Introduction

Moyamoya disease (MMD) is a rare, chronic cerebrovascular occlusive condition of unknown etiology. The cerebrovascular hemodynamic impairment of MMD, produced by thickening and cellular proliferation in the vessel wall, most often leads to progressive, bilateral stenosis or occlusion of the intracranial internal carotid arteries together with an abnormal expansion of small vessels distal to the occlusion. Its characteristic appearance on cerebral angiogram, resembling a “puff of smoke,” was first described by Takeuchi and Shimizu in 1957 [29]. The initial presentation of MMD can be transient ischemic attacks, cerebral infarction or intracranial hemorrhage in adults or children, although it has been reported that children and non-Asian adults more often present with transient motor disturbances [5, 7, 10, 12, 17, 23, 36, 37]. Among United States adult case series, the clinical features variously include motor deficits, speech disturbances, headache, and seizure [5, 10, 23]. Surgical revascularization is the most frequent interventional procedure for MMD, but specific indications for treatment in adults are not well established [10, 21, 28].

Limited neurocognitive data from case and small-series investigations of pre-surgical adult patients with MMD have not shown consistent patterns, but rather demonstrate neuropsychological impairments in varying domains often consistent with stroke location. Specific impairments have been reported in processing speed [3, 24], visuo-spatial abilities [13, 24], verbal memory [13] and executive functions [15].

Although the initial neurological presentation and neuroradiological features of MMD have been described, the neuropsychological presentation of adult patients diagnosed with MMD is infrequently reported and has yet to be well characterized. Through the combined experience of three large university hospitals in the United States, we sought to examine the frequency and extent of cognitive impairment in MMD and describe more fully the neurocognitive profile of North American patients with adult-onset MMD.

Methods

Pre-surgical moyamoya adult patients over 18 years of age were referred for neuropsychological evaluation from 1996 to 2008 at three United States Medical Centers: Columbia University, University of Illinois at Chicago, and University of Texas Southwestern Medical Center at Dallas. The charts of patients diagnosed with MMD at the Universities of Illinois and Texas were retrospectively reviewed. Cases from Columbia University were enrolled in a prospective study of neurocognition in MMD. Cases were excluded if they had a history of developmental disorder or any neurological condition independently known to cause cognitive impairment.

Measures

Neurocognitive functioning was assessed using standard neuropsychological tests, with neurocognitive data converted to z-scores based on published, age-corrected normative means. These instruments were the same across all three centers with the exception of those evaluating verbal memory, which was assessed with either one of two tests, the Hopkins Verbal Learning Test or the California Verbal Learning Test, yielding similar indices: immediate recall, delayed recall, and recognition [4, 8]. Tests measured attention (Wechsler Adult Intelligence Test—III (WAIS-III) Digit Span and Arithmetic [31]), processing speed (WAIS-III Digit Symbol-Coding [31] and Trail Making Test Part A [26]), visuo-spatial skills (WAIS-III and Wechsler Abbreviated Intelligence Scale (WASI) Block Design [30, 31]), language (Boston Naming Test [14], Animal Fluency [18], and the Controlled Oral Word Association Test (COWAT) [2]), and executive functions (WAIS-III and WASI Similarities [30, 31], Trail Making Test Part B [26], and the Wisconsin Card Sorting Test (WCST) [11]). In addition, manual dexterity (Grooved Pegboard test [19]) and grip strength (Hand Dynomometer [27]) were evaluated. Depressive symptoms were assessed using either the Beck Depression Inventory-Second Edition or the Centers for Epidemiological Studies Depression scale [1, 25]. Cognitive domain scores were derived by averaging the z-scores within each domain for patients who received all of the domain measures: patients who did not receive all the domain measures were excluded from the analysis.

Cognitive dysfunction was defined by previously-established criterion based on performance deviation from the published normative means of each test in the six cognitive domains: Attention/Working Memory, Speed of Processing, Visual Construction, Memory, Language, and Executive Function [33]. Patients were categorized as having cognitive impairment [Domain Dysfunction (DD)] if ≥2 of 6 cognitive domains were >1.5 standard deviations (SD) below the normative mean OR if ≥1 of 6 cognitive domains were >2 SD below the normative mean. We also report the percent of patients with cognitive dysfunction below the >1.5 and >2 SD level for each test and cognitive domain.

Results

Of the 29 pre-surgical moyamoya adult patients, 14 (48%) were from Columbia University, 10 (35%) were from University of Illinois at Chicago and 5 (17%) were from the University of Texas Southwestern Medical Center at Dallas. MMD was angiographically verified in 28 cases (97%); the other case was diagnosed with MRA.

Average age at presentation was 40 years (39.9 ± 11.2; range 20–65). Seventy-nine percent were right handed, and gender was predominantly female (62%). Race/ethnicity was 59% Caucasian, 10% African-American, 10% Hispanic, and 21% Asian. Average education was 14 years (13.89 ± 2.74; range 8–20) and estimated intelligence, available for 86% of subjects (25/29), was in the average range (98.7 ± 17.2) [22, 32]. Initial clinical presentation leading to the neuropsychological evaluation was stroke in 22 patients (76%) of which 21 were ischemic; TIA in 5 patients (17%); and syncope in 2 patients (7%). Presenting symptoms were predominantly lateralized motor and sensory (62%) and none presented with other localizing cognitive syndromes such as aphasia or apraxia (Table 1).

Table 1 Presenting symptoms
Full size table

Means and SDs for results of the individual neuropsychological tests by domain are presented in Table 2. On individual tests, the poorest mean group performances were on delayed verbal memory (Delayed List Recall −1.3 SD below the mean), language (COWAT −1.2 SD below the mean), and executive function (Trail Making Part B −1.3 SD below the mean).

Table 2 Standardized results (z-scores) on neuropsychological measures and domains
Full size table

Of the 29 participants, 20 (69%) demonstrated neurocognitive dysfunction based on the DD criteria, with 62% of patients meeting both DD criteria. The number of patients with performance below the 1.5 and 2 SD levels is reported in Table 3. Functions for which patients most frequently had more pronounced impairment (below the 2 SD criterion) were processing speed (29% on Trail Making Part A), verbal memory (31% on Delayed List Recall), language (26% on COWAT) and executive function (25% on Trail Making Part B). Subjects’ endorsements of depressive symptoms indicated that 36% had minimal symptoms, 36% had mild symptoms, 12% had moderate symptoms, and 16% reported severe symptoms. Depressive symptoms did not account for patients’ cognitive dysfunction, since all correlations between depression scores and cognitive variables were nonsignificant.

Table 3 Percent of sample with cognitive dysfunction at the 1.5 and 2.0 SD levels
Full size table

The clinical information for each patient, including age, stroke history, medical comorbidities, medications, MRI and cerebral angiogram results, as well as test results reaching the impairment criteria of 1.5 and 2.0 SD below the normative mean, is presented in Table 4. The majority of patients, 83%, had a history of stroke, 17% (5/29) were taking antiepileptic drugs, 10% (3/29) were taking benzodiazapines, 31% had medical comorbidities that are risk factors for stroke (diabetes and hypertension), and 19 of the 28 patients (68%) had evidence of bilateral disease on cerebral imaging. As expected, a large proportion of the patients (86%) had a bilateral disease pattern on angiographic studies and 80% of those patients with a unilateral pattern did not demonstrate a lateralized pattern of deficits. Moreover, patients had a similar incidence of impairment regardless of stroke history: 84% of patients with and 75% of patients without stroke history had impairment on testing (P = 0.553).

Table 4 Patient clinical information
Full size table

Discussion

At initial clinical presentation and prior to any neurosurgical intervention, two-thirds (66%) of MMD patients had cognitive dysfunction. This is more than twice the rate previously reported [15]. In our MMD patient sample, dysfunction was found on tests of memory, verbal fluency, processing speed, and executive functions, whereas attention and verbal reasoning were not affected. Our findings are consistent with other studies that demonstrate cognitive dysfunction in adult moyamoya patients but unique in showing that many adult patients with moyamoya have widespread neurocognitive dysfunction across two levels of impairment.

Other causes of cognitive and motor dysfunction such as presence of stroke, medication effects or medical co-morbidities were examined but did not account for the extent of impairment in the subjects. The concordance of cognitive and motor deficits with ischemic disease, for example, was variable: some patients had impairments with no focal ischemia while other patients had ischemic disease without deficits. Furthermore, the relatively young age of our patient sample suggests that atherosclerotic cerebrovascular disease was not a confounding etiology. In fact, only nine cases had atherosclerotic risk factors (hypertension or diabetes), making it more likely that the strokes were caused by hemodynamic failure from the moyamoya itself.

Medications, particularly anti-epileptic drugs (AEDs) and the benzodiazepine clonazepam could have contributed to dysfunction on testing. However, it is the long term use of multiple, older AEDs in epilepsy patients that have been shown to cause the greatest cognitive impairment [20]. All but one of those on AEDs were on AED monotherapy and only one was on a first-generation AED.

The pattern of deficits demonstrated in this sample from three university hospitals appears less attributable to focal stroke and more typical of diffuse small vessel disease possibly caused by chronic hypoperfusion. The global nature of cerebral dysfunction was further supported by the frequent bilateral abnormality in manual strength and dexterity not accounted for by frank ischemic events. The deficits in processing speed, verbal fluency, and executive functions are similar to those found in vascular cognitive impairment, in which bihemispheral dysfunction is also expected [9]. Impairments in manual dexterity and cognitive flexibility have also been associated with greater subcortical white matter hyperintensities found in vascular cognitive cases [35]. Executive functions mediated by the frontal lobe, typically affected in vascular cognitive impairment, were also affected in this sample of moyamoya cases as well as in reports by others [3, 13, 15, 24].

A state of chronic hemodynamic compromise arising from long-term hypoperfusion, we believe, is the likely mechanism underlying the bilateral disease and deficits demonstrated in this moyamoya patient group. Cerebral hypoperfusion associated with cognitive dysfunction has been well-documented in MMD patients using perfusion-weighted MRI, with the most severe effects in the deep watershed territory of the white matter, in contrast to the relatively better perfusion in the occipital lobes, basal ganglia, brain stem and cerebellum [34]. Kim et al. [16] evaluated a group of adult, first-ever ischemic stroke moyamoya patients with single photon emission computed tomography (SPECT) and found that perfusion defects were larger than MRI ischemic lesions. Their pattern of abnormal cerebral blood flow in the anterior circulation suggested that the etiology of the ischemic lesions was predominantly hemodynamic compromise.

A link between chronic cerebral hypoperfusion and reversible neuronal dysfunction is supported by reports of revascularization resulting in increased cerebral blood flow and improvements in cognition [6]. In a case of moyamoya with preoperative bilateral hemispheral hypoperfusion on perfusion MR, extracranial–intracranial arterial bypass to the right cerebral hemisphere increased MR perfusion to normal levels in both hemispheres and resulted in improvements in right hemisphere neurocognitive functions [13].

There are several limitations to our study. First, our patient sample is subject to a referral bias: in two of the three institutions, the cases are representative of clinical referrals and do not reflect the neurocognitive profile of consecutive cases seen at the institution. Therefore, our reported sample may be biased toward patients with the greatest neurocognitive impairments. On the other hand, cognitive dysfunction may be generally under-represented in reports of moyamoya patients. In a recent report by Karzmark et al., an atypically conservative criterion was employed that required at least half of all cognitive tests to be 1 SD below the mean to meet the cut-off for mild impairment. Second, our data did not include perfusion studies that would have facilitated the assessment of cerebral hypoperfusion and its relationship to cognitive dysfunction in this sample. Third, our sample was not sufficiently powered to detect potential differences in race, gender or institution. Additional cases will be needed to examine the effects of epidemiological and clinical variables on cognition. Finally, there were some minor differences in the neurocognitive assessment measures used across the three medical centers, although attempts were made to equalize comparable results by evaluating outcomes by functional domains. Future multi-site studies should employ a battery to prospectively assess the broad range of functions potentially affected in moyamoya disease based on the assumed underlying mechanisms. Prospective evaluations should include serial structural and perfusion imaging. The inclusion of functional outcomes will provide validating information as to the impact of neurocognitive dysfunction as well as the impact of revascularization on daily functioning.

In conclusion, we found that prior to neurosurgical intervention, a large proportion of patients with moyamoya disease demonstrate disruption of neurocognition in a broad range of functions, particularly those functions mediated by subcortical and frontal systems. The lack of relationship between areas of ischemic lesions and neurocognitive dysfunction suggests the possibility of chronic hypoperfusion, rather than focal stroke, as a primary cause of the cognitive impairment. Our findings also emphasize the importance of performing neurocognitive evaluations to assess more comprehensively the full clinical impact of MMD; stroke-free moyamoya disease may not be a silent disorder when cognitive functioning is taken into consideration. Cognitive dysfunction may be an indication for earlier intervention with reperfusion procedures that can salvage ischemic neurons, restore cognition, and perhaps prevent new stroke onset.