7
European Journal of Radiology 83 (2014) e49–e55 Contents lists available at ScienceDirect European Journal of Radiology jo ur nal ho me page: www.elsevier.com/locate/ejrad Reproducibility of high-resolution MRI for the middle cerebral artery plaque at 3 T Wan-Qun Yang a , Biao Huang a,, Xin-Tong Liu b , Hong-Jun Liu a , Pei-Jun Li a , Wen-Zhen Zhu c,∗∗ a Department of Radiology, Guangdong Academy of Medical Sciences, Guangdong General Hospital, Guangzhou, Guangdong 510080, China b Department of Neurology, Guangdong Academy of Medical Sciences, Guangdong General Hospital, Guangzhou, Guangdong 510080, China c Department of Radiology, Tongji Hospital, Tongji Medical College, Huazhong University of Science & Technology, Wuhan, Hubei 430030, China a r t i c l e i n f o Article history: Received 7 August 2013 Accepted 8 October 2013 Keywords: High-resolution magnetic resonance imaging Atherosclerosis Plaque analysis Middle cerebral artery a b s t r a c t Purpose: To assess the reproducibility of HR-MRI for the identification of MCA atherosclerotic plaque components and quantification of stenosis. Materials and methods: Seventy-three consecutive subjects who initially had ischemic stroke or asymp- tomatic MCA stenosis (>50%) were enrolled in the study. All subjects were scanned using 3.0 T MRI. Two independent readers reviewed all images and one reader reevaluated all images four weeks later. The tissue components of plaques were analyzed qualitatively and the vessels were quantitative measured. Results: HR-MRI displayed the artery wall and lumen clearly. The intra-observer reproducibility was excellent for the identification of plaques (kappa [] = 0.96; 95% CI: 0.83–1.04) and contrast enhance- ment ( = 0.89; 0.78–0.95); it was substantial for intra-plaque hemorrhage ( = 0.79; 0.57–0.96) and the fibrous cap ( = 0.65; 0.42–0.86). The inter-observer reproducibility was excellent for plaques ( = 0.92; 0.73–1.06), substantial for contrast enhancement ( = 0.80; 0.65–0.93), intra-plaque hemor- rhage ( = 0.68; 0.47–0.92) and moderate for the fibrous cap ( = 0.58; 0.44–0.79). Both intra-observer and inter-observer reproducibility were excellent for quantitative vessel, lumen and wall measurements with intraclass correlation coefficients ranging from 0.91 to 0.97 and 0.87 to 0.96, respectively. However, vessel and wall areas and the intervals defined by the Bland–Altman plots were wide in comparison to the mean. Conclusions: The identification of MCA atherosclerotic plaque components and the quantification of ves- sel and lumen measurements are reproducible. The reproducibility is overall acceptable. HR-MRI may provide a useful tool for clinical risk evaluation in MCA atherosclerosis. © 2013 Elsevier Ireland Ltd. All rights reserved. 1. Introduction Intracranial large-artery atherosclerotic stroke, especially in the middle cerebral artery (MCA), is a common cause of ischemic stroke in African, Asian, and Hispanic populations. In Chinese populations, estimates of the responsibility of intracranial stenosis in causing stroke are about 33–50% [1]. Digital subtraction angiography (DSA), CT angiography (CTA) and MR angiography (MRA) are the most commonly used tools to detect intracranial artery atherosclerotic Corresponding author at: Department of Radiology, Guangdong Academy of Medical Sciences, Guangdong General Hospital, #106 Zhongshan 2nd Road, Guangzhou, Guangdong 510080, China. Tel.: +86 20 83827812 50352; fax: +86 20 83870125; mobile: +86 013822157591. ∗∗ Corresponding author. Tel.: +86 27 83663258; mobile: +86 013886018612. E-mail addresses: [email protected] (B. Huang), [email protected] (W.-Z. Zhu). stenosis. However, these imaging modalities only show luminal narrowing and fail to show vascular wall characterization. Reports have shown that the risk of ischemic stroke is predicted more by plaque components than by the degree of luminal narrowing [2,3]. In recent years, many high-resolution-MRI (HR-MRI) studies have been used to depict the MCA wall and lumen, identify arterial atherosclerotic plaques, dissection and vasculitis [4], explore the plaque distribution characteristics and the remodeling [5,6], and help with understanding atherosclerosis pathophysiology [7–9]. In measuring atherosclerosis burden and plaque instability, HR- MRI can be a potential guide for identification of high-risk patients requiring more intensive treatments; it can also be used to monitor the effect of treatments [10]. These previous studies usually focused on case report and small samples. Reproducibility of HR-MRI for the identification of different MCA plaque components and lumen measurements has rarely been addressed. Furthermore, compara- tive studies between MCA imaging and pathology findings are very 0720-048X/$ see front matter © 2013 Elsevier Ireland Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ejrad.2013.10.003

Reproducibility of high-resolution MRI for the middle cerebral artery plaque at 3T

Embed Size (px)

Citation preview

Page 1: Reproducibility of high-resolution MRI for the middle cerebral artery plaque at 3T

Ra

WPa

b

c

a

ARA

KHiAPM

1

miesCc

oGf

z

0h

European Journal of Radiology 83 (2014) e49– e55

Contents lists available at ScienceDirect

European Journal of Radiology

jo ur nal ho me page: www.elsev ier .com/ locate /e j rad

eproducibility of high-resolution MRI for the middle cerebralrtery plaque at 3 T

an-Qun Yanga, Biao Huanga,∗, Xin-Tong Liub, Hong-Jun Liua,ei-Jun Lia, Wen-Zhen Zhuc,∗∗

Department of Radiology, Guangdong Academy of Medical Sciences, Guangdong General Hospital, Guangzhou, Guangdong 510080, ChinaDepartment of Neurology, Guangdong Academy of Medical Sciences, Guangdong General Hospital, Guangzhou, Guangdong 510080, ChinaDepartment of Radiology, Tongji Hospital, Tongji Medical College, Huazhong University of Science & Technology, Wuhan, Hubei 430030, China

r t i c l e i n f o

rticle history:eceived 7 August 2013ccepted 8 October 2013

eywords:igh-resolution magnetic resonance

magingtherosclerosislaque analysisiddle cerebral artery

a b s t r a c t

Purpose: To assess the reproducibility of HR-MRI for the identification of MCA atherosclerotic plaquecomponents and quantification of stenosis.Materials and methods: Seventy-three consecutive subjects who initially had ischemic stroke or asymp-tomatic MCA stenosis (>50%) were enrolled in the study. All subjects were scanned using 3.0 T MRI. Twoindependent readers reviewed all images and one reader reevaluated all images four weeks later. Thetissue components of plaques were analyzed qualitatively and the vessels were quantitative measured.Results: HR-MRI displayed the artery wall and lumen clearly. The intra-observer reproducibility wasexcellent for the identification of plaques (kappa [�] = 0.96; 95% CI: 0.83–1.04) and contrast enhance-ment (� = 0.89; 0.78–0.95); it was substantial for intra-plaque hemorrhage (� = 0.79; 0.57–0.96) andthe fibrous cap (� = 0.65; 0.42–0.86). The inter-observer reproducibility was excellent for plaques(� = 0.92; 0.73–1.06), substantial for contrast enhancement (� = 0.80; 0.65–0.93), intra-plaque hemor-rhage (� = 0.68; 0.47–0.92) and moderate for the fibrous cap (� = 0.58; 0.44–0.79). Both intra-observerand inter-observer reproducibility were excellent for quantitative vessel, lumen and wall measurements

with intraclass correlation coefficients ranging from 0.91 to 0.97 and 0.87 to 0.96, respectively. However,vessel and wall areas and the intervals defined by the Bland–Altman plots were wide in comparison tothe mean.Conclusions: The identification of MCA atherosclerotic plaque components and the quantification of ves-sel and lumen measurements are reproducible. The reproducibility is overall acceptable. HR-MRI mayprovide a useful tool for clinical risk evaluation in MCA atherosclerosis.

© 2013 Elsevier Ireland Ltd. All rights reserved.

. Introduction

Intracranial large-artery atherosclerotic stroke, especially in theiddle cerebral artery (MCA), is a common cause of ischemic stroke

n African, Asian, and Hispanic populations. In Chinese populations,stimates of the responsibility of intracranial stenosis in causing

troke are about 33–50% [1]. Digital subtraction angiography (DSA),T angiography (CTA) and MR angiography (MRA) are the mostommonly used tools to detect intracranial artery atherosclerotic

∗ Corresponding author at: Department of Radiology, Guangdong Academyf Medical Sciences, Guangdong General Hospital, #106 Zhongshan 2nd Road,uangzhou, Guangdong 510080, China. Tel.: +86 20 83827812 50352;

ax: +86 20 83870125; mobile: +86 013822157591.∗∗ Corresponding author. Tel.: +86 27 83663258; mobile: +86 013886018612.

E-mail addresses: [email protected] (B. Huang),[email protected] (W.-Z. Zhu).

720-048X/$ – see front matter © 2013 Elsevier Ireland Ltd. All rights reserved.ttp://dx.doi.org/10.1016/j.ejrad.2013.10.003

stenosis. However, these imaging modalities only show luminalnarrowing and fail to show vascular wall characterization. Reportshave shown that the risk of ischemic stroke is predicted moreby plaque components than by the degree of luminal narrowing[2,3]. In recent years, many high-resolution-MRI (HR-MRI) studieshave been used to depict the MCA wall and lumen, identify arterialatherosclerotic plaques, dissection and vasculitis [4], explore theplaque distribution characteristics and the remodeling [5,6], andhelp with understanding atherosclerosis pathophysiology [7–9].In measuring atherosclerosis burden and plaque instability, HR-MRI can be a potential guide for identification of high-risk patientsrequiring more intensive treatments; it can also be used to monitorthe effect of treatments [10]. These previous studies usually focused

on case report and small samples. Reproducibility of HR-MRI forthe identification of different MCA plaque components and lumenmeasurements has rarely been addressed. Furthermore, compara-tive studies between MCA imaging and pathology findings are very
Page 2: Reproducibility of high-resolution MRI for the middle cerebral artery plaque at 3T

e al of R

dtt

itsiaharf

2

2

pmtitadPsMtosod(as

2

HgcatfrpTmsd2hfba2pcdGr

50 W.-Q. Yang et al. / European Journ

ifficult to conduct. Hence, it has practical significance to evaluatehe reproducibility of in vivo MRI for the recognition and quantita-ive measurements of MCA atherosclerotic plaque.

The purpose of this study was to assess the intra-andnter-observer reproducibility of high-resolution MRI for the iden-ification and quantification of MCA atherosclerotic plaque andtenosis and to compare the difference of plaque character-zation and vessel wall properties between symptomatic andsymptomatic atherosclerotic MCA using this technique. Weypothesized that the identification of MCA plaque componentsnd the quantification of vessel using HR-MRI are reproducible. Theeproducibility is overall acceptable during clinical practice and forurther research.

. Clinical materials and methods

.1. Subjects

High-resolution MRI in atherosclerotic stenosis of the MCA wasartial of an ongoing multicenter prospective study analyzing theultimoding diagnosis of acute ischemic stroke and assessing

he prognostic value. Patients with the following features werencluded: (1) MCA stenosis ≥50% at the M1 segment on conven-ional angiography; (2) without any other major cause of stroke;nd (3) ≥2 risk factors (hypertension, diabetes mellitus, hyperlipi-emia, cigarette smoking, and obesity) for atherosclerotic disease.atients were considered symptomatic if they had an ischemictroke or transient ischemic attack in the territory of the stenoticCA within the previous 30 days. Patients were considered asymp-

omatic if they did not have a history of cerebrovascular eventsr if an ischemic event occurred in a vascular distribution out-ide the stenotic MCA. The exclusion criteria were: (1) evidencef cardioembolism; (2) complete MCA occlusion; (3) arteritis; (4)issection; (5) after intra-arterial or intravenous thrombolysis; and6) poor image quality secondary to motion artifacts. The study waspproved by the local Medical Ethics Committee and all patientsigned informed consent forms.

.2. Magnetic resonance imaging protocol

All subjects were imaged with a 3.0 T MRI Scanner (Signa ExciteD, GE Healthcare, Milwaukee, WI, United States), with a peakradient strength of 50 mT/m, using a standard 8-channel headoil. A standardized stroke MR protocol was performed including

routine MRI (T1WI, T2WI, and DWI) and three-dimensional (3D)ime-of-flight (TOF) MRA. The lesion site on the MCA M1 segmentor HR-MRI scanning was determined by a neuroradiologist, aftereviewing both the MRA and traditional MR images. The HR-MRIrotocol included three sequences. First, pre- and post-contrast1WI image scanning used a double inversion recovery measure-ent of the black blood two-dimensional (2D) fast spin echo

equence parameters: TR equal to two heart beats (1200–2000 ms,epending on heart rate), TI/TE 3000/50 ms, echo train length (ETL)2, band width 14.7, and number of average (NEX) 4. Next, theigh resolution T2WI image scanning was acquired also using

ast-spin echo sequence parameters: TR/TE 3000/50 ms, ETL 50,and width 62.5, and NEX 8. Both acquisitions were run with

field of view of 13 cm, matrix size of 256, slice thickness of

mm, and spacing of 0.2 mm. The scan direction was parallel orerpendicular to the M1 segment. Three minutes later, for the post-ontrast T1-weighted imaging, the contrast agent (gadopentetateimeglumine, Gd-DTPA, Magnevist; Bayer Schering Pharma, Berlin,ermany) was injected at an intravenous dose of 0.1 mmol/kg (flow

ate: 2 mL/s).

adiology 83 (2014) e49– e55

2.3. Image review

All images were reviewed by two experienced neuroradiolo-gists independently blinded to the patient’s clinical findings andhistory on a digital picture archiving and communication (PACS)workstation. The quality of each image from the four sequences(3D-TOF, T2WI and pre-and post-contrast T1WI) for every sub-ject was graded on a three-point scale according to the methoddescribed by Ryu et al. [9] based on the conspicuity of the ves-sel margins, lumen and the wall architecture. The grades were: (1)grade 1 was that outer boundary of artery and lumen could not beidentifiable; (2) grade 2 was that outer boundary and/or lumen waspartially obscured; and grade 3 was that wall architecture depictedin detail and outer boundary and lumen could be clearly identified[9]. Subjects in which image quality was grade 1 were excludedfrom the study. The MR images of the qualifying MCA artery wereassessed using a unified form and published criteria [2]. Plaqueswere defined if there were markedly eccentric or focal wall thick-enings, at which the thickest point of the wall was estimated to be>200% the thinnest part on visual inspection of T2-weighted images[6,7]. Wall thickening was identified if the wall was abnormal oninspection but did not reach the criteria that defined plaque.

2.4. Tissue components (qualitative assessment)

The variety of tissue components was identified based onprevious postmortem MCA, carotid artery, and coronary arterypathological control studies [2,11]. Both recent and fresh intra-plaque hemorrhages were identified as hyperintense areas on T1WIand 3D-TOF images (>150% of the signals of adjacent gray mat-ter). The fibrous cap appeared as a bright band adjacent to thelumen in MCA atherosclerosis on T2W images. Both intra-plaquehemorrhages and fibrous cap were considered present if they wereobserved on at least one imaging slice. All plaque signal intensitieswere compared with the adjacent brain parenchymal gray matteron each sequence. The degree of plaque enhancement was eval-uated qualitatively via comparison of the pre- and post-contrastcross-sectional T1W images.

2.5. Quantitative analysis

The quantitative measurements were performed at the nar-rowest lumen location on cross-sectional HR-MRI T2W images,including vessel area, lumen area, reference vessel area, and thewall area. In order to calculate the cross-sectional area of the ves-sel and lumen, two regions of interest of the outer vessel wallboundary and the lumen were manually drawn in the same slice(Fig. 1). The outer vessel wall boundary was traced along the inter-face between the vessel wall and meninges (or cerebrospinal fluid).The wall area was calculated by subtracting the lumen area fromthe outer-wall boundary area. The degree of MCA stenosis on HR-MRI was stated as the following: degree of stenosis = (1 − narrowlumen area/reference lumen area) × 100% [12]. The reference MCAsegment site was selected, with the nonoccluded lumen preferablyclose to the stenotic segment. If a proximal reference site was alsoabnormal, the neighboring distal site could be used instead.

All the qualitative assessment and quantitative measurementdata from both readers were used to calculate the inter-observerreproducibility. To assess intra-observer reproducibility, reader 1reevaluated all images which were presented in a different order

four weeks after the first reading session. For further comparisonof the plaque characterization and vessel wall properties betweenthe symptom group and asymptom group, the measurement wasaveraged and the tissue component differences between the firstset of reader 1 and reader 2 were solved by consensus.
Page 3: Reproducibility of high-resolution MRI for the middle cerebral artery plaque at 3T

W.-Q. Yang et al. / European Journal of Radiology 83 (2014) e49– e55 e51

Fig. 1. Visualization of intra-plaque hemorrhage and quantitative measurement of the lumen for symptomatic middle cerebral artery stenosis using HR-MRI. A 56-year-oldmale patient with infarction in right basal ganglia. (A) Sagittal reconstruction of 3D-TOF source image, (B) sagittal T1W image, (C) sagittal T2W image and (D) axial T1Wi measb steno

2

c((uc9iamacmtadFstc

3

(ohts

f

mage show a plaque with obvious hyperintensity signal (arrow). (E) Quantitativeoundary and the lumen are manually drawn in the same slice. (F) MRA shows the

.6. Statistical analysis

Statistical analysis was performed with SPSS 13.0 statisti-al software. Agreements were assessed by calculating kappa�) for dichotomous data and intraclass correlation coefficientsICC) with a one-way random effect for intra-observer contin-ous variables and a two-way random effect for inter-observerontinuous variables. All agreement parameters had calculated5% confidence intervals (CIs). Values were graded accord-

ng to the method proposed by Landis and Koch (<0.0 = poorgreement; 0.0–0.2 = slight agreement; 0.21–0.40 = fair agree-ent; 0.41–0.60 = moderate agreement; 0.61–0.80 = substantial

greement; 0.81–1.0 = almost perfect) [13]. Intraclass correlationoefficient values that were ≥0.75 were considered excellent agree-ent [14]. Besides, we analyzed the level of agreement by plotting

he differences between the two area measurements against theverages of the two area measurements according to the methodescribed by Bland and Altman (i.e. Bland–Altman plots) [15].inally, qualitative data comparison was conducted with Chi-quare test. A t-test was used to quantitative data comparison. Awo-sided ̨ level of 0.05 was used to infer the statistically signifi-ant difference.

. Results

Between March 2010 and February 2013, seventy-three subjects46 males and 27 females) were enrolled in the study. Median agef the subjects was 64 years (range: 30–84 years). Among them 38ad a stroke, 3 had a transient ischemic attack and 32 were asymp-

omatic. The median time from stroke onset to MRI examination inymptomatic patients was 10 days (range, 4–17).

A total of 65 plaques were found on HR-MRI. Plaques wereound in each of the 41 symptomatic cases (41/41) and 24 of

urement of the vessel and lumen. Two regions of interest of the outer vessel wallsis in the right MCA (arrow head).

the asymptomatic cases (24/32). These were displayed as eitherfocal or eccentric wall thickening with homogenous (28 cases)or heterogeneous (37 cases) intensity. Within the 65 plaques, 11hemorrhage cases (Fig. 1) and 55 fibrous cap cases (Fig. 2) wereobserved. The intra-observer reproducibility was substantial for thefibrous cap (� = 0.65) and for intra-plaque hemorrhage (� = 0.79);inter-observer reproducibility was moderate for the fibrous cap(� = 0.58), and substantial for intra-plaque hemorrhage (� = 0.68)(Table 1).

Seven (10.8%) patients did not receive T1 contrast enhancementimages after performing plain scan because of kidney failure (n = 3)or too bad condition to continue the examination (n = 4). Finally,there were 58/65 patients who received T1 contrast enhance-ment images; 40 were from symptomatic cases and 18 werefrom asymptomatic cases. Focal, irregular, and eccentric vesselwall enhancement was present in 43 of the 58 patients withatherosclerotic disease (Fig. 2); 39 patients had enhancementonly in the vessel supplying the acute ischemic injury area, andfour patients did not have evidence of ischemic injury in theterritory supplied by the vessels. Intra-observer agreement wasalmost perfect (� = 0.89) and inter-observer agreement was sub-stantial (� = 0.80) for the identification of contrast enhancement(Table 1).

The frequency of intra-plaque hemorrhage and contrastenhancement in symptomatic group was significantly higher thanthat in the asymptomatic group. However, the difference in thefrequency of fibrous cap between two groups did not reach sta-tistical significance (Table 2). Intra-observer and inter-observerreproducibility for quantitative area measurements were excellent,

with the ICC ranging from 0.91 to 0.97 and 0.87 to 0.96, respectively(Table 1). Bland–Altman plots showed small absolute differencesin intra-observer (Fig. 3) and inter-observer (Fig. 4) measure-ments. However, for the vessel and wall area measurements,
Page 4: Reproducibility of high-resolution MRI for the middle cerebral artery plaque at 3T

e52 W.-Q. Yang et al. / European Journal of Radiology 83 (2014) e49– e55

Fig. 2. Eccentric narrowing and enhancement of symptomatic middle cerebral artery stenosis. A 78-year-old female patient with infarction in right corona radiata and cortex.( e infarn hancea

tttao

TI

TM

a

A) MRA shows severe narrowing of the right M1 segment (arrow head). (B) Multiplon-enhanced T1WI HR-MR image shows wall abnormality (arrow). (E) Contrast-enn active atherosclerotic plaque (arrows). (F) Normal wall at a reference site.

he Bland–Altman plots showed a wide interval compared with

he mean. The lumen area, wall area and degree of stenosis inhe symptomatic group were significantly higher than that in thesymptomatic group. However, there was no significant differencef vessel area between two groups (Table 2).

able 1ntra-observer and inter-observer reproducibility.

Plaque Intra-observer

Components Agreement � (95% CI)

Hemorrhage 97% 0.79 (0.57–0.Fibrous cap 84% 0.65 (0.42–0.Enhancement 92% 0.89 (0.78–0.

Quantitative measurements Intr

Vessel area 0.95 (0.92–0.97)

Lumen area 0.97 (0.95–0.98)

Wall area 0.96 (0.35–0.98)

Reference vessel area 0.91 (0.85–0.94)

able 2orphology analysis of MCA atherosclerotic plaque components and stenosis.

Symptomatic MCA stenosis (n = 41)

Plaque 41

Hemorrhage 11

Fibrous cap 37

Enhancement 39(39/40)a

Vessel area 8.64 ± 2.59 mm2

Lumen area 2.86 ± 2.18 mm2

Wall area 5.78 ± 3.12 mm2

Reference vessel area 6.59 ± 2.25 mm2

Degree of stenosis 55.86 ± 27.40%

40 patients received T1 contrast enhancement images in symptomatic group, while 18

cts are observed in the distribution of stenotic right MCA on DWI. (C) T2WI and (D)d T1W image in cross-section shows eccentric right MCA enhancement, indicating

4. Discussion

Our study demonstrates the capability of in vivo HR-MRI todelineate MCA wall structure. In our study, inter- and intra-observer agreements were excellent for the identification of

Inter-observer

Agreement � (95% CI)

96) 94% 0.68 (0.47–0.92)86) 77% 0.58 (0.44–0.79)95) 87% 0.80 (0.65–0.93)

aclass coefficient correlation (95% CI)

0.87 (0.78–0.93)0.96 (0.94–0.98)0.91 (0.85–0.94)0.90 (0.84–0.94)

Asymptomatic MCA stenosis (n = 32) P

24 0.0010 0.00218 0.1544(4/18)a <0.001

7.88 ± 2.07 mm2 0.2064.67 ± 2.42 mm2 0.0023.21 ± 2.93 mm2 0.0017.43 ± 2.47 mm2 0.15938.52 ± 21.21% 0.005

patients in asymptomatic group.

Page 5: Reproducibility of high-resolution MRI for the middle cerebral artery plaque at 3T

W.-Q. Yang et al. / European Journal of Radiology 83 (2014) e49– e55 e53

F en ar

pAoiferaoc

rt[MFoaMsoaalvi

ig. 3. Bland–Altman plots of intra-observer reproducibility for vessel area (A), lum

laques and substantial to excellent for contrast enhancement.s expected, intra-observer reproducibility was better than inter-bserver reproducibility. The results were similar to those observedn previous studies. Li et al. analyzed 49 MCAs of 48 patients andound both intra-observer and inter-observer reproducibility wasxcellent for plaques and moderate for wall thickening [7]. Theseesults indicated the ability of HR-MRI to show the definite wallbnormality and reliable results [16]. However, the inter- and intra-bserver agreements were moderate to substantial for the fibrousap and intra-plaque hemorrhages.

In our study, the reproducibility of MCA intra-plaque hemor-hage was substantial for intra- and inter-observer agreements;his was not as good of an outcome as in carotid artery studies17,18]. We inferred some possible reasons: (1) The incidence of

CA intra-plaque hemorrhage was lower than the carotid artery.or example, Xu et al. found that the occurrence rate of high signalsn T1-weighted fat-suppressed images (HST1) was 10.1% (11/107)nd the occurrence rate between symptomatic and asymptomaticCAs was significantly different within high-grade MCA steno-

is [19]; (2) The complicated T1WI hemorrhage signal intensityften generated an intensity overlap with the lipid core; (3) Thetherosclerotic MCA had much less wall area (mean, 5.8 mm2) than

n atherosclerotic carotid artery (mean, 93.4 mm2) [20]. Withinimited wall space, it was difficult to identify a relatively lowerolume of intra-plaque hemorrhage. Even so, the identification ofntra-plaque hemorrhage in intracranial atherosclerotic arteries is

ea (B), wall area (C), and reference vessel area (D) at the level of the M1 segment.

still reliable and it can be used for risk stratification of ischemicevents.

In a recent study addressing whether MRI-defined plaque vul-nerability predicted cerebrovascular events, a protective fibrouscap in MCA atherosclerosis was documented as a bright bandappearing adjacent to the dark lumen on T2-weighted images [21].In the present study, a fibrous cap was observed for most cases(55/65) within plaque and showed moderate to substantial repro-ducibility, which was higher than the reproducibility seen with thecarotid artery. One possible reason was the lower incidence of MCAintra-plaque hemorrhage and calcification. The difficulties in dis-tinguishing a fibrous cap from the lipid component in plaques withhemorrhages and calcifications have been previously documented.Disruption caused by the overlap between the signal intensity ofhemorrhages and other components as well as calcification adja-cent to the lumen could explain these difficulties. We were goingto further identify the fibrous cap status, but decided against itbecause of the difficulty in defining the “thick,” “thin”, and “rup-ture” parameters for the smaller-sized fibrous cap and the limitedMCA spatial resolution. Our results highlighted that much remainsto be done to improve the ability of HR-MRI to confidently identifyfibrous cap status.

It has been reported that gadolinium contrast agent use canhelp differentiate between intracranial atherosclerotic plaques,inflammation, and other wall pathologies [4]. Gadolinium con-trast has also showed great value in detecting vulnerable plaque

Page 6: Reproducibility of high-resolution MRI for the middle cerebral artery plaque at 3T

e54 W.-Q. Yang et al. / European Journal of Radiology 83 (2014) e49– e55

F en ar

tMaadImcd

tMmodotqptdbtusi

ig. 4. Bland–Altman plots of inter-observer reproducibility for vessel area (A), lum

hrough enhancement of carotid arteries, basilar arteries, andCAs [22–25]. In this study, HR-MRI has displayed focal, irregular

nd eccentric vessel wall enhancement both in symptomatic andsymptomatic groups. We found substantial to excellent repro-ucibility relating to the identification of contrast enhancement.

n the absence of pathological specimens, good reproducibility isore significant in clinical practice, because contrast agent use

an improve the signal-to-noise and contrast-to-noise ratios foretection and characterization of atherosclerotic plaques.

In our study, the reproducibility was excellent for the quan-ification of vessel and lumen areas, in keeping with previous

CA studies. Although ICC suggested an almost perfect agree-ent, the Bland–Altman plot showed a relatively wide interval

f agreement compared with the mean. This suggested possibleiscrepancies between two observers. However, both the intra-bserver and inter-observer variability was still acceptable. Inhis study, high-resolution cross-sectional T2WI was selected foruantitative analysis because the contrast between the lumen andlaque, outer wall boundary, and CSF was higher than that seen inhe other three sequences. Most importantly, T2WI was useful foristinguishing the vessel lumen in the presence of calcified plaqueecause CSF is hyperintense and calcium is hypointense. In addi-

ion to delineating the vessel wall on T2WI sequences, our studysed flow compensation in slice direction and the inferior spatialaturation band to minimize pulsation artifacts and to suppress thenflow blood signal intensity.

ea (B), wall area (C), and reference vessel area (D) at the level of the M1 segment.

One limitation of our study was the lack of in vivo MCA wallhistology for comparison. The wall structure analysis was hypoth-esized, but not confirmed by histology. Consequently, no pathologicconfirmation data could be used in our study; we may have foundgood agreement despite both readers misinterpreting images.However, we carefully used the criteria which were previously pub-lished. In addition, we did not use a tool to automatically analyzeimages for quantitative measurements. However, such tools havenot been thoroughly validated for detecting and the measuring allplaque components. Finally, there are several well-described limi-tations to the use of statistics which could influence interpretationof our results. Very high (or low) prevalence results in high levels ofexpected agreement; consequently, the value of statistics is oftenlow despite nearly perfect agreement as we observed for hemor-rhage on T1W images. Observer bias, in the systematic differencebetween two observers in interpreting an image, is a disagreementwith important practical implications, but it is not separately iden-tified by statistics.

HR-MRI can display the artery wall and lumen at the same time,providing more information about MCA atherosclerotic diseasecompared with 3D TOF MRA. Taken together, our results indicatethat MRI can be used quite confidently in clinical studies strat-

ifying the risk of stroke and in studies evaluating the effect ofdrugs on atherosclerotic plaque size. Further prospective studiesare required to longitudinally monitor the progression or regres-sion of atherosclerotic plaque using HR-MRI.
Page 7: Reproducibility of high-resolution MRI for the middle cerebral artery plaque at 3T

al of R

5

acfpHM

C

A

F2uo

R

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

W.-Q. Yang et al. / European Journ

. Conclusions

Feasibility and reproducibility of HR-MRI for identifying MCAtherosclerotic plaque components is generally acceptable. Plaqueharacterization and vessel wall properties on HR-MRI were dif-erent between symptomatic and asymptomatic MCA stenosis. Theresence of vulnerable plaque is closely related to stroke. Hence,R-MRI may provide a useful tool for clinical risk evaluation inCA atherosclerosis.

onflict of interest

There is no conflict of interest statement.

cknowledgements

This work was partially supported by Chinese National 12thive-year Science and Technology Supporting Program under grant011BAI08B10, National Natural Science Foundation of Chinander grant 81271560 and Medical Scientific Research Foundationf Guangdong Province under grant A2012011.

eferences

[1] Wong LK. Global burden of intracranial atherosclerosis. Int J Stroke: Offi J IntStroke Soc 2006;1(3):158–9.

[2] Degnan AJ, Gallagher G, Teng Z, et al. MR angiography and imaging for theevaluation of middle cerebral artery atherosclerotic disease. Am J Neuroradiol2012;33(8):1427–35.

[3] Mazighi M, Labreuche J, Gongora-Rivera F, et al. Autopsy prevalence ofintracranial atherosclerosis in patients with fatal stroke. Stroke 2008;39(4):1142–7.

[4] Swartz RH, Bhuta SS, Farb RI, et al. Intracranial arterial wall imagingusing high-resolution 3-tesla contrast-enhanced MRI. Neurology 2009;72(7):627–34.

[5] Huang B, Yang WQ, Liu XT, et al. Basilar artery atherosclerotic plaques distri-

bution in symptomatic patients: a 3.0 T high-resolution MRI study. Eur J Radiol2013;82(4):e199–203.

[6] Xu WH, Li ML, Gao S, et al. In vivo high-resolution MR imaging of symptomaticand asymptomatic middle cerebral artery atherosclerotic stenosis. Atheroscle-rosis 2010;212(2):507–11.

[

adiology 83 (2014) e49– e55 e55

[7] Li ML, Xu WH, Song L, et al. Atherosclerosis of middle cerebral artery: evaluationwith high-resolution MR imaging at 3 T. Atherosclerosis 2009;204(2):447–52.

[8] Chung GH, Kwak HS, Hwang SB, et al. High resolution MR imaging inpatients with symptomatic middle cerebral artery stenosis. Eur J Radiol2012;81(12):4069–74.

[9] Ryu CW, Jahng GH, Kim EJ, et al. High resolution wall and lumen MRI of themiddle cerebral arteries at 3 tesla. Cerebrovasc Dis 2009;27(5):433–42.

10] Shi M, Wang S, Zhou H, et al. Wingspan stenting of symptomatic middle cerebralartery stenosis and perioperative evaluation using high-resolution 3 Tesla MRI.J Clin Neurosci 2012;19(6):912–4.

11] Chen XY, Wong KS, Lam WW, et al. Middle cerebral artery atherosclerosis:histological comparison between plaques associated with and not associatedwith infarct in a postmortem study. Cerebrovasc Dis 2008;25(1–2):74–80.

12] Samuels OB, Joseph GJ, Lynn MJ, et al. A standardized method for measuringintracranial arterial stenosis. Am J Neuroradiol 2000;21(4):643–6.

13] Landis JR, Koch GG. The measurement of observer agreement for categoricaldata. Biometrics 1977;33(1):159–74.

14] Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater reliability.Psychol Bull 1979;86(2):420–8.

15] Bland JM, Altman DG. Statistical methods for assessing agreement between twomethods of clinical measurement. Lancet 1986;1(8476):307–10.

16] Klein IF, Lavallee PC, Touboul PJ, et al. In vivo middle cerebral artery plaqueimaging by high-resolution MRI. Neurology 2006;67(2):327–9.

17] Turan TN, Bonilha L, Morgan PS, et al. Intraplaque hemorrhage in symptomaticintracranial atherosclerotic disease. J Neuroimaging 2009;21(2):159–61.

18] Touze E, Toussaint JF, Coste J, et al. Reproducibility of high-resolution MRIfor the identification and the quantification of carotid atherosclerotic plaquecomponents: consequences for prognosis studies and therapeutic trials. Stroke2007;38(6):1812–9.

19] Xu WH, Li ML, Gao S, et al. Middle cerebral artery intraplaque hemorrhage:prevalence and clinical relevance. Ann Neurol 2012;71(2):195–8.

20] Yuan C, Beach KW, Smith Jr LH, et al. Measurement of atherosclerotic carotidplaque size in vivo using high resolution magnetic resonance imaging. Circu-lation 1998;98(24):2666–71.

21] Niizuma K, Shimizu H, Takada S, et al. Middle cerebral artery plaque imagingusing 3-Tesla high-resolution MRI. J Clin Neurosci 2008;15(10):1137–41.

22] Ma N, Lou X, Zhao TQ, et al. Intraobserver and interobserver variability for mea-suring the wall area of the basilar artery at the level of the trigeminal ganglionon high-resolution MR images. Am J Neuroradiol 2010;32(2):29–32.

23] Skarpathiotakis M, Mandell DM, Swartz RH, et al. Intracranial atheroscle-rotic plaque enhancement in patients with ischemic stroke. Am J Neuroradiol2013;34(2):299–304.

24] Vergouwen MD, Silver FL, Mandell DM, et al. Eccentric narrowing and enhance-

ment of symptomatic middle cerebral artery stenoses in patients with recentischemic stroke. Arch Neurol 2011;68(3):338–42.

25] Kawahara I, Morikawa M, Honda M, et al. High-resolution magnetic resonanceimaging using gadolinium-based contrast agent for atherosclerotic carotidplaque. Surg Neurol 2007;68(1):60–5.