Childhood‐onset cerebellar ataxia in Japan: A questionnaire‐based survey

Abstract Objective The diagnosis of childhood‐onset cerebellar ataxia (CA) is often challenging due to variations in symptoms and etiologies. Despite the known regional differences in the prevalence of etiologies underlying CA, the frequency and characteristics of CA in Japan remain unclear. We conducted a questionnaire‐based survey to identify the clinical characteristics of childhood‐onset CA in the Japanese population. Materials and Methods Questionnaires were sent to 1,103 board‐certified pediatric neurologists in Japan from 2016 to 2017. The primary survey requested the number of patients with CA under care, and the follow‐up secondary questionnaire requested additional clinical characteristics of the patients. Results The primary survey obtained 578 responses (response rate, 52.4%) on 385 patients with CA, including 171 diagnosed and 214 undiagnosed cases (diagnostic rate, 44.4%). The most frequent etiology was dentatorubropallidoluysian atrophy (DRPLA), followed by mitochondrial disorders and encephalitis. The secondary survey obtained the clinical characteristics of 252 cases (119 diagnosed and 133 undiagnosed cases). Multiple logistic regression analysis revealed that a younger age at onset, hearing issues, and short stature were associated with a higher risk of remaining undiagnosed with CA in Japan. Conclusions The diagnostic rate of childhood‐onset CA in the current study was comparable to those reported in other countries. The high prevalence of autosomal dominant ataxia, especially DRPLA, was a signature of CA in Japan. These data offer insights into the characteristics of childhood‐onset CA in the Japanese population.


| INTRODUC TI ON
The prevalence rates and etiologies of childhood-onset cerebellar ataxia (CA), which defines a heterogeneous group of disorders, vary across regions (Musselman et al., 2014). For example, studies have shown that dentatorubropallidoluysian atrophy (DRPLA) is more frequent in Asia and Portugal compared to the rest of Europe and the Mediterranean (Coutinho et al., 2013;van de Warrenburg et al., 2014).
The information of childhood-onset CA in Japan is limited. A study including both adult-and childhood-onset SCAs in Japan reported that sporadic cases accounted for approximately double the number of hereditary cases and that, among hereditary SCAs, the autosomal dominant (AD) type was more common than the autosomal recessive (AR) type; this finding is contrast to that observed in the Caucasian population, in which Machado-Joseph disease/SCA3, SCA6, and DRPLA are the most common types (Tsuji, Onodera, Goto, & Nishizawa, 2008). Furthermore, Friedreich ataxia with GAA repeat expansion, the most common form of hereditary SCAs among Caucasians, has never been reported in the Japanese population (Maruyama et al., 2002). However, no studies thus far addressed childhood-onset CA in Japan.
Due to the diverse symptoms and pathogenesis along with variations in etiologies across regions, the diagnosis of childhood-onset CA is often challenging and might be facilitated by a better understanding of the prevalence of underlying etiologies. Thus, surveillance is necessary to update the epidemiology of childhood-onset CA.
In the current study, we conducted a survey to collect information on childhood-onset CA in Japan by sending questionnaires to pediatric neurologists. We also compared the clinical features between patients with and without CA diagnosis to further delineate the clinical presentation of childhood-onset CA.

| MATERIAL S AND ME THODS
This cross-sectional questionnaire-based survey was approved by the ethics committee of the National Center of Neurology and Psychiatry in Tokyo, Japan (approval number A2016-070). The primary objective was to determine the number of patients with childhood-onset CA in Japan and the prevalence rates of underlying etiologies. The secondary objective was to elucidate the clinical characteristics of childhood-onset CA.
Questionnaires were sent via postal mail to 1,103 pediatric neurologists who were certified by The Japanese Society of Child Neurology as the majority of the patients with childhood-onset CA in Japan were presumed to be under the care of these specialists.
Furthermore, assessment of the neurological findings including ataxia by the pediatric neurologists was considered to be reliable.
The questionnaires were sent in two phases. The primary survey requested the number and diagnosis of patients with CA as a symptom who were under the care of the pediatric neurologist.
Information on patients with CA who did not receive a definitive diagnosis was also requested. The secondary survey requested further clinical information of the patients, including disease onset, sex, family history, and clinical course (progressive or nonprogressive), but did not include personal information that could specify the individual.
The primary survey took place from November 2016 to February 2017. The secondary survey was sent to those physicians who agreed to participate in the secondary survey among those who initially responded to have at least one patient with CA under care. All data were anonymized by the responding physicians, and the answers to the questionnaires were sent back by postal mail.
The inclusion criteria were as follows: disease onset between birth and 18 years of age, presence of cerebellar atrophy or hypoplasia confirmed by magnetic resonance imaging, and presence of ataxia.
The data extracted from the questionnaires were transcribed into an excel chart and categorized into two groups: one group comprising patients without a definitive diagnosis (undiagnosed group) and one group comprising patients with a diagnosis of a specific disorder (diagnosed group). The number of patients with specific diagnoses was determined in the diagnosed group.
In addition, multiple logistic regression analysis was performed to assess the ability of independent variables among symptoms, signs, and disease course to predict the definitive diagnosis of CA in the standard medical setting of Japan. A stepwise approach was adopted for variable selection with inclusion and exclusion of criteria with a p value of .15. The C-statistic and the Hosmer-Lemeshow test were used to measure the goodness of fit for the logistic regression model. A p value of <.05 was considered as statistically significant in all analyses. Statistical analysis was performed with SAS ver. 9.4 (SAS Institute).  Figure 2). The responses were obtained from university hospitals, general hospitals, special functioning hospitals (neurology, pediatrics, or rehabilitation), regional medical care support hospitals, clinics, healthcare centers, and others (included research institutes and unanswered). The majority of the reported patients were under care of one of the following: university hospital, general hospital, or special functioning hospital ( Figure 3).

| RE SULTS
In the secondary survey sent to the 150 pediatric neurologists, detailed clinical information on 263 patients, including 125 diagnosed and 138 undiagnosed patients, was obtained from 113 physicians responding to the survey (response rate, 75.3%). After examination by the authors, 11 patients, including six diagnosed and five undiagnosed patients, were excluded because they did not fulfill the inclusion criteria of the current study. Therefore, information on 252 patients, including 119 diagnosed and 133 undiagnosed patients, was included in the final analysis. The patients' current age ranged from 1 to 55, and there were 201 individuals who were under age 20. Table 1 summarizes the diagnoses of the patients with childhoodonset CA identified via the primary survey. Of the 171 diagnosed cases obtained by the primary survey, the most frequent diagnosis was DRPLA (n = 30), followed by mitochondrial disorders (n = 20), encephalitis (n = 16), Joubert syndrome and related disorders (n = 14), and ataxia telangiectasia (AT) and related disorders (n = 14).
Among these, the AD SCAs (n = 37) were more predominant than the AR SCAs (n = 21). The most frequent AD SCA was DRPLA (n = 30), followed by SCA29 (n = 4). Among the AR SCAs, AT and related disorder (n = 14) were the most common, followed by ataxia with oculomotor apraxia type 1 (AOA1; n = 5). Other AR SCAs such as Marinesco-Sjögren syndrome and infantile-onset spinocerebellar degeneration, albeit observed, were less common.
The secondary survey provided more detailed information, and 119 diagnosed patients (Table A1 in Appendix 1) identified via the secondary survey were analyzed to elucidate age of onset F I G U R E 1 The flowchart of data collected via the primary and secondary survey. Questionnaires were sent to 1,103 board-certified pediatric neurologists in Japan. The response rates were 52.3% in the primary survey and 75.3% in the secondary survey. A total of 385 cases, including 171 diagnosed and 214 undiagnosed cases, with cerebellar ataxia were reported, with a diagnostic rate of 44.4% F I G U R E 2 The prefectures of Japan which the responses were obtained. Black indicates at least one patient was present in the prefecture, and gray indicates responses were obtained, but 0 patient present. Gray prefectures are Fukushima, Fukui, Hiroshima, Tokushima, Wakayama, Kochi, Ehime, and Oita, from north to south for childhood-onset CA in Japan (Table 2). Briefly, among the 119 patients, the ages of disease onset were <1 year (infantile-onset), 1-5 years (early childhood-onset), and >6 years (middle childhood to adolescence onset) in 56 (48.3%), 34 (29.3%), and 26 (22.4%) patients, respectively, whereas information on age at onset was missing in three patients. In the infantile-onset group, the most common underlying disease was Joubert syndrome and related disorders (n = 11), followed by SCA29 (n = 4), mitochondrial diseases (n = 4), and CASK-related disorders (n = 4). In the early childhood-onset group, the most common etiology was DRPLA (n = 8), followed by AT (n = 7), and mitochondrial diseases (n = 7). In the middle childhood to adolescent-onset group, the most common diagnosis was DRPLA as well (n = 8), followed by encephalitis (n = 5). Of note, among the patients with the frequently observed hereditary diseases, the mean ages at onset were 6.4 (±5.0), 0.2 (±0.20), 2.5 (±1.6), and 3.9 (±4.0) years for DRPLA, SCA29, AT and related disorder, and AOA1, respectively.
Assessment of the 119 cases with a focus on family history revealed that 35 patients (30.2%) had a positive family history, whereas the remaining 81 patients (69.8%) had a negative family history in the diagnosed group. No information on family history was available for three patients. Among those with a positive family history, DRPLA (n = 15) was the most common, whereas mitochondrial diseases (n = 11), AT (n = 9), and Joubert syndrome (n = 7) were the most common among the patients with a negative family history (Table 2).
Among the diagnosed patients in the secondary survey (n = 119), 54 patients (45.4%) exhibited a progressive clinical course, 64 patients (53.8%) exhibited a nonprogressive course; there was one patient with missing information on their clinical course. Of the cases with the progressive course, the most frequent etiologies were DRPLA (n = 18), AT and related disorder (n = 9), and mitochondrial diseases (n = 7). In contrast, among the patients with a nonprogressive course, the most common cause was Joubert syndrome and related disorders (n = 12), followed by mitochondrial diseases (n = 7) and encephalitis (n = 5; Table 2).
Albeit not statistically significant, negative family history and lack of motor developmental delays tended to be associated with failure of a definitive diagnosis (Table 3). The C-statistics was 0.70, and the p value of Hosmer-Lemeshow test was .76, indicating a good model fitting.
We further examined diagnostic tests that contributed to definitive diagnosis and found that genetic analysis, brain imaging, and testing of body fluids (blood and cerebrospinal fluid) were required in 73.1%, 49.6%, and 17.6% of the patients with a definitive diagnosis of childhood-onset CA, respectively. Of note, multiple test modalities were necessary to confirm the diagnosis in some patients.

F I G U R E 3
The charts showing the types of medical facilities which the responses were obtained. The facilities with no patient included more clinics, and lower number of university hospitals compared to the facilities with at least one patient. The majority of the facility with at least one patient consisted of university hospitals, general hospitals, and special functioning hospitals TA B L E 1 Summary of the 171 diagnosed cases collected via primary survey Among the undiagnosed cases, 45 patients underwent genetic analyses including whole-exome sequencing (WES; n = 10) and disease-specific genetic analysis (n = 14). No precise information regarding genetic analyses was available for the remaining 21 patients.
The most common brain imaging finding that contributed to the diagnosis was molar tooth sign (vermian hypoplasia) in Joubert syndrome and related disorders. Other brain imaging findings which contributed to the diagnosis in the diagnosed group were diffuse brain atrophy in DRPLA, basal ganglia abnormalities in mitochondrial  AT and related disorders, and Joubert syndrome and related disorders were the most common etiologies of childhood-onset CA in Japan.

| D ISCUSS I ON
Given that the population of Japanese children younger than 20 years is 21,580,000 (e-Stat, 2019, January 20) and that 201 patients younger than 20 years were included in the current study, the estimated overall prevalence of children with CA in Japan is at least 0.93 per 100,000. Systematic reviews and meta-analyses to assess the frequency of childhood-onset CA reported the overall prevalence of ataxia in Europe was no more than 26 per 100,000 children (Musselman et al., 2014), which is higher than our findings, which might be attributable to differences in prevalence rates across countries or differences in methods used across studies.
Studies suggest the significance of age at onset, clinical progression (progressive or nonprogressive), and family history for the definitive diagnosis of childhood-onset CA (Fogel & Perlman, 2007;Singer, Mink, Gilbert, & Jankovic, 2015). The average age at onset for AT in the current study (2.5 ± 1.6 years) is lower than the previously reported onset age of above 5 years (Singer et al., 2015). Furthermore, the onset age for AOA1, which follows a similar clinical course as AT, was reported as approximately 7 years (Fogel & Perlman, 2007), which was 3.9 ± 4.0 years in the current study, indicating that the onset of AT and AOA1 might be earlier in the Japanese population. DRPLA is a well-known AD SCA with an early childhood-onset (Fogel, 2012;Singer et al., 2015). Studies assessing adult-onset as well as childhood-onset DRPLA reported the age of onset for DRPLA as 36.9 ± 18.0 years in Japan (Al-Maawali et al., 2012;Maruyama et al., 2002); however, the current study indicated that the onset age for childhood-onset DRPLA was 6.4 ± 5.0 years.

Assessment of the clinical progression in the study cohort revealed
that SCAs were the most common etiology among patients with a progressive course, whereas Joubert syndrome and related disorders was the most common cause among the nonprogressive cases. Previous studies implicated that ataxias with congenital or nonspecific causes, including cerebral palsy, tended to exhibit a static course; an acute progressive course might result from intoxications, infections, or inflammatory processes; and chronic progressive ataxia might result from genetic disorders (Fogel, 2012;Singer et al., 2015). These findings are in agreement with the current study findings, suggesting that genetic disorders should be considered in chronic progressive cases and that congenital malformations should be included in the differential diagnosis for nonprogressive cases in patients with childhood-onset CA in Japan. However, it should be noted that certain genetic disorders with extremely slow progression might be misdiagnosed as nonprogressive CA.
Among the SCAs with a positive family history, DRPLA was the most common cause; however, there were four patients with DRPLA who had no family history even though a positive family history is a common feature of AD SCAs. In the current study, SCA29 (n = 4) was the second most common cause of AD SCAs, with three patients exhibiting no family history.
In the current study, the mean onset age of 2.3 ± 3.5 years and positive family history rate of 18.8% within the undiagnosed group were similar to those reported in cohorts from Canada, demonstrating that the majority (82%) of undiagnosed cases presented at ages younger than 2 years and comprised a smaller population (  Note: Odds ratios for other than "age at onset" are described as "yes/no". Sex was not selected with the stepwise procedure in the logistic model. Abbreviations: CL, confidence limit; SD, standard deviation. Multiple regression analysis using clinical features as variables indicated that younger age of onset, hearing issues, and short stature were risk factors for not receiving a definitive diagnosis for childhood-onset CA in Japan. Patients with C10ORF2 mutation, pontocerebellar hypoplasia type 7, Marinesco-Sjögren syndrome, DRPLA, MECP2 mutation, or mitochondrial disorders were included in the diagnosed individuals who had young age onset, with either hearing issue or short stature. Therefore, these disorders may be first considered as differential diagnosis if not have been assessed. Further, AR SCAs are generally associated with clinical findings outside the nervous system (Fogel & Perlman, 2007;Singer et al., 2015). Hearing issues and a short stature might be unrelated to the nervous system, and the undiagnosed cohort of the current study might have included patients with AR SCAs.
Another speculation might be that individuals with childhood-onset CA harboring hearing issues or a short stature might be representing a new disease entity with these signs that is yet to be identified.
In AD SCAs, genetic panel testing screening is a mainstream diagnostic approach for SCA types 1, 2, 3, 6, 7, and 17, and DRPLA (van de Warrenburg et al., 2014). Several recent studies demonstrated the utility of WES for the diagnosis of childhood-onset CA, especially AR SCAs, a rapidly expanding group following the implementation of this technology (Fogel, 2012;Ohba et al., 2013;Sasaki et al., 2015;Sawyer et al., 2014;van de Warrenburg et al., 2014). Given the possibility that there might be patients with AR ataxias in the undiagnosed group, the diagnostic rate might improve with the implementation of WES for all patients with no definitive diagnosis.
One notable limitation of the current study was the administration of the survey to a limited number of physicians and facilities.
Nonetheless, to best of our knowledge, this is the first study to investigate patients with childhood-onset CA in Japan, providing new insights on the prevalence and etiologies and the feasibility of potential predictors of diagnosis in these patients.

ACK N OWLED G M ENTS
We thank all pediatric neurologists who spared their time to respond to the questionnaire. We also acknowledge The Japanese Society of Child Neurology for providing the list of board-certified pediatric neurologists in Japan. This study was funded by the Intramural Research Grant (30-6) for Neurological and Psychiatric Disorders of NCNP. The funding organizations had no role in the design or the conduct of the study. The funding organization also had no role in the collection, management, analysis, interpretation of data, preparation, review, or approval of the manuscript.

CO N FLI C T O F I NTE R E S T
The authors have no conflict of interests to declare.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.