Frequency of rare recessive mutations in unexplained late onset cerebellar ataxia

Sporadic late onset cerebellar ataxia is a well-described clinical presentation with a broad differential diagnosis that adult neurologists should be familiar with. However, despite extensive clinical investigations, an acquired cause is identified in only a minority of cases. Thereafter, an underlying genetic basis is often considered, even in those without a family history. Here we apply whole exome sequencing to a cohort of 12 patients with late onset cerebellar ataxia. We show that 33 % of ‘idiopathic’ cases harbor compound heterozygous mutations in known ataxia genes, including genes not included on multi-gene panels, or primarily associated with an ataxic presentation. Electronic supplementary material The online version of this article (doi:10.1007/s00415-015-7772-x) contains supplementary material, which is available to authorized users.


Introduction
Adult onset cerebellar ataxia poses a considerable diagnostic challenge. Initial investigations focus on detecting degenerative, toxic, structural and inflammatory etiologies which together underlie around a third of cases [1]. Thereafter, molecular investigations for a monogenic basis of disease are often undertaken despite 80 % of patients having no relevant family history [2].
Current molecular investigations for sporadic cases echo that of familial forms, beginning with testing for trinucleotide repeat disorders, such as the spinocerebellar ataxias (SCA1, 2, 3, 6, 7 and 17), dentatorubral pallidoluysian atrophy (DRPLA) and Friedreich's ataxia (FDR) in most centres [1]. However, this approach fails to identify a molecular diagnosis in 87-98 % of late onset sporadic cases [1,3], and subsequent investigations are undertaken on a gene-by-gene basis, often at considerable time and expense.
The difficulty in establishing monogenic forms of disease using this approach is increasingly challenging given that at least 60 causative ataxia genes are reported [4]. Recent studies have therefore utilized next generation sequencing focusing on infantile or juvenile onset cases [5], or adult onset ataxia with a demonstrable family history [4]. Only two studies have described sub-sets of patients with sporadic onset adult disease, despite it being a major form of ataxia, and suggested that a molecular diagnosis can be reached in *10 % of cases [4,6]. Given this, we applied whole exome sequencing to a cohort of individuals with sporadic late onset ataxia.
Blood genomic DNA was fragmented, exome enriched and sequenced (Nextera Rapid Exome Capture 37 Mb and HiSeq 2000, 100 bp paired-end reads). In-house bioinformatic analysis included alignment to UCSC hg19, using BWA as aligner and GATK to detect SNV and INDELS across all samples using standard filtering parameters according to GATK Best Practise Recommendations [7] (see supplementary methods). Further analysis was performed on variants with a minor allele frequency \0.005 in several reference databases and 302 unrelated in-house controls (see supplementary methods). Rare heterozygous, homozygous and compound heterozygous variants were defined, and protein altering and/or putative 'disease causing' mutations as predicted by at least three out of four software programmes were included. Pathogenicity was defined in accordance with American College of Medical Genetic guidelines (see supplementary methods). Genes known or suggested to cause ataxia as a primary or secondary phenotype in humans from two suggested clinical panels [4,8] together with additional genes in which ataxia may result as part of the phenotype (list-supplementary methods) were assessed for variants according to the above criteria, and confirmed by Sanger sequencing (supplementary methods).
Variants were defined using a priori criteria: (1) confirmed pathogenic: dominant disorders-variant previously shown to cause ataxia in humans; recessive disorderseither 2 variants previously shown to cause ataxia in humans; or 1 pathogenic variant with a second variant predicted to affect protein function by at least 3 of 4 prediction algorithms (SIFT, Polyphen2, Mutation Taster, LRT), or through frameshift or truncation. (2) Probable pathogenic: dominant and recessive disorders-variants in known genes causing ataxia in humans and predicted to affect protein function by at least three of four prediction algorithms; (3) uncertain significance: dominant and recessive disorders-variants predicted to affect protein function with weak evidence that gene alteration causes ataxia in humans.
The study was granted ethical approval from a Research Ethics Committee based in the North of England.

Population
Twelve Caucasian individuals of British origin (5 male) with no known consanguinity were included (Table 1). Mean age at disease onset was 46.7 years (SD 11; range 30-70 years). Mean disease duration was 16.6 years (SD 6.9; range 6-30 years). For one patient, the disease duration fell within the range expected for multi-system atrophy [9]. This patient had a normal DaTscan and autonomic function tests. Three individuals had CSF examination with negative oligoclonal bands. Five had nerve conduction studies; two of which were abnormal. Detailed clinical features and the results of clinical investigations are shown in Table 1.

Diagnosis
We identified previously described pathogenic mutations in four of the 12 (33 %) patients in our cohort. All were present on confirmatory Sanger sequencing. No probable pathogenic variants were identified and variants of uncertain significance were found in an additional two cases (17 %). Findings are summarised in Table 2.

Discussion
We identified confirmed or probable pathogenic variants causing sporadic late onset ataxia in four patients (33 %) in our cohort. These findings are comparable to childhood/ adolescent ataxia using targeted sequencing panels (40 %) [4] and whole exome sequencing (27 %) [5]. They are also significantly higher than previous data for adult onset cases using either panels or whole exome (both *10 %) [4,6].
We detected pathogenic variants in SPG7, SYNE1 and ANO10 (previously published by Balreira et al. [10]). Fogel et al. [6] also identified pathogenic variants in these genes (SPG7 (n = 2), SYNE1 (n = 3) and ANO10 (n = 1). The clinical features of these patients appear relatively homogenous between their and our study, with pure cerebellar ataxia beginning above the age of 40 for ANO10 and SYNE1 cases, and a more heterogeneous age of onset (\20-50) with additional neurological features including spasticity and a polyneuropathy in SPG7 cases [6]. Therefore, pathogenic mutations in these genes appear to be an important and frequently identified cause of late onset sporadic ataxia.
We used whole exome sequencing (WES) rather than targeted next generation 'panels', and it remains a contentious issue as to which is more appropriate in the investigation of neurogenetic disorders. WES enables greater genome coverage, and hence detection of pathogenic mutations in genes not considered as having ataxia as a primary phenotype. Our results highlight this as SPG7 was not covered by one ataxia panel [4], SYNE1 by another [8], and ANO10 was not included in either panel. WES however, may result in detection of unexpected findings such as pathogenic mutations predisposing to cancer or neurodegenerative disease, which must be considered and included in appropriate consent procedures. It must also be noted that neither WES nor targeted panels are appropriate to screen for genomic rearrangements or trinucleotide repeat sequences. Determining pathogenicity can be challenging for heterozygous variants without a family history of disease and additional living family relatives for segregation analysis. In our cohort, we found heterozygous variants in SLC33A1 and PLEKHG4 in single cases (Table 2). Heterozygous mutations in SLC33A1 have been associated with spastic paraplegia (SPG42) with ataxia in a single family, and likewise, missense mutations in PLEKHG4 have been implicated in dominant late onset forms of spinocerebellar ataxia in Japanese individuals. Despite the rarity and putative pathogenicity of the variants in our patients, the lack of data to test segregation makes attributing pathogenicity difficult. As NGS begins to develop larger variant datasets in rare diseases it is vital to share such data through collaborative projects which may aid pathogenicity confirmation through the identification of the same or related variants in unrelated families with a similar phenotype.
In conclusion, we have demonstrated that application of WES to a cohort of unrelated individuals following exclusion of common trinucleotide repeat disorders establishes a molecular cause of disease in a third of cases. These findings have significant implications for clinical practise.