Collateral Circulation in Moyamoya Disease

Introduction

Moyamoya disease (MMD) is a chronic cerebrovascular disorder characterized by progressive stenosis or occlusion of the intracranial internal carotid artery (ICA) and its proximal branches and involves the development of a basal collateral network.¹ The natural history and potential prognostic factors of MMD are not well known, and the clinical presentation varies, especially in adults.² Therefore, predicting the risk of stroke, determining the surgical indication, and choosing the optimal treatment time are highly important for patients with MMD.

The cerebral collateral circulatory system is the subsidiary network of vascular channels that stabilize the cerebral blood flow when the principal arteries fail.³ MMD has a chronic, irreversible, and progressive natural course, and the progressive stenosis and occlusion of the intracranial principal artery reflect not only the process of disease progression but also the process by which collateral circulation is established. Relatively limited attention has been devoted to its relationship with recurrent stroke events in MMD patients, although many studies have indicated that collateral circulation status is closely related to clinical outcomes in acute ischemic stroke patients.^4,5 The Suzuki staging system is a time-oriented classification of MMD that has been widely applied to evaluate disease progression. However, because the same morphological Moyamoya vasculopathy (Suzuki stage) may present with different clinical characteristics,^6–8 we hypothesized that the individualized level of collateral circulation among different patients may be a major contributor to this phenomenon.

In this study, we sought to establish a new grading system for MMD that is based on the collateral circulation and Suzuki stage, considering the disease progression and associated collateral compensation and to elucidate its value in assessing the clinical features and hemodynamics of the disease, guiding surgical indication, and predicting stroke events.

Methods

The authors declare that all supporting data are available within the article and its online-only Data Supplement.

Subjects

The ethics committee of our institution granted ethics approval of this study (ky-2018-6-59) and waived written informed consent. In total, 301 consecutive confirmed MMD patients who were treated at our department between January 2014 and December 2016 were enrolled in the study. Patients included in this study met the following criteria: (1) diagnosed with bilateral MMD by digital subtraction angiography (DSA) according to the Japanese guideline published in 2012, with exclusion of Moyamoya syndrome;^1,9,10 (2) no history of prior bypass surgery; (3) no history of systemic diseases, such as atherosclerosis, terminal carcinoma, and immune system disease; and (4) no presence of atherogenic risk factors, such as hypertension, diabetes mellitus, dyslipidemia, heavy smoking, and heavy drinking (such patients were excluded because of the unclear influences of those factors on the collateral circulation).¹¹

Clinical Data Collection

Clinical information, including sex, age, clinical manifestations, neuroimaging findings, hemodynamic status, neurological deficits, perioperative complications, and follow-up outcome, was recorded. The major clinical symptoms on admission were divided into the following 2 subgroups: (1) ischemic symptoms, including mild ischemic symptoms, such as transient ischemic attack and headache, and severe ischemic symptoms, such as imaging-confirmed infarction; and (2) imaging-confirmed hemorrhage, including intraventricular hemorrhage (IVH) and intraparenchymal hemorrhage (IPH). A symptomatic MMD hemisphere was defined as (1) a hemisphere with a positive history of recurrent transient ischemic attack, infarction or hemorrhage or (2) a hemisphere with persisting neurological symptoms, such as hemiparesis, sensory deficits, or aphasia.¹² Persistent headache that could not be clearly localized to only one hemisphere was attributed to both hemispheres. Patients were excluded if the details of the clinical event could not be confirmed or if the symptoms did not match the imaging findings. All the clinical symptoms or the disease type were determined by neurosurgeons with at least 5 years’ experience in clinical practice. The hemorrhage was certified by computed tomography, and the infarction was certified by magnetic resonance imaging (MRI). Dynamic susceptibility contrast–MRI was used to evaluate the regional hemodynamic status of the patients after they were admitted to our department.

MMD Collateral Grading System

All patients underwent DSA, including injection of both internal carotid arteries and both vertebral arteries, as well as assessment through the late venous phase to evaluate the collateral flow from all possible sources. Lateral and anteroposterior views of each artery injection were collected. The angiographic collateral grade was evaluated with the system described subsequently. All angiography images were reviewed by 2 experienced readers (Drs Han and Zhao) who were blinded to the clinical details. Any differences in their observations were resolved by consensus.

The grading score was obtained based on the collateral circulation and ranged from 1 to 12 as follows.

1. The leptomeningeal system includes the following 3 parts of the collateral networks according to the anatomy extent of pial collateral blood from the posterior cerebral artery (PCA) territory to the anterior cerebral artery (ACA) territory and the middle cerebral artery (MCA) territory on the delayed venous phase^13,14: (1) pPCA→ ACA (the parieto-occipital branch of the posterior cerebral artery anastomoses to the ACA) was defined as retrograde flow extending to the ACA territory from the pPCA or posterior pericallosal artery; blood supply to the cortical border zone between the ACA and PCA territory was assigned a score of 1 (Figure 1A), and blood supply over the Central Sulcus via the posterior pericallosal artery was assigned a score of 2 (Figure 1B); (2) the anterior temporal branch of the PCA anastomoses to the temporal branch of the MCA was assigned a score of 1 (Figure 2A); (3) pPCA→ MCA (pPCA anastomoses to the MCA); if the retrograde flow extended in superficial vessels only (M4 segment of MCA), a score of 1 was assigned (Figure 2B); if the retrograde flow extended into Sylvian fissure (M3 segment of MCA), a score of 2 was assigned (Figure 2C); and if the flow extended to the occlusion (the reconstituted vessels at the distal end of the occlusion within the M1 or proximal M2 segments), a score of 3 was assigned (Figure 2D). The leptomeningeal system scores the sum of the 3 parts above, and the absence of leptomeningeal anastomoses was assigned a score of 0.

2. Regarding the basal brain perforators, they were occasionally partially overlapped with the Moyamoya vessels, and differentiating among these perforators was difficult. Both of the perforators and internal cerebral artery (ICA) reflect the blood supply of the anterior collateral circulation. Thus, we defined the expansion of this collateral route by using the Suzuki stage as follows: scores of 6 to 1 corresponded to Suzuki stages 1 to 6.

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

The scoring criteria are illustrated in Figure 3 (detailed descriptions were summarized in Table I in the online-only Data Supplement). The 3 collateral circulation status grades in MMD were defined as follows: scores of 1 to 4 were defined as a poor collateral status (grade I), scores of 5 to 8 were defined as a fair collateral status (grade II), and scores of 9 to 12 points were defined as a good collateral status (grade III).

• Download figure
• Download PowerPoint

Hemodynamic Examination

The cerebral hemodynamic status was assessed by a dynamic susceptibility contrast–MRI examination using a MAGNETOM Skyra 3T MRI scanner (Siemens, Germany) following previously described methods.¹⁵ The acquired dynamic susceptibility contrast–MRI images were processed using a post-processing workstation (Syngo Via 20, Siemens) and were analyzed with MR Neuro-Perfusion software. Regions of interest (ROIs) were manually drawn in the MCA territory and upper cortex of the cerebellum away from the infarcted areas.^16,17 Each of the cortical ROIs consisted of a series of 3- to 5-cm diameter circles along the cortical rim of MCA territories and 1-cm diameter circles of the cortical cerebellum. The territorial ROIs were averaged over all ROIs acquired from 3 slices within a specific territory. The mean relative cerebral blood volume (CBV), mean relative cerebral flow (CBF), and time to peak (TTP) values of the ROIs in each hemisphere were calculated. Ratios of the CBV and CBF (the target regions to the control cerebellum region) were designated relative CBV and relative CBF, respectively, and a difference in the TTP between the target regions and the control region (cerebellum) was designated as the TTP delay.

Treatment and Follow-Up

Encephaloduroarteriosynangiosis was performed in all 301 enrolled patients within 2 weeks after DSA, and the opposite hemisphere was usually revascularized 3 months after the first surgery if required. Postoperative DSA was routinely performed 6 to 9 months after bilateral encephaloduroarteriosynangiosis. The effect of revascularization through encephaloduroarteriosynangiosis was graded according to the following system described by Matsushima et al.¹⁸: Grade 1 was defined as the revascularization of <1/3 of the MCA distribution, grade 2 as the revascularization of 1/3 to 2/3 of the MCA distribution, and grade 3 as the revascularization of >2/3 of the MCA distribution. The neurological function status of each patient at the time of admission and follow-up was assessed using a modified Rankin Scale by neurosurgeons with at least 5 years’ experience in clinical practice.

Follow-up was conducted on a per-hemisphere basis start from the encephaloduroarteriosynangiosis on that hemisphere to the end of this study (October 2017). Follow-up was conducted at 3 and 6 months and annually after surgery by clinical visit and telephone interview. The end point was the occurrence of infarction or bleeding in the symptomatic hemispheres. End point determination was made by an independent, experienced physician, blinded to the DSA data, based on review of the clinical and imaging data.

Statistical Analyses

A Kruskal-Wallis H test was used to compare the continuous variables. A Pearson χ² test was applied to compare the categorical variables between the groups. A Spearman rank test was used to verify correlations. The Kaplan-Meier method was utilized to test for statistical significance using a log-rank test. A receiver operating characteristic (ROC) curve and a subsequent analysis of the area under the curve were applied to evaluate the symptom and the performance of predicting postoperative stroke. Interobserver variability for the assessment of collaterals between the 2 observers was assessed using κ statistics. All analyses were performed with SPSS software, version 24.0 (International Business Machines Corp, Almond, NY), and P<0.05 was considered statistically significant.

Results

Patient Characteristics

A total of 301 patients were enrolled in this study, including 105 children (<18 years old) and 196 adults (>18 years old). Among them, 132 (43.9%) patients were males, and 169 patients were women. The median age was 28±16 years (range, 4 to 65 years). A total of 364 symptomatic hemispheres were analyzed. Transient ischemic attack was the most common initial symptom (135 hemispheres, 37.1%). Other symptoms included infarction (133, 36.5%), intracranial hemorrhage (48, 13.2%), and headache (48, 13.2%; Table II in the online-only Data Supplement).

Association Between MMD Collateral Grading and Clinical Characteristics

According to the new MMD collateral grading system, there were 134, 143, and 87 hemispheres in the poor, fair, and good collateral grade groups, respectively (Table).

Among the ischemic MMD patients (315 hemispheres), 65.8% (75 of 114) of grade I hemispheres, 35.7% (45 of 126) of grade II hemispheres, and 15.8% (12 of 76) of grade III hemispheres presented with infarction. Patients who presented with infarction were more likely to have lower collateral grades than patients who presented with mild ischemia symptoms (P<0.001; Z=3.079; Figure 4A). The ROC curve demonstrated that patients who presented with infarction were more likely to score <8 points, with the highest combination of sensitivity (88.0%) and specificity (44.1%). The area under the ROC curve was 0.763 (95% CI, 0.711–0.816; P<0.001; Figure 4B).

• Download figure
• Download PowerPoint

Among the hemorrhagic MMD patients (49 hemispheres), 10.0% (2 of 20) of grade I hemispheres, 35.3% of grade II hemispheres (6 of 17), and 81.8% of grade III hemispheres (9 of 11) presented with IVH. Patients who presented with IPH were more likely to have lower collateral grades than patients who presented with IVH (P=0.017; Z=1.540; Figure 4C). The ROC curve demonstrated that hemorrhage patients who presented with IVH were more likely to score >8 points, with the highest combination of sensitivity (87.1%) and specificity (64.7%). The area under the ROC curve was 0.812 (95% CI, 0.684–0.940; P<0.001; Figure 4D).

The modified Rankin Scale distribution on admission in each grade also showed that the lower grades were correlated with worse neurological deficits (P<0.001; r_s=−0.420; Table).

Association Between MMD Collateral Grading Scores and Hemodynamic Data

In total, dynamic susceptibility contrast–MRI was performed on 223 symptomatic hemispheres, and 41 hemispheres were excluded because of severe cerebellar infarction on MRI; 182 symptomatic hemispheres were ultimately analyzed. The mean values of TTP delay in collateral grades I, II, and III were 5.72±2.59, 3.66±1.48, and 2.37±1.15, and more severe TTP delay was correlated with lower collateral score (P<0.001; r_s=−0.574; Figure I in the online-only Data Supplement). The mean values of relative CBV ratio in collateral grades I, II, and III were 1.60±0.60, 1.67±0.72, and 1.85±0.72. The mean values of relative CBF in collateral grade I, II, and III were 1.19±0.48, 1.31±0.58, and 1.43±0.52. A lower collateral score was correlated with a worse relative CBV ratio (P=0.016; r_s=0.178) and a worse relative CBF ratio (P=0.002; r_s=0.224).

Association Between MMD Collateral Grading and Therapeutic Outcome

Encephaloduroarteriosynangiosis was performed in 348 of the 364 symptomatic hemispheres, and the postoperative angiographic findings showed that a better Matsushima grade was correlated with a lower collateral grade of the 303 ischemic hemispheres (P=0.011; r_s=−0.146). In contrast, no correlation with the collateral grade was observed in the 45 hemorrhagic hemispheres (P=0.872; Table).

During the perioperative period, critical complications (9 infarction, 1 hemorrhage) occurred in 10 hemispheres. Lower grades were correlated with higher perioperative complication rates (P=0.005; r_s=−0.152; Table).

During an average postoperative follow-up time of 27.7±9.3 months, stroke occurred in 18 hemispheres (18 infarctions) of ischemic MMD patients and in 2 hemispheres (1 infarction, 1 bleeding) of hemorrhagic MMD patients (no one dead). Lower MMD collateral grades were also correlated with worse follow-up neurological outcomes within the modified Rankin Scale score (P<0.001; r_s=-0.255; Table). A total of 14 strokes in the follow-up occurred in the poor grade, 4 in the fair grade, and 2 in the good grade. The Kaplan-Meier estimate of follow-up stroke showed that lower grades were correlated with higher follow-up stroke rates (P=0.007; Figure 5A). The ROC curve demonstrated that scores of ≤4 points predicted a higher follow-up stroke rate, with a sensitivity and specificity of 70.0% and 64.9%, respectively. The area under the ROC curve was 0.696 (95% CI, 0.645–0.744; P=0.002; Figure 5B).

• Download figure
• Download PowerPoint

Interobserver Agreement

The 2 observers exhibited agreement in the grading of Moyamoya vessels (κ=0.762; P<0.001), pial collaterals from the pPCA to the ACA territory (κ=0.764; P<0.001), pial collaterals from the anterior temporal branch of the PCA to the temporal branch of the middle cerebral artery territory (κ=0.906; P<0.001), and pial collaterals from the pPCA to the MCA territory (κ=0.626; P<0.001).

Discussion

In this study, we proposed a new MMD collateral grading system with scores ranging from 1 to 12 points that correlated with the severity of clinical symptoms, the hemodynamic status, and the therapeutic outcome. Ischemic MMD patients who have more seriously ischemic symptoms were more prone to score <8 points (fair and poor collateral status). Hemorrhage patients who presented with IVH were more prone to score >8 points. Patients who scored ≤4 points predicted a higher follow-up stroke rate and worse prognosis.

Hypoperfusion confers an increased susceptibility to ischemia, and hemodynamic abnormalities are a mechanism of infarction, especially external watershed infarction.¹⁹ In this study, we assumed that cerebral collateral circulation reflects the intracranial blood supply and hemodynamic status and that it might match with the symptoms to some extent. According to our study, collateral grades were significantly correlated with symptoms, particularly in ischemic MMD patients: patients who presented with infarction were more likely to score <8 points, but those who presented with mild ischemia symptoms were more likely to score >8 points. This finding may be attributed to the cerebral collateral circulation, which plays a pivotal role in the pathophysiology of cerebral ischemia. Successful compensatory collateralization is considered a defensive measure against ischemic and hemorrhagic stroke in MMD.^6,20 The collateral circulation, especially the leptomeningeal system, plays the most important role in the collateral supply of the ischemic cortex in the ACA and MCA territories in MMD patients.²¹ Although collateral circulation is important for MMD, no systematic grading method for analyzing and assessing MMD has been established thus far. In this study, we consider the leptomeningeal system (including the collateral source of the PCA and the posterior pericallosal artery) the superficial-meningeal system’s collateral circulation and the basal brain perforators and Moyamoya vessels the parenchymal system’s collateral circulation. We attempted to capture the entire scope of collateral circulation and MMD progression via the proposed collateral grading system and used this system to assess the intracranial blood supply.

Hemodynamic evaluation further confirmed our finding. MMD collateral grade scores were consistent with the hemodynamic parameters calculated by MR perfusion. Thus, the MMD collateral grade may reflect the severity of cerebral ischemia. A close relationship exists among the severity of clinical symptoms, the hemodynamic status, and collateral formation.

In hemorrhagic MMD patients, the different collateral grades are significantly correlated with the hemorrhagic type: those who presented with IVH were more likely to score >8 points, whereas those who presented with IPH were more likely to score <8 points. Previous studies have reported that the dilatation and abnormal branching of the anterior choroidal artery are strong predictors of hemorrhagic events,²² whereas other studies have revealed that collateral anastomoses from choroidal arteries and periventricular vessels are responsible for IVH, and the bleeding points are most likely to be distributed posteriorly of the lateral ventricle. Therefore, good intracranial collateral circulation from the PCA or choroidal arteries may be correlated with IVH.^23,24 IPH is mainly encountered in patients who score <8 points, which is likely because of the rupture of weak basal brain perforator vessels under unusually increased hemodynamic stress because of the poor collateral pathway. The results also revealed that IVH and IPH in MMD patients may have different pathogeneses for hemorrhage.

Although the pathogenesis of MMD is poorly understood, revascularization surgery is currently believed to be the most successful therapy for preventing the progression of clinical symptoms. Therefore, determining the surgical indication and choosing the optimal treatment time have always been core problems in designing a therapy plan. In this study, we found that a score of 8 points is an important critical value for clinical symptoms. We hypothesize that 8 points may be the threshold for impaired cerebrovascular reserve capacity. Patients who score <8 points may represent a state of misery perfusion with insufficient compensation, and a more aggressive surgical intervention may need to be considered. Considering MMD is a progressive disease, a surgical treatment for symptomatic hemisphere with a score >8 points is still necessary but is also relatively safe. For asymptomatic or mildly symptomatic patients who had been undergoing conservative treatment, bypass surgery might be necessary if the symptoms are aggravating, cerebral blood perfusion is decreasing, or collateral circulation status is getting worse.

We think that ischemic patients with a lower collateral grade have an urgent need for surgery; however, the risk of perioperative stroke would be high, although revascularization would be effective in these patients. MMD patients with ≤4 points (poor collateral circulation status) are in an extreme ischemic status with rapid progression. The preoperative intracranial collateral status is an important factor predictive of perioperative complications and the long-term prognosis after surgery. Patients with poor collateral status have a higher perioperative and follow-up stroke rates than patients with good collaterals, but they still have better prognosis than MMD patients conservatively treated.^25–27 Thus, revascularization surgery may still be recommended for the patients who score ≤4 points, but more attention should be given to these patients during the perioperative and follow-up periods. At the same time, more rigorous, randomized, and controlled design are needed to prove our point.

Patients with worse intracranial collateral status (severe ischemia) would have a better angiographic outcome with revascularization. A lower MMD collateral grade, which represents a greater degree of cerebral ischemia, consequently demonstrates a stronger demand to stimulate the formation of extracranial collateral circulation. Regardless, the MMD collateral grading score was significantly correlated with the Matsushima grade in the ischemic patients but not the hemorrhagic patients, which was likely due to a more complicated hemodynamic status and destruction of the blood-brain barrier after hemorrhage. By further analyzing the effect of encephaloduroarteriosynangiosis with postoperative stroke, we found that most of the postoperative strokes occurred in 16 hemispheres of the good revascularization group (Matsushima grades 2–3) and 4 hemispheres of the poor revascularization group (Matsushima grade 1). Among the hemispheres with good revascularization (197 hemispheres), the postoperative strokes occurred in hemispheres with collateral scores ≤4 points more than in hemispheres with collateral scores >4 points (P=0.004; Table III in the online-only Data Supplement). This result further suggests that the poor preoperative collateral status is an important predictive factor of postoperative stroke, even in patients with good revascularization by encephaloduroarteriosynangiosis.

The duro-cortical system, which consists of all dural branches of the ICA, external carotid artery, and vertebrobasilar artery, is too complex to score in a collateral grading system because of the numerous and scattered anastomosed types with cortical leptomeningeal. In our study, dural-cortical collaterals supplying blood to the cortex were present in 35.7% (130/364) of the hemispheres, though the compensation from the middle meningeal artery to the important functional cortical area in the MCA territory²⁸ (such as Broca area, Wernicke area, or the motor cortex of the frontal lobe) was only 9.3% (34/364). A rapid and effective anastomosis may not be able to form¹⁸ due to the cerebrospinal fluid layer that usually separates the dura and brain.

Limitations

Our study has several limitations. First, all patients were enrolled from a single-center, and a potential selection bias towards regions and races may have occurred. Second, because the components of the duro-cortical system are too complex to generalize and classify, this system was excluded from our collateral grading system. Cases with extraordinarily well-formed transdural collaterals may confound the results of the prediction of clinical symptoms and prognosis using our collateral grading system. Third, the value of cerebral hemodynamics determined by perfusion MRI is relative and not absolute, and further technical refinements are required to enhance its diagnostic value. Our study is a retrospective study, and how the intracranial collateral system is activated and why MMD patients have individual collateral development remain unknown.

Conclusions

The intracranial collateral status is correlated with the severity of the clinical symptoms and the therapeutic prognosis. The proposed new MMD collateral grading system can reflect the status of intracranial collateral circulation, aid in the evaluation of the severity of cerebral ischemia and hemorrhagic type, determine the appropriate surgical indication, assess the surgical risk, and predict prognosis. Larger samples and longer follow-up studies will be needed to prove its value.

Acknowledgments

We thank the individuals who contributed to the study or article preparation but did not fulfill all the criteria of authorship.

Footnotes