Systematic review of autosomal recessive ataxias and proposal for a classification

The classification of autosomal recessive ataxias represents a significant challenge because of high genetic heterogeneity and complex phenotypes. We conducted a comprehensive systematic review of the literature to examine all recessive ataxias in order to propose a new classification and properly circumscribe this field as new technologies are emerging for comprehensive targeted gene testing. We searched Pubmed and Embase to identify original articles on recessive forms of ataxia in humans for which a causative gene had been identified. Reference lists and public databases, including OMIM and GeneReviews, were also reviewed. We evaluated the clinical descriptions to determine if ataxia was a core feature of the phenotype and assessed the available evidence on the genotype-phenotype association. Included disorders were classified as primary recessive ataxias, as other complex movement or multisystem disorders with prominent ataxia, or as disorders that may occasionally present with ataxia. After removal of duplicates, 2354 references were reviewed and assessed for inclusion. A total of 130 articles were completely reviewed and included in this qualitative analysis. The proposed new list of autosomal recessive ataxias includes 45 gene-defined disorders for which ataxia is a core presenting feature. We propose a clinical algorithm based on the associated symptoms. We present a new classification for autosomal recessive ataxias that brings awareness to their complex phenotypes while providing a unified categorization of this group of disorders. This review should assist in the development of a consensus nomenclature useful in both clinical and research applications.


Background
The classification of the hereditary ataxias has represented a challenge for decades due to the large heterogeneity of clinical presentations and the important overlap between different pathologies [1]. The first to propose a global classification for this group of disorders was Greenfield in 1954, whose classification was based on pathoanatomical findings [2]. This was followed by Harding's classification in 1983, which regrouped the ataxias according to age of onset, as a proxy for mode of inheritance, and clinical findings [3]. Although this clinical classification had merit, it quickly became overshadowed by a nomenclature based on gene discoveries within each specific type of ataxia starting with ATXN1 in Spinocerebellar ataxia 1 in 1993 [4] and FXN in Friedreich ataxia [5]. Since then, over 40 genes have been discovered in the dominant ataxias and as many in recessive ataxias [6].
One of the main challenges in the study of recessive ataxias is the difficulty to properly circumscribe which disorders belong to the field of hereditary ataxias and which belong to other disease categories. Indeed, ataxia is a cardinal symptom in cerebellar disorders, but may also be a presenting symptom of hereditary spastic paraplegias, hereditary polyneuropathies, neurodevelopmental disorders, and mitochondrial diseases, for example. Concurrently, recessive ataxias often manifest with complex phenotypes, even more so than their dominant counterparts, and may present diverse associated features including neuropathy, pyramidal and extrapyramidal involvement, oculomotor abnormalities, cognitive involvement, seizures, retinopathy, hypogonadism, and many others. This explains the high variability in the list of included disorders in recent literature reviews on recessive ataxias [7,8].
Nevertheless, the advent of next generation sequencing techniques requires to properly determine which disorders belong to each disease category in order to design thoughtful targeted panels and facilitate the interpretation of whole exome and whole genome sequencing data. Indeed, targeted panel sequencing is a highly effective method for the diagnosis of neurological disorders, but it requires insightful categorization of disease phenotypes to respond to the specific needs of clinicians [9,10]. Similarly, the interpretation of unknown variants in the analysis of whole exome or whole genome sequencing data poses a significant challenge for clinicians who must determine if the gene is associated with the suspected disease category and if the phenotype correlates with what has previously been described. As next generation sequencing techniques become increasingly available and the ability to detect DNA repeat expansion diseases improves [11], the proper classification of diseases will represent a useful tool in the interpretation of test results. Hence, this calls for a systematic effort to review recessive diseases in which ataxia is a prominent feature in order for experts in the field to collectively determine which disorders should be included in a recessive ataxia classification.
Therefore, the purpose of this article is to review the literature on recessive diseases presenting with ataxia in order to present a new classification. The goal is to bring together experts for the development of a much-needed consensus that fulfills research and clinical needs.

Methods
We conducted a systematic review to identify articles relevant to the classification of autosomal recessive ataxias. We searched Pubmed and Embase from inception to September 2016 in order to identify original articles on disorders presenting with ataxia. The search strategy was large and targeted both recessive and sporadic ataxias, since recessive inheritance may appear sporadic in certain circumstances (full search strategy is provided in Additional file 1). We also reviewed reference lists of relevant articles and public databases including OMIM and GeneReviews to identify other relevant articles.
We reviewed the titles and abstracts of all identified references to select original articles on recessive forms of ataxia in humans for which a causative gene was identified. We evaluated the articles from a clinical perspective to determine if cerebellar ataxia was a prominent feature in the reported patients or rather a secondary finding in other movement or multisystem diseases. Diseases reporting only on cerebellar atrophy or cerebellar malformations without any clinical consequence were not included. For each listed disorder, we reviewed the evidence for a genotype-phenotype association using the US National Human Genome Research Institute guidelines [12]. Major considerations included the exclusion of previously described genes, the number of unrelated individuals described with similar genotype-phenotype correlations, the evidence of segregation with the disease, the absence of the variant in large control cohorts, and the presence of biochemical or animal-model functional validation. For the primary ataxias, we identified two relevant references from different research groups when possible. All relevant articles were fully reviewed to be included in this classification of recessive ataxias.  Identified disorders were classified in three categories: the first included the primary autosomal recessive ataxias, the second included other movement or multisystem recessive diseases that have prominent ataxia, and the final group was composed of recessive disorders that may occasionally present with ataxia, but where ataxia is a secondary feature.
We also developed a clinical algorithm for the primary recessive ataxias based on the most frequent phenotype and cardinal symptoms associated with each disorder.
The objective of this algorithm is to rapidly summarize the main discriminatory features between different ataxias to serve in a clinical setting, but also as a pedagogical and research tool.
Results 3750 references were identified through the literature search in Pubmed and Embase, and 49 additional references were identified through reference lists or public databases. After removal of duplicates, 2354 references   [128,129] HSP hereditary spastic paraplegia, LBSL leukoencephalopathy with brainstem and spinal cord involvement and lactate elevation, SPARCA1 spectrin-associated autosomal recessive cerebellar ataxia type 1, SPAX spastic ataxia, SPG spastic paraplegia were reviewed on the basis of title and abstract. Finally, 130 articles were selected on the basis of the aforementioned criteria and completely reviewed to be included in this qualitative analysis (Fig. 1). The proposed new list of autosomal recessive ataxias is presented in Table 1 in chronological order of gene discovery. The disorders included in this list were evaluated as having a relatively predominant cerebellar involvement compared to the involvement of other neurologic and non-neurologic systems. Table 2 presents the other complex motor or multisystem disorders that have prominent ataxia. Finally, Table 3 presents disorders that may occasionally present with ataxia, but where ataxia is a secondary feature. Certain decisions were made in the elaboration of this classification. Notably, abetalipoproteinemia (ABL) and Refsum disease were not included in the list of primary recessive ataxias, but rather in the list of complex disorders that have prominent ataxia. Indeed, despite their important Friedreich-like neurological picture, these disorders are primary lipid metabolism disorders with multisystem involvement. Moreover, ataxic disorders that are allelic to other movement disorders, especially spinocerebellar ataxias and hereditary spastic paraplegias, were assigned to the second category to avoid any confusion with the primary recessive ataxias. The MARS2-linked autosomal recessive ataxia with leukoencephalopathy (ARSAL/ SPAX3) was not included because the genetic evidence was deemed insufficient [13]. Finally, some disorders described only in single families were included, despite this being a factor for weaker genetic evidence, if other major considerations were met; this was indicated in the list.
The primary recessive ataxias were also organized in a clinical algorithm (Fig. 2) according to the presence of key clinical clues, which include the presence of sensorimotor involvement, cognitive impairment, spasticity, and oculomotor abnormalities.

Discussion
We present a new classification for the autosomal recessive ataxias. This classification should allow for better Optic atrophy [143,144] ISSD infantile sialic acid storage disease categorization of recessive disorders presenting with ataxia with a clear separation between the primary recessive ataxias and disorders that may present with ataxia as an associated feature but belong to other disease categories. We also provided a clinical algorithm as a tool for diagnostic, learning, and research purposes. This comprehensive classification will allow for improved genetic diagnosis by targeted next generation sequencing applications as the ability to detect DNA repeat expansion diseases is quickly becoming a reality with prospects of treatment in the future [11,14,15]. As compared to previously published reports on this subject [7,8], we systematically reviewed the literature to evaluate the available evidence on the diseaseassociated genes in order to include all disorders presenting with a predominant cerebellar ataxia phenotype. The systematic review methodology with a structured data search and comprehensive evaluation of all references allowed for a complete evaluation of the literature regarding disorders presenting with ataxia to ensure that all potentially relevant disorders were included in this classification. Nevertheless, some methodological elements were not applicable to the task at hand. For example, two references were selected for each primary recessive ataxia, and articles that provided evidence for a separate genetic basis with a clinical corollary of ataxia were preferred. Therefore, some articles that provided only detailed clinical description were not included. Moreover, inclusion criteria were clearly defined but there remained a place for interpretation to determine if cerebellar ataxia was a core feature of the phenotype and if the genotype-phenotype association was convincing. Thus, the classification of individual disorders between the three groups, i.e. as a recessive ataxia, a complex disorder with predominant ataxia or a disorder where ataxia is a secondary feature, remains a subjective appreciation and is open for discussion by a dedicated task force in order to reach a consensus. Finally, the search strategy was designed to be as sensible as possible, but ataxia is a frequent symptom in neurology, and it is possible that other ataxia-associated disorders could be considered for inclusion.
Important challenges remain to be addressed. First, the nosology of recessive ataxias is still highly confusing. Contrary to the dominantly inherited spinocerebellar ataxias, no universal acronym was adopted in the field of recessive ataxias, such that disorders were named based on the author who first described them, on regions of high prevalence, or according to clinical presentation. In the last few years, the term spinocerebellar ataxia, autosomal recessive (SCAR) was used to designate novel recessive ataxias, but this nomenclature did not include the previously described and most frequent ataxias. Moreover, as SCAR assignation was based on locus discovery, some of the included SCARs do not correspond to an identified gene. The term SPAX has also been used to designate ataxias with a strong spasticity component, irrespectively of their mode of inheritance. Recently, the International Parkinson and Movement Disorder Society Task Force for Nomenclature of Genetic Movement Disorders recommended a nomenclature with a gene suffix in order to overcome the shortcomings of the numbered locus system, which include erroneously assigned loci, the mingling of causative and risk factor genes, unconfirmed causative associations, and inconsistent phenotypic correlations [16]. These concerns are justified, although numbered naming systems present definite advantages for ease of use and proper delineation of the field. The nomenclature of recessive ataxias should be discussed by a dedicated task force of international experts in order to develop a naming system that reflects the complexity of the recessive ataxia phenotypes while allowing convenient clinical use.
Finally, large phenotypic variability exists between patients from different families and even from a single family with the same mutated gene, depending on the type of mutation and on its location in the gene. Other factors that affect age at onset and clinical course probably include the presence of modifier genes and environmental exposures. Hence, one could argue that the paradigm of one gene-one disease presented here does not reflect all the phenotypic variability observed, and could as well be replaced by the concept of one patientone disease as we identify new genetic and environmental prognostic features that characterise more precisely the age at onset, evolution, and response to treatment. Such developments are likely to modify our understanding of genetic disorders and of their classification.

Conclusion
We present herein a classification of the autosomal recessive ataxias based on a systematic review of the literature. This work should serve as a framework for scientific discussion in order to bring together experts for the establishment of a much-needed consensus in this field.  Availability of data and materials The dataset of records screened for publication generated during this study can be obtained using the search strategy provided in the additional file.

Additional file
Authors' contributions MB designed the search strategy, conducted the systematic review, and drafted the manuscript. CJK, GAR, and ND provided essential intellectual input and revised the manuscript. All authors read and approved the final manuscript.

Competing interests
The authors declare that they have no competing interests.

Consent for publication
Not applicable.
Ethics approval and consent to participate Not applicable.