It is not always easy to determine the true epidemiological features of a specific disease that has a low incidence. Most published studies depend on patient data obtained from selected large hospitals or compiled from numerous previously reported literature. Consequently, these studies have many inherent selection biases. A nationwide all-inclusive study, which appears to be an ideal approach to exclude bias, is not always feasible as the registration of patients suffering from diseases with a low incidence takes several years; furthermore, many omissions in registration can take place because of the weak incentives provided to doctors and patients. Indeed, the epidemiological data of many neurological diseases that appear even in textbooks may suffer from serious selection biases. Moyamoya disease is one such typical neurological disease that has a low incidence. The present epidemiological data for this disease has been obtained mainly from select community and university hospitals and may suffer from a serious selection bias.1^–4

In addition, the development of a brain check-up system in Japan revealed many asymptomatic cerebrovascular diseases,5 6 including cases of familial occurrence of moyamoya disease. This means that the previous epidemiological data may have underestimated the number of cases.

However, moyamoya disease has been designated as an intractable disease by the Japanese Ministry of Health, Labour and Welfare, which is fortunate from the point of view of disease tracking. Patients who are officially registered with moyamoya disease are qualified to receive a medical allowance from the government.7 Therefore, nearly all patients are considered to have a strong incentive to seek registration. Hokkaido is the second largest island in Japan, and the data of patients with moyamoya disease have been well documented for the past several years, including demographic and radiological data. Therefore, analysis of patient registry data in Hokkaido may help to reveal the true features of moyamoya disease. In this study, all-inclusive continuous data from a well-limited area in Japan are analysed to reveal the contemporary epidemiological features of moyamoya disease.

PATIENTS AND METHODS

Geographical facts

Hokkaido is the second largest island in Japan, with an area of 78 000 km²; it accounts for 22% of the area of Japan. This prefecture has a population of 5.63 million, which is 4.4% of the total Japanese population. In addition, although Hokkaido is a large island, it has only one prefectural government; Japan has 47 prefectures in total (fig 1). The other major islands are divided into many prefectures. Thus, Hokkaido presents us with an ideal opportunity to obtain sufficient data from a relatively isolated area that represents a single prefecture.

Patient registration

This survey was based on the certification system of registered intractable diseases of the Ministry of Health, Labour and Welfare in Japan from 1979. Each prefectural government is required to certify, register and provide financial support to patients with moyamoya disease as it has been officially designated an intractable disease by the Japanese government.7 Patients who were suspected to have moyamoya disease in hospital could make an application to the prefectural government to be granted radiological assessment, including magnetic resonance imaging (MRI), magnetic resonance angiography (MRA) and/or conventional cerebral angiography data, in order to confirm the particular vascular changes associated with moyamoya disease, including basal moyamoya vessels. These applications were evaluated by the Commission of Certification by specialists in each prefecture. The applications were sorted into the following classifications: definite moyamoya disease, probable moyamoya disease (unilateral moyamoya disease) or improbable cases (eg, atherosclerosis without moyamoya vessels or quasi-moyamoya disease related to other diseases) based on the criteria prepared by the Research Committee on Spontaneous Occlusion of the Circle of Willis (moyamoya disease) in Japan.8

The subjects surveyed in this study were patients with definite and probable moyamoya disease, as certified by the Japanese Ministry of Health, Labour and Welfare in Hokkaido from 2002 to 2006. We analysed newly certificated data: onset age for symptomatic cases and detected age for asymptomatic cases, gender, clinical presentation at onset, familial disease from 2002 to 2006, and the number of already registered patients in 2005.

Analysis

In this study, the types of clinical findings were divided into four subgroups: (a) ischaemia, including infarction and transient ischaemic attacks; (b) haemorrhage; (c) no symptoms—asymptomatic patients were identified by the brain check-up system or by administration of screening examinations to individuals with familial history; and (d) other symptoms, including headaches, seizures and involuntary movements. Furthermore, the following were investigated: (1) detection rate, prevalence and incidence of ischaemia and haemorrhage; (2) gender differences; (3) age distribution at onset or detection; (4) type of clinical findings at onset; and (5) proportion of family history. As far as age and gender were concerned, the data in this study obtained from the Hokkaido area were standardised based on data of the whole Japanese population at 2005. This figure was the most accurate data based on the national census in past years. Prevalence was calculated using all of the registered patients at 2005. The method of statistical analysis used was Chi-square test.

Protection of privacy act

In this study, all data were obtained, under agreement, from data from the Department of Health and Welfare of the Hokkaido Prefecture Government. All data used in this study were irretrievably unlinked to personal information.

RESULTS

Detection rate, incidence and prevalence

A total of 283 new applications for disease certification were submitted during the 5 years from 2002 to 2006. There were 592 cases from individuals who had already been diagnosed with the disease in 2005. The new applications comprised 233 definite cases, 34 probable cases and 16 improbable cases that did not meet the criteria for valid diagnosis. Therefore, a total of 267 newly enrolled cases were investigated in this study. The 60 definite cases (22.5%) were assessed by MR examination without cerebral angiography. The prevalence of risk factor within these applications was as follows: hypertension, 0.3%; diabetes, 0.3%; hyperlipidaemia, 0%; and systemic atherosclerotic disease; 0.7%. The detection rate of the disease per year was 0.94 patients per 100 000 people (95% confidence interval (CI): 0.71 to 1.24), and the prevalence was 10.5 patients per 100 000 people (95% CI: 9.44 to 11.7). These figures are higher than those previously reported (table 1). The incidence of ischaemia concerned with the disease was 0.53 patients per 100 000 people-years, and the incidence of haemorrhage was 0.2 patients per 100 000 people-years.

Gender differences

The ratio of female to male patients was 2.18 (table 1). However, no significant gender differences were observed, although this figure appears to be slightly higher than that of 1.8, as reported previously.2

Age distribution

The percentage of patients under 10 years of age at onset was 15.1% (table 1). This percentage was significantly lower than that of 47.8%, as reported previously (table 1).2 Figure 2 shows the age distribution of the new patients. The highest peak of detection rate was observed at 45–49 years and a smaller peak was observed at 5–9 years. In female patients, two definite peaks in age distribution were observed—a higher peak at 45–49 years and a smaller peak at 5–9 years; this was similar to the pattern of total age distribution. In male patients, this two-peak pattern of age distribution was not clear. It appears that the age distribution pattern was mainly influenced by the female patients. This two-peak pattern was also observed in the previous report.2 However, the main peak was shifted towards the adult age group.

Types of clinical findings

The percentage of haemorrhage cases was 21.0% and these cases showed only one peak at 35–39 years (figs 3, 4). The percentage of cases with ischaemia was 57.4%; these cases showed two peaks—one at 5–9 years and the other at 45–49 years. The percentage of asymptomatic cases was 17.8% and had two small peaks at 5–9 years and adult age.

In the disease pattern by gender, the ratio of ischaemia in females was significantly smaller than that in males (53.0% vs 65.9%; p<0.05). Female patients showed a higher incidence of haemorrhage (22.2%) and asymptomatic patterns (20.5%) than male patients (19.5%, 12.2%). These data were not significant (p = 0.62 and p = 0.10, respectively). In the disease pattern by age, 78.4% of the patients below 10 years of age had ischaemia compared with 53.5% of the patients of 10 years and older (p<0.01). For haemorrhage, 2.7% of the younger group versus 24.3% of the older group (p<0.01) were affected.

Familial occurrence

Familial history was observed in 15.4% of patients. This figure was higher than that of 10.0%, as reported previously (table 1).2 In the younger age group, 37.8% patients had familial moyamoya disease, which was significantly higher than that in the older age group (12.2%) (p<0.01).

DISCUSSION

The epidemiological features of moyamoya disease have been reported several times in Japanese literature.1^–4 However, as shown in table 1, there are several differences in the epidemiological features of moyamoya disease between the data from previously and our study. The results produced from our study might not exclude latent regional bias, although this bias seems to be small enough to neglect. However, in order to minimise this regional bias, the age and gender of this data obtained from Hokkaido were adjusted to those of the whole Japanese population. As far as race/ethnicity is concerned, all data in this study were obtained from Japanese individuals. The results of this study revealed higher detection rates and prevalence compared with previous studies, as well as a peak shift in detection rate from children to adults, and a change in the type of clinical findings and higher familial occurrence.

Higher detection rate and prevalence

The higher detection rates and prevalence of moyamoya disease disclosed in this study do not necessarily indicate that the values of these two important epidemiological parameters have actually increased. Indeed, the increase in the registered number of adult patients with moyamoya disease has resulted in an increase in the values of detection rate and prevalence. These higher figures than previous reports2 9 probably reflect the availability of appropriate diagnostic tools and the brain check-up system that has been extensively developed in Japan. On the contrary, this figure could be lower than the actual figure as this system can register fatal cases due to moyamoya disease. In any case, the higher detection rate and prevalence reflects the actual features of moyamoya disease.

Peak shift from children to adults

One of the well-known specific features of moyamoya disease is its two-peak pattern of age distribution and its higher incidence in childhood in Japan. This study also revealed a two-peak pattern similar to that reported in previous papers.1–4 10 This study, however, revealed that the higher peak observed in adults, particularly in female patients, is more prominent than in children. This result is similar to the data of Uchino et al., which is a multi-race/ethnicity study in non-Asian countries.9 It is conceivable that this difference is caused by the method of data sampling. Most of the previous surveys were conducted by using questionnaires on the epidemiological features of moyamoya disease. This method was affected by sampling hospitals. Previous surveys had an inclination to use mainly university hospitals and public general hospitals. In this study, however, there were 33.3% patients from large hospitals and 15.4% from university hospitals. The other 51.3% patients were from numerous neurosurgical, neurological or paediatric specialised small hospitals or clinics. The selection of hospitals might lead to an overestimation of the number of psediatric patients and an underestimation of the number of adult cases in the past literature.

Changes in types of clinical findings and familial occurrence

Minor changes in the types of clinical findings, including an increase in haemorrhage onset, is not always remarkable as it simply reflects an increased number of adult patients with moyamoya disease in this study. However, with regards to clinical findings, the most remarkable difference observed in this study was an increase in the number of patients with asymptomatic moyamoya disease. These patients accounted for 18.0% of the total number of patients. The driving force for the elevated number of asymptomatic cases is believed to be the brain check-up system in Japan5 6 and the knowledge of the familial occurrence of moyamoya disease. These factors were responsible for the identification of asymptomatic patients and indicate that many patients with asymptomatic moyamoya disease have still not been identified. The familial occurrence rate is 15.4% in this study and was higher than that reported previously.10^–12 This increase is also a reflection of the screening of the families with a history of moyamoya disease.

CONCLUSION

The detection rate, incidence and prevalence of moyamoya disease are higher than previously reported. The two-peak pattern was recognised as previously reported. However, the peak of detection rate in adult patients was higher than that in paediatric patients. In addition, it was revealed that patients with asymptomatic moyamoya disease are not always rare in Japan.