Added Value of Vessel Wall Magnetic Resonance Imaging in the Differentiation of Moyamoya Vasculopathies in a Non-Asian Cohort
Background and Purpose—Although studies have evaluated the differential imaging of moyamoya disease and atherosclerosis, none have investigated the added value of vessel wall magnetic resonance imaging (MRI). This study evaluates the added diagnostic value of vessel wall MRI in differentiating moyamoya disease, atherosclerotic-moyamoya syndrome (A-MMS), and vasculitic-MMS (V-MMS) with a multicontrast protocol.
Methods—We retrospectively reviewed the carotid artery territories of patients with clinically defined vasculopathies (moyamoya disease, atherosclerosis, and vasculitis) and steno-occlusive intracranial carotid disease. Two neuroradiologists, blinded to clinical data reviewed the luminal imaging of each carotid, evaluating collateral extent and making a presumed diagnosis with diagnostic confidence. After 3 weeks, the 2 readers reviewed the luminal imaging+vessel wall MRI for the presence, pattern and intensity of postcontrast enhancement, T2 signal characteristics, pattern of involvement, and presumed diagnosis and confidence.
Results—Ten A-MMS, 3 V-MMS, and 8 moyamoya disease cases with 38 affected carotid segments were included. There was significant improvement in diagnostic accuracy with luminal imaging+vessel wall MRI when compared with luminal imaging (87% versus 32%, P<0.001). The most common vessel wall MRI findings for moyamoya disease were nonenhancing, nonremodeling lesions without T2 heterogeneity; for A-MMS eccentric, remodeling, and T2 heterogeneous lesions with mild/moderate and homogeneous/heterogeneous enhancement; and for V-MMS concentric lesions with homogeneous, moderate enhancement. Inter-reader agreement was moderate to substantial for all vessel wall MRI characteristics (κ=0.46–0.86) and fair for collateral grading (κ=0.35). There was 11% inter-reader agreement for diagnosis on luminal imaging when compared with 82% for luminal imaging+vessel wall MRI (P<0.001).
Conclusions—Vessel wall MRI can significantly improve the differentiation of moyamoya vasculopathies when combined with traditional imaging techniques.
Moyamoya vasculopathy, which is divided into moyamoya disease (MMD) and moyamoya syndrome (MMS), is a steno-occlusive process of the carotid termini, proximal middle cerebral artery, and anterior cerebral artery with development of robust compensatory collaterals at the base of the brain.1 MMS may arise secondary to many underlying disease processes, including sickle cell anemia, neurofibromatosis 1, radiation therapy, congenital syndromes, intracranial atherosclerotic disease (A-MMS), and vasculitis (V-MMS).1 It is important to differentiate MMD from other causes of MMS because treatment may differ drastically.2,3 Specifically, the first-line therapy for symptomatic MMD is surgical revascularization,3 whereas the Stenting and Aggressive Medical Management for Preventing Recurrent Stroke in Intracranial Stenosis (SAMMPRIS) trial2 showed that aggressive medical management is the first-line therapy for a patient with high-grade (70%–99%) atherosclerotic stenosis. Similarly, V-MMS should not be surgically treated, but rather treatment should focus on the underlying inflammatory process.
Diagnostic evaluation typically focuses on luminal imaging and clinical features; however, there is a significant overlap in luminal patterns of disease between MMD and MMS. Collaterals are compensatory4 and similar collateralization can be seen in A-MMS or MMD,5,6 and depending on the stage of disease, collaterals may not be seen in MMD.7 In addition, both MMD and MMS can present with either bilateral or unilateral disease.3 The clinical presentation may not allow for definitive differentiation, especially in adult presentations of moyamoya vasculopathy.
There have only been a few studies that evaluated differences in vessel wall magnetic resonance imaging (MRI) appearances between atherosclerosis and MMD.8–10 This study is the first to assess the added value of a multicontrast vessel wall MRI protocol in addition to luminal imaging for differentiation of MMD, A-MMS, and V-MMS.
Materials and Methods
After Institutional Review Board Approval, consecutive patients who had undergone intracranial arterial wall imaging from May 2013 to November 2015 were included from a prospectively maintained database. A neurointerventionalist (M.R.L.) reviewed the cases to determine the pattern of luminal disease and selected cases with Moyamoya vasculopathy (steno-occlusive disease of the carotid terminus, A1 and M1 segments). Two stroke neurologists (K.J.B. and A.d.H.) reviewed the clinical information and imaging reports of pattern of disease, while blinded to vessel wall MRI information and clinical diagnosis, and categorized the vasculopathies as vasculitis, atherosclerosis, reversible cerebral vasoconstriction syndrome, or MMD. If there was disagreement in the diagnosis, a third stroke neurologist (D.L.T.) arbitrated.
Patients were scanned on a 3T Siemens Trio MR scanner (Siemens Healthcare, Erlangen, Germany). The imaging protocol included high-resolution multi-planar T1 (0.4×0.35 in-plane resolution; slice thickness, 2 mm; repetition time/echo time, 1000/10 ms; and time, 36 s per slice) pre and post contrast, T2 (in-plane, 0.4×0.4; slices, 1 mm; repetition time/echo time, 3550/72 ms; and time, 9.3 s per slice), and 3D SPACE T2-weighted (0.6×0.6 mm in-plane resolution; slice thickness, 0.6 mm; repetition time/echo time, 2400/80 ms; slices, 64; and time, 10:20 minutes) sequences. More detailed imaging parameters can be found in a previous publication.11
Two independent raters (D.K.H. and W.A.C.), blinded to clinical and vessel wall MRI data, reviewed consecutive luminal studies performed for the moyamoya vasculopathy subjects. The luminal studies included any digital subtraction catheter angiographic, computed tomography angiography, or magnetic resonance angiography study that had been performed before the vessel wall MRI. Each intracranial internal carotid territory was evaluated independently with a brief written description of luminal disease pattern on the contralateral side to provide an idea of the pattern of global disease. The raters evaluated the presence and site of luminal disease, collateral development on a 3-point scale (0=no collaterals, 1=moderate collaterals, and 2=robust collaterals), presumed diagnosis (MMD, V-MMS, or A-MMS), and their confidence of diagnosis on a 4-point Likert-like scale (0=equivocal, 1=60% confidence, 2=75% confidence, and 3=>90% confidence). Raters were allowed to defer diagnosis when confidence was equivocal. After a 3-week washout period, the raters independently reviewed vessel wall and luminal data together for each internal carotid artery with an expanded written description, including both luminal and vessel wall patterns of disease on the contralateral side while blinded to patient clinical data. The raters evaluated the pattern of lesion arterial wall involvement (eccentric and concentric), presence of enhancement (y/n), the degree of enhancement (none=equivalent to normal wall, mild=less than the enhancement of the infundibulum, and moderate=greater than equal to the enhancement of the infundibulum),12 pattern of enhancement (homogeneous was defined as complete enhancement of the lesion, heterogeneous enhancement was incomplete lesion enhancement, and focal enhancement was defined as a point or short linear region of lesional enhancement),11 outward remodeling (positive if the outer wall area are qualitatively greater than the proximal normal segment), T2 lesion appearance (heterogeneous, not heterogeneous), presence of juxtaluminal T2 hyperintense band, presumed diagnosis, and confidence in the diagnosis (using the same 4-point scale). We used a qualitative assessment for outward remodeling to show that in individual cases, this can be evaluated by the radiologist and that it can be assessed and contribute to clinical differentiation in individual patients.
Continuous and categorical variables were summarized as mean±SD and count (percentage), respectively. Diagnostic accuracy was computed as the percentage of cases where the rater’s presumed diagnosis matched the final clinical diagnosis. Any cases classified as equivocal by the rater, without a presumed diagnosis, were considered an incorrect diagnosis. Diagnostic accuracy was compared between the luminal imaging only session and the luminal imaging+vessel wall MRI session using McNemar test. Inter-rater agreement for each individual luminal imaging and vessel wall MRI finding was summarized as the percentage of cases, where the readers gave the same rating (percent agreement) and unweighted Cohen κ. Inter-rater agreement for the presumed diagnosis was summarized as percent agreement, with equivocal cases always being considered a disagreement, even if both raters provided an equivocal diagnosis. Percent agreement was compared between luminal imaging and luminal imaging+vessel wall MRI sessions using McNemar test. Individual luminal imaging and vessel wall MRI findings were compared between the A-MMS and MMD groups using the χ2 trend test (for collateral grade) or the standard χ2 test (all other findings). There were too few vasculitis patients for comparison.
Throughout the analysis, the left and right sides were treated as separate—although not independent—observations, as the raters only viewed and rated one side at a time during the review. For the analysis of diagnostic accuracy and prevalence of individual imaging findings, each rater’s assessment was also treated as a separate but not independent observation. This approach of pooling the 2 raters treats them evenly and results correspond to an average of the 2. This is a more efficient use of data than analyzing only 1 rater or each rater separately. To avoid double counting, all cases and treating all observations as independent, all hypothesis tests (McNemar and χ2 tests) were conducted as permutation tests, where all observations from the same patient were permuted together.13 Clustering all observations from the same patient like this accounts for the dependence between these observations and does not inappropriately inflate the effective sample size; 95% confidence intervals for Cohen κ were computed using the nonparametric bootstrap with the percentile method, also resampling by patient.12 All statistical calculations were conducted with the statistical computing language R (version 3.1.1; R Foundation for Statistical Computing, Vienna, Austria). Throughout, 2-tailed tests were used with statistical significance defined as P<0.05.
Clinical Diagnoses and Characteristics
One hundred forty-eight consecutive vessel wall MRI cases with luminal imaging were reviewed, with 21 found to have moyamoya vasculopathy on luminal imaging. Of these, 10 had A-MMS, 8 had MMD, and 3 had V-MMS (all of which were varicella vasculitis). Patient clinical and demographic information are summarized in Table 1. Of note, all patients with A-MMS and none of the MMD or V-MMS patients had ≥2 vascular risk factors. Of the 42 hemispheres imaged with luminal imaging, a total of 38 pathological carotid arterial territories were evaluated (2 per patient, with 4 segments determined to be normal on luminal imaging) by 2 independent readers for a total of 76 ratings. Luminal imaging examinations performed within each disease group are also listed in Table 1. All cases underwent clinical evaluation based on current diagnostic guidelines.3 All cases who lacked digital subtraction catheter angiography either fulfilled proposed criteria that obviated the need for digital subtraction catheter angiography, or had clinical diagnoses that ruled out a diagnosis of MMD.
Vessel Wall MRI Characteristics
There were significant differences in the vessel wall MRI appearances of A-MMS, V-MMS, and MMD (Table 2). Typical findings are shown in Figure. A-MMS most frequently had eccentric, outward remodeling lesions that were heterogeneous on T2-weighted vessel wall MRI and had a juxtaluminal hyperintense band. All A-MMS lesions showed a mild or moderate degree of enhancement that was homogeneous or heterogeneous. In comparison, MMD showed noneccentric, nonremodeled lesions, without T2 wall signal heterogeneity or a juxtaluminal T2 hyperintense band and rarely enhanced. When enhancement was present with MMD, the lesions were concentric and showed homogeneous, mild enhancement. V-MMS showed concentric moderately enhancing lesions without outward remodeling or T2 lesion heterogeneity.
Added Value of Vessel Wall MRI
On the basis of luminal imaging alone, readers made the correct diagnosis (of A-MMS, MMD, or V-MMS) in 24 of 76 evaluations (32%). When luminal imaging and vessel wall MRI were both available, readers were significantly more accurate with 66 of 76 (87%) correctly diagnosed (P<0.001) (Table 3). The improvement in diagnostic accuracy was similar for each rater individually (rater 1: 29%–89%, P<0.001; rater 2: 34%–84%, P<0.001). Improvement in diagnostic accuracy with the addition of vessel wall MRI was seen in A-MMS (32% versus 82%, P=0.009), MMD (37% versus 90%, P=0.016), and V-MMS (13% versus 100%, P=0.25), although there were only 3 vasculitis patients (8 readings). Of the 52 readings with an incorrect diagnosis based on luminal imaging alone, 43 (83%) were correctly reclassified with the addition of vessel wall MRI. One reading correctly diagnosed by luminal imaging as MMD was then incorrectly reclassified by luminal imaging+vessel wall MRI as A-MMS. There were 9 readings with an incorrect diagnosis by both luminal imaging and luminal imaging+vessel wall MRI, of which 7 had a clinical diagnosis of A-MMS (misdiagnosed as MMD or equivocal) and 2 MMD (misdiagnosed as A-MMS or equivocal).
Readers’ confidence rating was the same (4 of 76 or 5.3%) or higher (94.7%) in all cases of the luminal imaging+vessel wall MRI review compared with the corresponding luminal imaging alone review. On the Likert scale, the average increase in confidence was 2.4 when vessel wall MRI+luminal imaging correctly reclassified a case, 1.3 when a case was diagnosed correctly by both luminal imaging alone and luminal imaging+vessel wall MRI, and 1.6 when a case was incorrectly diagnosed by both luminal imaigng alone and luminal imaging+vessel wall MRI.
Inter-reader agreement on the luminal imaging and vessel wall MRI findings is summarized in Table 4. Agreement was moderate or substantial for most findings, with collateral grade having fair agreement (58% of cases, Cohen κ=0.35). Inter-reader agreement on diagnosis was 11% (4 of 38) based on luminal imaging (all uncertain diagnoses were counted as disagreements) and 82% (31 of 38) using luminal imaging+vessel wall MRI (P<0.001).
We report the first study using a multicontrast vessel wall MRI protocol for the differentiation of MMD from A-MMS and V-MMS. In addition, this is the first study to assess the added value of vessel wall MRI over luminal imaging alone in moyamoya vasculopathy differentiation. This study shows that a multicontrast vessel wall MRI protocol with luminal imaging can differentiate between MMD and MMS because of atherosclerosis or vasculitis significantly better than luminal imaging alone. We found there was a moderate to substantial agreement in all vessel wall MRI characteristics studied, indicating that vessel wall MRI characteristics can be consistently evaluated in moyamoya vasculopathy. There was only fair agreement in the assessment of luminal imaging of collaterals. The likelihood of a correct diagnosis in the setting of moyamoya vasculopathy significantly increased when vessel wall MRI was combined with luminal imaging (from 31.6% to 86.8%, P<0.001), and this increase was significant for MMD and A-MMS. Although the increase in diagnostic accuracy was not statistically significant for vasculitis, given the study was underpowered to detect such a difference, the accuracy increased from 12.5% to 100% with the inclusion of vessel wall MRI. In addition, rater confidence increased with direct visualization of vessel wall abnormalities.
Historically, angiographic imaging has served as the reference standard for differentiation of MMD from A-MMS, with differentiating features considered to be prominent moyamoya collaterals and bilaterality of carotid terminus steno-occlusive disease. However, limiting the diagnostic accuracy of luminal imaging are the observations that collaterals are compensatory, frequently visualized in A-MMS,5,6 only present in the intermediate stages of MMD evolution,7 and both MMD and MMS can be unilateral or bilateral. In this study, we found distinctive patterns of vessel wall MRI appearance in MMD, A-MMS, and V-MMS that improve diagnostic accuracy over angiography and other forms of luminal imaging alone. MMD most commonly showed no postcontrast vessel wall enhancement and absence of eccentric wall thickening, no outward remodeling, lack of heterogeneous T2 wall signal, and absent juxtaluminal T2 hyperintense band. When wall enhancement was seen in MMD (13%), it was mild, concentric, and homogeneous enhancement, which differed in appearance from both A-MMS and V-MMS. A-MMS showed eccentric and outwardly remodeling wall thickening with heterogeneous lesion T2 signal, a juxtaluminal T2 hyperintense band, and heterogeneous or homogeneous, and mild or moderate lesion enhancement. V-MMS showed concentric, moderate, homogeneously enhancing lesions without T2 lesion heterogeneity, juxtaluminal T2 hyperintense band, nor outward remodeling.
There have been a few studies that have compared MMD with atherosclerosis on vessel wall MRI. Kim et al8 compared 12 MMD and 20 atherosclerosis patients using vessel wall MRI and found that MMD typically showed noneccentric, nonenhancing lesions with wall shrinkage, whereas atherosclerosis showed eccentric, enhancing, and outward remodeling lesions. Yuan et al10 compared 21 MMD and 44 atherosclerosis subjects, and found that on vessel wall MRI, MMD typically shows concentric wall thickening with homogeneous signal when compared with eccentric, heterogeneous atherosclerotic lesions. These studies agree with our findings that atherosclerotic lesions show outward remodeling and heterogeneous wall signal, but in addition, we showed atherosclerosis differs from MMD by having a T2 hyperintense juxtaluminal band and either homogeneous or heterogeneous lesion enhancement, whereas MMD typically does not show appreciable wall thickening and rarely enhances. Ryoo et al9 assessed 25 MMD, 7 probable MMD, and 16 patients with atherosclerosis on vessel wall MRI, and found that 90.6% of MMD and probable MMD typically showed circumferential wall enhancement at the carotid termini with wall shrinkage, whereas atherosclerotic lesions typically showed eccentric wall enhancement and outward remodeling. This study’s findings differ from the above-mentioned studies and our own in regard to the frequency of wall enhancement in MMD. In our study, we found that 13% of MMD lesions show mild, circumferential enhancement, which is a substantially lower rate. This discrepancy may be because of the heterogeneity of the genetic background of MMD14,15 and that this study may have captured a subtype of MMD with differences in pathophysiology. Ryoo et al9 evaluated Korean patients with MMD, whereas our MMD population was either white or Hispanic; the 2 populations have different genetic susceptibility with different biochemical pathways.16 Another possibility is that MMD or subtypes of MMD are transiently inflammatory,17 and Ryoo et al’s9 study captured these patients during this intermediate stage.
There are several limitations to this work. First, this was a retrospective study. Second, the luminal imaging modalities performed were heterogeneous; however, best practice guidelines were used for imaging performance and patient care, and the approach to image review best approximates clinical practice at many institutions. Third, the sample size is limited because moyamoya vasculopathy is uncommon, especially MMD and V-MMS. Fourth, there is no histology confirmation of diagnosis, and in the setting of MMD and A-MMS, the only gold standard is autopsy evaluation, which was not used in this study. We rather relied on the expert review of 3 stroke neurologists, who had access to the comprehensive diagnostic workups, as a reference standard. Further multicenter investigation with inclusion of genotypic and phenotypic analysis would be helpful to better classify the imaging appearances of MMD, better define its subtypes, and illuminate disease pathophysiology and progression.
Vessel wall MRI improves diagnostic accuracy and diagnostic confidence in the differentiation of MMD from A-MMS and V-MMS compared with luminal imaging alone. The differentiation is important as treatments for each disease entity differ significantly. Care needs to be taken when making conclusions based on a limited number of cases, and this study requires replication in a larger independent cohort. However, if further confirmed in larger studies, the diagnostic algorithm and criteria for MMD may be revised for improved diagnostic accuracy in addition to potentially limiting invasive diagnostic tests.
Sources of Funding
The study was supported by the National Institutes of Health 1R56NS092207-01 grant fund from the National Institute of Neurodegenerative Disorders and Stroke.
D.S. Hippe received grants from GE and Philips Healthcare grants unrelated to the current work. Dr Hatsukami obtained grants from Philips Healthcare grants unrelated to the current work. Dr Yuan received grants from Philips Healthcare grants unrelated to the current work. He is a member of Philips Radiology Medical Advisory Network. The other authors report no conflicts.
- Received March 1, 2016.
- Revision received May 3, 2016.
- Accepted May 11, 2016.
- © 2016 American Heart Association, Inc.
- Derdeyn CP,
- Chimowitz MI,
- Lynn MJ,
- Fiorella D,
- Turan TN,
- Janis LS,
- et al
- 3.↵Research Committee on the Pathology Treatment of Spontaneous Occlusion of the Circle of Willis: Health Labour Sciences Research Grant for Research on Measures for Infractable Diseases. Guidelines for diagnosis and treatment of moyamoya disease (spontaneous occlusion of the circle of willis). Neurol Med Chir. 2012;52:245–266.
- Togao O,
- Mihara F,
- Yoshiura T,
- Tanaka A,
- Noguchi T,
- Kuwabara Y,
- et al
- Tanaka M,
- Sakaguchi M,
- Kitagawa K
- Ryoo S,
- Cha J,
- Kim SJ,
- Choi JW,
- Ki CS,
- Kim KH,
- et al
- Mossa-Basha M,
- Hwang WD,
- De Havenon A,
- Hippe D,
- Balu N,
- Becker KJ,
- et al
- Davison AC,
- Hinkley DV
- Milewicz DM,
- Kwartler CS,
- Papke CL,
- Regalado ES,
- Cao J,
- Reid AJ
- Shoemaker LD,
- Clark MJ,
- Patwardhan A,
- Chandratillake G,
- Garcia S,
- Chen R,
- et al
- Masuda J,
- Ogata J,
- Yutani C