Article Text

Download PDFPDF

Research paper
Regional differences in incidence and patient characteristics of moyamoya disease: a systematic review
  1. Rachel Kleinloog,
  2. Luca Regli,
  3. Gabriel J E Rinkel,
  4. Catharina J M Klijn
  1. Utrecht Stroke Centre, Rudolf Magnus Institute of Neuroscience, Department of Neurology and Neurosurgery, University Medical Centre Utrecht, The Netherlands
  1. Correspondence to R Kleinloog, Utrecht Stroke Centre, Rudolf Magnus Institute of Neuroscience, Department of Neurology and Neurosurgery, G.03.232, University Medical Centre Utrecht, Heidelberglaan 100, Utrecht 3584 CX, The Netherlands; r.kleinloog{at}


Background and purpose Moyamoya disease (MMD) is a rare cause of stroke, initially described in Japan. In other countries, incidences and presenting symptoms may differ from those in Japan. The literature on regional differences in incidence and patient characteristics of MMD was systematically reviewed.

Methods Medline, EMBASE and CINAHL were searched for population based studies on MMD published between January 1969 and January 2011. From studies that met predefined inclusion criteria, information was extracted on incidence and patient characteristics. Incidences with corresponding 95% CIs if possible were calculated and descriptive statistics for patient characteristics were used.

Results 8 studies were included: three from Japan, one each from Taiwan and China and three from the USA. Incidences per 100 000 patient years ranged in Japan from 0.35 to 0.94 (95% CI 0.69 to 1.19), in the USA from 0.05 (−0.04 to 0.12) in Iowa to 0.17 (−0.06 to 0.40) in Hawaii and were 0.41 (0.28 to 0.54) in Nanjing, China and 0.02 (0.003 to 0.04) in Taiwan. Female to male ratio ranged from 1.1 (0.9 to 1.5) in Nanjing to 2.8 (1.2 to 6.1) in Iowa. Proportions with intracerebral haemorrhage as the initial presentation were 56% in China, 52% in Taiwan, 29% in Hawaii, 21% in Japan and 10% in Iowa. Patients with childhood onset presented most often with ischaemia (>75%) in all regions.

Conclusions MMD incidence was higher in Japan and China than in Taiwan and North America and presenting symptoms showed regional differences, which are thus far unexplained. Population based data on MMD in Europe are lacking.

Statistics from


Moyamoya disease (MMD) is a progressive cerebrovascular disorder with occlusion or stenosis of the distal internal carotid artery or its proximal branches, and an extensive network of cerebral collaterals at the base of the brain.1 In a minority of patients the posterior cerebral artery may also be affected.2 3 Patients typically present with symptoms of cerebral ischaemia or intracerebral haemorrhage (ICH) as a child or during young adulthood.2 3 Epidemiological surveys in Japan have provided an insight into the incidence of MMD in Japan.4 5 MMD has now been reported to occur in patients from different ethnic backgrounds all over the world,6 but the incidence in regions other than Japan appears lower. Epidemiological studies of MMD in regions other than Japan are scarce. In addition to regional differences in incidence, it has been suggested that patient characteristics differ between regions.7–12

The purpose of this study was to systematically review the regional differences in incidence and patient characteristics of MMD.


Search strategy and selection criteria

We searched Medline, EMBASE and CINAHL for population based studies reporting on MMD from January 1969 to January 2011 with different combinations of the following keywords (or their synonyms) ‘moyamoya’, ‘epidemiology’, ‘population’ or ‘incidence’ (table 1). The starting year of 1969 was chosen because in this year the disease was given its name, moyamoya.1 Articles found by these searches were screened by reviewing title and abstract. After identification of relevant articles, we searched for additional publications in the references lists, related articles and citation lists. We included studies that: (1) allowed calculation of the incidence of MMD in the investigated region or gave insight in their calculation method; (2) reported at least the age and sex of the patients; and (3) included a description of case finding methods. Papers published in Dutch, English, German, French, Italian or Spanish were included. We used the definition of MMD as stated by the Research Committee on Spontaneous Occlusion of the circle of Willis (Moyamoya Disease) in Japan.13 According to this definition, patients with bilateral disease are considered to have definite MMD whereas patients with unilateral vasculopathy are classified as having probable MMD. If the classical features of moyamoya on angiogram are found in combination with other disease entities (such as haematological disorders, genetic syndromes or radiation injury),3 the patient is diagnosed with moyamoya syndrome (MMS).

Table 1

Search strategy

Data extraction

Two authors independently reviewed all relevant articles (RK reviewed all articles and the second reviewer was either LR, GJER or CJMK) and extracted the following items: region, case finding methods, number of cases with MMD, probable (unilateral) MMD and MMS, total population at risk per year, reported incidence of MMD, sex distribution, age of onset distribution, family history, ethnicity, mode of presentation (ischaemic (including transient ischaemic attacks), haemorrhagic, asymptomatic, other) and case fatality. Disagreement between the two reviewers was resolved in a consensus meeting.

Data analysis

MMD incidence with 95% CIs was calculated by dividing the number of patients with definite MMD by the total population at risk per year. If patients with probable MMD or patients with MMS could not be separated from patients with definite MMD, the incidence with 95% CI was calculated for these groups combined. Subsequently, we calculated differences in incidence between regions with 95% CIs. We analysed patient characteristics for all patients with MMD and for adults and children separately with descriptive statistics and calculated differences with 95% CIs.


Search results and case finding methods

The literature search yielded a total of 387 articles of which eight met our inclusion criteria (figure 1). Five articles described the incidence of MMD in Asia, of which three were in Japan,4 5 14 one was in Taiwan15 and one was in Nanjing, China.16 Three studies reported the incidence of MMD in four states of the USA: California and Washington,17 Iowa18 and Hawaii.19 There were no studies on the incidence of MMD from other continents. Case finding methods were highly variable between the different regions (table 2). None of the studies took a random sample of the whole population. Three studies sent out questionnaire surveys to identify MMD patients, with response rates of 66.6% in 1994 and 56.8% in 2003 in Japan and 70% in Hawaii.4 5 19 One study relied on the certification system of registered cases in Japan and one study relied on their stroke database to select all patients in the mid-west of the USA.14 18 Two studies searched the discharge database for cases diagnosed with MMD.16 17 In Taiwan, patients were collected from seven major medical centres.15 The numbers of patients with MMD were small in most of the regions (except Japan), despite long study periods. Diagnostic criteria by Fukui (or earlier versions)13 were used in seven of eight regions (table 2).4 5 14–16 18 19 One region did not have access to digital subtraction angiograms for all patients and relied on the International Classification of Diseases-9.17 Two studies gave their estimate of incidence of MMD with 95% CIs.14 16

Figure 1

Flowchart of the search methods.

Table 2

Incidence of moyamoya disease according to region


For MMD, the incidence in Japan (Hokkaido)14 was about 20–40 times higher than in Taiwan (difference in incidence 0.81 per 100 000; 95% CI 0.52 to 1.05)15 and Iowa (difference 0.80 per 100 000; 95% CI 0.54 to 1.05)18 and twice as high as in Nanjing (difference in incidence 0.43 per 100 000; 95% CI 0.15 to 0.72),16 as shown in table 2. In Hawaii, the incidence was about five times lower than in Japan (difference 0.67 per 100 000; 95% CI 0.34 to 1.00).14 19 The difference in incidence between Hawaii and Taiwan (0.44 per 100 000; 95% CI −0.09 to 0.38) and between Hawaii and Iowa (0.13 per 100 000; 95% CI −0.12 to 0.37) did not reach statistical significance.15 18 19

Two studies reported on ethnicity of patients with MMD within their study cohort in a certain region.17 19 In Hawaii, the incidence in the Japanese population was 0.34 per 100 000, in Chinese 0.55, in Hawaiians 0.19, in Caucasians 0.05 and in Filipinos 0.06.19 In California, the combined incidence of MMD, probable MMD and MMS, was 0.06 per 100 000 person years for Whites, 0.28 for Asian Americans, 0.13 for African Americans and 0.03 for Hispanics.17

Sex and age

In all studies, MMD was more common in women (table 3), with a female to male ratio varying from 1.1 (95% CI 0.9 to 1.5) in Nanjing16 to 2.8 (95% CI 1.2 to 6.1) in Iowa.18

Table 3

Patient characteristics of moyamoya disease according to region

In six studies, two age peaks were reported: one in childhood and one in early adulthood.4 5 14 16–18 In the studies from Taiwan15 and Hawaii,19 no age peak was found in childhood. The age peak in childhood ranged from 1–10 years in Iowa to 10–14 years in Japan, and the peak in adulthood from 31–40 years in Taiwan to 55–59 years in California and Washington.4 5 14–19

In the five regions for which we could extract the proportions of patients presenting in childhood or adulthood, 74–89% of all patients presented as adults.14–16 18 19

Family history

Five studies reported on family history of MMD.4 5 14–16 In Japan, 10–15% of patients had a positive family history for MMD,4 5 14 in Taiwan none15 and in Nanjing 1.5%.16

Presenting symptoms

Six studies gave information on the mode of presentation of MMD (table 3).14–19 In Japan, Iowa and Hawaii, ischaemia was the most prevalent mode of presentation.14 18 19 In comparison with other regions, the proportion of patients presenting with ICH in Taiwan was significantly higher than that in Hawaii (difference 24%, 95% CI 18 to 46), Japan (difference 31%, 95% CI 20 to 43) and Iowa (difference 42%, 95% CI 27 to 57).14 15 18 19 Also, in Nanjing, the proportion of patients presenting with ICH was significantly higher than that in Hawaii (difference 27%, 95% CI 8 to 48), Japan (difference 35%, 95% CI 27 to 43) and Iowa (difference 46%, 95% CI 33 to 59).14 16 18 19 Data on the distribution of mode of presentation in Washington and California were not sufficient to compare this region with others.17 In Japan, 18% and in Nanjing, 4% of patients were asymptomatic at the time of detection of MMD.14 16

Five studies provided information on the mode of presentation for patients with adulthood onset and childhood onset MMD separately (table 3).14–16 18 19 In children, the presenting symptoms were most often attributable to ischaemia in all regions and in adults in most regions, except for Taiwan and Nanjing where adults more often presented with ICH.15 16

Case fatality

Case fatality was reported in only one study. In patients with definite MMD from Hawaii, no deaths were recorded in 21 patients with a mean follow-up of 1.8 years (with a range from no follow-up to 20 years of follow-up).19


This review shows that the incidence of MMD in Japan is about 20–40 times higher than that in Taiwan and Iowa, USA, two times higher than in Nanjing, China and about five times higher than in Hawaii, whereas no information is available for other regions. Women are more often affected than men in all regions. The predominant mode of presentation was ischaemia in all regions, both in children and in adults, except in Taiwan and Nanjing where adults more often presented with ICH.

Our finding that the incidence in Hawaii was higher than in other states in the USA is most likely explained by the high proportion of people of Japanese and Chinese descent living in Hawaii.19 This reinforces the observation that persons of Asian descent (especially Japanese and Chinese) have a predilection for developing MMD. In contrast with these findings, the incidence in Taiwan,15 another region in Asia, was comparable with that in California, Washington and Iowa.17 18 This low incidence in Taiwan compared with other regions in Asia might be explained by underestimation of the true incidence of MMD in Taiwan due to incomplete case finding.15 In the Taiwanese study, patients with MMD were retrieved from the seven major medical centres, but it is unclear from the report whether MMD patients might also have been treated at other centres, thus escaping detection for the study.15 Underestimation of the actual incidence may also have occurred in the non-Asian countries due to lack of awareness of the disease in these populations.17–19

The incidence in Japan increased between 19944, 20035 and 2002–200614 because in recent years a significant proportion of asymptomatic patients have been identified through screening programmes.

We found no studies on the incidence of MMD in other regions, including Europe. An estimated incidence of MMD in Europe of 0.3 per department per year has been reported based on a questionnaire sent to 160 European departments of neurology, paediatric neurology and neurosurgery.21 The incidence of MMD, probable MMD and MMS in Switzerland was found to be 0.023 in a patient series of 67 surgically treated patients presenting to three Swiss centres during the period 1993–2009.22 Because this study included only surgically treated patients, was based on data from only three Swiss centres and also included patients referred from abroad, it did not fulfil the predefined inclusion criteria of our review.

Our finding that the majority of patients, especially those with childhood onset, present with symptoms of ischaemia in most regions is supported by several hospital based cohort studies.8–12 23 Similar to the relatively high proportion of patients with MMD presenting with ICH in Taiwan and Nanjing,15 16 a high incidence of ICH in adult onset of MMD of 62.4% was reported in a hospital based Korean cohort8 and in Japan (based on registered cases up to 1994).24 It is unclear why in Taiwan and Nanjing, China adults more often present with ICH in comparison with Hokkaido, Japan and regions outside Asia. Asymptomatic presentation of MMD was only reported from Nanjing, China and Japan.14 16 This can be explained by wide availability of radiological screening through brain check-up systems in these regions.

Patients from Japan had a positive family history of MMD in 15% of cases.4 5 14 In combination with the high incidence of MMD in Japan, this suggests a genetic contribution to MMD. Several linkage studies in Japanese patients suggested chromosome 17q25 as a region of interest,25 and a strong association with the RFN213 gene on this chromosome was found in a recent genome wide association study in 72 Japanese MMD patients versus 45 controls.26 Recently, mutations in the ACTA2 gene were found to be associated with a variety of vascular diseases, including thoracic aortic aneurysms and dissections (TAAD), premature coronary artery disease and also MMD in American patients of Northern European descent and a positive family history for TAAD and MMD.27 In a recent study of 39 European MMD patients without a family history of MMD or TAAD, there were no mutations in the ACTA2 gene except in one patient with a new mutation in an exon of ACTA2.28 The same authors found in a small study, totalling 40 patients and 68 controls from Europe, an association with a single nucleotide polymorphism on chromosome 1p13.3,29 with a known association with atherosclerotic disease,30 and two other single nucleotide polymorphisms on chromosome 5q31-q32 and 19q13.1 that play a role in atherosclerosis, vascular growth and transformation processes with MMD.31 32

This systematic review has several limitations. Firstly, the studies applied different definitions of definite, probable MMD and MMS, and not all studies discriminated between these categories. It is unlikely that the variation in definitions of probable MMD had a large influence on the results as the relative proportion of probable MMD was low.4 5 14 19 Secondly, the definition of adulthood onset differed between the studies, with a cut-off age of ≥10 years in the studies from Japan and China,14 16 >15 years in Taiwan15 and ≥16 years in Iowa.18 Thirdly, despite our extensive literature search, we found only eight articles that met our inclusion criteria and the number of patients in most of the regions was small, which led to wide CIs. Finally, the case finding methods in the different articles varied considerably. Some studies used questionnaires to find all patients with MMD in their region.4 5 19 Because the response rates to these questionnaires were between 56.8% and 70%, the incidence in these studies may have been underestimated.4 5 19 Other studies used hospital discharge databases to find patients,16–19 which may have led to inclusion of patients with an incorrect diagnosis and exclusion of patients who were not admitted to hospital. Also, MMD patients with presentation other than ICH or ischaemic stroke can go undetected when using stroke databases.18 Nevertheless, it appears unlikely that the observed increased incidence in Japan in comparison with other regions can be attributed to the variation in case finding methods only as this difference was also apparent in the two Japanese studies that were based on questionnaires and thus may have underestimated the incidence.4 5 The differences in patient characteristics found in our review of population based studies are comparable with those found when comparing clinical characteristics described in hospital based series of patients with MMD, which underscores the validity of our data.

In conclusion, information on incidence and patient characteristics of MMD in regions outside Japan remains limited. Population based studies from Europe are not available. Reliable estimates of incidence of MMD and patient characteristics are needed to obtain insight into the impact of this relatively rare disease and its genetic and environmental determinants. In Europe, a registry of patients with MMD and MMS should be set up to facilitate further research.



  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.