Moyamoya Disease

Abstract

Moyamoya disease (MMD) is a rare idiopathic steno-occlusive disease of the internal carotid arteries resulting in collateralization from new fragile vessels at the base of the brain termed “moyamoya” vessels. It most commonly affects patients of Asian descent, but has been observed in all ethnic groups. The age of onset fits a bimodal distribution in the 5–9- and 40–44-year age groups. MMD most often presents with stroke-like symptoms in children and bleeding in adults. The treatment for MMD involves revascularization surgery to replenish blood flow to the diseased areas of the brain. Revascularization can be performed by “direct” bypass where an external vessel is anastomosed to a brain vessel to replenish flow immediately or by “indirect” means where a vascular donor tissue is surgically planted on or near the surface of the brain to encourage formation of new vasculature. Outcomes for moyamoya are favorable with revascularization surgery showing decreased rates of stroke reported in patients after successful revascularization. In this chapter we present the pathophysiology, epidemiology, classification, workup, treatment, and outcomes for MMD.

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Introduction

Moyamoya disease (MMD) is a bilateral steno-occlusive process that selectively affects the supraclinoidal segment of the internal carotid artery (ICA). MMD was originally described in 1957 as an idiopathic steno-occlusive disorder of the carotid arteries (Takeuchi and Shimizu 1957). The name “moyamoya” comes from the Japanese word meaning “hazy” or “puff of smoke” and describes the angiographic appearance of both dilated collateral vessels and new vessels that form to supply ischemic regions of the brain resulting from progressive carotid stenosis (Kono et al. 1990; Scott and Smith 2009; Suzuki and Takaku 1969). The classification of moyamoya is based on angiographic anatomy and clinical history of associated diseases. Specific angiographic findings include: bilaterality versus unilaterality, absence or presence of an associated systemic or local condition associated with moyamoya, and the involvement of the supraclinoid carotid (C1–C2 segment) versus distal anterior circulation or posterior circulation vessels (Table 1). The term “moyamoya disease” is specifically used to describe idiopathic, bilateral steno-occlusive carotid disease in the supraclinoid carotid division (Fukui 1997; Suzuki and Takaku 1969), whereas “moyamoya syndrome” is used to describe the appearance of bilateral or unilateral moyamoya vessels in the setting of an associated, underlying disease state (Table 2) or in the case of unilateral idiopathic carotid disease (Scott and Smith 2009). “Atypical moyamoya” is a term used by many authors to describe moyamoya vessel formation related to non-carotid steno-occlusive disease or in cases with associated aneurysms or pseudoaneurysms (Adams et al. 1979; Grabel et al. 1989). Some authors have suggested that this classification system generates confusion, prompting a proposal to use the term moyamoya disease for the idiopathic, bilateral disease state and angiographic moyamoya for all other cases of moyamoya associated with another disease state, isolated unilateral disease, and disease within other vascular distributions (Natori et al. 1997). However, this simplified categorization has not been widely adopted to date.

Table 1 Conditions associated with moyamoya syndrome

Table 2 Classification of moyamoya

Epidemiology and Genetics

The prevalence of MMD is greatest in the Eastern Asian population with preponderance for individuals of Japanese descent. In Japan the prevalence of MMD has been estimated to be as high as 10.5 per 100,000 population with an incidence of 0.35–0.94 per 100,000 (Kuriyama et al. 2008; Wakai et al. 1997). In North America the incidence was previously reported as one quarter to one fifth that of Japan at 0.086 per 100,000 population (Uchino et al. 2005), but more recent evidence suggests a higher prevalence of 0.57/100,000 (Starke et al. 2012). Within the North American population, there is a higher incidence of MMD among patients of Asian and African descent where ethnicity-specific prevalence is 4.6 and 2.2 times higher than Caucasians, respectively (Uchino et al. 2005). There is a 2:1 female to male ratio among moyamoya patients (Wakai et al. 1997).

There is a bimodal distribution in age at diagnosis in the moyamoya population. Wakai et al. found that the early peak is in the pediatric population at age five and the secondary peak is in the 45–49-year-old range with 48 % of patients diagnosed at age <10 years (Wakai et al. 1997). A study undertaken 11 years following the publication of the Wakai data in Japan, however, found that the proportion of pediatric patients diagnosed has dropped to 15 % with a compensatory increase in adult diagnoses between ages 45 and 49 (Baba et al. 2008).

In 15 % of moyamoya cases there is a familial link (Yamauchi et al. 1997). Pedigree analysis has revealed an autosomal dominant inheritance pattern in families with MMD (Mineharu et al. 2006). Genetic linkage analyses have been undertaken by a number of groups that have led to the identification of distinct 5 chromosomal regions linked to MMD (Achrol et al. 2009). The primary candidate arising from linkage analysis has been at chromosome position 17q25.3. The original finding arose from a cohort of 55 Japanese familial MMD patients who underwent linkage analysis by Mineharu et al. (2008). Subsequent analyses by Liu et al. and Kamada et al. have focused on a mutated copy of the gene RNF213 (Kamada et al. 2011; Liu et al. 2010, 2011). Knockdown of RNF213 in zebra fish led to irregular vessel development suggesting a possible role for RNF213 mutations in the genesis of MMD. In a small nucleotide polymorphism study focused on genes encoding growth hormones and cytokines in Caucasian patients with MMD, Roder et al. found errors in the promoter region of platelet-derived growth factor receptor beta (PDGFR-β) and in the first exon of transforming growth factor beta (TGF-β) (Roder et al. 2010). There has not been a strong link between RNF213 and MMD in non-Asian patients, opening the possibility that the genetic origins of the disease are different between the two cohorts. Mutations in smooth muscle alpha actin (ACTA2) can predispose to developing MMD as well as premature coronary artery disease and thoracic aortic disease (Guo et al. 2009). Recently Xq28 deletions removing MTCP1/MTCP1NB and BRCC3 have been shown to cause a type of X-linked familial moyamoya syndrome (Miskinyte et al. 2011). Pathophysiology of MMD initially presents with a decrease in the outer diameter of the supraclinoidal ICA. This is associated with accumulation of smooth muscle cells and fibrous tissue within the intima, a loss of elastic lamina, and regression of the media with stenosis of the lumen (Yamashita et al. 1983). As the lumen decreases in caliber, thrombosis may occur in the vessel lumen. This process is progressive and continues until the vessel has occluded.

Moyamoya vessels appear to form in response to chronic brain ischemia related to carotid narrowing and occlusion. There is an elevation of proangiogenic factors released into the CSF in moyamoya patients, including hypoxia inducible factor 1, vascular endothelial growth factor, basic fibroblast growth factor, TGF-β, hepatocyte growth factor, and the matrix metalloproteinases (Achrol et al. 2009; Kuroda and Houkin 2008). Moyamoya vessels are histologically abnormal demonstrating dilatation at the origin, disordered or fragmented elastic lamina, fibrin deposition in the proximal orifice, loss of vessel wall media, and microaneurysm formation (Yamashita et al. 1983). These features contribute to the risk of bleeding and thrombus formation that characterize the presenting symptoms in MMD. Over time collateral circulation from the leptomeningeal vasculature may develop and provide improved flow to the affected hemisphere. When flow is restored moyamoya vessels may regress, as is the case after revascularization surgery.

Clinical Presentation

Patients with MMD have four primary presentations: ischemic symptoms, hemorrhage, headache, and as an incidental finding. Pediatric patients tend to present most often with stroke or transient ischemic attacks (TIAs) . TIAs commonly occur in children during periods of intense crying where hyperventilation leads to hypocapnia, intracranial vasoconstriction, and decreased flow in the setting of impaired cerebrovascular reactivity. Focal motor deficits, dysarthria, aphasia, and syncope are the most common deficits noted. Additional signs in children may include involuntary movements or visual deficits. These deficits resolve when the patient ventilates normally. Rarely pediatric patients can present with movement disorders related to ischemia and infarction that creates imbalance in the cortical–subcortical–ganglionic–thalamic–cortical motor circuitry (Pandey et al. 2010).

Adults may present with hemorrhage, ischemia, or a combination of both. TIAs in adults may be related to hypoperfusion in the setting of dehydration or hypotension that occurs in situations such as the perioperative period for major surgeries or extreme athletic events like marathon running. Adults may accumulate clinically silent strokes that result in cognitive dysfunction or neuropsychiatric disorders that slowly become apparent over time (Festa et al. 2010; Vermeer et al. 2003).

Hemorrhage is the presenting sign in over one half of Asian adults with MMD, although Caucasian MMD adults tend to present more commonly with ischemic symptoms (Gross and Du 2013; Nishimoto 1979; Saeki et al. 1991; Yonekawa et al. 1997). Hemorrhage is typically intracerebral in the basal ganglia or thalamus or intraventricular. These bleeds most likely arise from fragile neovessels in the lenticulostriate distribution where microaneurysm formation and abnormalities in vessel wall elastin distribution increase the potential for bleeding. Intraventricular bleeding most likely represents rupture of one of these vessels directly into the ventricular wall. Hemorrhagic conversion of ischemic strokes is a secondary cause of intraparenchymal hemorrhage in adult patients. Rarely does MMD present with subarachnoid hemorrhage (SAH). Subarachnoid blood is generally from flow-related saccular aneurysms that tend to form in the vertebrobasilar circulation in response to elevated flow through posterior communicating artery collaterals. However, SAH can occasionally occur in MMD in the absence of circle of Willis aneurysms.

Headache is a symptom that increasingly leads to the diagnosis of MMD, as patients with headache are more likely to undergo MRI imaging in their workup. Patients experience migraine-like headaches that are recurrent and can be daily in nature. The cause of these headaches is thought to be stimulation of nociceptive pain receptors in the dura in response to collateral vessel formation. Alternatively, migrainous headaches may arise from spreading depression that can be generated by cortical ischemia. Patients with headache who undergo imaging may have evidence of prior stroke on MRI. A majority of patients have resolution of headaches following revascularization; however, improvements in headache are not universal, and new headaches have been reported in the post-revascularization period that were not previously noted.

The increasing use of MRI and CT scanning for routine surveillance of brain health and in cancer screening has led to an increase in incidentally found, asymptomatic moyamoya patients. Evidence of silent infarction on MRI is often an indication to intervene and offer revascularization surgery. Asymptomatic patients with no evidence of prior stroke represent a challenge in management because MMD may progress or stall in adult patients. Therefore, best medical management and close observation may be undertaken in asymptomatic cases without MR infarcts, unless markedly reduced cerebral blood flow and impaired hemodynamic reserve or steal phenomenon is present, which may be an indication for revascularization.

Diagnosis and Staging of Moyamoya Disease

Moyamoya is a progressive disease that initiates with carotid narrowing and progresses to occlusion and exuberant formation of moyamoya vessels supplied by dural–pial and ethmoidal collaterals to the affected hemisphere. The condition is ultimately limited with a resolution of moyamoya vessels once the exclusive external carotid arterial supply of the hemisphere is established. The Suzuki staging system is a useful grading system for moyamoya, dividing the disease state into six stages from initiation to resolution (Suzuki and Kodama 1983; Suzuki and Takaku 1969), although it has not been definitively shown that progression through the Suzuki stages is inexorable. Table 3 summarizes the Suzuki staging system, and Fig. 1 summarizes representative angiography of each Suzuki stage. We have also identified a late stage of angiographic MMD progression characterized by severe stenosis of the entire cervical ICA that represents a secondary change in the artery related to poor distal outflow. The angiographic tapered appearance of the cervical ICA origin artery is referred to as the “bottle neck sign” and should not be confused for a dissection (Fig. 1g) (Khan et al. 2011; Yasaka et al. 2006).

Table 3 Suzuki staging of moyamoya

Fig. 1

Representative internal carotid (a–e) and external carotid (f) angiograms demonstrating Suzuki staging of moyamoya. (a) Stage I disease with narrowing of the supraclinoid internal carotid artery. (b) Stage II disease demonstrating dilatation of the middle cerebral artery and early formation of basal moyamoya vessels. (c) Stage III disease with occlusion of the middle and anterior cerebral artery but patent posterior communicating artery and further formation of moyamoya vessels. (d) Stage IV disease with intensification of moyamoya vessels and loss of the posterior communicating artery. (e) Stage V disease demonstrating diminishing moyamoya vessels as external collaterals increase in prominence. (f) External carotid angiogram Stage VI disease demonstrating loss of moyamoya vessels and collateral supply via the external carotid artery. (g) The angiographic tapered appearance of the cervical ICA origin artery referred to as the “bottleneck sign” should not be confused for a dissection (Reprinted with modification from “Classification and imaging of moyamoya phenomena. Moyamoya Disease: Diagnosis and Treatment” with permission from Thieme)

Imaging of Moyamoya

The choice and order of imaging modality in the workup of MMD depends upon the patient’s presentation. When presenting with hemorrhage, an anatomical imaging protocol that uses computed tomography (CT) followed by dual subtraction angiography (DSA) is useful to diagnose and characterize the location and grade of disease and to determine an acute treatment plan, if needed (Fig. 2). When presenting with ischemic symptoms, however, including stroke and transient ischemic attack, patients may initially be imaged with MRI to define regions of infarction using diffusion-weighted imaging (DWI) MRI, to anatomically localize moyamoya vessels, and to image carotid stenosis using MRI angiography (MRA) (Fig. 3). If the etiology of strokes based on anatomical imaging is consistent with flow-related ischemia (e.g., infarction in watershed distributions) or the source of ischemic symptoms cannot be determined, imaging measures of cerebral blood flow (CBF) and cerebrovascular reserve using functional imaging techniques may be appropriate to define brain regions at risk of stroke (Fig. 4). Impaired cerebrovascular reserve is considered by some groups to be an indication for intervention with revascularization procedures (Fierstra et al. 2011; Mikulis et al. 2005).

Fig. 2

Atypical moyamoya presenting with intraventricular hemorrhage. (a) Acquired CT 1 week following ictus demonstrates hypodensity in the left frontal region (*) and subacute hematoma in the left occipital horn. (b) and (c) AP and lateral angiogram demonstrate occlusion of the middle cerebral artery (MCA) with reconstitution of the MCA via moyamoya collateral. (d) Enhanced T1 MRI demonstrates a subacute to chronic infarct based on gadolinium enhancement (*). This correlates to the left frontal hypodensity in (a). (e) and (f) Axial FLAIR MRI demonstrates chronic changes in the left frontal lobe related to prior stroke and subacute hematoma in the occipital and temporal horns of the left lateral ventricle with evidence of ventriculomegaly. (b, c, f) Arrowheads demonstrate an aneurysm, another atypical feature that has formed on the reconstituted MCA perforators distal to the moyamoya vessels; this is the likely source of hemorrhage (Reprinted from “Classification and imaging of moyamoya phenomena. Moyamoya Disease: Diagnosis and Treatment” with permission from Thieme)

Fig. 3

This case of moyamoya disease was discovered due to transient ischemic attacks characterized by episodic left-sided weakness. (a) and (b) Right AP and lateral internal carotid angiograms demonstrating stage III moyamoya disease. (c) Right external carotid angiogram showing partial reconstitution of cortical flow. (d) Hypercapnic cerebrovascular reactivity (CVR) study performed with BOLD MRI revealed bilateral impairment of cerebrovascular reactivity (blue brain) worse in the right hemisphere. At a later date the patient suffered a retroperitoneal hematoma and hypovolemic shock, after which she experienced fixed arm and leg weakness. Diffusion MRI revealed a right cortical stroke (e) with enhancement after gadolinium infusion on T1 MRI (f). The right hemisphere also exhibited the “ivy” sign (with leptomeningeal enhancement) on gadolinium-enhanced T1 (f) (Reprinted from “Classification and imaging of moyamoya phenomena. Moyamoya Disease: Diagnosis and Treatment” with permission from Thieme)

Fig. 4

This patient underwent bilateral encephaloduroarteriosynangiosis (EDAS) for early-stage moyamoya disease. (a) Moyamoya demonstrated on left internal carotid artery angiogram. (e) Hypercapnic cerebrovascular reactivity (CVR) map performed 1 year after EDAS demonstrates good cerebrovascular reserve. (b) Left external carotid artery angiogram shows minimal reconstitution of cerebral blood flow following EDAS. The patient went on to experience intermittent episodes of speech stalling and word finding difficulty. (f) Repeat CVR demonstrates impaired cerebrovascular reserve in the left hemisphere. Because the superficial temporal artery was utilized for the prior EDAS, an external carotid artery supplying free omental flap was laid on the left hemisphere. (c) and (d) Lateral and AP external carotid angiogram with omental flap in place and reconstituting flow to the left hemisphere. (g) CVR study demonstrates normal cerebrovascular reserve following omental flap placement. (h) T1-weighted MRI shows fat of the omental flap laid on the left cerebral hemisphere. (e, f, g) MR perfusion (Reprinted from “Classification and imaging of moyamoya phenomena. Moyamoya Disease: Diagnosis and Treatment” with permission from Thieme)

Anatomical Imaging

Computed Tomography (CT)

In the acute setting of stroke or hemorrhage, CT is generally the first imaging modality utilized (Fig. 2). Intraventricular, intraparenchymal, and subarachnoid hemorrhage are readily demonstrated with CT and can guide the emergent treatment of the patient (e.g., determining need for cerebrospinal fluid diversion or clot evacuation). In addition to these acute findings, CT can demonstrate chronic ischemic changes visualized as hypodensities in both subcortical and cortical regions, pointing to prior or ongoing flow-related versus embolic ischemic insults.

Computed Tomographic Angiography (CTA)

Modern CTA techniques provide high-resolution images of the cerebral vasculature and are best for imaging larger parent arteries where carotid stenosis and occlusion can be readily visualized. In stage II disease, dilated parent vessels and leptomeningeal collaterals become apparent on CTA. Prominent moyamoya vessels can be visualized in the basal ganglia and cortex; however, these are not universally visible. In late-stage disease (stage V and VI) CTA demonstrates the prominent external carotid collateral circulation and loss of flow, stenosis, and/or loss of major cerebral artery divisions. CTA provides excellent 3-D imaging of the cerebral vasculature and can reveal small aneurysms or pseudoaneurysms in atypical moyamoya (Kawaguchi et al. 1996a). The lack of temporal resolution with CTA results in a “snapshot” of the vasculature that can be biased toward the arterial or venous phase of cranial circulation. Thus, the extent of moyamoya may be under- or overestimated based on the scan acquisition time relative to the mean transit time of contrast through the affected area, resulting in a snapshot of the lesion at a more arterial or venous phase of perfusion (Romero et al. 2009). In the postoperative period CTA can be utilized to determine bypass patency (Fig. 5d, e, Fig. 6f, g).

Fig. 5

Moyamoya syndrome with unilateral internal carotid artery occlusion and reconstitution of cerebral flow via external carotid collaterals (angiogram (a, b) demonstrates supraclinoid ICA occlusion). The patient was symptomatic with transient motor ischemic attacks and had an impaired cerebrovascular reactivity study (c). The patient underwent superficial temporal artery (STA) to middle cerebral artery bypass on the right side with resolution of symptoms. CT angiography was used to follow graft patency ((d) post-op CTA demonstrates flow through STA graft, and (e) 3-D CTA demonstrates a patent STA graft) (Reprinted from “Classification and imaging of moyamoya phenomena. Moyamoya Disease: Diagnosis and Treatment” with permission from Thieme)

Fig. 6

Cerebrovascular reserve (CVR) studies are useful in diagnosing chronic ischemia in moyamoya and following treatment effect. Patient 1 (upper row) was diagnosed with moyamoya disease with impaired CVR in bilateral hemispheres (a, blue brain). This patient underwent bilateral encephaloduroarteriosynangiosis (EDAS) to treat the disease (b, right external carotid angiogram postoperatively; (c) left external carotid artery angiogram postoperatively). (d) The patient underwent follow-up CVR 1 year following surgery and had normal cerebrovascular reserve. Patient 2 (lower row) was diagnosed with moyamoya and impaired CVR in the right hemisphere (e). The patient underwent right superficial temporal to middle cerebral artery anastomosis with good reconstitution of flow in the right middle cerebral artery (f, g). The patient returned for postoperative CVR and was noted to have normal cerebrovascular reserve (h). (a, d, e, f) MR perfusion (Reprinted from “Classification and imaging of moyamoya phenomena. Moyamoya Disease: Diagnosis and Treatment” with permission from Thieme)

Magnetic Resonance Imaging (MRI)

MRI is a powerful imaging modality in moyamoya to visualize both the vascular and parenchymal sequelae of the disease. Typical MRI studies include multiple sequence acquisitions that are specific to the various features of moyamoya. Carotid stenosis and occlusion may be observed on T2-weighted imaging as a change in caliber or the loss of carotid flow void. Furthermore, MRI-based angiography using gadolinium or arterial spin labeling techniques can detect carotid stenosis or occlusion. Moyamoya vessels are apparent as flow voids in the basal cisterns and the basal ganglia, representing dilated vessels of the anastomotic networks that evolve from dural–pial connections and dilated lenticulostriate vessels, respectively. Contrast-enhanced T1-weighted and fluid attenuation inversion recovery (FLAIR) MRI can demonstrate leptomeningeal enhancement termed the “ivy sign,” representing pial collaterals and/or congestion of the leptomeninges (Fig. 2f) (Maeda and Tsuchida 1999).

Both acute and chronic signs of ischemia and stroke are best imaged with MRI. In the acute setting diffusion restriction on DWI MRI indicates acute stroke (Fig. 3). Gadolinium enhancement on T1-weighted imaging is commonly observed in the subacute progression of stroke. In the chronic stage, stroke can be visualized as regions of cortical and subcortical atrophy and ex vacuo ventricular dilatation on T2-weighted images. FLAIR images demonstrate hyperintensities in regions of chronic strokes where atrophy and gliosis occur in both cortical regions and white matter.

Stroke identified in “watershed” regions between vascular distributions is indicative of hemodynamic insufficiency in the distal vasculature, whereas embolic strokes are visible in parent vessel and end artery distributions. Differentiating the etiology of ischemic injury in moyamoya is critical in the selection of subsequent diagnostic tests and ultimately in determining therapy; in this regard MRI is an important early modality in the workup of ischemic presentations of moyamoya.

MRI is also useful in the long-term follow-up of moyamoya patients as it permits surveillance of carotid steno-occlusive disease in the ipsilateral and contralateral sides, images bypass patency, and allows surveillance of the parenchyma to judge effect on tissue and to identify new or evolving infarcts. A recent study has demonstrated that patients with effective revascularization via superficial temporal artery to middle cerebral artery (STA–MCA) bypass have increased cortical thickness on follow-up MRI, a potential imaging biomarker for therapeutic effect (Fierstra et al. 2011).

Catheter Angiography

The “puff of smoke” appearance from which the name moyamoya derives was originally observed on carotid angiography. Moreover, the Suzuki staging system was based on this modality (Fig. 1) (Suzuki and Takaku 1969). Hence, angiographic imaging remains the gold standard diagnostic test for moyamoya; however, some authors have suggested that a combination of noninvasive tests may be adequate to make a diagnosis and treatment decisions (Bacigaluppi et al. 2009). Modern catheter-based angiographic techniques (digital subtraction angiography, DSA) provide high spatial resolution images through the arterial and venous phases of contrast injection as well as 3-D imaging. DSA permits selective imaging of the internal carotid, vertebral artery, and external carotid distributions to specifically characterize the extent of carotid occlusion and the origin of collateral supply in moyamoya. Other atypical vascular abnormalities associated with moyamoya, such as aneurysms and rarely arteriovenous malformations, are best visualized on DSA as the images are of high quality and magnified views of suspicious regions are possible (Grabel et al. 1989; Kawaguchi et al. 1996a; Nakashima et al. 1998).

Imaging of Cerebral Blood Flow and Cerebrovascular Reserve

Functional vascular imaging modalities give indirect and direct measures of both CBF and metabolic function in the brain. In moyamoya these data drive treatment decisions for revascularization therapy in both symptomatic ischemic patients and asymptomatic patients, as impaired hemodynamics are predictive of stroke risk (Lee et al. 2009; Ogasawara et al. 2002).

Blood flow is measured by introducing a tracer (generally a contrast agent) that can be tracked through the cerebral circulation serially over time. Mean transit time (MTT) and an estimation of cerebral blood volume (CBV) can be derived from the curve defining tracer concentration in a region of interest over time (Jackson 2004). Relative CBF can then be calculated using central volume theorem (CBF = CBV/MTT) and quantitatively expressed as mL/100 g tissue/min based on estimations from known blood flow values. Blood flow maps are generated by applying this method to each pixel in an axial image and assigning color lookup table values to each. Diminished CBF is thought to correlate to parent vessel stenosis in moyamoya and may be apparent in ischemic regions, indicating a need for treatment using revascularization techniques (Neff et al. 2006).

Both direct measures of tissue metabolism and measures of vascular autoregulatory function, an indirect measure of metabolic demand, are useful in guiding treatment decisions for patients with early ischemic symptoms or in prognostication of asymptomatic cases. Metabolic demand in the brain is measured noninvasively as the oxygen extraction fraction (OEF) using positron emission tomography (PET). OEF is derived by injecting radiolabeled oxygen (¹⁵O₂) intravenously and measuring the proportion of oxygen that is extracted from the blood by the tissue. In ischemic tissue there is an elevation of the OEF as more oxygen is extracted from red blood cells to meet equivalent metabolic demands in the setting of decreased CBF.

Vascular autoregulation in the brain occurs at the level of the arteriole where tissue demand for oxygen is detected as local changes in pH, nitric oxide signaling, and other paracrine mediators and results in changes in smooth muscle contraction/relaxation with resultant increase or decrease in vascular resistance and blood flow. Under normal conditions arterioles can either contract or relax in response to a stimulus to alter CBF as required. This is known as cerebrovascular reactivity (CVR) . However, in the case of moyamoya, if persistent oligemia occurs, the arterioles are persistently, maximally dilated to maximize blood flow and cannot react to further hypoxic/hypercarbic stimuli, known as “exhaustion of cerebrovascular reserve.” Patients in this state are at higher risk for ischemic symptoms and may go on to stroke (Ogasawara et al. 2002). To test CVR, CBF measurements are taken before and after a hypercapnic challenge induced by a dose of acetazolamide, a carbonic anhydrase inhibitor, by adjusting end-tidal PCO₂ (Mandell et al. 2008) or by breath holding. In normal patients there is an increase in CBF to the hypercarbic challenge compared to baseline, whereas in patients with exhausted cerebrovascular reserve, there is no change between baseline and challenge CBF or even a paradoxical decrease in CBF referred to as “steal.”

Computed Tomographic Perfusion (CTP) and MRI Perfusion (MRP)

CTP utilizes an iodinated contrast agent to track the transit of blood through the cerebral vasculature. MRP employs gadolinium contrast agents or endogenous contrast generated by arterial spin labeling to do the same. CBF is derived from the MTT and CBV that are derived from the perfusion curve as discussed above. These methods suffer from variability related to changes in blood pressure, hematocrit, motion, and susceptibility artifact, and the selection of arterial input function is difficult in patients with bilateral stenoses.

Xenon CT

Xenon is a diffusible gas that is radio-opaque. In this study patients inhale a gas mixture of O₂ and stable xenon. Xenon enters the brain tissue by diffusion. The accumulation of xenon correlates to blood flow and is used to derive estimates of CBF. Xenon CT can be undertaken at baseline and with acetazolamide or breath holding challenges to calculate CVR. However, xenon can occasionally cause potential side effects including vomiting, headache, and rarely convulsions or respiratory failure and may present difficulties with some patients.

Positron Emission Tomography (PET)

PET scanning utilizes ¹⁵O₂ as described above to measure OEF. This method can also provide direct, quantitative measures of CBF using H₂ ¹⁵O and CBV using C¹⁵O, making it a powerful tool. PET is limited by availability and by the short half-life of ¹⁵O, requiring a cyclotron to produce radiopharmaceuticals in the near vicinity.

Single-Photon Emission CT (SPECT)

In the SPECT methodology a radioactive tracer such as ¹²³I–IMP ARG or ^99mTc-HMPAO is injected intravenously. The contrast agent crosses the blood–brain barrier and is temporarily fixed in the tissue. The amount of fixed tracer correlates with blood flow. During this interval gamma emissions are measured and used to approximate blood flow. In moyamoya two sets of measurements are made to calculate cerebral reactivity: first is a baseline CBF followed by CBF under acetazolamide challenge separated by a sufficient time interval to allow washout of the tracer.

Blood Oxygen Level-Dependent (BOLD) MRI

BOLD MRI is a functional technique based on detecting the difference between the field inhomogeneity created by deoxyhemoglobin (dHb) within red blood cells in the microvasculature and the relatively homogenous field of adjacent parenchyma (Logothetis and Pfeuffer 2004). The BOLD sequence detects susceptibility caused by dHb; hence, signal drops when dHb concentrations increase. After neuronal firing there is a brief drop in BOLD signal related to increased oxygen extraction and elevated dHb; however, seconds later there is an autoregulatory elevation in local CBF, and BOLD signal increases. In this way, BOLD detects neuronal activity indirectly.

The BOLD technique has been adapted for CVR imaging by acquiring images before and after a hypercapnic challenge utilizing a rebreathing circuit designed to adjust and maintain specific end-tidal CO₂ (Blockley et al. 2011; Mandell et al. 2008; Vesely et al. 2001). In response to a hypercapnic challenge, regions of the brain with normal CVR induce a regional increase in blood flow and a resultant increase in BOLD signal, whereas regions with exhausted CVR and chronic, maximal vascular dilatation cannot respond with an elevation in CBF, and BOLD signal drops due to a relative elevation of dHb (Conklin et al. 2010; Heyn et al. 2010). CBF in these regions may drop under these conditions due to the “steal” phenomenon where blood flow is shunted away from oligemic regions toward adjacent, dilated vascular beds. This method provides a noninvasive, nonradioactive methodology to diagnose, grade, and follow-up patients undergoing revascularization therapy (Figs. 4, 5, and 6) (Fierstra et al. 2011).

Treatment of Moyamoya Disease

Medical Treatment

Medical management of MMD focuses on prevention of hemodynamic ischemia and possibly thromboembolic events related to systemic hypotension or intracranial vasoconstriction due to hypocarbia. Antiplatelet therapy with daily dosing of acetylsalicylic acid (ASA) is the most common treatment prescribed, although its efficacy in preventing strokes for MMD has never been proven. The principle of antiplatelet therapy is to prevent the formation of microthrombi at sites of vessel stenosis and at the origin of moyamoya vessels where turbulent flow and flow constriction can create a hypercoagulable state. Microthrombi may produce small emboli that occlude distal end arterioles and create symptoms of TIA as well as radiographic evidence of infarction, although hemodynamic insufficiency is considered the main mechanism of ischemia in MMD. There is no evidence that more potent antiplatelet agents such as clopidogrel (Plavix) are beneficial in MMD. Anticoagulation is not recommended in moyamoya patients due to hemorrhage risk.

Moyamoya patients are at risk to develop hypertension due to concomitant renal artery disease (Buerki and Steinlin 2013). In these cases stepwise titration of antihypertensives is required as sharp drops in blood pressure can result in brain ischemia and transient or permanent symptoms of ischemia. Moyamoya patients are at highest risk for these complications due to a loss of cerebrovascular reactivity and an inability to compensate for systemic hypotension. Similarly, when a patient with MMD undergoes general anesthesia for any surgical procedure, care must be taken to avoid sudden or sustained drops in blood pressure that could result in brain ischemia. Likewise, hypocapnia must be avoided intraoperatively as intracranial vessel constriction can result in oligemia and stroke in the impaired regions of the brain supplied by moyamoya collaterals.

In the case of moyamoya syndrome with an underlying cause, secondary stroke prevention related to correcting the underlying pathology is also very important. Patients with sickle cell disease may require recurrent blood transfusions (Russell et al. 1984) or even a bone marrow transplant. Patients with hyper or hypothyroidism require treatment of their underlying condition and supplementation of thyroid hormone.

Outcomes for medical therapy in moyamoya have not been widely reported. This is because recurrent symptoms are frequent and often lead to surgical therapy. For instance, in a retrospective review of moyamoya patients treated surgically or medically, Hallemeier et al. found that patients with bilateral symptomatic MMD had an 82 % incidence of recurrent symptoms within 5 years of first symptom onset with best medical management (Hallemeier et al. 2006). Therefore, medical management must be accompanied by close follow-up and in many cases should be used as a bridge to definitive surgical therapy. In cases of asymptomatic, incidental moyamoya, initial medical therapy may be appropriate; however, these patients tend to deteriorate after 5 years of observation (Hallemeier et al. 2006). Therefore, the decision to intervene and prevent ischemic events must be considered. Evidence of silent infarction on MRI and evidence of progressive stenosis on serial angiography are indicators that surgical therapy should be considered to prevent symptomatic progression.

Surgical Treatment

The surgical treatment of MMD is focused on restoring flow to the ischemic hemisphere. Restoring flow reduces the risk of future ischemia by enhancing collateral blood flow and restoring cerebrovascular reserve, providing protection from oligemia during periods of hypotension and hypoxemia. Restoring blood flow in the affected hemisphere halts and may eventually reverse the production of fragile angiogenic vessels, leading to a lower hemorrhage risk. Surgical restoration of blood flow can be achieved using “direct” or “indirect” methods or a combination of both. Direct methodologies reestablish blood flow by bypassing the ICA using an external carotid artery vessel or conduit from the extracranial circulation to perform an anastomosis to a distal vessel, usually an M3 or M4 branch of the MCA, in the affected hemisphere (Figs. 7 and 8). Indirect methodologies apply vascularized tissue to the surface of the brain thus permitting angiogenesis that establishes vessels from the graft to the pial surface of the affected hemisphere over months to years (Figs. 9 and 10).

Fig. 7

Each step of the direct bypass surgery. (a) Starting in front of the tragus within the hairline, the STA is mapped for 7–8 cm. (b) The STA is prepared under the surgical microscope, leaving an approximately 8 mm tissue cuff around the vessel. (c) A 6 × 6 cm craniotomy is performed over the sylvian fissure, and (d) the size of the M4 branches of the MCA are evaluated at high magnification. (e) The largest M4 branch is chosen as recipient. The STA is fish mouthed and the wall of the MCA cut using microscissors, removing a tiny elliptical piece of the superior wall. Under high magnification a bypass between the STA and MCA is performed using 10-0 interrupted sutures. (f) After completing the direct anastomosis, the STA with its cuff of soft tissue is laid on the cortical surface to induce an additional indirect revascularization (Reprinted from Neurosurgery Clinics of North America, Guzman R and Steinberg GK (2010) Direct Bypass Techniques for the Treatment of Pediatric Moyamoya Disease 21(3): 568 with permission from Elsevier)

Fig. 8

Technique for direct bypass using superficial temporal artery (STA) to middle cerebral artery (MCA) bypass . The STA has been isolated and prepared for bypass. (a) A suitable recipient M4 vessel is exposed by opening the arachnoid and measured to ensure caliber is sufficient for anastomosis (13 mm). (b) The graft is positioned, temporary clips are placed proximally and distally, and indigo carmine blue dye is applied to the STA and MCA to improve visualization. (c) An arteriotomy is made in the recipient MCA vessel, and the donor STA vessel is cut to match the proximal and distal orifices of the anastomosis. The toe stitch is then placed using 10-0 monofilament. (d) The heel stitch is placed next. (e) The front of the anastomosis is completed using multiple interrupted sutures. (f) The back of the anastomosis is then closed using multiple sutures. The anastomosis is flushed with sterile saline before being completely closed. (g) Temporary clips are removed; the anastomosis is checked visually with the intraoperative Doppler and with the transonic flow meter for patency and quantitative flows. (h) Patency of the graft is verified using indocyanine green (ICG) fluorescence angiography under the microscope

Fig. 9

Indirect revascularization using encephaloduroarteriomyosynangiosis (EDAMS) . (a) The M4 branches of the middle cerebral artery (MCA) are exposed by opening the overlying arachnoid. Measurements in this case revealed the largest recipient M4 artery diameter 0.6 mm and very fragile; therefore an indirect bypass was selected. (b) The superficial temporal artery (STA) with a cuff of the adventitia and soft tissue is laid on the brain. (c) The temporalis muscle is brought into the field and fastened to the dural edges, resulting in direct contact of the temporalis muscle with the brain and overlying the STA that is visible under the elevated edge of the muscle. (d) The closure is completed by suturing the remaining edge of the muscle to the dural edge

Fig. 10

Indirect bypass using an omental onlay graft . (a) The graft has been harvested laparoscopically and tunneled subcutaneously from the abdomen by skip incisions and is laid on the surface of the brain. (b) The omentum lies on the surface of the brain in the subdural space. (c) T1-weighted MRI demonstrating the hyperintense fat graft in the subdural space on the right side. (d) AP carotid angiogram demonstrating a region in the frontoparietal cortex without flow. (e, f) AP and lateral angiogram of the celiac trunk demonstrating omental arterial flow in the subcutaneous space and supplying blood flow to the previously oligemic region of frontoparietal cortex

Direct Revascularization

Direct revascularization is preferred over indirect because blood flow is restored immediately (Figs. 7 and 8). Therefore, when considering which methodology to use, it must first be determined if an adequate donor vessel for anastomosis is present. An angiogram can be used to visualize the diameter of the STA in the parietal and frontal branches. One or both of these branches can be used as a donor vessel. Due to impairments in flow that hinder angiography, however, preoperative visualization of the recipient distal MCA vessels can be unreliable, leaving intraoperative visualization and exploration of the M3 and M4 branches of the MCA necessary to select an optimal recipient vessel.

When an adequate donor vessel cannot be identified, a secondary vessel or conduit must be used. Situations where the STA is not available include patients with prior trauma, prior craniotomy, idiopathic occlusion or agenesis, and prior failed bypass. In these cases the occipital artery may be considered for posteriorly positioned hemispheric dysfunction. The occipital artery is technically more cumbersome to dissect than the STA due to its tortuous course. Another less common maneuver is to anastomose the proximal STA “stump” to the MCA using a saphenous interposition graft or radial artery graft as a conduit. Proximal external carotid artery or extracranial ICA bypasses using a conduit are also possible. However, these large-diameter grafts generally carry higher flow than the STA leading to a greater risk of hyperperfusion or hemorrhage. Matching the distal diameter to recipient vessels in these types of larger vessel grafts is technically difficult and carries occlusion risks associated with graft stenosis. Therefore conduit grafts are not recommended in MMD.

STA–MCA bypass is the standard direct revascularization procedure used in MMD (Figs. 7 and 8). The Stanford University protocol for this revascularization procedure requires patients to stay on ASA leading up to surgery. The patient is placed under general anesthesia with great care not to induce hypotension during induction. For many years neurophysiologic monitoring was performed throughout the procedure using somatosensory evoked potentials, motor evoked potentials, and electroencephalography. However, after performing about 1,000 direct bypasses with <3 % perioperative stroke rate and intraoperative electrophysiological monitoring not predictive of these, we now only use a frontal scalp EEG grid to confirm burst suppression after propofol infusion for the short M4 occlusion period. Patients are placed under moderate hypothermia – in smaller patients by passive cooling with a cooling blanket and in obese patients with intravascular cooling catheters. Hypothermia (33 °C) is maintained until the anastomosis is complete. The STA is mapped using ultrasound preoperatively. The STA, including a cuff of vascular fascia around it, is dissected out from 1 cm above the zygomatic root to the distal end of the selected parietal or frontal branch. The vessel is left in situ and allowed to flow until moments before the anastomosis start. The temporalis is split and retracted leaving the vessel intact. Two burr holes are carefully drilled under the proximal and distal end of the vessels at the edge of the wound. A circular frontotemporoparietal craniotomy measuring about 6 cm diameter is turned below the vessel by connecting the burr holes and taking care not to interrupt or damage the STA. The craniotomy is sized to expose the sylvian fissure.

With the sylvian fissure exposed, a recipient vessel is identified. Typically the recipient vessel is an M4 branch of the MCA arising from the sylvian fissure. The angular branch of the MCA is often the ideal recipient. A branch oriented perpendicular to the surgeon is ideally positioned for the anastomosis. The recipient vessel is dissected from the arachnoid over a 1 cm length. In some cases a tiny perforating artery arising from the recipient is coagulated and divided to facilitate temporary occlusion and arteriotomy of the M4 vessel. A high visibility background is placed behind the vessel. The recipient vessel is then covered with a cotton pledget containing papaverine or nicardipine while the donor vessel is prepared.

The donor vessel is prepared by first placing a temporary aneurysm clip on the proximal end and then occluding and transecting the distal end at the wound edge. The vessel is flushed with heparinized saline. The distal vessel is then trimmed to an appropriate length to allow the graft to sit without tension or torsion in the subdural space. The adventitia is dissected from the distal donor artery 1 cm from the distal end to provide a clean vessel for anastomosis. The diameter of the distal orifice of this vessel is matched to the recipient vessel by making an angular cut across the distal vessel to create a “fish mouth.” The donor vessel is then brought into the surgical field so that it rests close to the recipient vessel in an orientation suited to perform the anastomosis. Indigo carmine blue dye is used to highlight the vessel edges of both donor and recipient vessels. The diameter of the donor (STA) and recipient (MCA) vessels should be matched when possible. Anastomosis of grafts less than 0.8 mm in diameter is technically difficult and carries a higher risk of occlusion than in larger vessels; however, reliable anastomosis of STA and MCA vessels down to 0.6 mm is possible by experienced surgeons (Guzman et al. 2009).

The recipient vessel is prepared by placing temporary aneurysm clips proximal and distal to the planned opening. Prior to temporary occlusion, hypothermia is confirmed and the patient is place in burst suppression using propofol. A diamond-shaped arteriotomy in the superficial vessel wall is performed to match the opening of the fish mouth on the distal donor vessel. The recipient vessel is flushed with heparinized saline. The anastomosis is then undertaken by stitching the heel and toe (proximal and distal corners of the diamond opening) of the anastomosis using 10-0 monofilament interrupted stitches. The back wall and then front wall of the anastomosis are then stitched using interrupted or running 10-0 monofilament. Prior to the final stitch insertion, the anastomosis is irrigated with heparinized saline. Temporary clips are then removed from the recipient vessel followed by the donor vessel. Minor bleeding may be corrected by applying pressure with a small piece of cotton, Surgicel, or Gelfoam at the anastomosis site. Additional interrupted sutures may also be used, but care must be taken to avoid stenosis of the anastomosis. In the case of major bleeding, temporary clips may need to be returned. Doppler ultrasound is used to confirm flow in the graft. Indocyanine green (ICG) fluorescence angiography can be used to demonstrate flow in the graft and determine directionality of flow. Last, Charbel transonic flow probes (Transonic Systems, Inc.) may be used to determine quantitative blood flow and directionality of flow in the donor and recipient vessels. The donor artery with surrounding fascia is then laid on the pial surface to induce an additional indirect bypass.

With the anastomosis patency confirmed, the craniotomy can be closed. The proximal burr hole is enlarged to accommodate the entry of the STA graft with the cuff of the adventitia, which provides additional indirect revascularization over time. The craniotomy is fastened with titanium plates, and the temporalis muscle and skin are closed. Graft patency can be checked intermittently by confirming proximal STA flow with the Doppler ultrasound.

Variations on the STA–MCA bypass have been suggested. A smaller craniotomy can be utilized to minimize incision size and tissue trauma. However, this approach may not identify an optimal recipient vessel as fewer vessels will be visualized. Also, the long-term development of indirect blood flow through craniotomy edges and along the cuff of the adventitia and surrounding soft tissue is theoretically decreased through a smaller craniotomy. A “double-barreled” STA bypass, where both parietal and frontal branches of the STA are used to create anastomoses in two distinct vascular territories, has been discussed (Duckworth et al. 2013). Although this method creates more flow than a single anastomosis, the second anastomosis increases operative and anesthetic times and requires a more extensive incision than the standard single STA branch bypass. Further work to characterize improvements in blood flow and patient outcomes compared to single vessel bypass is required.

Indirect Revascularization

Indirect revascularization methods rely on the inherent ability of the ischemic brain to drive neoangiogenesis from vascular tissues exposed to the brain. Similarly, leptomeningeal collateral flow is observed in late-stage moyamoya cases that have progressed. The mechanism underlying this effect has not been completely elucidated. In moyamoya patients there is an increase in serum vascular endothelial growth factor, monocyte chemoattractant protein 1, matrix metalloproteinase 9, and platelet-derived growth factor and an increase in fibroblast growth factor in the spinal fluid, which are either markers of a proangiogenic state or directly contribute to angiogenesis (Kang et al. 2010; Lim et al. 2006; Yoshimoto et al. 1996). It is likely that the ischemic brain regulates the production of these factors; however, the chemoattractant molecule that directs new vessels to the pial surface has not yet been identified. Therefore, indirect revascularization methods require a vascularized tissue to be placed in contact or near the pial surface of the brain to allow the formation of vessels from the vascular donor tissue to the pial surface. This process takes a minimum of 3 months to occur and may continue to provide new vessel formation over months to years as the primary occlusive disorder progresses. Children have a greater likelihood of developing new vessels from donor tissue when compared to adults (Mizoi et al. 1996). This may be due to higher levels of proangiogenic factors expressed or a better innate ability to form new vessels in response to ischemia in younger patients. Because of this difference, direct bypasses are preferred in adult patients.

Donor tissues for indirect revascularization include the temporal artery and its surrounding adventitia and soft tissue, temporalis muscle, pericranium, galea, dura, and omentum. There are several reported techniques for performing indirect bypasses with one or more of these tissues. Here we summarize the most common indirect techniques and common combinations of techniques that have been used to improve efficacy.

Indirect revascularization is generally thought to be safer than direct revascularization. Indirect procedures are usually shorter than direct bypasses, thus providing decreased exposure to potential anesthesia risks such as hypotension, hypoxemia, and hyper-/hypocarbia. These complications are at highest risk of occurring at the beginning and end of anesthesia; thus, careful monitoring is required at these times. Another theoretical advantage of indirect revascularization is that no temporary vessel occlusion is required. However, the ischemia induced by a short temporary occlusion of an M4 vessel is minimal. Thus, the risks associated with indirect procedures may not be significantly less than for direct procedures. Multiple burr holes have been used extensively for indirect revascularization. This practice is most common in the pediatric population but is also described in adults (McLaughlin and Martin 2013). In this procedure a scalp incision is planned posterior to the STA in a bicoronal fashion. Multiple burr holes are then drilled in offset rows over the affected hemisphere. The dura at the base of each burr hole is opened to expose the pia. The pia may also be opened over cortical vessels, if encountered, to enhance the opportunity for new vessel formation. Dural leaflets may be resected to provide exposure of the pia to the galea and pericranium. Alternatively, the dura is inverted, placing the outer, vascular dural surface against the pia. The inner dural layer will not permit vessel formation if it is against the pia. Many surgeons leave the pericranium in place on the surface of the skull and open pericranial leaflets to perform burr holes. The pericranial leaflets are then laid over the burr hole, contacting the pial surface to increase the likelihood of revascularization.

Proponents of multiple burr hole surgery have reported excellent results. Sainte-Rose and colleagues reported that 14 of 14 children treated had no further symptoms of transient ischemia (Sainte-Rose et al. 2006). Only one permanent complication occurred in this study: an iatrogenic shunt infection requiring shunt revision. Subgaleal CSF collections were common in this group but could be managed conservatively. Although the procedure is used less frequently in adults, Kawaguchi et al. reported excellent results in 10 adult patients treated with multiple burr holes, with revascularization occurring in all cases; however, there was no revascularization at the site of two burr holes where subdural effusion impaired vessel ingrowth (Kawaguchi et al. 1996b). In their study 1 patient developed a postoperative TIA and progressive MCA stenosis that required an additional burr hole procedure. Six patients in this study presenting with TIAs were followed closely with cerebrovascular reactivity imaging that showed improvements in CBF and reactivity in all cases as well as resolution of moyamoya vessels in the lenticulostriate distribution in 4 patients. However, in their summary of 143 pediatric patients treated by indirect revascularization, Scott and colleagues commented that burr hole procedures provided revascularization that was focal as well as variable in extent and quality (Scott et al. 2004). They suggest that using an additional indirect revascularization method, such as pial synangiosis, could provide further, more extensive revascularization. In their experience the addition of multiple burr holes to pial synangiosis was of little or no benefit, and they have since abandoned multiple burr hole surgery altogether.

The most common indirect revascularization technique is encephaloduroarteriosynangiosis (EDAS) . In this procedure the STA is exposed through an incision placed on top of the artery. The artery is carefully dissected out with a cuff of soft tissue along an 8–10 cm length. The vessel is left in situ with blood flow proximally from the artery origin and out distally into the scalp. The vessel is retracted to the side, and a craniotomy is performed. The dura is opened under the length of the vessel, and the pia overlying fissures and cortical vessels is generally opened in multiple locations under the vessel. Then the vessel is placed on the brain surface. The soft tissue cuff is sewn to the dural edges to secure the vessel in place. Pial synangiosis uses a similar technique but opposes the vessel directly to the surface of the brain by suturing the STA to the edges of a linear opening in the pia using interrupted 10-0 monofilament.

A common variation to EDAS is encephaloduroarteriomyosynangiosis (EDAMS , Fig. 9). In EDAMS a wider craniotomy is performed to maximize exposure of the brain. The temporalis muscle is brought into the craniotomy anterior or posterior to the arterial entry. The STA is laid over the surface of the brain as previously described; however, in EDAMS the muscle is laid over the STA and over the surface of the brain. The dural margins are resected maximally to allow the muscle to be sutured into place on the remaining dural edges, securing the muscle in place on the surface of the brain. The bone flap is then replaced over the muscle and arterial tissue by creating a slot-shaped orifice at the base of the craniotomy that can accommodate the additional tissue passing through the opening.

Vascularized pericranial grafts have been used to revascularize the brain (Kuroda et al. 2010). In these cases the pericranium is preserved by separating it from the galea as a vascularized flap. A craniotomy is then performed under the pericranial flap region. The dura is opened and a portion of dura is resected to expose the brain. The pia can be opened in multiple locations prior to placement of the graft. The pericranium is then brought in and sutured on the dural edges to close the dura and leave the pericranium against the brain surface. This technique is particularly useful in the anterior skull to cover the frontal lobes or to place a vascular graft in the interhemispheric fissure. Similarly the temporalis muscle can be used to perform a myosynangiosis without EDAS in cases where the STA is unavailable.

The dura itself can be used as donor tissue to supply flow from the middle meningeal arterial system to the brain (Dauser et al. 1997). The outer layer of dura is vascularized by the meningeal vessels, whereas the inner layer is avascular. Therefore, to promote revascularization from the dura, the outer dural surface must be opposed to the brain surface. This is achieved by undertaking dural inversion where a dural sleeve containing a meningeal vessel is elevated as a pedicled flap based on the origin of the meningeal artery supplying it. The dura is then carefully twisted and turned over to apply the outer surface to the brain. The flap can then be sutured into place along the dural edges. Great care must be taken with this procedure to avoid loss of blood flow due to twisting of the meningeal artery.

When the donor temporalis muscle is unavailable due to prior surgery, the omentum can be used (Karasawa et al. 1993). This technique requires surgical isolation of the omentum with an arterial and venous pedicle. The free flap is transplanted to the surface of the brain with an arterial anastomosis between the proximal STA and the omental gastroepiploic artery and a venous anastomosis between the omentum and superficial temporal vein. Patients require antiplatelet therapy after this procedure. Graft failure related to thrombosis of arterial or venous anastomoses is possible in this procedure. A tunneled, vascularized omental flap has also been used to overcome issues related to pedicled flaps. At Stanford we have recently modified this approach, using laparoscopy to harvest and tunnel a vascularized omental flap in pediatric patients (Navarro et al. 2013). The flap is tunneled subcutaneously using skip incisions to bring the flap into the cranial wound. The omental flap is pliable and can easily cover an entire hemisphere or even both hemispheres and can also be inserted into the interhemispheric fissure. The omental flap is secured to the dural edges at the periphery of the cranial opening. An adequate opening in the base of the bone flap must be made to allow the omental flap to pass intracranially without occlusion of arterial or venous flow. Follow-up angiography using selective injections of the celiac trunk demonstrates excellent flow through the omental circulation to the brain. Omental grafts provide extensive revascularization with restoration of cerebrovascular reactivity based on CBF studies.

Combinations of indirect and direct revascularization techniques have been described in single procedures to increase exposure of the brain to donor tissues. Pial synangiosis can be easily added to EDAS, EDAMS, dural, pericranial, and galeal flap procedures to increase exposure of the brain to the STA or vessels within the donor tissue (Kim et al. 2003; Kinugasa et al. 1994; Scott et al. 2004). Similarly, multiple burr holes can be added during any indirect procedure adjacent to the craniotomy to increase the potential for indirect revascularization (Patel et al. 2010). The addition of EDAS or temporal muscle grafts to STA–MCA procedures can provide indirect blood flow in the periphery of the cranial opening in addition to immediate blood flow through the STA (Houkin et al. 1997). The Stanford experience using a direct revascularization technique that includes a cuff of soft tissue on the STA donor vessel as well as a wide craniotomy results in additional indirect blood flow arising over time that supplements direct flow through the STA (Guzman et al. 2009).

Natural History and Outcomes

There is scant randomized, controlled data comparing medical and surgical management of MMD. The tendency in clinical practice is to treat MMD, as it is known to be progressive. We therefore have little data related to the natural history of the disease. Furthermore, clinical consensus supports revascularization for MMD, which makes a trial of medical management or conservative management compared to surgery difficult to justify. That said, understanding the natural history of MMD is crucial to evaluating the efficacy of surgical management. We look to case cohorts to extrapolate these data. For instance, Kuroda et al. studied a 40-patient cohort with asymptomatic MMD in Japan (Kuroda et al. 2007). In this cohort, 34 patients were not treated with early surgery. There were strokes in 20 % of cases and impaired cerebral reactivity in 40 % of cases at presentation. Over a mean follow-up period of 43.7 months, these 34 patients experienced an annual stroke rate of 3.2 % with 3 TIAs, 1 ischemic stroke, and 3 intracranial hemorrhages. There were 5 patients with angiographic progression of stenosis, and, with progression, 4 of 5 patients had ischemic symptoms. This small cohort study with limited follow-up provides compelling data that supports the notion that MMD is progressive in nature and at relatively high risk for symptomatic progression.

Choi et al. conducted a retrospective cohort study in both surgically and medically managed patients with symptomatic MMD in South Korea (Choi et al. 1997). They identified 52 patients managed medically with a mean follow-up of 67 months. In patients presenting with ischemic symptoms, 55 % went on to have recurrent symptoms during the follow-up period. In children the rate of recurrent ischemic symptoms was 70 %. This may relate to the fact that only pediatric patients in this study had angiographic progression (primarily from unilateral to bilateral disease) during the short follow-up period and pediatric patients were more likely to have Suzuki stage 3–4 disease, whereas adults were more likely to have stage 4–6 disease. The authors noted that Suzuki stage 3 and 4 patients were the most likely to have recurrent symptoms. An additional 16 patients (31 % of the total cohort) had evidence of new asymptomatic infarction or hemorrhage on follow-up imaging. In patients presenting with hemorrhage, only 11.7 % of nonsurgically treated patients had a recurrent hemorrhage, and none died. Furthermore, patients presenting with hemorrhage had similar, improved clinical scoring at the last follow-up independent of surgical or medical management and despite having patients with worse scoring in the nonsurgical cohort at initial presentation. This observation raises concerns about the value of revascularization in the setting of a hemorrhagic presentation. Until recently the benefit of revascularization surgery for preventing future hemorrhage in MMD patients presenting with hemorrhage was controversial. However, a recently completed landmark, prospective Japanese study randomizing 80 hemorrhagic MMD patients (16–65 years of age) into direct surgical revascularization versus medical management (no antiplatelet agents used in either group) with 5 years of follow-up showed surgical revascularization reduced recurrent bleeding and improved outcome (Miyamoto et al. 2014). The rebleed rate was 2.7 %/year in the surgical group versus 7.6 %/year in the medical group (P = 0.042) and for all morbidity: surgery 3.2 %/year versus medical 8.2 %/year (P = 0.048).

A recent cohort study of 125 symptomatic adult and pediatric MMD patients from Henan Province, China, included 100 patients with medical management with a mean follow-up of 28.8 months and a mean age of 31.6 years (Cui et al. 2013). There was a bimodal peak in presentation at 5–9 years of age and at 40–44 years of age. In this cohort 64 % presented with ischemia, 20 % with hemorrhage, and 16 % with headache, choreiform movements, or as incidental findings. Recurrent risk of ischemic symptoms during follow-up was 42 % in this group. In the hemorrhagic cohort there was a 28 % rebleed rate in the conservatively managed cohort associated with a 26 % mortality rate at the time of recurrent hemorrhage. Interestingly, there was no re-hemorrhage in the small revascularized patient cohort (three patients who presented with an initial bleed) included in this study. This study supports the notion that recurrent ischemic symptoms are frequent in untreated MMD patients. The risk of hemorrhage observed in this study was significant and was associated with a high mortality rate.

Hallemeier et al. analyzed a retrospective cohort of 34 adult American patients with symptomatic MMD having a median follow-up of 5.1 years and a median age of 42 years (Hallemeier et al. 2006). In this cohort there was a 65 % incidence of recurrent ischemic symptoms after initial ischemic presentation and an 82 % incidence of recurrent ischemia in patients with bilateral disease at presentation. The incidence of new ischemic symptoms following surgical revascularization was 17 %. Asymptomatic patients with angiographic moyamoya experienced a 27 % incidence of ischemic symptoms following diagnosis. This analysis suggested a statistically significant advantage of surgical revascularization in symptomatic patients where the absolute difference in event rate was 48 %, but not in asymptomatic patients where the absolute difference was 10 %.

The Stanford experience with revascularization surgery is among the most extensive globally having performed 1,100 revascularization procedures in over 700 patients. The most recent published summary of detailed outcomes data in this cohort includes 264 patients who underwent 450 revascularization procedures with 4.9 years mean follow-up (Guzman et al. 2009). This analysis demonstrated a preference for direct bypass procedures at Stanford with 95 % of adults and 76 % of pediatric patients undergoing a direct STA–MCA bypass. There was a 3.5 % morbidity rate and 1.5 % permanent neurological morbidity observed in this study. This study was not powered to derive meaningful observations on outcomes in patients who presented with hemorrhage; however, the authors note that fragile moyamoya collaterals regress after revascularization, opening the possibility that revascularization could reduce the risk of hemorrhage from these abnormal vessels. Modified ranking scores improved in 71 % of patients studied, with patients reporting improved thinking and cognition. Although no significant improvement in measures of cognitive function was observed in this cohort, measurements of intellectual quotient in patients without cognitive impairment preoperatively were stable or improved after surgery. Future studies are required to determine the effect of revascularization on cognitive function. It is plausible that cognitive function may improve given the observation that cortical thinning in MMD patients is reversed after revascularization (Fierstra et al. 2011).

Summary

MMD is a rare steno-occlusive disorder of the carotid arteries that leads to abnormal collateral vessel formation to compensate for oligemia. MMD causes neurologic impairment secondary to brain ischemia, stroke, and bleeding. While medical therapy can decrease risk for ischemic complications, surgical revascularization is the mainstay therapy. Direct revascularization procedures are favored because they restore blood flow immediately, whereas indirect revascularization requires 3–6 months or more, to take effect. Outcomes of revascularization surgery are favorable with minimal risk.

In the future the underlying genetic causes of MMD will be identified. This information will lead to targeted therapies that may prevent or reverse the occlusion of the carotid artery. Likewise, an understanding of how moyamoya vessels form may identify targets related to stabilizing or reinforcing moyamoya vessels to prevent rupture and thrombosis.

There is little doubt that management of ischemic MMD using surgical revascularization is necessary and relatively safe compared to the presumed natural history of this disease. While the benefit of revascularization for hemorrhagic MMD has been questioned in the past, a recently completed randomized trial comparing surgical to medical management of hemorrhagic MMD demonstrated reduced recurrent bleeding and improved outcomes with direct surgical revascularization.