (17).

In vivo MR imaging is able to depictoverall plaque calcification with goodsensitivity and specificity (9). Further-more, a strong correlation between MRand histologic area measurements hasbeen achieved (r � 0.74, P � .001) (9).However, juxtaluminal calcification canbe more difficult to identify than calcifi-cation that is deeper within the plaque,because juxtaluminal calcification ap-pears as a hypointense signal on TOFand T1-, intermediate-, and T2-weighted images and is not distinguish-able from the lumen on black-blood T1-,intermediate-, and T2-weighted images.However, the addition of bright-bloodTOF imaging facilitates identification ofjuxtaluminal calcification, in that it is

Figure 5

Figure 5: Coronary angiography in a 53-year-old man with exertional chest pain. AA � ascending aorta,LA � left atrium, LV � left ventricle, PA � pulmonary artery, RV � right ventricle, RVOT � right ventricularoutflow tract. (a) Coronary MR angiogram (left) along left main and left anterior descending coronary arteriesand corresponding conventional coronary angiogram in right anterior oblique projection (right) indicate se-vere stenosis at bifurcation of left anterior descending and left circumflex coronary arteries involving the leftmain coronary artery (solid arrows) and more distal focal stenosis of left circumflex coronary artery (dottedarrows). (b) Coronary MR angiogram (left) in coronal orientation along course of right coronary artery andcorresponding conventional angiogram in left anterior oblique projection (right) indicate two stenoses ofproximal (solid arrows) and middle (dotted arrows) right coronary artery. (Reprinted, with permission, fromreference 61.)

REVIEW: MR Imaging of the Vulnerable Atherosclerotic Plaque Saam et al

Radiology: Volume 244: Number 1—July 2007 71

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

Glistening Yellow Plaques at AngioscopyYellow plaques, particularly glisteningones, may indicate a large lipid-necroticcore and a thin fibrous cap, suggesting ahigh risk of rupture. However, becauseplaques in different stages can be yellowand 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 ofangioscopic observations. Angioscopy isan intravascular imaging method thatachieves high spatial resolution. How-ever, the information is limited to theluminal surface, the technique is inva-sive and highly operator dependent, andit is currently not approved by the Foodand Drug Administration for use in theUnited States. Obviously, MR imagescannot depict the color of the plaque’ssurface. However, as mentioned earlierin the section Thin Cap with LargeLipid-Necrotic Core, invivo MR imagingis able to depict lipid-necrotic cores andthe status of the fibrous cap, both fea-tures of angioscopically glistening yel-low plaques.

Intraplaque HemorrhageExtravasation of red blood cells or ironaccumulation in plaque may representplaque instability and promote plaqueprogression (3,79). The mechanismthat results in intraplaque hemorrhagesis not completely understood: Constan-tinides (80) and others (81) have sug-gested that hemorrhage into a plaqueoccurs from cracks or fissures that orig-inate at the luminal surface, whilePaterson (82) has proposed that in-traplaque hemorrhage is secondary torupture of the vasa vasorum, a commonfeature of advanced lesions with plaquerupture and luminal thrombi.

In vivo MR imaging is able to depictcarotid intraplaque hemorrhage (Fig

3a) with good sensitivity and moderateto good specificity (13,20,83–85). Fur-thermore it has been shown that MRimaging can help differentiate betweenhemorrhages of different ages (83) withthe use of modified criteria for cerebralhemorrhage. Moody et al (84) showedthat T1-weighted images of the carotidarteries can be used to accurately iden-tify histologically confirmed complicatedplaque with hemorrhage or thrombus.

Though intraplaque hemorrhage islisted as a minor criterion, recent arti-cles (79,86,87) suggest a greater role ofhemorrhage in plaque destabilizationthan was previously thought. Results ofa study of 24 patients (79) who had diedsuddenly of coronary causes suggestedthat hemorrhage contributed to thedeposition of free cholesterol, infiltra-tion by macrophages, and enlargementof the necrotic core, thereby represent-ing a potent atherogenic stimulus. Con-sistent with this notion, a recent pro-spective longitudinal MR investigation(87) of 31 patients showed that hemor-rhage into a carotid plaque acceleratedplaque progression in a period of 18months. The percent change in wall vol-ume (6.8% vs 0.15%, P � .009) and

lipid-necrotic core volume (28.4% vs5.2%, P � .001) was significantlyhigher in the hemorrhage group than inthe control group (Fig 7). Further, pa-tients with intraplaque hemorrhage atbaseline showed a far greater suscepti-bility to repeat plaque hemorrhages(87). Therefore, intraplaque hemor-rhage may represent a critical transi-tion, promoting the conversion of a sta-ble to an unstable plaque phenotype(86).

Endothelial DysfunctionStudies have shown that endothelialdysfunction is associated with coronaryheart disease and stroke (88,89). Fur-thermore, vulnerable plaques with ac-tive inflammation and oxidative stressare likely to be associated with impairedendothelial function (3).

Although in vivo MR imaging is notable to depict the endothelium directly,MR and a number of other imaging mo-dalities can be used to measure arterialstiffness noninvasively (90). Arterialstiffness and endothelial dysfunctioncommonly coexist in patients at in-creased risk of cardiovascular disease(90), for example in patients with diabe-

Figure 6

Figure 6: Left: Transverse MR images demonstrate usefulness of a multisequence protocol. TOF image(21.0/3.2) reveals two distinct areas (arrows A and B) with irregular lumen boundaries that are not well definedon T1-weighted (T1W, 550/9.0), intermediate-weighted (IMW, 1528/28), and T2-weighted (T2W, 1528/56)images, consistent with juxtaluminal calcification. Right: Histologic specimens demonstrate that materialintruding into the lumen contains calcified nodules. Boxes A and B correspond to areas labeled on TOF image.

REVIEW: MR Imaging of the Vulnerable Atherosclerotic Plaque Saam et al

72 Radiology: Volume 244: Number 1—July 2007

tes (91) and in smokers (92). Indeed,some studies in children have directlyrelated increased stiffness with im-paired endothelial function (93,94).

Although MR imaging can help as-sess various stiffness parameters suchas vascular distensibility, flow measure-ments, and pulse wave velocity by usingtrue fast imaging with steady-state freeprecession (95) and gradient-recalled-echo pulse sequences with a velocity-encoding gradient for phase-contrastMR (96,97), only a few such investiga-tions have been performed in this field(90). This is due to the large number ofother imaging modalities, such as Dopp-ler US or applanation tonometry, whichalso allow evaluation of arterial stiffness(90). However, a recent study showedthat MR imaging offers insights not oth-erwise possible (90) with regard to de-scribing an age-related increase in pulsewave velocity in the proximal aorta rel-ative to that in the distal aorta (97). Inanother MR investigation (95) in whichcentral vascular distensibility was as-sessed, pulse wave velocity and flow-mediated dilatation showed global im-pairment of brachial, carotid, and aorticvascular function in young smokers.

Another indirect MR measure of theendothelial function is to measure thetransfer constant Ktrans by using dy-

namic contrast-enhanced MR imaging(48). This parameter quantifies the in-creased vascularity, permeability, andextracellular water content typically as-sociated with the inflammatory re-sponse.

Outward (Positive) RemodelingMany nonstenotic lesions undergo “ex-pansive,” “positive,” or “outward” re-modeling: namely, compensatory en-largement that permits growth in over-all plaque size without reduction inluminal area. Several authors (98,99)have suggested that such remodeling isa potential surrogate marker of plaquevulnerability. MR imaging is ideallysuited for serial and noninvasive assess-ment of arterial remodeling, becauseMR provides information about the lu-men and arterial wall simultaneously.

Clinical trials of MR imaging to ex-amine the results of statin therapy onplaque progression or regression havedemonstrated a greater effect on vesselwall area than on changes in lumen di-mensions, which suggests a remodelingprocess on the vessel wall. Corti et al(100) performed serial carotid and aor-tic MR imaging in patients treated withsimvastatin and found that vessel wallarea and vessel wall thickness de-creased by 18% and 19%, respectively,

in carotid lesions and by 16% and 16%,respectively, in thoracic lesions at 24months (Fig 8). Lumen area increasedby only 5% after 24 months of statintherapy. Recently, Yonemura et al(101) reported the results of a prospec-tive, randomized, open-label trial to elu-cidate the effects on atheroscleroticplaques in the thoracic aorta of atorva-statin dosages of 20 mg/d versus 5mg/d. Their results indicated that the20 mg/d dose reduced maximal vesselwall thickness and vessel wall area of tho-racic aortic plaques (12% and 18%,respectively; P � .001), whereas 5 mg/ddid not (�1% and �4%, respectively).Lumen area increased 5% in the high-dose group and 0% in the low-dosegroup.

In a study that included six healthysubjects and six subjects with nonsignifi-cant coronary artery disease (10%–50%diameter reduction on conventional an-giogram), Kim et al (102) introduced anoninvasive MR imaging method to dem-onstrate expansive remodeling in coro-nary arteries. Free-breathing three-di-mensional black-blood coronary MR im-aging with isotropic resolution depictedan increased coronary vessel wall thick-ness with preservation of lumen size inpatients with nonsignificant coronary ar-tery disease, consistent with expansivearterial remodeling. A recent study (103)in 15 volunteers demonstrated the feasi-bility of coronary wall MR imaging withfree-breathing and breath-hold two-di-mensional black-blood turbo spin-echosequences at 3 T. Although measure-ments of coronary wall thickness andarea and lumen diameter and area wereconsistent with previous MR measure-ments at 1.5 T, the authors concludedthat further improvement in resolutionand image quality is required to enabledetection and characterization of coro-nary plaques (103).

Summary, Clinical Perspectives, andConclusion

Noninvasive MR imaging has the poten-tial to help identify, directly or indi-rectly, most of the consensus criteriafor the identification of vulnerableplaque in larger vessels by using a vari-

Figure 7

Figure 7: Representative transverse T1-weighted MR images (800/9.3) of the progression of atherosclero-sis with intraplaque hemorrhage. Right carotid artery was imaged (a) at baseline and (b) 18 months later. Lu-men area (�) decreased and wall area (arrowheads) increased in each section on b. Bif � bifurcation, CCA �common carotid artery, ECA � external carotid artery, ICA � internal carotid artery. (Reprinted, with permis-sion, from reference 87.)

REVIEW: MR Imaging of the Vulnerable Atherosclerotic Plaque Saam et al

Radiology: Volume 244: Number 1—July 2007 73

ety of proposed image-acquisition andcontrast enhancement techniques (seeTable 2). Thus far, there is not a singleMR technique that can be used to assessall the vulnerable plaque criteria, and itremains to be seen which imaging ap-proach, if any, will be commonly used inclinical practice. While some of the pro-posed image acquisition techniques relyon one (11) or two (26,36,85,101) MRsequences, others are based on the ac-quisition of multiple contrast-weightedimages (9,10,104).

The number and types of sequencesused 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 MRsequences (100,105). To evaluate plaquecomposition and morphology in one im-aging session, however, most studiesrely on multiple MR sequences(9,10,104). Black-blood T1-, intermedi-ate-, and T2-weighted images with fat(8) and flow suppression (106) enablevisualization of the full vessel wall andare needed to characterize the majorplaque components, such as a lipid-necrotic core, calcification, and a loosefibrous matrix (9). Bright-blood TOFimages are needed for evaluation of thestatus of the fibrous cap (15,28) andidentification of calcified nodules (seeFig 6) (20). Other sequences, such ascontrast-enhanced T1-weighted images(22,27) and dynamic contrast-enhancedtwo-dimensional spoiled gradient-re-called-echo images (48) are useful forquantifying the size of the lipid-necroticcore (22) and evaluating plaque inflam-mation (47,48). Furthermore, the useof sophisticated contrast agents, such asultrasmall superparamagnetic iron ox-ides (53,54) and fibrin-targeted (41,42)and �v3-integrin–targeted (55) nano-particles offers the promise of in vivotargeted imaging of the plaque (7).

The ability to distinguish stable fromvulnerable plaques may ultimately per-mit identification of patients with lowversus those with high risk of cardiovas-cular complications. Furthermore, invivo MR imaging has potential as a toolfor screening for atherosclerosis and forserial monitoring of disease progressionor regression. It is important to note

that the consensus document (3,4) cri-teria that define the vulnerable plaqueare based on histologic studies of culpritplaques. Histologic examination canonly capture information regarding theplaque from a single time point. Thus,little is known about the evolution ofvulnerable plaques, and noninvasive se-rial MR imaging of such plaques can,therefore, provide useful insights intotheir natural history and might help toidentify plaques that are on a trajectoryof evolution toward a vulnerable state.

In a recent prospective in vivo MRimaging study (107) that used T1-, T2-,and intermediate-weighted images andTOF images, the authors investigatedthe association between carotid plaquecharacteristics and subsequent ische-mic cerebrovascular events in 154 sub-jects who initially had an asymptomatic50%–79% carotid stenosis seen at du-plex US. During a mean follow-up of38.2 months, 12 carotid cerebrovascu-lar events occurred. Cox regressionanalysis indicated that plaques with in-traplaque hemorrhage (hazard ratio:

4.7; 95% confidence interval: 1.6, 14.0;P � .004), larger mean intraplaquehemorrhage area (hazard ratio for 10-mm2 increase: 2.4; 95% confidence in-terval: 1.4, 4.1; P � .008), thin or rup-tured fibrous cap (hazard ratio: 9.4;95% confidence interval: 2.1, 42.1; P �.001), larger maximum percentage lip-id-rich–to-necrotic core ratio (hazardratio for 10% increase: 1.4; 95% confi-dence interval: 1.1, 1.9; P � .01), andmaximum wall thickness (hazard ratiofor 1-mm increase: 1.6; 95% confidenceinterval, 1.1, 2.1; P � .007) were asso-ciated with the cerebrovascular events.

Although initial findings of this pro-spective study (107) were promising,several challenges remain. Most of thein vivo MR studies in humans cited inthis article were based on analysis ofdata from relatively large human ves-sels, such as the carotid arteries. Toachieve similar results in the muchsmaller coronary arteries, substantialadvances in temporal and spatial resolu-tion are necessary. This may be accom-plished with improvements in pulse se-

Figure 8

Figure 8: Bar graphs show changes in atherosclerotic vessel wall dimensions (in square millimeters) afterstatin treatment: mean vessel wall area (top) and mean lumen area (bottom) at baseline (BL) and after 6, 12, 18,and 24 months of simvastatin treatment for aorta (left) and carotid arteries (right). ANOVA � analysis of vari-ance, error bars � standard error of the mean. (Reprinted, with permission, from reference 100.)

REVIEW: MR Imaging of the Vulnerable Atherosclerotic Plaque Saam et al

74 Radiology: Volume 244: Number 1—July 2007

quence design and MR hardware (eg,higher field strength, new coil designs).Therefore, prospective studies in largepopulations are needed to determinethe predictive value of MR-based as-sessment of vulnerable plaque featuresfor subsequent ischemic events.

Acknowledgment: We thank Andrew An Ho,MA, for editing this manuscript.

References1. World Health Organization. World Health Sta-

tistics 2005. http://www.who.int/healthinfo/statistics/whsatsdownloads/en/index.html.Published 2005. Accessed April 18, 2007.

2. Weinberger J. Diagnosis and prevention ofatherosclerotic cerebral infarction. CNSSpectr 2005;10:553–564.

3. Naghavi M, Libby P, Falk E, et al. Fromvulnerable plaque to vulnerable patient: acall for new definitions and risk assessmentstrategies—part I. Circulation 2003;108:1664–1672.

4. Naghavi M, Libby P, Falk E, et al. Fromvulnerable plaque to vulnerable patient: acall for new definitions and risk assessmentstrategies—part II. Circulation 2003;108:1772–1778.

5. Mistretta CA. Relative characteristics ofMR angiography and competing vascularimaging modalities. J Magn Reson Imaging1993;3:685–698.

6. Saam T, Kerwin WS, Yuan C. Vascular im-aging. In: Wnek GE, Bowlin GL, eds. Ency-clopedia of biomaterials and biomedical en-gineering. New York, NY: Dekker, 2004.

7. Choudhury RP, Fuster V, Badimon JJ, FisherEA, Fayad ZA. MRI and characterization ofatherosclerotic plaque: emerging applicationsand molecular imaging. Arterioscler ThrombVasc Biol 2002;22:1065–1074.

8. Yuan C, Mitsumori LM, Beach KW, Mara-villa KR. Carotid atherosclerotic plaque:noninvasive MR characterization and iden-tification of vulnerable lesions. Radiology2001;221:285–299.

9. Saam T, Ferguson MS, Yarnykh VL, et al.Quantitative evaluation of carotid plaquecomposition by in vivo MRI. ArteriosclerThromb Vasc Biol 2005;25:234–239.

10. Clarke SE, Hammond RR, Mitchell JR, RuttBK. Quantitative assessment of carotidplaque composition using multicontrastMRI and registered histology. Magn ResonMed 2003;50:1199–1208.

11. Serfaty JM, Chaabane L, Tabib A, Cheval-lier JM, Briguet A, Douek PC. Atheroscle-

rotic plaques: classification and character-ization with T2-weighted high-spatial-reso-lution MR imaging—an in vitro study.Radiology 2001;219:403–410.

12. Shinnar M, Fallon JT, Wehrli S, et al. Thediagnostic accuracy of ex vivo MRI for hu-man atherosclerotic plaque characteriza-tion. Arterioscler Thromb Vasc Biol 1999;19:2756–2761.

13. Toussaint JF, LaMuraglia GM, Southern JF,Fuster V, Kantor HL. Magnetic resonanceimages lipid, fibrous, calcified, hemor-rhagic, and thrombotic components of hu-man atherosclerosis in vivo. Circulation1996;94:932–938.

14. Cai JM, Hatsukami TS, Ferguson MS, SmallR, Polissar NL, Yuan C. Classification ofhuman carotid atherosclerotic lesions within vivo multicontrast magnetic resonanceimaging. Circulation 2002;106:1368–1373.

15. Hatsukami TS, Ross R, Polissar NL, YuanC. Visualization of fibrous cap thicknessand rupture in human atherosclerotic ca-rotid plaque in vivo with high-resolutionmagnetic resonance imaging. Circulation2000;102:959–964.

16. Yuan C, Kerwin WS. MRI of atherosclerosis.J Magn Reson Imaging 2004;19:710–719.

17. Virmani R, Kolodgie FD, Burke AP, Farb A,Schwartz SM. Lessons from sudden coro-nary death: a comprehensive morphologi-cal classification scheme for atheroscleroticlesions. Arterioscler Thromb Vasc Biol2000;20:1262–1275.

18. Stary HC, Chandler AB, Dinsmore RE, et al.A definition of advanced types of atheroscle-rotic lesions and a histological classification ofatherosclerosis: a report from the Committeeon Vascular Lesions of the Council on Arte-riosclerosis, American Heart Association.Circulation 1995;92:1355–1374.

19. Burke AP, Weber DK, Kolodgie FD, FarbA, Taylor AJ, Virmani R. Pathophysiologyof calcium deposition in coronary arteries.Herz 2001;26:239–244.

20. Yuan C, Mitsumori LM, Ferguson MS, et al.In vivo accuracy of multispectral magneticresonance imaging for identifying lipid-richnecrotic cores and intraplaque hemorrhagein advanced human carotid plaques. Circu-lation 2001;104:2051–2056.

21. Yuan C, Kerwin WS, Ferguson MS, et al.Contrast-enhanced high resolution MRI foratherosclerotic carotid artery tissue char-acterization. J Magn Reson Imaging 2002;15:62–67.

22. Cai J, Hatsukami TS, Ferguson MS, et al. Invivo quantitative measurement of intact fi-brous cap and lipid-rich necrotic core size

in atherosclerotic carotid plaque: compari-son of high-resolution, contrast-enhancedmagnetic resonance imaging and histology.Circulation 2005;112:3437–3444.

23. Parker DL, Yuan C, Blatter DD. MR angiog-raphy by multiple thin slab 3D acquisition.Magn Reson Med 1991;17:434–451.

24. Mitsumori LM, Hatsukami TS, FergusonMS, Kerwin WS, Cai J, Yuan C. In vivoaccuracy of multisequence MR imaging foridentifying unstable fibrous caps in ad-vanced human carotid plaques. J Magn Re-son Imaging 2003;17:410–420.

25. Trivedi RA, King-Im J, Graves MJ, et al.Multi-sequence in vivo MRI can quantifyfibrous cap and lipid core components inhuman carotid atherosclerotic plaques. EurJ Vasc Endovasc Surg 2004;28:207–213.

26. Kramer CM, Cerilli LA, Hagspiel K, DiMa-ria JM, Epstein FH, Kern JA. Magnetic res-onance imaging identifies the fibrous cap inatherosclerotic abdominal aortic aneu-rysm. Circulation 2004;109:1016–1021.

27. Wasserman BA, Smith WI, Trout HH III,Cannon RO 3rd, Balaban RS, Arai AE. Ca-rotid artery atherosclerosis: in vivo mor-phologic characterization with gadolinium-enhanced double-oblique MR imaging—ini-tial results. Radiology 2002;223:566–573.

28. Yuan C, Zhang SX, Polissar NL, et al. Iden-tification of fibrous cap rupture with mag-netic resonance imaging is highly associ-ated with recent transient ischemic attackor stroke. Circulation 2002;105:181–185.

29. Falk E, Shah PK, Fuster V. Coronary plaquedisruption. Circulation 1995;92:657–671.

30. Wentzel JJ, Aguiar SH, Fayad ZA. VascularMRI in the diagnosis and therapy of the highrisk atherosclerotic plaque. J Interv Cardiol2003;16:129–142.

31. Toussaint JF, Southern JF, Fuster V, Kan-tor HL. Water diffusion properties of hu-man atherosclerosis and thrombosis mea-sured by pulse field gradient nuclear mag-netic resonance. Arterioscler Thromb VascBiol 1997;17:542–546.

32. Bradley WG Jr. MR appearance of hemor-rhage in the brain. Radiology 1993;189:15–26.

33. Chu K, Kang DW, Yoon BW, Roh JK. Dif-fusion-weighted magnetic resonance in ce-rebral venous thrombosis. Arch Neurol2001;58:1569–1576.

34. Yoshikawa T, Abe O, Tsuchiya K, et al.Diffusion-weighted magnetic resonance im-aging of dural sinus thrombosis. Neuroradi-ology 2002;44:481–488.

35. Taber KH, Hayman LA, Herrick RC, Kirk-

REVIEW: MR Imaging of the Vulnerable Atherosclerotic Plaque Saam et al

Radiology: Volume 244: Number 1—July 2007 75

patrick JB. Importance of clot structure ingradient-echo magnetic resonance imagingof hematoma. J Magn Reson Imaging 1996;6:878–883.

36. Corti R, Osende JI, Fayad ZA, et al. In vivononinvasive detection and age definition ofarterial thrombus by MRI. J Am Coll Car-diol 2002;39:1366–1373.

37. Paydarfar D, Krieger D, Dib N, et al. In vivomagnetic resonance imaging and surgicalhistopathology of intracardiac masses: dis-tinct features of subacute thrombi. Cardiol-ogy 2001;95:40–47.

38. Kampschulte A, Ferguson MS, Kerwin WS,et al. Differentiation of intraplaque versusjuxtaluminal hemorrhage/thrombus in ad-vanced human carotid atherosclerotic le-sions by in vivo magnetic resonance imag-ing. Circulation 2004;110:3239–3244.

39. Saam T, Cai J, Ma L, et al. Comparison ofsymptomatic and asymptomatic atheroscle-rotic carotid plaque features with in vivoMR imaging. Radiology 2006;240(2):464–472.

40. Yu X, Song SK, Chen J, et al. High-resolu-tion MRI characterization of human throm-bus using a novel fibrin-targeted paramag-netic nanoparticle contrast agent. Magn Re-son Med 2000;44:867–872.

41. Flacke S, Fischer S, Scott MJ, et al. NovelMRI contrast agent for molecular imaging offibrin: implications for detecting vulnerableplaques. Circulation 2001;104:1280–1285.

42. Botnar RM, Perez AS, Witte S, et al. In vivomolecular imaging of acute and subacutethrombosis using a fibrin-binding magneticresonance imaging contrast agent. Circula-tion 2004;109:2023–2029.

43. Botnar RM, Buecker A, Wiethoff AJ, et al.In vivo magnetic resonance imaging of cor-onary thrombosis using a fibrin-binding mo-lecular magnetic resonance contrast agent.Circulation 2004;110:1463–1466.

44. Spuentrup E, Buecker A, Katoh M, et al.Molecular magnetic resonance imaging ofcoronary thrombosis and pulmonary em-boli with a novel fibrin-targeted contrastagent. Circulation 2005;111:1377–1382.

45. Hansson GK. Inflammation, atherosclero-sis, and coronary artery disease. N EnglJ Med 2005;352:1685–1695.

46. Bassiouny HS, Sakaguchi Y, Mikucki SA, etal. Juxtalumenal location of plaque necrosisand neoformation in symptomatic carotidstenosis. J Vasc Surg 1997;26:585–594.

47. Kerwin W, Hooker A, Spilker M, et al.Quantitative magnetic resonance imaginganalysis of neovasculature volume in ca-

rotid atherosclerotic plaque. Circulation2003;107:851–856.

48. Kerwin WS, O’Brien KD, Ferguson MS,Polissar N, Hatsukami TS, Yuan C. Inflam-mation in carotid atherosclerotic plaque: adynamic contrast-enhanced MR imagingstudy. Radiology 2006;241(2):459–468.

49. Lin W, Abendschein DR, Haacke EM. Con-trast-enhanced magnetic resonance angiog-raphy of carotid arterial wall in pigs. JMagn Reson Imaging 1997;7:183–190.

50. Aoki S, Aoki K, Ohsawa S, Nakajima H,Kumagai H, Araki T. Dynamic MR imagingof the carotid wall. J Magn Reson Imaging1999;9:420–427.

51. McCarthy MJ, Loftus IM, Thompson MM,et al. Angiogenesis and the atheroscleroticcarotid plaque: an association betweensymptomatology and plaque morphology. JVasc Surg 1999;30:261–268.

52. Schmitz SA, Taupitz M, Wagner S, Wolf KJ,Beyersdorff D, Hamm B. Magnetic resonanceimaging of atherosclerotic plaques using su-perparamagnetic iron oxide particles. J MagnReson Imaging 2001;14:355–361.

53. Ruehm SG, Corot C, Vogt P, Kolb S, Deba-tin JF. Magnetic resonance imaging of ath-erosclerotic plaque with ultrasmall super-paramagnetic particles of iron oxide in hy-perlipidemic rabbits. Circulation 2001;103:415–422.

54. Kooi ME, Cappendijk VC, Cleutjens KB, etal. Accumulation of ultrasmall superpara-magnetic particles of iron oxide in humanatherosclerotic plaques can be detected byin vivo magnetic resonance imaging. Circu-lation 2003;107:2453–2458.

55. Winter PM, Morawski AM, Caruthers SD,et al. Molecular imaging of angiogenesis inearly-stage atherosclerosis with alpha(v)beta3-integrin-targeted nanoparticles. Cir-culation 2003;108:2270-2274.

56. Goldstein JA, Demetriou D, Grines CL,Pica M, Shoukfeh M, O’Neill WW. Multiplecomplex coronary plaques in patients withacute myocardial infarction. N Engl J Med2000;343:915–922.

57. Asakura M, Ueda Y, Yamaguchi O, et al.Extensive development of vulnerableplaques as a pan-coronary process in pa-tients with myocardial infarction: an angio-scopic study. J Am Coll Cardiol 2001;37:1284–1288.

58. Rapp JH, Saloner D. Current status of ca-rotid imaging by MRA. Cardiovasc Surg2003;11:445–447.

59. Huston J III, Fain SB, Wald JT, et al. Ca-rotid artery: elliptic centric contrast-en-hanced MR angiography compared with

conventional angiography. Radiology 2001;218:138–143.

60. Kim WY, Danias PG, Stuber M, et al. Cor-onary magnetic resonance angiography forthe detection of coronary stenoses. N EnglJ Med 2001;345:1863–1869.

61. Nederkoorn PJ, van der Graff Y, HuninkMG. Duplex ultrasound and magnetic reso-nance angiography compared with digitalsubtraction angiography in carotid arterystenosis: a systematic review. Stroke 2003;34:1324–1332.

62. Ropers D, Regenfus M, Wasmeier G,Achenbach S. Non-interventional cardiacdiagnostics: computed tomography, mag-netic resonance and real-time three-dimen-sional echocardiography—techniques andclinical applications. Minerva Cardioangiol2004;52:407–417.

63. Weber OM, Pujadas S, Martin AJ, HigginsCB. Free-breathing, three-dimensional cor-onary artery magnetic resonance angiog-raphy: comparison of sequences. J MagnReson Imaging 2004;20:395–402.

64. Deshpande VS, Shea SM, Laub G, Simon-etti OP, Finn JP, Li D. 3D magnetization-prepared true-FISP: a new technique forimaging coronary arteries. Magn ResonMed 2001;46:494–502.

65. Bornert P, Stuber M, Botnar RM, et al.Direct comparison of 3D spiral vs. Carte-sian gradient-echo coronary magnetic reso-nance angiography. Magn Reson Med 2001;46:789–794.

66. Katoh M, Stuber M, Buecker A, GuntherRW, Spuentrup E. Spin-labeling coronaryMR angiography with steady-state free pre-cession and radial k-space sampling: initialresults in healthy volunteers. Radiology2005;236(3):1047–1052.

67. Botnar RM, Stuber M, Danias PG, Kiss-inger KV, Manning WJ. Improved coronaryartery definition with T2-weighted, free-breathing, three-dimensional coronaryMRA. Circulation 1999;99:3139–3148.

68. Stuber M, Botnar RM, Danias PG, et al.Double-oblique free-breathing high resolu-tion three-dimensional coronary magneticresonance angiography. J Am Coll Cardiol1999;34:524–531.

69. Young GR, Humphrey PR, Nixon TE, SmithET. Variability in measurement of extracra-nial internal carotid artery stenosis as dis-played by both digital subtraction and mag-netic resonance angiography: an assessmentof three caliper techniques and visual impres-sion of stenosis. Stroke 1996;27:467–473.

70. North American Symptomatic Carotid End-arterectomy Trial. Methods, patient char-

REVIEW: MR Imaging of the Vulnerable Atherosclerotic Plaque Saam et al

76 Radiology: Volume 244: Number 1—July 2007

acteristics, and progress. Stroke 1991;22:711–720.

71. European Carotid Surgery Trialists’ Collab-orative Group. MRC European Carotid Sur-gery Trial: interim results for symptomaticpatients with severe (70–99%) or with mild(0–29%) carotid stenosis. Lancet 1991;337:1235–1243.

72. Ota H, Takase K, Rikimaru H, et al. Quan-titative vascular measurements in arterialocclusive disease. RadioGraphics 2005;25:1141–1158.

73. Morasch MD, Gurjala AN, Washington E,et al. Cross-sectional magnetic resonanceangiography is accurate in predicting de-gree of carotid stenosis. Ann Vasc Surg2002;16:266–272.

74. Dodds SR. The haemodynamics of asym-metric stenoses. Eur J Vasc Endovasc Surg2002;24:332–337.

75. Kim DY, Park JW. Computerized quantifi-cation of carotid artery stenosis using MRAaxial images. Magn Reson Imaging 2004;22:353–359.

76. van Bemmel CM, Viergever MA, NiessenWJ. Semiautomatic segmentation and ste-nosis quantification of 3D contrast-en-hanced MR angiograms of the internal ca-rotid artery. Magn Reson Med 2004;51:753–760.

77. Kodama K, Asakura M, Ueda Y, YamaguchiO, Hirayama A. The role of plaque rupturein the development of acute coronary syn-drome evaluated by the coronary angio-scope. Intern Med 2000;39:333–335.

78. Takano M, Mizuno K, Okamatsu K,Yokoyama S, Ohba T, Sakai S. Mechanicaland structural characteristics of vulnerableplaques: analysis by coronary angioscopyand intravascular ultrasound. J Am CollCardiol 2001;38:99–104.

79. Kolodgie FD, Gold HK, Burke AP, et al.Intraplaque hemorrhage and progression ofcoronary atheroma. N Engl J Med 2003;349:2316–2325.

80. Constantinides P. Plaque fissuring in hu-man coronary thrombosis. J AtherosclerRes 1966;6:1–17.

81. Davies MJ, Thomas AC. Plaque fissuring:the cause of acute myocardial infarction,sudden ischaemic death, and crescendo an-gina. Br Heart J 1985;53:363–373.

82. Paterson JC. Capillary rupture with intimalhemorrhage as a causative factor in coro-nary thrombosis. Arch Pathol 1938;25:474–487.

83. Chu B, Kampschulte A, Ferguson MS, et al.Hemorrhage in the atherosclerotic carotid

plaque: a high-resolution MRI study. Stroke2004;35:1079–1084.

84. Moody AR, Murphy RE, Morgan PS, et al.Characterization of complicated carotidplaque with magnetic resonance directthrombus imaging in patients with cerebralischemia. Circulation 2003;107:3047–3052.

85. Cappendijk VC, Cleutjens KB, HeenemanS, et al. In vivo detection of hemorrhage inhuman atherosclerotic plaques with mag-netic resonance imaging. J Magn Reson Im-aging 2004;20:105–110.

86. Virmani R, Kolodgie FD, Burke AP, et al.Atherosclerotic plaque progression and vul-nerability to rupture: angiogenesis as asource of intraplaque hemorrhage. Arterio-scler Thromb Vasc Biol 2005;25:2054–2061.

87. Takaya N, Yuan C, Chu B, et al. Presence ofintraplaque hemorrhage stimulates progres-sion of carotid atherosclerotic plaques: ahigh-resolution magnetic resonance imagingstudy. Circulation 2005;111:2768–2775.

88. Halcox JP, Schenke WH, Zalos G, et al.Prognostic value of coronary vascular endo-thelial dysfunction. Circulation 2002;106:653–658.

89. Targonski PV, Bonetti PO, Pumper GM,Higano ST, Holmes DR Jr, Lerman A. Cor-onary endothelial dysfunction is associatedwith an increased risk of cerebrovascularevents. Circulation 2003;107:2805–2809.

90. Oliver JJ, Webb DJ. Noninvasive assess-ment of arterial stiffness and risk of athero-sclerotic events. Arterioscler Thromb VascBiol 2003;23:554–566.

91. De Vriese AS, Verbeuren TJ, Van deVoorde J, Lameire NH, Vanhoutte PM. En-dothelial dysfunction in diabetes. Br J Phar-macol 2000;130:963–974.

92. Zeiher AM, Schachinger V, Minners J.Long-term cigarette smoking impairs endo-thelium-dependent coronary arterial vaso-dilator function. Circulation 1995;92:1094–1100.

93. Martin H, Hu J, Gennser G, Norman M.Impaired endothelial function and in-creased carotid stiffness in 9-year-old chil-dren with low birthweight. Circulation2000;102:2739–2744.

94. Tounian P, Aggoun Y, Dubern B, et al.Presence of increased stiffness of the com-mon carotid artery and endothelial dysfunc-tion in severely obese children: a prospec-tive study. Lancet 2001;358:1400–1404.

95. Wiesmann F, Petersen SE, Leeson PM, etal. Global impairment of brachial, carotid,and aortic vascular function in youngsmokers: direct quantification by high-reso-

lution magnetic resonance imaging. J AmColl Cardiol 2004;44:2056–2064.

96. Mohiaddin RH, Firmin DN, Longmore DB.Age-related changes of human aortic flowwave velocity measured noninvasively bymagnetic resonance imaging. J Appl Physiol1993;74:492–497.

97. Rogers WJ, Hu YL, Coast D, et al. Age-associated changes in regional aortic pulsewave velocity. J Am Coll Cardiol 2001;38:1123–1129.

98. Smits PC, Pasterkamp G, Quarles van Uf-ford MA, et al. Coronary artery disease:arterial remodelling and clinical presenta-tion. Heart 1999;82:461–464.

99. Varnava AM, Mills PG, Davies MJ. Rela-tionship between coronary artery remodel-ing and plaque vulnerability. Circulation2002;105:939–943.

100. Corti R, Fuster V, Fayad ZA, et al. Lipidlowering by simvastatin induces regressionof human atherosclerotic lesions: 2 years’follow-up by high-resolution noninvasivemagnetic resonance imaging. Circulation2002;106:2884–2887.

101. Yonemura A, Momiyama Y, Fayad ZA, etal. Effect of lipid-lowering therapy withatorvastatin on atherosclerotic aorticplaques detected by noninvasive magneticresonance imaging. J Am Coll Cardiol 2005;45:733–742.

102. Kim WY, Stuber M, Bornert P, KissingerKV, Manning WJ, Botnar RM. Three-di-mensional black-blood cardiac magneticresonance coronary vessel wall imaging de-tects positive arterial remodeling in pa-tients with nonsignificant coronary arterydisease. Circulation 2002;106:296–299.

103. Koktzoglou I, Simonetti O, Li D. Coronary ar-tery wall imaging: initial experience at 3 tesla. JMagn Reson Imaging 2005;21:128–132.

104. Cappendijk VC, Cleutjens KB, Kessels AG,et al. Assessment of human atheroscleroticcarotid plaque components with multi-sequence MR imaging: initial experience.Radiology 2005;234(2):487–492.

105. Lima JA, Desai MY, Steen H, Warren WP,Gautam S, Lai S. Statin-induced cholesterollowering and plaque regression after 6 monthsof magnetic resonance imaging-monitoredtherapy. Circulation 2004;110:2336–2341.

106. Yarnykh VL, Yuan C. T1-insensitive flow sup-pression using quadruple inversion-recovery.Magn Reson Med 2002;48:899–905.

107. Takaya N, Yuan C, Chu B, et al. Associationbetween carotid plaque characteristics andsubsequent ischemic cerebrovascular events:a prospective assessment with MRI—initialresults. Stroke 2006;37(3):818–823.

REVIEW: MR Imaging of the Vulnerable Atherosclerotic Plaque Saam et al

Radiology: Volume 244: Number 1—July 2007 77